The Columbia University High-Beta Tokamak

HBT-EP Archives 2002 - present.

2002

Wednesday December 18 03 58 36492-36513 External Field

During the run I was working with 'high beta' shots that I usually use in my 3/1 experiments. I was trying to develope a model discharge that would be stable to low frequency 3/1 oscilations with the aluminum shells fully inserted and unstable when they're retracted. Due to certain problems with the CAMAC racks and Crunch cabinet I didn't get much run time. Never the less I got several good shots. The data is being analyzed. Regards, Mikhail

Wednesday December 18 05 32 36514-36542 FB setup shots

Developed a "long lived" saturated RWM target mode that lasts about 2 to (sometimes even) 3 ms near the end of the shot. These plasmas were made with the Al shells inserted to suppress higher m/n=4/1 and 3/1 RWMs and 2/1 tearing type modes also. I was successful in bringing the q=2 surface external with a Ip ramp style shot. These shots have an Ip ramp of 2 MA/sec during the first half of the shot and then Ip rolls over somewhat at ~15kA during the last half of the shot when the kink is excited. The kink is of m/n=2/1 helicity instead of our more typical 3/1 variety. The q=2 surface goes out into the vacuum at ~ 4.5ms. The shots typically disrupted at about 8 to 9ms (see shot^#35642 as an example). Feedback next.

Thursday December 19 05 38 36543 ICRH

Disappointingly, the large oscillations on the loop voltage now appear not to be related to the plasma after all. Today, in contrast to last week, the oscillation was there even in the absence of plasma. Last week we had one shot, 36486, without plasma, and the loop voltage oscillation was absent. However, today, we took several non-plasma shots, with and without banks firing, 36562, 36564, 36565, where the loop voltage clearly oscillated during the RF. On the bright side, the RF B probe is now absolutely calibrated. The RF B field at the probe is about 1 Gauss, within expectations for vacuum. No clear coupling was seen today, ie. vacuum B fields were basically the same as those when the plasma was around. The calibration factors have all been worked into the MDSPlus tree, so the f_bprobe tag now directly reflects the RF B-field in Gauss. Cory developed some good discharges that would hang around 92-91 cm for a couple of milliseconds. Nonetheless, Q was only observed to dip during inboard disruptions. We did not make significant progress on the triple probe, for lack of time, mostly.

Thursday December 19 05 43 -36603 ICRH Addendum

Two things to add to the report: We just discovered that by mistake, we ran today with the 10% helium mixture instead of the 20% hydrogen. So no chance of heating today. The other thing: I forgot to put in the last shot number of the day. It was 36603.

Friday December 20 04 20 none

Steve Paul was in today and reinstalled a new fiber optic bundle and optics to carry light from the machine to the screen room, as a consequence of that and some work on TS triggering we did not run today. Steve expects a factor of roughly 4 increase in collected light. He also stated he has made progress on fixing the systematic errors in the measurement through calibration improvements. Since the machine will not be running next week we took the opportunity to pump on and bake the Mach and Biased probe bellows and opened them up to the chamber. They should be ready for experiments when we run again. The bake is on.

2003

Friday January 10 04 25 Rivera null none

The lower 12 turn coil has been installed. The OH has been dismantled to facilitate installing the lower 21 turn coil.

Wednesday January 22 05 26 Maurer 36606-36619 shake-down

Finished getting the machine back in order(insulation, TF cooling oil, etc...) today after installation of majority of VF feeback coils during the past two weeks. Took initial OH and VF vacuum shots going in increments up to typical maximun runtime voltages (e.g. VFEL=200V, OHEL=250V). Observed no problems with new coilsets. Ready for TF and plasma shots tomarrow. Since we are a day behind in getting the machine back together we are planning on moving the run schedule back a day. I will continue tomarrow and check out the TF and get first plasmas and Mikhail will then run on Friday.

Thursday January 23 06 10 Maurer 36620-36656 Clean-up

First plasmas were taken today since prior to Xmas and the new VF coil installation. Plasmas started out short (~2-3ms lifetime) but rather quickly started to last out to ~6ms after about ten shots. Tomarrow we should be able to get some real data. New mounts for the fan array SXR amplifier box were measured out and are being machined. They are needed because of the top middle new VF feedback coil location. The array will hopefully be up and running early next week after the table is installed and an extension is added to the copper bellows (Jim & John). Work on the TS triggering system and the new and repaired u-port adapters also piggy-backed on the run today (Alex & Royce).

Monday January 27 05 26 Mikhail Shilov 36657-36722 External Field

During Fri, Jan 24 and morning of Mon, Jan 27 runs I developed a reproducible discharge type which doesn't have low frequency external kink oscillations in the first phase of the discharge with the aluminum shells all the way in and q*

Monday January 27 05 28 Mikhail Shilov 36657-36722 External Field

During Fri, Jan 24 and morning of Mon, Jan 27 runs I developed a reproducible discharge type which doesn't have low frequency external kink oscillations in the first phase of the discharge with the aluminum shells all the way in and q* less ~ 3. In the afternoon of Mon, Jan 27 I applied external 3/1 corotating, static, static with 180 degree phase flip field to those kind of plasmas. The data is yet to be analyzed. Here is some good natural shots: 36664, 36668, 36671, 36688, 36691, 36703, 36720.

Thursday January 30 06 47 Maurer/Paul 36786-36833 Rotation & TS

Ran some calibrations of the new collection optics to start the day. Working out kinks with new collections optcs (for increased light collection). Ran in 10% and 100% He discharges. Small light levels seen on 100% He discharge cases. Currently not understood why light level is so low. Checking various possibilities. TS triggering progress made. Scope and laser triggered off of jorway pulse stopped pretriggering of scope. Need to AC couple the scope inputs to be able to bump the volts per division to measure TS scattered light. (Alex)

Wednesday February 5 05 31 Cates 36835-36943 Smart Shell RWM Feedback

On Tuesday, several hours were spent testing feedback preamps, and repairing small problems in the system. In the afternoon (shots 36855-65), some target current driven plasmas with 3/1 RWM induced disruptions were tested with and without feedback. Wednesday these current ramp shots were used to do a disruption study with feedback gains changed between full gain, half gain and quarter gain, in order to supplement previous gain studies with only full gain and 0.1 gain. I was also able to test the feedback shots (shots 36878-36891, 36939-43) with these gains against a smaller current ramp which is unstable for a longer period of time, hopefully allowing better observation of mode amplitude growth. As a side note, I have found Mikhail's improved solid_tok code (which now runs every shot automatically) to be a very helpful tool for run time analysis and correlating predicted mode stability with diagnostic signals.

Thursday February 6 05 24 Pedersen 36944-36985 ICRH/triple probe

We exercised the ICRF today and tested the triple probes with some new circuitry. We also took more data with the rf magnetic pickup probe, which was calibrated a while ago. ICRF worked as usual (hardware working fine, but no plasma heating) after a filter replacement. New triple probe circuitry allowed us to take useful data. The ion saturation currents were lower than expected and we changed some resistors to increase the signal levels. We took several good sets of data on the scope; they were stored on floppy and are being analyzed. We can extract a plasma density out of these measurements, and we expect it to be low - how low remains to be seen. Floating potentials were on the order of -40 volts (so Te=10 eV). A higher noise level was observed during the ICRF pulse, but it seems that the signals can still be extracted. RF pickup probe: We noticed small bumps (increases) in the signals when plasmas were inboard limited, but only for those plasmas which were still healthy (the current was still rising and the plasma had not disrupted). Examples: 36978, 83, 84 A typical increase was from 1.1 Gauss to 1.3 Gauss, noticeable but not dramatic.

Thursday February 6 09 33 Maurer 36896-37035 Rotation

... the second shift

Ran plasmas for equilibrium reconstructions for Youhong. Got well centered discharges with essentially constant edge q and major radius over a ~2msec(from about 2 to 4ms during the shot) period with limited MHD that should be good for the VALEN eddy current calculations and Tokamac runs. S. Paul came in for a late night run. We fixed the previous run's problems with signal level and then ran a series of discharges where we systematically varied the puff time to investigate changes in measured intensity levels and effects on the observed rotation rates. More analysis needed, but there were incouraging correlations between mini and major disruptions and the ion velocity being driven to zero. The ions are inferred to rotate in the direction of Ip. Also, TS triggering work was carried out during the ICRH run. There are still some noise issues to sort out with the AC coupled scope inputs. No recognizable scattered light was observed when we did take shots with the plasma. (Alex)

Tuesday February 11 06 08 Pedersen/Maurer etc 37036-45 VF Feedback

The feedback coils are all connected and polarities have been checked. The total series resistance is 32 mOhm, as expected. We pulsed the OH and the VF and looked at the induced voltages in the feedback system. These voltages peak at 2-4 kV. The VF start and the OH breakdown spike each contribute at that level but with opposite polarities. The OH spike induced voltages are roughly in line with expectations, but the VF start was not expected to induce as much voltage as it does. In other words, the coupling between the old VF and the feedback coils must be larger than expected. We consider the voltage level safe for the LANL power supply, which is a 10 kV power supply that will be connected to the FB system through a large transformer.

Thursday February 13 06 16 Maurer/Pedersen none Plasma position feedback

Good news and ...more good news: First, the previously reported induced voltages in the FB systems are off by a factor of 10. The scope probe used was actually a 1:100 divider rather than a 1:1000 divider. Hence the induced voltages were actually on the order of 200 V, which is much more in line with expectations. Nicolai calculated coupling coefficients as follows: Feedback coils - OH coils: c=0.0065 (design value 0.0025) Feedback coils - VF coils: c=0.027 (design value 0.018) No gross discrepancy, in other words. Second, we pulsed the Feedback system with the one LANL power supply. It worked very well with current rise from zero to about 200 A in about 100 microseconds. Data was recorded with the scope for a more precise analysis. The cos theta coil pickup will be recorded on a later run day.

Wednesday February 19 09 07 Nick none

The run-summary page is working again. Upgrading to 10.2.4, rewrote the httpd file without PHP support which is necessary for this script to work.

Wednesday February 19 05 17 Maurer fixing SXRs...

Yesterday and today were spent re-installing the SXR fan array which has been off the machine since the top middle VF feedback coil was installed about two weeks ago. The amplifier box, new table and extended bellows are now setup. The cabling was run differently to help minimize icrh pickup (hopefully) in the future. Initially there were 7 "bad" channels. All appear to be working except for channels 14 and 16 which are open circuit at the vacuum feedthrough. These two channels have been bad for some time. It is not known whether this is due to a bad vacuum side connection or if the diodes are bad (usually diodes fail short when the junction fails). The other channels (1,2,10, 13,15) had either faulty amplifiers or cabling problems that have been fixed except for a small oscillation on 13 and 15. We should be able to get an emissivity centroid out of the array as a check on open loop jog experiments with the new VF feedback coil.

Friday February 21 11 00 Pedersen/Maurer 371??-37206 Radial feedback/triple probe

We did several open loop tests using the LANL into the power supply and operated the triple probe as an ion saturation probe. Open loop tests The open loop test shots were 37190-37206. We found a significant pickup on the cos theta coil when pulsing the FB system. There is a simple proportional pickup but there also seems to be a smaller pick up that looks like it decays exponentially. Not sure about what to do about that one. We have enough data to nail down the calibration factor for the simple linear coupling, and can then proceed to get a believable major radius signal. The LANL worked very well, giving up to 300 A, but the signal generator used to drive the LANL waveforms gave us some problems. It is too early to conclude anything about the efficacy of the new feedback system, since the major radius measurement is contaminated by pickup. The SXR fan array may provide useful independent measurement of the radial position. We can report a preliminary analysis at the Monday meeting.

The triple probe was operated in ion saturation with a 300 V battery to ground, with a 24 Ohm resistor used to measure the ion saturation current. Signals were much better than when operated as a triple probe and the densities are higher too. I suspect that one of the probe leads/connections or the triple probe box is flaky, and I would discard the data taken last week on that basis. One probe tip seemed flaky at first but then started working, the other two were fine. Densities for outboard limited shots start out in the low 10^¹¹ cm^^-3 range and gradually increase to about 5*10¹¹ after about 5 ms. These numbers assume a constant 10 eV electron temperature and that other calibration constants are also accurate. They are a factor of 5 larger than the inboard limited plasma densities reported with the triple probe circuit. Signals are bursty but follow the overall evolution of the plasma well. We need to do inboard limited discharges next. At present, a single probe tip is being digitized in the north rack and a calibrated density (assuming Te=10 eV) is in the tree with the tag ne_inboard

Tuesday February 25 05 21 Maurer MC feedback

Reconfigured crunch amp box and input sensors today for mode control feedback using new gap coils. Checking polarity of signals and magnitude of feedback gains. Initial plasma shots with feedback taken. More of the same tomorrow.

Thursday February 27 05 45 Maurer/Pedersen 37265-37266 VF feedback

Todays run objective was to continue VF open loop jog experiments with the new VF feedback coil. Unfortunantly we did not get very far. The LANL charging supply failed at the start of the run day. The day was spent toubleshooting the supply(Nick has ordered the required parts to fix it). To be able to run we have currently swapped in the charging supply for the other LANL amplifier. Its installation is nearly complete (one dump relay is not working). It should be up and ready tomorrow morning. A calibration error was discovered in analyzing triple probe data. The calculated density form isat measurements is now approximately ~2e18 per cc on the inboard high field side edge plasma. This value is much more in line with other previous outboard Langmuir probe measurements. Also, mode control feedback data was taken yesterday. Analysis is underway. The new u-port adapter checked out OK and is currently installed in the icrh rack. Reminder: Tomorrow we going up to Argon to re-align Thomson scattering.

Friday February 28 05 16 Maurer/Klein

HBT-EP was up to ~200 Torr of Argon today in order to align Thomson scattering. We are currently pumped down using one cryo to 2e-6 Torr with the base pressure comprised of Argon and H2O. The bake has been turned on. Both cryos were regen'ed during the up to Argon. The TS align had some problems. The stray light level appears to be enhanced over the previous alignment in November. We can discuss this further at the Monday meeting. The LANL supply is back up and running. All the dump-interlocks are now working on the charging supply.

Monday March 3 05 56 Liu 37267-37300 clean up

1. shots 37271-37275 are vacuum shots. When all Krytrons were on and took OH only shot, there was current pickup (about 100-200A) in VF Rogowski coil. However, when VF Krytrons were off, there was no pickup in VF Rogowski coil. This is consistend with prof. Muael's prediction. 2. shots 37276-37300 are clean up shots. plasma lasts 6ms.

Wednesday March 5 06 46 Maurer/Paul 37308-37355 rotation

Doppler rotation data taken using the Thomson scattering viewport to measure the perpendicular plasma He emission. The measured wavelength shift is constant to within 2/100ths of an Angstrom (equivalent to +/- 1km/sec rotation). This constant value will be used to set the baseline of the measurement. There is a baseline shift due to temperature drift that is in the process of being calibrated out now that other systematic errors have been removed. Analysis underway. In taking off the TS collection optics to mount the rotation optics for todays run, a burn spot on the Nikon collection lens was noticed. This might be effecting the position alignment and calibration. Please stop in tomorrow Alex so we can talk. Lastly, we had problems initializing the crates at the end of the day. An unknown RMA error in the action monitor was a sympton. We were unable to correct the problem. cua900 will be rebooted tonight and we will try again tomorrow. This could delay the start of the VF feedback run Thurs.

Thursday March 6 05 27 Maurer/Pedersen 37363-373425 Radial Feedback open loop

Today we did the open loop tests of the radial position control system. It was a very successful day. With the aluminum shells at 8 cm, the plasma moved 2.5 cm in 1.5 msec when we applied a 4 msec square pulse of 250-300 A to the FB coil system. Positive current made the plasmas move inward, negative currents made the plasma move outward. Inward moving plasmas would disrupt early. With the shells at 4 cm, the plasmas were observed to move approximately the same amount but it took about twice as long to get there. The response looks approximately like an exponential decay, consistent with a resistive decay. The outward movements were confirmed by the soft x-ray centroid, which also moved outward. The inward moving plasmas did not have a clear signature on the soft x-ray centroid, presumably because we have several bad channels on the inboard side. Some problems with the LANL control prevented us from doing more fancy waveforms, and will need to be fixed before closed loop feedback can be attempted.

Wednesday March 12 07 07 Paul/Maurer/Cates 37429-37497 Rotation Diagnostic Comissioning

Run was devoted to gaining confidence in the toroidal rotation measurements. Known from earlier runs that the screenroom temperature slightly affects the diagnostic, even though the diagnostic housing is temperature controlled to within 0.2 degrees C. 37429-37445 optimize shots for duration and minimum MHD

37446-37452 track down noise appearing as isolated spikes in rotation diagostic

37453-37464 align detector to look perpendicular to plasma flow; take reproducible shots and record dependence on screenroom temperature

37465-37497 switched to normal toroidal view to look at rotation:

conclusion: 1. once the helium in the plasma is ionized (after 2 msec), the measured value of toroidal velocity is within 1 km/sec of zero.

2. plasma seen to spin up steadily in the toroidal direction up to 6 km/sec in counter-clockwise (looking from above). Plasma velocity abruptly drops to zero (in < .5 msec) about 1 msec before the end of the shot. Plasma always disrupts subsequent to the locking of the plasma rotation. Sometimes the rotation will drop to near zero, but the plasma recovers and spins up again, only to lock a second time and disrupt fatally.

3. For most shots, plasma rotation increases over the lifetime of the shot, never reaching a steady value, when the locking and disruption occurs. On a few, the rotation appears to saturate. On such shots the SXR signal also saturates and rolls over. On the more common shots, SXR increases steadily and drops suddenly just prior to the disruption.

4. I am fairly confident in the results from the rotation diagnostic once the helium is ionized. The next step is to use various techniques, beginning with the bias probe to alter the plasma rotation and see the effect.

Thursday March 13 04 39 Maurer 37498-37514 Feedback & Setup

Today we ran plasmas in preparation for mode control feedback and started to set up the bias probe and mach probe for future experiments. The capacitor bank for the bais probe was run through a dummy load at the end of the day. It appears to be fine. I had a few problems with the pwr supply and a flakey trigger before it worked. We will re-route cabling tomorrow. We should have both probes in the edge plasma tomorrow(that was the goal for today). As a side note, the LANL unit was tested again on Tuesday with its repaired charging supply. The output is railed at 250A(as it was in last weeks expt) when no input is applied! For the initial VF feedback tests of Feb and early March we ran the unit about once a week for a 6-8 hour period. This output is possibly caused by an offset drift somewhere after the input stage that needs to be tweaked or a bad driver tube on the hot deck. We're flying somewhat blind without schematics for the amplifiers so it may take some time to figure out the actual cause. The output tubes appear to be OK in the sense that they turn on and draw the same filament current. Our attempts at actual position feedback will be stalled until this is fixed.

Tuesday March 25 05 21 Maurer 37588-37627 MC Feedback

Todays run goal was suppression of disruptions induced by 3/1 RWMs using the new mode control coil set in order to compare performance of the new system with smart shell feedback. I had great difficulty getting good RWMs to suppress in making the plasmas. The 3/1 external mode observed today was a short bursty fluctuation (almost delta function like) giving a fast kink that only caused minor disruptions. I am using similar bank settings, etc. that we have used in the past to make good RWMs for the smart shell experiments. Varying the Ip ramp rate and initial q after plasma startup did not have a significant effect. Shilov has agreed to run Friday instead of tomorrow, so I will continue with these experiments on Wednesday!

Thursday March 27 05 15 Pedersen 37640-683 ICRH

We ran the ICRH today. Not much progress to report. Most of the day was spent developing discharges that move inboard without disrupting but the run was cut short because of a water leak that is being fixed now. The ICRF was fired on a small number of shots and appeared to behave as usual.

Thursday April 24 05 42 Maurer 37940-37960 biased probe

Ran plasmas in D2 today. Checked out new triple probe circuitry. The new box is working fine (first actual triple probe operation!). Measured Te~12ev in the high field side edge and densities (a few 10^¹² per cc) were similar to those that were measured when the probe was measuring just isat previously. Some minor wiring needs to be done and then we should be able to install the box in the north rack and have it as a day to day diagnostic in the tree (Pedersen). Ran biased probe shots later in the day. Had some problems with the floating potential measurement that need to be fixed. So the run was not too productive. More tomorrow. Ended the day by installing and testing the differential amps in the north rack for real time Ro in the control room for position control. One of the four chs checked out ok, but the other three had clipped square wave outputs with a sine wave input. Not sure why. I will pull the box out and we'll have to test it on the bench to see why. It was working previously. An aside on run summaries... As a group we have been somewhat forgetful as of late in writing up run summaries (myself included). To help in jogging our memories to do so Thomas and I thought it would be useful to remind run operators that writing the summaries is not optional. It is mandatory (similar to the checkout list that must be filled out prior to running hbtep). In the future please submit forgotten run reports the next day or asap. Thanks.

Friday April 25 05 43 Maurer 37961-37967 setup for VF feedback

Progress was made on getting real time Ro in the control room today. (I thought this would take the morning, but it consumed the whole day) Pulled the differential amps for real time Ro in the control room out of the north rack this morning and bench tested them. The dip swtiches for the gain stage must have not been been switching properly. I had set them for unity gain yesterday and I must not have pushed down hard enough to engage the switch on the three that were not working. The clipping observed yesterday was due to them still being set for high gain (G=1000). They work fine now. I changed the input integrator on the board to have an RC time of a hundred microseconds so the west rack signals are not so highly attenuated but still low pass filtered for noise.

Signals from the north rack and then west rack to the control room checked out ok through the whole isolation system. The differential amp gain was adjusted so that the "system" gain was ~1 at 1kHz. Plasmas were then made with the actual Ip and cos1theta signals form the west rack going over to the north rack and then via the iso-network to the control room.

Good news: the data base signals for Ip and Ro look uneffected (recall that the everything looked like a flux loop when previously connected). Not so good news: the observed signals in the control room have features of being correct during the bias and startup phases of plasma operation, but then saturate and are pinned at 4.5V throughout the rest of the discharge. Perhaps there is an extra gain I measured incorrectly.

Tuesday April 29 05 08 Rivera/Maurer none LANL fix

Today we made some progress with fixing the LANLs. The LANL hotdecks were taken apart and tested. It looks like one side of the push-pull output now works. Some history: Previously, problems were fixed in the output stage of the low voltage feedback inputs(input trigger and feedback signal electronics). This had minor effects on the railed LANL output. A bad power supply was then found in one of the hot deck drivers and replaced. After which we were then unable to generate any output at all. Today the trigger and feedback signals were measured at the input to the hot deck and checked out OK. A mosfet that drives the hot deck driver tube on the A-side did not have the same feedback signal superimposed on it as on the B-side (A and B are the push-pull sides). This might still be a culprit in the way it was not working before. We can not say exactly because more than one thing has been fixed prior to being able to measure the output LANL current. We then carefully re-assembled the hot decks and powered the system up. The output is now a half wave rectified version of the input sine wave (with no large offset!). It looks like one side of the push pull configuration is now working. The lack of output after the first hot deck fix was probably due to bad/missed connections or the driver tubes weren't seated properly when it was re-assembled (we think).

Wednesday April 30 04 55 Maurer 37968-37977 biased probe

Ran plasmas in 10% Helium for biased probe rotation studies. Fixed the floating potential measurement box and reinstalled it for the days run. Still could not get a good Vf measurement. The digitizer signal looked essentially like a flux loop. I thought this was due to pickup from my extra long cable runs to the south and then north racks since we've moved the probe 180 degrees around the machine to the microwave interferometer section. Previously when the probe was at the old mach probe stand by the south rack I had no such difficulties with the measurement. After inserting the probe more deeply into the plasma to see some type of plasma potential fluctuation signiture on top of the flux loop pickup on the signal it still looked like a flux loop alone. I then checked continuity to the probe tip mechanically in two different ways. It appears that the feedthrough connections to the probe tip have opened up since it was installed. I believe I'll have to pull it behind the gate valve and check the internal connections out at atmospheric pressure. Too bad. This is now probably a good time to switch the positions of the internal magnetic probe and biased probe stands since I am bringing the bias probe head up to air ( so the new internal magnetic probe is at a chamber section with stainless shells instead of aluminium). We can start this in the background while people run Thurs and Fri.

Thursday May 1 05 05 Cates 37978-38020 RWM Smart Shell Feedback

The Smart Shell Feedback configuration was setup and tested for a RWM run using non disrupting RWM unstable plasmas. shots 37992-38007 were discharges with moderate current ramps which had sizable RWM modes and then large tearing modes later in the shot. Feedback with full gain and coverage was applied to these shots in an attempt to see changes in mode growth rate and rotation frequency. The rest of the shots were an attempt to get a plasma with a mild current ramp which was unstable for a longer period of time, but the plasma was moving around a lot in major radius so I was unable to get a better target plasma to apply feedback to.

ROTATION DIAGNOSTIC

I used the 10% He gas mix and Steve Paul's rotation diagnostic was online - there may have been some offset problems due to the screenroom being warm at the start of the run (shots 37981-37985). For a few shots (eg 38007,38015) the trend of the rotation signal was slowing down with time - the opposite of the normal temporal evolution - this may be due to a changing major radius, but is worth checking out because it is the first time we have seen this trend.

MAJOR RADIUS FEEDBACK

Dave was able to get real time signals of the plasma current and Cos Theta rogowski coils to the control room for use with the dsp.

Thursday May 8 10 43 Cates 38045-38064 RWM Traget Plasma Development

In the morning the Ebara cryo pump adsorber was replaced and the pump was regenerated. The pump temperature was cycling from 8K to ~40K before the change - this symptom was resolved with the new adsorber. Afternoon run was an attempt to create fast startup (high beta) discharges with unstable RWM. I had trouble getting a fast startup and could not drive the initial plasma current above 8 kA. Something seems to have changed in the breakdown. I was running in 100% D2 - last week I attributed this breakdown effect to the introduction of Helium, but maybe something else is going on. Thomas tested the recently modified triple probe box, but did not get good diagnostic signals. The connection of the Cos theta coil to the north rack and control room for use by the dsp for real time radius measurements had to be disconnected because it interfered with the normal digitized signal by introducing a large pickup.

Friday May 9 11 24 Shilov 38067-39106 External Field

I continued with 3/1 external field experiments with current ramps. Running in 100% D2, I experienced the same problem of low initial current as reported by C.Cates in the previous run summary. Raising puff time to 1.7 ms seemed to fix the problem. Something may be wrong with the puff system. I'll check it tomorrow. Data from static and rotating external field application on current ramps was gathered in the afternoon. Still needs to be analyzed.

Tuesday May 13 01 49 Rivera none LANL fix

The LANL amplifier is working again. The output looks the same as it did when first turned on ~2 months ago. However, it should be recalled that at that time there had been an offset in the feedback signal due to a failing tube in the ouput stage. A problem that was not addressed with this fix.

Thursday May 15 01 27 Shilov 38107-38158 External Field

(note: Shilov is referring to krytrons when he writes ignitrons -Nick)

As I expected, the recent problems with plasma start up was due to a half-dead battery in the puff unit. After replacing the battery I was able to reduce the puff time 2 times compared to what I had to set it in the previous run. The reproducibility also improved. Some good data with rotating external field was collected. Later in the day I started having problems with VF CB bias ignitrons (both) firing spontaneously. Nick suggested it was due to ignitrons' age and cut the 'keep alive' wires from the ignitrons. It seemed to fix the problem with the spontaneous firing but produced another problem which was that the bank wouldn't fire ~20% of the shots. That was pretty annoying. The end.

Thursday May 15 04 29 Maurer none fixing cryo

Cates' run was postponed today so we could pump and purge the ebara cryo He supply to see if that stops the temperature cycling that started Monday afternoon (Shilov ran with only the cti cryo Wed). The adsorber that we replaced last Wednesday appears to have only temporarily solved the problem we were having. If this does not fix the problem the seals are mostly likely bad in the expander head and will either have to be replaced or a rebuilt pump head purchased.

Monday May 19 04 36 stillits 38163-38174 Ip signals

Ip raw signals from North rack were tested in DSP. Unfortunately, the output signal from the DSP did not give the correct profile for the plasma current because the raw data received from North rack was influenced by noise (similar to a flux loop). We have to check it out tomorrow (Tuesday).

Thursday May 22 10 35 Maurer none cryo repair

Wednesday was spent taking off the ebara cryo to be sent out and rebuilt. It was FedEx'ed by the end of the day. The gate valve on that pump stand has a through leak at atmospheric pressure. This brought the base pressure in the chamber up to a few times 1e-7Torr while we were taking the cryo off. We blanked off the 10" gate valve and pulled a rough vacuum behind it to stop the through leak. The base pressure in the chamber is now back down to what it was prior to taking the cryo off at 1e-8Torr.

Tuesday June 3 04 43 Pedersen/Stillits/Rivera none LANL/Radial feedback

Today's objective was to control the LANL power supply (driving the feedback coils) directly from the DSP in open loop configuration. We were partly successful: We were able to control the LANL with the DSP but the signals were noisy and distorted to the extent that this problem will need to be fixed. We narrowed the problem down to the optical isolation output having trouble driving the input on the LANL when the LANL is gated on. We are investigating why. Otherwise, the LANL was working well (except tripping too often), there were minimal current offsets, when driven directly by a signal generator. Also, the optical isolation network seems to be fully functional except for its inability to drive the LANL input. Signals looked clean when input into a scope at both 1MOhm and 50 Ohm.

Monday June 9 02 54 meeting time

The weekly HBTEP status meeting shall forthwith be held at 9:30am Monday mornings in the control room.

Thursday June 12 10 24 Muarer none intalling cryo

The rebuilt cryo arrived yesterday and was re-installed in the afternoon. It was cooled down late yesterday and was valved open to the chamber this morning. Base pressure is back down to 6.6 e-9Torr with the rebuilt cryo head a 8K. Shilov is making plasmas today.

Friday June 13 10 45 Shilov 38276-38314

Clean up shots. Plasmas OK

Monday June 16 05 19 Pedersen/Stillits/Rivera none LANL/Radial control

We used the afternoon to check out the LANL some more. A reference signal was reconnected following a conversation between Dave Nadle and Nick Rivera. This did not substantially improve the signals, which were rather noisy. We then tried to use a 1.5 V battery to generate an input signal to the LANL certifiably free of ground loops. The LANL basically ignored this signal and put out a current of -5 Amps regardless of the polarity of the battery, and whether it was hooked up at all, and even when it was connected in parallel with a large capacitor. I now suspect that the input on the LANL is malfunctioning. On the bright side, 5 A of offset is 40 times better than it used to be. An arc inside the LANL stopped the run and will be checked out tomorrow.

Wednesday June 18 10 30 Shilov 38315-38384 RWM, External Field

I spent the morning developing reproducible target discharge with q* approaching 3 slowly from below. It was successfully achieved after a while. Later in the day I applied static and phase flip external fields to the plasmas. The results look promising. The data is yet to be analyzed.

Monday June 23 03 04 Cates 38423-38454 RWM - IP Ramp

Friday 6/20 - Discharge development in an attempt to get a plasma with a RWM which grows slow enough to observe growth rates without disrupting the plasma. By the end of the run, I was able to get plasmas which would work well for a systematic feedback study. These shots (38444-38454) had large 3/1 modes which grew and saturated in .25-.75 ms with frequencies of 4-7 kHz - excellent targets for our smart shell feedback system.

Tuesday June 24 05 29 Cates 38455-38490 RWM Smart Shell Feedback

I continued making the mild current ramp discharges which looked promising from last week. After finding a fairly reproducible shot, I applied full gain smart shell feedback to the plasmas. The feedback clearly alters the rotation frequency of the mode, and sometimes reduces the amplitude. I will need to check the mode structure and growth rates more carefully. I am looking at the RWM in the first 4 ms of the shot - but these plasmas do last to 8-9 ms with significant SXR sawteeth and saturated tearing modes in the second half of the shot.

Tuesday June 24 09 03 Shilov 38385-38422 External Field

I continued with 3/1 external field experiments on the target discharge developed previous day ( ex. Shot ^#38371 ). The naturally growing RWM was observed with a short unstable time window between ~2.3 3 ms. The plasmas lasted for ~ 3-4 ms. I applied static external field with phase flips at 2.5ms and 3 ms. After analyzing the data, it looked like the RWM mode is unstable around q ~ 2.8 and stable above that value which could be related to the error of the edge q measurement. Unfortunately the data collected wasnt sufficient to make any conclusion on the difference between plasma field phase rearrangement time at 2.5 and 3 ms. I will continue with the experiment this week.

Thursday June 26 11 51 Shilov 38492-38542 External field

Continued with 3/1 external field experiments with the previously developed discharge. Took a lot of phase flip shots with different flip times. The data is being analyzed.

Friday June 27 02 01 Alex Klein n.a. T.S. calibration

HBT-EP was filled to 170 Torr nitrogen. The non-laser wavelength channels showed a signal when laser was fired through the chamber, due to Raman scattering. I checked to note absence of breakdown in laser beam at this pressure. Collection optics were carefully aligned to maximize signal, then I proceeded to pump down and to collect sets of 5 data sets (for averaging) on scope at several pressures. The signals appear to depend linearly on fill pressure (good). Complete dissappearance of signals at pressures below 1 Torr show absence of stray light leakage into non-laser wavelength channels. Ease and accuracy of interpreting T.C. gauge for nitrogen is a plus. Ease of aligning collection optics in the absence of stray light is a great plus.

A minus: Chamber appears to be water logged; it remains to be seen how dirty of an ordeal a nitrogen calibration is. Will analyze todays data and write short report on Thomson Scattering system.

Monday June 30 05 04 cates/stillits 38542-38559 cleanup & ip-signal testing

Cleanup runs were performed. Ip-signals from north rack: signals are looking much better (having almost the right profile), but the signals are too constant - looks almost like a step-function (which gives a linearly increasing plasma current). We have to check that tomorrow.

Tuesday July 1 05 56 Cates 38560-38619 Clean Up

Clean Up shots taken to recover from up to nitrogen for Raman scattering calibration. Current Ramp discharges are fairly consistent, although the loop voltage is still around 10 V for a standard 10kA plasma (compared to around 5 V last week). Royce has been testing the triple probe: We tested the triple probe's electronics circuitry box today using the oscilloscope. Channel V1 and V3 performed as expected. V2 which will indicate the temperature is still saturating the scope so we are reducing the floating potential from 1/5 Vo to about 1/10 or 1/20 to get a complete measurement; Overall a successful test. Remi has noticed a large magnetic pickup on the soft xray tomography channels which is present even when diodes are disconnected and amplifiers are off - he is testing the system to see if he can improve the pickup.

Thursday July 3 01 48 Cates 38620-38656 CLean Up

On Wed. I tried to reproduce earlier RWM discharges, but the machine was not quite clean enough to get reproducible results. As a reference the Loop voltage for 10kA current ramp shots was down to 7-8 V. So half way through the day I switched to taking high current, outboard limited plasmas which lasted as long as possible in an attempt to scrape the limiters clean. These shots improved through the day. Several diagnostics were tested during the day: Yuhong discovered that the flux loops have been disconnected for the month of June (shots 38257-38647). Remi was able to improve the magnetic pickup on the soft xray tomography channels by finding and removing a ground loop in the copper box. The triple probe electronics box was tested by Royce and signals were very noisy.

Tuesday July 8 05 18 Cates 38691-38723 RWM Current Ramps

This morning I took about 15 shots with high current and edge q near 2 in order to help clean up the outboard limiters. These shots were also good for testing the SXT channels. In the afternoon I switched back to mild current ramps in order to produce observable RWM growth rates. I was able to get good target plasmas with Ip and minor radius increasing monotonically and q edge slowly dropping below 3 to induce a 3/1 RWM at around 2-3 ms. See shot 38719 as an example. Unfortunately about the same time that I was able to get reproducible plasmas - the South Rack digitizer signals became flaky and I had serial highway errors almost every shot when storing the South rack data (the feedback sensor coil data). The south rack is being cooled properly and I am not sure what is causing the serial highway errors. Royce, Remi and Nicolai were working on the triple probe, sxt and vf feedback diagnostics respectively throughout the day - no big news for any of those systems.

Thursday July 10 09 58 Shilov 38724 - 38777 Ext. Field

I started developing a good discharge for RWM external field experiment in the morning. In the afternoon I started getting reproducible plasmas suitable for the experiment. Around that time the same serial highway error appeared and persisted as it did the day before. After a while it was in every shot. So no signals from the south rack could be digitized. I completed the run.

Friday July 11 05 01 Cates 38823-28870 RWM Smart Shell Feedback

First I examined some shots Mikhail has worked on with edge q going from 2.6 to above 3. After checking to make sure the feedback system was in good shape (one board needed a minor repair - signal 5BS1 was bad) I took full gain feedback shots on these discharges. The best shots are 38848-38859. I then switched back to the more standard current ramp shots I have worked with in the last couple weeks in order to complete a gain scan with feedback. Unfortunately the South rack serial highway problems began around 3:15. I did a few tests trying to eliminate this problem and found out that the problem does not go away after an hour with the south rack crate off, that it is not the crate controller or u-port adapter. It must be a problem with the camac crate itself, it did not occur until the crate had been on for about 6 hours, and did not go away in an hour. The plasmas were well behaved with loop voltages around 5V and the feedback system is working well, so it was a disappointment to quit early due to camac problems.

Tuesday July 15 06 08 Cates 38897-38960 RWM Smart Shell Feedback

I completed a gain scan with relative gain values of 1, .5 and .25 on current ramp plasmas with non-disruptive RWM's.

example shots:

38901 fb off

38902 fb on full gain

38934 fb 0.5 gain

38950 fb 0.25 gain

The plasmas were fairly reproducible and the feedback was working well. -Alex took Thomson scattering data - the laser fired flawlessly - the signal strengths were a little weak - which could be explained by lower density and temperature of plasma during current ramps. -Royce continued testing the triple probe circuitry and Remi tested soft x-ray tomography amplifiers. -Nicolai and Dave tested some buffer amps for getting coil current signals to the control room for use in the radial position feedback system.

Wednesday July 16 12 30 Alex Klein

Hello all: The following is a draft of the Thomson Scattering poster abstract to be submitted to APS. Any comments, suggestions, words of approval, etc. would be greatly appreciated. Thanks very much, Alex

"Thomson Scattering on the HBT-EP Tokamak" A. Klein, D.A. Mauer, M.E. Mauel, G.A. Navratil, C. Cates, M. Shilov,??? Thomson Scattering can provide accurate, non-intrusive measurements of the electron density and temperature in a plasma at a spatially and temporally localized point. Thus, this diagnostic represents one of the most attractive methods for measuring essential plasma parameters. We have revived the existing Thomson Scattering system on HBT-EP and improved the resolution and calibration of the instrument. The system is based on a design in use at DIII-D (Carlstrom, et al, Rev. Sci. Instr. 61, 2858, 1990), and uses a 800 mJ pulsed Nd-YAG laser and a five channel interference filter polychromater. Absolute system calibration is achieved using Raman scattering in nitrogen (Yamada, et al, Rev. Sci. Instr. 74, 1675, 2003), because stray light levels have become excessively high and the Raman technique is more accurate than calibrartion based on Raleigh scattering alone. The scattering system is triggered at key times during the 10 msec tokamak discharge; the obtained measurements are compared with conductivity temperature data for several plasma configurations.

Wednesday July 16 06 15 Cates 38961-39009 RWM Smart Shell Feedback

In the morning I took some more feedback data with reduced feedback gain to complete yesterday's gain scan. In the afternoon I ran the feedback at full gain with only the 10 midplane coils activated. I was able to get several good shots with feedback. examples:

38987 fb on

38986 fb off

Unfortunately I lost about two hours due to yet another camac thermal problem - a crate in the west rack has a bad fan on its power supply, and the power supply would shut off every few minutes due to overheating. The crate was stable the rest of the afternoon after adding an external fan to the power supply case. Nicolai and Dave successfully tested a buffer amp to fan out the vfc and ohc signals to the control room for radial position feedback.

Sunday July 20 11 00 Shilov 39010-39133 Ext. Field

I continued with 3/1 external field experiments on "high beta" plasmas. Collected some good phase flip data on Thursday. On Friday I was struggling to get the same reproducible discharge but the plasma s were really disruptive and I couldn't sustained then for more than 4 ms. After a while I realized that the TF bank setting was changed from 6.1 to 5.7 kV some time from Thursday evening to Friday morning without indicating that in the data base. After raising it to regular value the plasmas got much better but unfortunately the south rack camac error appeared at the same time. So I had to finish the run.

Tuesday July 22 04 58 Alex Klein 39137-39195 T.S. parameter space

Rather than run Thomson scattering shots yesterday afternoon, the entire day today was used for that purpose. Yesterday, the biased probe and the internal magnetic probe were installed instead. Todays runs: D. Maurer spent a good bit of time trying to create low temp. high n plasmas, to see the T.S. signals clearly with those kinds of plasmas (Ch. 2,3,4 should vary greatly in the 1-10 eV range). A few shots appear to show low temps according to T.S. data. Fill pressure does correlate with higher n according to T.S. data. Ch. 4 gives generally very hefty signal (up to 250 mV) and correlates well with density for a range of temperatures. The afternoon was devoted to try to achieve max temp and moderate densities, to try to see a signal on channel 5, which is supposed to be sensitive to higher temperature plasmas. Unfortunately, could not see any signal in Ch. 5. Will analyze all data from today more thoroughly later.

Tuesday July 29 05 38 Maurer 39299-39333 cleanup

Cleanup shots only were taken today. As mentioned at the Monday meeting the base pressure was an order of mangitude too high after the weekend bake out and had a N2/O2 ratio characteristic of air. A blown circuit braker and bad pump on a manifold that has a high pressure leak caused the problem. The pump was replaced yesterday and the pressure came down to 1e-8Torr. The manifold that it was connected to contains the limiter piece that we had difficultly sealing after our last up-to-air. This is most likely the location of the leak although we did not leak check separate sections of the manifold to prove it. The first twenty shots taken today were cold with no SXRs and Vloop around 10V. By the end of the day I was seeing some SXRs and the loop voltage was coming back down to 7V. We probably need another morning of cleanup before we get good discharges.

Thursday July 31 04 40 stillits/shilov 39334-39362 testing of signal

The raw and semi-integrated cos1T-signals were finetuned. The signals match the signals on the HBTEP-Tree. PS. CTX has asked us to shot down their camac crate every evening.

Wednesday August 6 12 15 Shilov/S.Pauel 39377-39422 Rot. Diag/RWM

The objective of the experiments was to reproduce RWM shots I use in my external field experiments running with 10%He 90%D2 so the Princeton rotation diagnostic can be used to measure ion rotation. As I pointed out before the plasmas were more disruptive compared to those with 100% D2 and it was hard to maintain q* between 2 and 3. In several discharges which parameters were close to those of target plasmas rotation diagnostic showed pretty constant ion rotation rate of 2-4 km/s. Also speculation can be made that the ion rotation is slowed down by the plasma magnetic field fluctuations. Good shot to look at are:

39413

39422

39409

Also, as S. Pauel noted, the ion rotation measurement could be affected by the plasma moving inwards, which was necessary to maintain the desired q* profile. Since the view line of the rotation diagnostic is fixed, the He rotation was measured at different plasma minor radial.

On Friday 08/06 I tried reducing the He concentration from 10% to 5% by mixing gases from two bottles in the puff line. After the first attempt of mixing 100%D2 with 10%He and 90%D2. The rotation diagnostic radiation levels stayed the same. After the second attempt the levels went below the level required for rotation measurement. I assumed this technique of mixing wasnt working properly.

Wednesday August 6 04 00 Maurer/James/Klein 39432-39442 Probes/TS

Fixed offsets on the triple probe today. Fixed cabling and setup Mach probe. Should be able to piggy back and get Mach ignals in the tree tomorrow. Bad news: TS laser alignment was bumped. The laser was realigned through the aperatures on the optical table. Thomson signal was seen with plasma shots. Data analysis underway to see how much realigned signal magnitude differs from last weeks shots.

Thursday August 7 04 02 Maurer/James/Klein none setup

No plasmas made today.

Setup the capacitor bank, trigger and cabling for the bias probe. Finished setting up the Mach probe. In the process of installing the internal magnetic probe. Should have new data from all three tomorrow.

Friday August 8 05 28 Mauer/Liu/James 39447-39472 Diagnostic

D. Mauer tested the Bias Probe and data was taken. R. James positioned the Mach probe in the plasma (shot 39469), but the signals were weak. We will investigate more next week to improve the signl strength. Y. Liu hooked up the Hall internal magnetic probe system. The whole system is now operational! But there are some things that need to be checked -- channel two is noisy compared to channels one and three; channel four has some problems with the amplifier.

Monday August 11 04 13 stillits/cates 39475-39483 vfc/R - signals

The VFC-signal was fine tuned.

Output-signals for major radius, R, was obtained. The maximum difference in major radius from the MDS-scope and the output-signal on the scope is 3-4 mm.

Monday August 11 06 24 James/Cates 39484-39486 Triple Probe

Just a few shots taken. C. Cates gave an instructional session on how to affect changes to the plasma by adjusting various parameters. Unfortunately, we ran out of time. These 3 shots are an initial attempt to produce cold/inboard limited plasmas, ideal for testing the triple probe. We will continue to produce these cold plasmas plus hot, sparse, and dense plasmas for more testing in subsequent runs.

Friday August 15 11 52 Cates --- Power Outage

Due to the power outage yesterday the vacuum chamber went up to a maximum pressure of 2.2 mTorr. It quickly pumped down to 8e-8 Torr with the turbo pump, and is mostly water. Both cryo pumps have been regenerated and are now cooling. The ignitron cooling lines burst this morning when power was restored and flooded the basement. The water is mopped up and should be ok by Monday. The servers (cua900, beta, sherwood) are back up and running normally.

Tuesday August 19 04 36 Maurer/Rivera none none

No plasmas today. TF cooling oil system problems. The pump on the TF cooling oil turned on briefly and then tripped its 20A circuit breaker while getting ready to run. In inspecting the unit the pump motor is drawing 33A(the breaker appears OK). The starting and running capacitors both checked out OK. The bearings on the unit also seem OK. The compressor part of the unit turns on OK. The motor/pump unit does sound like its running loud compared to previous operation.

We are not sure if it is the pump itself or the motor that is causing problems at this point. We will remove the motor and pump unit tomorrow and test the motor to sort out what is going on.

Wednesday August 20 10 45 M. Shilov/ S. Paul 39382-39422, 39537-39550 Princeton Rotation

Ion rotation rate in the shots similar to high beta 3/1 shots stayed ~ 3 km/s. Natural mode rotation in these shots with 10%He is somewhat less and stays ~ 3-4 kHz whereas the mode rotation in 100% D2 shots is ~ 5-6 kHz.Also S. Paul noted that due to the plasma major radius shift by ~ 1-1.5 cm the rotation measurement could be taken for different minor radia and could not represent the plasma rotation dynamics. Good shots to look at: 39409 & 39412 & 39422 Ion rotation rate starts building up ( upto 8 ms/ in the shots with higher current ramp at ~ 4 ms when significant increase in SXR signals is observed and sawtoothing starts.Ex. Shot: ^#39538

Wednesday August 20 01 25 Rivera none

The motor was taken out of the TF cooling unit this morning and tested. It is broken. A new motor has been ordered ($350) and sent next day from a distribuor from NJ. If it gets shipped today we will have it Thursday, otherwise it should arrive Friday for installation. We are unable to run until this problem is fixed.

Thursday August 21 04 20 Maurer 39554-39574 clean up

The new motor for the TF chiller was installed this morning. The TF oil pump is working fine now.

Clean up shots taken in the afternoon. I had about 0.25V of SXRs on the central fan array channels by the end of the run. The machine cleaned up in about 15 shots.

The basement AC is broken. This caused serial highway errors at the start of the run. As a temporary fix until the AC is working, opening the basement rack door and using a fan to cool the rack got rid of the CAMAC errors. Shilov is running tomorrow.

Wednesday August 27 02 14 Maurer 39593-39621 Biased probe

Biased probe experiments were run all day Tues and Wed morning. The probe is up and running. The problems that I had last time with a fraction of the bank voltage appearing on the probe prior to gating the scr appear to be caused by leakage current through the scr itself. I removed a capacitor on the probe side of the scr that was put in place to attenuate high frequency noise on the floating potential measurement. The voltage that now appears on the probe prior to gating only charges the capacitance of the probe shaft which is small and quickly discharges through the plasma during the breakdown phase. I can run this way now, but we'll spec out a new scr for future use.

The shots taken were essentially probe clean-up discharges. With V_probe=175V the probe current was maximum of about 60A and changed significantly during the shot. The probe was inserted 2cm into the plasma edge. The plasmas would typically be pushed in by the VF when the probe was pulsed on, most likely due to impurites coming off of the probe itself. I think I will probably need another 1/2 day of cleanup shots.

Wednesday August 27 07 06 Liu 39622-39649 calibrate hall probe

All channels of internal magnetic probe are currently working well. Thanks Nick! VF only shots were taken to check all the channels. From those VF only shots, it's obvious that the vacuum chamber delays the soaking through of the vertical field a lot. A few TF only shots were taken to calibrate the hall sensors and more analysis will come out later on. Plasma shots show that the hall sensor signals are pretty quiet, which are within the bit error (~1Gauss). However, I couldn't see the magnetic field fluctuation, which is about 5 Gausses seeing from SMPs.

Thursday August 28 03 06 stillits/rivera xxxx Lanl

The lanl is working ! Excellent job, Nick !

We triggered the lanl with pulses (from the dsp) of different polarities and durations, and the lanl did what it was supposed to do.

Tuesday September 2 04 48 stillits 39664-39726 Testing Lanl ouput-signal

Lanl's output-signal is on the Tree (hbtep::top.camac.north:L8210_3:input_2). We pulsed (with a square wave) the lanl by the use of the dsp. We did not get perfect, but ok, output-signals. The spikes on the scope (in the control room), however, are gone ! The maximum difference between the output-signal on the mds-scope and the scope (in the control room) is 20-30 A.

Thursday September 4 04 34 Maurer/Liu/Stillts 39746-39768 Bias probe

Ran more bias probe dischages with the probe tip inserted 4cm into the plasma and biased at 150V into electron stauration. Still cleaning up the tip during the first 10 or so shots. Later in the day I saw the first evidence of driving MHD mode rotation to zero and then in the reverse direction. This phase did not last long, maybe 1/2 to 1msec before the modes would re-accelerate in the opposite (natural rotation) direction. There appear to be amplitude effects while this occurs that can be looked at to see what the correlation is between mode rotation and mode amplitude.

Facilites workmen finally stopped by and looked at the AC in the bank room. They need a part to fix the unit and said they will be back on Monday to finish the repair.

Nicolai worked on radial position control signals during the run and took more shots afterward. They seem to be OK after some problems yesterday with one of the buffer amps.

Yuhong also did some Hall probe shots at the end of the day to see how running more than one detector off of a single battery effects noise levels.

Thursday September 4 04 42 stillits/maurer 39726-39743 Signals. From yesterday:

We had hoped to take some open-loop shots, but we experienced some problems with the cos1T-signal. However, this problem was solved today.

Friday September 5 02 17 Maurer none

No shots today. H20 shutdown in Mudd Bldg for plumbing work. Regen'ed Ebara cryo since the compressor was turned off.

Saturday September 6 02 16 Maurer/Liu 39786-39820 Biased probe

Continued biased probe experiments from Thursday. Fixed a pretriggering problem and then slowly increased the bias probe voltage over the course of the run to V_bprobe=250V. The probe appears to be relatively clean now. I was able to reverse mode rotation from the induced Er. In some cases this did appear to effect the external mode amplitude. Hall probe data taken also.

Monday September 8 05 19 stillits/maurer 39821-39860 testing of signals

The cos1T- and Ip- channels on the diff. ampl. have been fixed and are now working. We are getting slightly better values for the major radius; the max. deviation is 3 mm (compared to the mds-scope).

Tuesday September 9 06 25 pedersen/stillits 39861-39884 open- and closed-loop tests

Open-loop tests (39862-39868) were conducted with success.

Moreover, closed-loop tests (39883-39884) were conducted. The desired position was R = 93 cm. The lanl was triggered after 1.4 ms, and the closed-loop algorithm gave the correct lanl-current. In both closed-loop runs the plasma moved in to R = 93 cm after 2ms before it moved out to R ~ 93.5 cm. So, fine-tuning of the feedback coefficients needs to be conducted to improve the feedback system.

Wednesday September 10 06 34 Maurer/Pedersen/Stillits 39885-39951 Biased probe/VFFB

Plasmas were ran today for closed loop VFFB tests and biased probe rotation experiments.

The biased probe voltage was systematically scanned from 100V up to 400V in a series of current ramp discharges that caused the q=3 surface to go external at about 2ms. The induced rotation appears to stabilize the m=3 mode as it goes out into the vacuum. More data analysis needed though. Also, in monitoring the triple probe density signal there was a systematic density increase with increasing applied bias voltage from 100V to 250V on the probe. Unfortunantly the triple probe stopped working during the run, most likely due to a blown chip, so I was not able to see the increase up to the 400V level. The probe current appears to saturate somewhat once the voltage is up past 150V at 50 to 60A. Tomorrow I will run plasmas with 10% He and the Doppler rotation diagnostic.

We made several closed-loop runs today (39891-39906) with different desired positions of the major radius (91, 92, and 93 cm). Some runs were more successfull than others. The biggest problem with the closed-loop runs seems to be the spikes in the dsp-calculation of the major radius, but we have fixed that by using an interpolation of the major radius. Moreover, we also experienced problems with triggering of the DSP; that also needs to be fixed. We will continue the closed-loop runs tomorrow.

Thursday September 11 05 37 Maurer 39952-39991 Biased probe RWM expts

Discharges were made today in 10% He 90% D2 to study the effects of biased probe driven ExB rotation on m=3 RWMs.

I setup a control shot similar to Shilov's with an initial edge q of about 2.7 slowly moving inward in major radius. The edge q would go through the 3 surface at about 3ms. This gave me a burst of m=3 RWM that lasted from 2 to 3ms of typically 3G max amplitude and frequency around 2 to 5kHz. To try and keep the RWM stability evolution due to the evolving current profile the same on a shot to shot basis I kept the start up bank settings the same and varied the OHEL a little on a shot to shot basis to always keep the q=3 surface entering the plasma edge at 3ms.

I gated the Bprobe on at different times during the wall mode burst with 100V or 200V on the tip. At the lower voltage I was able to torque the RWM frequency to zero and then down to -5 or sometimes -10kHz. The mode was still present at these rotation speeds. There were periods where the RWM amplitude appears modulated and correlated with frequency modulations as the mode rotation changed due to the probe.

The most interesting effects were observed at 200V on the tip. Gating the probe on at 1.5ms just prior to the usual onset of the RWM at 2ms spins the plasma enough to essentially suppress the RWM. What MHD is there before the probe is turned on (start up modes/noise) appear to rotate up -20kHz or so. More data analysis needed. To see this effect look at shots 39986 or 39987. A control shot is 39971 to see the typical RWM burst with no probe voltage. Unfortunantly, the Bprobe had some arcing problems at 200V that I don't understand, so I don't have too many shots like these yet. There also was a systematic change at 200V on the rotation diagnostic (although somewhat small maybe?). I will continue the run tomorrow once I cure the arcing problem.

Monday September 15 04 16 Maurer 39998-4007

Attempted to fix the bias probe by conditioning the tip with 1.5KV at 5mA power supply. The probe tip to ground resistance was 85kOhms before conditioning. After high-potting, the resistance to ground was about 5MOhms measured using the 500V meter. Vacuum shots with the probe tip at 350V had the correct RC decay. Plasma shots with the probe tip inserted 4cm into the plasma with 200V on the tip are still arcing unfortunantly. I will take the probe out tommorrow morning.

Good news: Triple probe electronics were re-installed and are up and running. Royce also finished bench tests of the differential amp for the Mach probe. We should be able to install it tomorrow.

Nicolai re-intalled his differential amp and tested out his signals. Everything is OK.

Tuesday September 16 05 38 stillits/maurer 40008-40030 testing of signals

Biased probe removed this morning. Arcing damage was found from the copper conductor to the stainless steel housing back by the mounting conflate flange. There did not appear to be substantial arcing or tracking on the Alumina itself from the Moly electrode. The probe is already under repair. We are shooting to have it fixed and re-installed by Friday.

The Hall probe saw its first kink mode fluctuations!

The signals were tested and are now working. Moreover, the new interpolated major radius calculation is also working (max. deviation is 1-2 mm) and without any significiant spikes.

Wednesday September 17 05 38 stillits 40031-40066 Testing of algoritm

I spent most of the day eliminating bugs in the feedback program. They were found and eliminated, and we got very nice and smooth output signals for the major radius. However, the output signal for the plasma velocity is not smooth enough, but that will be fixed tomorrow.

Thursday September 18 06 50 stillits 40067-40090 Testing algoritm

The new feedback algoritm with the velocity term added is working. The major radius has a max. deviation of 2-3 mm, and the calculated plasma velocity is close to the velocity in the scope. Thus, we should be ready to make the first closed-loop run tomorrow morning.

Friday September 19 05 18 stillits 40091-40114 cloased-loop runs and testing

40091-40108: testing of feedback algorithm with derivative gain. closed-loop runs:

40109-40110, we tried to stabilize the plasma at R=93 cm. It was partly successfull.

40111-40114 (except 40112, the dsp was not triggered), the desired position was R=92 cm (40111 is awesome !).

In these shots there are small deviations in the actual position, but they are only of a few milimeters. In future runs we will let the lanl be switched on for more than 3 msec because 3 msec is too short, as can be seen in some of the runs.

Monday September 22 06 45 stillits 40123-40146 cloased-loop runs and testing

Today, the lanl was turned on for 6ms (earlier it was 3 ms). Moreover, in all shots the desired major radius was R=93 cm.

We were able to stabilize the plasma at ~ 93 cm for almost 7 ms (shot number 40134). So, the lanl is working for 6ms.

We are experiencing problems shutting off the lanl when cos1T reaches a certain value. That needs to be fixed.

Corey fixed the major radius so that the major radius, which influences q and beta, is automatically recalibrated.

Tuesday September 23 06 36 Liu/Stillits 40147-40202 Feedback + Hall probe

We ran the feedback for R=93 cm (40147-40154), R=92 cm (40155-40165), and R=94 cm (40166-40178). Moreover, the duration of how long the lanl is switched on was increased from 6ms to 9 ms (from shot 40169). Because of that we were able to repeatedly create plasma of 7-8ms duration. However, the shots for R=94 cm are little too volatile, but that can be adjusted by lowering VF (the lanl was not triggered in shot 40170 and 40173). (Stillits)

Hall probe was checked for its low measurements on poloidal field. The reason is supposed to be the influence of the high toroidal field. I am trying to figure out if I can fix this problem. (Liu)

Wednesday September 24 05 21 stillits 40203-40229 feedback

Channel 2 (the Ip-signal) on the diff. ampl. was down today. So, shots 40203-40217 are useless.

In shots 40218-40222 we tried to stabilize the plasma at R=94 cm, and in 40222 we created a hot plasma that lasted for ~ 8.5 ms. In the rest of the shots we tried R=91 cm, but it was not a huge success. We were able to stabilize the plasma at R=91 cm for max. 1.5 ms (40223 and 40226). But it can probably be improved by changing some of the parameters.

Tuesday September 30 04 51 stillits 40285-40287 testing signals

We tested the Ip-signal (ch. 2). We are getting a signal now, but the signal looks like a ground-loop signal.

Wednesday October 1 05 13 Mauer/James 40288 - 40305 ICRH/Bias Probe

ICRH fired successfully today! We are still waiting for another crate power supply from Princeton or to receive one being repaired to get comprehensive data, but we did see the ICRH fire via an oscilloscope. We also saw pickups on the mach probe from ICRH which fires at about .002 sec. It looks like one of the channels is OOC on the Mach probe again so I am going to trouble shoot it tomorrow [James].

The bias probe is up and running again(only tested at low voltage yet). I had some problems with triggering that are fixed now at the start of the run. The SCR does appear to be going bad that is used to switch the bank voltage on to the probe tip. I was getting about 10V leakage on the tip prior to pulsing it with only 50V total on the bank. Probe cleanup shots tomorrow.

Thursday October 2 05 22 Maurer biased probe

Took biased probe cleanup shots today. The run was cut short because of problems with the TF bank. An eyebolt nut came loose and caused the one side of the rope dump to short the bank while it was being charged (the bank was at about 5kV). This boiled the local dump water resistor and blew its top off. The cupper sulfate H20 blown out of the first dump resistor appears to have then allowed a conductive path from the neighboring dump resistor, which was at whatever residual voltage was left on the bank, and then shorted the rest of the TF charge to ground burning an arc path in the wood shelf that the dump resistors sit on. The H20 has been cleaned up. We still need to cut out the carbonized material on the wooden shelf before we run again. This should be finished tomorrow morning.

Friday October 3 06 07 Maurer/James/Klein 40319 - 40350 Bias probe/ICRH

Tf repair was completed this morning(Nick and Jim). The bank was brought up to full voltage before lunch and tested OK. Plasmas were made in the afternoon and more biased probe clean-up shots taken. The probe tip was biased to 200V by the end of the day. It appears to still be a little dirty. More clean-up tomorrow at higher voltage.

The power supply for the ICRH crate was installed and is a permanent member of the serial highway! Took a few shots and worked out some triggering bugs. Now, the signals are in the tree and showing up on the scope but some of the inputs look switched (will diagnose this weekend). The Mach Probe is also installed and the digitizer for one of the channels is the problem and not the electronics; it will be switched to another open channel.[James]

Saturday October 4 01 45 Maurer 40352-40382 biased probe

Continued taking biased probe cleanup shots today. V_probe= 300V with I_probe=30 to 50A. Nothing exciting to report.

Tuesday October 7 05 34 Maurer/Stillits 40383-40422 Biased probe/VFFB

More bias probe clean-up shots and work on the VFFB signal problem today. First hints of a temperature increase on the triple probe were seen in some shots today. Not as dramatic as previously observed but an improvement from Saturdays run. Typical currents with 300V on the probe are about 50A. The probe appears to be clean enough to apply 100 to maybe 200V on the tip without disrupting the plasma. shots taken with 300Vs still cause the plasma to move in and disrupt, probably due to impurity influx. Hopefully the 200V case will be a useful ICRH target for tomorrows run.

The feedback system is up and running. Nick solved the problem we have had with ch. 2 on the dsp.

Wednesday October 8 06 36 Mauer/James/Stillits 40423-40461 ICRH/Feedback

It took about half he day to with the scope and signal generator to put the ICRH system back together again. Once that was figured out, we took a few shots before Shilovs Invited talk dry run. We were unable to perform the high density inboard limited plasmas with the working ICRF because of time restraints, but we were able to get some good baseline ICRF data while tuning the feedback system. If there is time, tomorrow afternoon we will try again with the bias probe (to get high density plasmas).

Subsequently, the feedback system has been fixed and all channels are operating normally. The feedback system (and most of the other diagnostics) was greatly affected by the ICRH discharge. Some sort of filter or shielding will have to be explored now that we are actively trying to couple the plasma.[James]

Monday October 13 05 22 James/Stillits 40557 - 40590 Testing (Triple Probe/Feedback)

Shots 40557-40579 Feedback shots to calibrate system with newly installed filters. ICRH Will fire tomorrow to test filters. [Stillits]

Shots 405780-40590

Tested the triple probe and mach probe electronics and they are working at spec. on the bench. When installed in the rack, density signal (triple probe) is still erratic and the right channel (mach probe) is flat/no signal. Concentrating on the triple probe (for the ICRH run on wed. and thurs.), I tested the cabling from the probe to the electronics with the high voltage meter and they were good. Next I bypassed the t-probe electronics and used the m-probes power supply, rogowski coils, and electronics/o-scope to test the raw t-probe signals. After 10 shots, the probe appears to be working normally, but more testing tomorrow so we hope to lock in a solution. Alex was able to prep Thompson Scattering and get signals into the o-scope. It is ready to go for tomorrows Testing. Thanks to Yuhong for hanging around and babysitting the new plasma shooters! [James]

Tuesday October 14 05 35 Klein/Stillits 40610-40630 Thomson Scattering

The plan was to create very hot plasmas which tend to be outboard limited, then use position control to (rapidly) push the plasma inwards into the field of view of the TS collection optics. In preparation yesterday, I (Klein) crudely aligned the laser. Should have done this more critically.

Today: Very healthy signals on all channels, including ^#5, which is supposed to come in to play for hot plasmas, was very encouraging. These signals were present for warm plasmas at r = >94 cm, surprising. Then we made hotter plasmas and pushed them inboards, as planned. Timed TS to coincide with r= 92 cm position. Signals looked the same. Finally, TS was fired after plasmas had long died: signals looked the same. Conclusion: laser beam is poorly aligned, scraping some edge or surface, overwhelming stray light is creeping into each channel. Will re-align laser and try again...

Tuesday October 14 05 50 stillits/klein/james 40592-40649 testing of signal + TS + IRCH

The feedback system is up and running again with the new filters installed. The max. deviation is only a few millimeters. That is also the case when ICRH is turned on. The problem we had getting the lanl to give a zero current when the semi-integrated signal for cos1T is less than zero has been fixed. So, we should be able to run the ICRH with feedback tomorrow.

Wednesday October 15 05 27 stillits/james 40652-40667 feedback + ICRH

40652-40655: The signals were tested with IRCH on.
40657-40669: The feedback was turned on and the plasma was positioned at 92cm.

40660-40662: The feedback was turned on and the plasma was positioned at 91.5cm.

40663: The feedback was turned on and the plasma was positioned at 92 cm. The IRCH was turned on at 2ms.

40663: The feedback was turned on and the plasma was positioned at 91.5 cm. The IRCH was turned on at 2ms.

Rough day at the shop for the triple probe. I was unable to diagnose the problem with the density channel. After making some repairs to the electronics and installing protection, the output was not correct. I was also unable to get the third channel to trigger on the scope and ran out of time before I could try the others. I will try those early tomorrow and if I cant get the probe to function as a triple probe, then I will run it as an Isat probe or use the mach probe electronics and use it as a double probe for ICRF testing with the Feedback system. [James]

Friday October 17 09 24 Pedersen/Stillits/James 40680-40733 ICRH

Today's objective was to do plasma-RF antenna coupling experiments using the radial feedback system to move the plasmas inboard in a controlled way, and the triple probe to confirm increased densities. Although we did not observe any clear signs of heating, we did see some antenna loading, so the run was successful overall.

Transient loading of the ICRF antenna was clearly observed on a number of shots, including on the order of 10 shots where the loading was clearly observed before the plasma disrupted (eg. 40692). Towards the end of the run day, a decent target discharge for further studies was developed.

Shots 40720 and 40723 are good examples of this discharge without ICRH. The radial feedback was preprogrammed to move the plasma from outboard limited to inboard limited for 1 msec, then back to being outboard limited. Ion saturation current from the inboard side triple probe (the box is still bad, hence no real density measurement) would increase by 30-80% during these jogs. On a number of shots we hit the plasma with ICRF during this phase. Most of these shots disrupted but a few survived (40728 and 29). These two plasmas were apparently not pushed far enough inboard to couple to the antenna. We will want to continue development of these discharges next week, hopefully with the Bdot probe and the triple probe box up and running.

Monday October 20 07 55 James 40734 - 40749 ICRH-RF Pickup Probe

Ran ICRF this afternoon to prepare for tomorrows run. I first used the o-scope to insure that the probe and transmission lines were operational. They did in fact produce a signal, though it seemed a little askew. I then tested multiple digitizers until I was able to get a signal. Finally I was able to reproduce a signal in the tree that resembled the signal in the scope, but not the typical RF B probe in absences of plasma; see shot ^#40749 (atypical) and compare to ^#36547 (typical). Also notice that the loop voltages are very different in these two shots, but I am not sure it could cause such a drastic difference in the RF B probe signal. In the morning I will run with a plasma and the ICRF and diagnose the problem if indeed there is one.

Monday October 20 10 32 Nick None

Since no runs were scheduled at this morning's meeting we used the time to modify the ICRH cabinet. The ICRH should not be run until I have time to inspect the cabinet for run readiness.

Tuesday November 11 04 31 pedersen/stillits 40751-40758 ICRH + feedback

We did not make any plasmas today because of problems with some of the U-port adapters.

Wednesday November 12 05 38 pedersen/stillits 40760-40802 ICRH + feedback.

We tried to create reproducible plasmas that behaved like preprogrammed in feedback:

1.8msec - 2.8msec: R = 93 cm

2.8msec - 3.8msec: R = 91.5 cm

3.8msec - disruption: R = 93 cm

At the end (shot^#: 40795-40802), we were finally able to create reproducible plasmas that behaved as preprogrammed in feedback. The ICRH was turned on in shot^# 40796, 40798, 40800, and 40802, but turned off in shot^# 40795, 40797, 40799, 40801.

However, we did not see any heating of the plasma that could be a consequence of ICRH. Next ime we will try to push the plasma farther inboard (to R ~ 90-91 cm).

Thursday November 13 05 23 pedersen/stillits 40803-40836 ICRH/radial feedback

We continued developing discharges for ICRH coupling using the radial position control system.

The ICRH rack appears to give serial highway problems so it was bypassed. We (Corey+Dave) suspect the power supply is faulty. The rack was bypassed, so no antenna loading measurements today.

On the positive side, the triple probe box works fine, and did so all day yesterday as well, except for some minor offsets on the temperature. Using the triple probe inboard density measurement, we developed discharges that have 3.5-5*10^¹⁸ m^^-3 density for about 1 msec as the plasma is shoved inboard by the radial control system (used in both closed loop and open loop feedback today). This is a factor of 2 improvement over typical inboard densities (2*10^18), and is close to inboard densities seen during inward disruptions (5 to 8*10^18). The ICRH was pulsed during some of these discharges. No clear effects of heating but we are somewhat limited by several diagnostics being offline (ICRH diagnostics were out due to CAMAC problems and the soft x-rays appear not to be working).

Further discharge development will be done when we have ICRH diagnostics available to confirm loading and coupling.

Friday November 14 04 10 stillits 40837-40877 ICRH + feedback

Shot^# 40837-40867:

We tried to creaty discharges (with 20% H2) with and without ICRH. We were able to obtain inboard edge densities around ~ 5e18 m^-3 (shot^# 40863, ICRH turned on at 3.5 msec), and we presumably saw an increase in temperature in shot^# 40855 (ICRH turned on at 3ms).

Shot^# 40868-40877 (Steve):

We tried with He 10% to create discharges that stayed at R = 92 cm for a considerable time. In the last shot we succeed in doing that.

Friday November 21 02 51 steve paul/stillits 40868-40922 rotation + feedback.

Friday December 5 07 37 Rivera/James/Mauer 40923 - 40937 ICRH

We tested the voltage dividers on the ICRF today via the oscilloscope and they performed as expected! Data from the run with voltage and current measurements is being analyzed. We are in the process of setting up the digitizer to make the dividers a permanent diagnostic measurement for the ICRH system. The CAMAC crate and power supply we ordered came in about mid-day and was immediately tested and installed. Preliminary tests show that it is operating normally. Shot ^# 40936 shows an ICRF only shot through the tree. The B-dot probe is not set up correctly, but the others are performing normally. Soon after this shot, we had some trouble with the dumps not responding to the reset controls in the control room. This problem fixed itself, but we are not sure of the cause or the fix. The air conditioners in the basement also froze causing the CAMAC racks to fail. We shut down in effort to let them cool off with external fans, but they were still over heated by the end of the day. Nick is working on the air conditioner repair, but until that is complete, we may have to revert back to shutting down during lunch and any other periods of extended down time between shots.

Friday December 5 12 24 STILLITS 40939-40941 testing software

Thanks to Mikhail, the software for the feedback system is now up and running.

John (study group) put the feedback current cable in the ceiling, and it is working.

Monday December 8 06 27 Maurer/Stillits/James 40949-40967 A little bit of everything

CAMAC problems from last week caused by ICRH rack, possibly due to the fiber optic cable to the u-port adapter. Everything is working OK now. Worked on circiuitry for capacitive dividers for ICRH. (James)

Shots taken to check out new control panel operation of radial position feedback setup writing text file to dsp, recompiling the program, and initializing the dsp fro an external trigger. So far so good! (Stillits)

Turned up gains for new RWM feedback. Checking out sensor and control coil signals. Real feedback on current ramps tomorrow. (Maurer)

S. Paul called and is planning on coming up Friday.

Tuesday December 9 05 15 Maurer/Stillits 40968-41007 testing feedback and software.

Checking, correcting and figuring out some errors in the RWM feedback wiring. Nothing important to discuss yet. (Maurer)

The software for the radial position feedback control system is up and running. The feedback system was successfully tested with new software and the new feedback-current cable. So, the feedback system is now operational!!! (Stillits)

Thursday December 11 10 33 Maurer 41007-41021 RWM feedback

Yesterdays run was cut short due to the same intermittent bank problem we had last week. The cap bank dumps were not reseting properly from the panic panel. Nick and I went to work on it this morning but the dumps worked fine. Again, the problem "fixed" itself. So we will have to wait until it breaks again(and stays broken long enough that we can diagnosis whats going on). When problems are about to occur the bias bank dump fails first. If you are taking shots the signature of this to look for is the OH bias bank controller not reaching full charge while the rest of the banks do(red light never goes on). The next time the dump button is hit on the panic panel the dumps will not reset. If this happens let Nick and I know so we can fix try and diagnosis whats going on.

Thursday December 11 12 01

Today at 1pm, I'll show how to use the feedback system.

Nicolai

Thursday December 11 03 59 james/pedersen/stillits 41025-41036 ICRH + feedback

We tried to create a discharge that behaved as a square-wave with respect to major radius. We succeeded in doing that when we got problems with the ICRH. The charging power supply in the basement is down.

Monday December 15 06 12 James 41069 - 41099 ICRF

It was slow running today. The CAMAC Serial Highway was down most of the morning and half the afternoon. Apparently there were some access violations that fixed themselves after Cory rebooted the system as the administrator. Some good news, the circuit breaker seems to be holding on between shots.

Had trouble with the B-probe as well. The L_8210 did not seem to be working. I replaced it with another and tested it with an fxn generator. Then tested again with new digitizer and both rectified and un-rectified signals. Neither signal was seen via the digitizer or the scope (turns out that the new L_8210 is not 1 mega-ohm, and a new 1mega-ohm will have to be found and installed, but the scope was set for 1 mega-ohm and it did not see a signal). I tested the resistance of the lead wire form the probe coil in the machine to the ICRH crate and there was none. There may be a problem with the wire, connections or the probe itself. I did just get signals from this probe after APS, so the failure is even more peculiar and recent. I checked the connector, and it was tightly fixed so I will have to probe deeper to find the solution.

I was unable to test the capacitive dividers at high voltage, but further analysis of the low voltage tests performed earlier last week show the frequency to be ~4.505 MHz.

Tuesday December 16 07 01 D. Maurer/S. Paul 41100-41174 bias probe with radial control

We were looking for fluid rotation effects with the bias probe turned on. Typical example without probe is 41124, where the natural rotation is about 5 km/sec in the ion drift direction (toroidally clockwise, when viewed from above HBTEP). Edge MHD measured by the shell probe increases from 5.2 msec onward on that shot, and the mode growth is coincident with the braking of the plasma. The plasma rotation slows to near zero by 5.9 msec, at which time the plasma disrupts.

For the most part, the probe seemed to make the plasma cold and difficult to control. Perhaps the reason is lack of conditioning as the probe was last used some time before the APS meeting. The best example of the bias probe's effect the rotation is 41145, where the ion flow was reversed, however that does not occur often. Shot 41142 had the bias probe turned on, yet the results were very similar to shot 41124, as described above.

The best correlate with high rotation appears to be a strong SXR signal. A good example is 41109 and 41127, where the edge rotation jumps following each sawtooth crash.

2004

Friday January 9 03 42 Alex Klein n.a. TS calibration

Filled chamber with nitrogen and performed Raman type Thomson Scattering system calibration. Laser was stil aligned through the chamber, needed no adjustment, near perfect burn spot on inboard side. Laser energy still 800 mJ. Inspected optics, all clean and good. Collection lens was aligned carefully. Took 5 shots at each pressure (from 150 to 5 Torr), swapped Ch. 3 and Ch. 5 electronics to calibrate Ch. 5 and repeated. Data was saved in spreadsheet form for more precise analysis. Now will have to calculate gain coefficients for each channel.

Water contamination appears to be minimal (RGA says 10^^-7 Torr, mostly nitrogen, after 1 hour pump down).

Wednesday January 14 04 06 Maurer 41175-41205 Clean-up

Clean-up shots taken after up to N2 for TS realignment and calibration. Plasmas lasting 5 to 6 msec by the end of the day. SXRs visible after only 10 shots or so. Will continue tomorrow.

Tuesday January 20 06 31 T. Pedersen, R.James, N. Stillits 41242 - 41257 ICRF

We were able to run ICRF (with no B probe) today. We took about 15 shots with the Feedback system on. We did see some signs of loading (comparing shots 41249 vacuum shot to 41256), but were again stopped short by what seems to be more CAMAC problems. This time in the VF current, and it seems to be intermittent. We will look into it more tomorrow. We are planning to try again on next Tuesday. It seems like we are on the right track with the high density plasmas moved into the antenna with the feedback system approach.

Monday January 26 04 16 Maurer 41258-41279 feedback

Took several VF only shots to start the day and check out the VF bank current monitor problems from last week. The digitizer/circuitry behaved correctly the whole run day. I will keep an eye on it this week to see if the problem re-occurs.

The rest of the day was spent taking set-up shots for the weeks feedback runs. I am starting to see some nice m=3 fluctuations as the edge q drops below 3. These are in shots that start with an initially low plasma current, have a rapid Ip ramp and send the q=3 surface into the vacuum anywhere from 2 to 4ms depending upon the bank voltages. I had some difficulty keeping the plasma centered and this would lead to some shot-to-shot variability. I plan to run tomorrow with the radial position control from the start to help keep the major radius from moving out and forcing the edge q too low. Todays run is cut short because I have to go in for my allergy shots at 5pm. I will continue tomorrow.

Tuesday February 17 07 49 Maurer 41477-41506 MC feedback

Todays run goal was to continue the feedback experiments of a week ago using the new mode control coil set with poloidal field sensors.

I had some difficulty during the whole run generating target RWMs to feedback on. Started to see small RWM fluctuations near the end of the day. Also had some difficulty breaking down late in the day. The D2 bottle is nearly empty (has been low for awhile) and I plan on changing it before tomorrows run. This might have lead to some of the problems getting a nice fluctuation to feedback on relative to last week where I was adjusted the puff time to change the RWM onset time with qstar. The feedback was turned on during a few shots to check the loop. Things seem to be OK. I will continue again tomorrow.

Wednesday February 18 05 28 Maurer 41507-41531 Feedback

Spent the day trying to setup plasmas for feedback. Still had problems with the breakdown off and on again. Changing the D2 bottle had small effect on the breakdown problems. As an aside the D2 manifold is showing its age. It has to be maxed out to just barely get the fill line pressure above 20 psig(where we usually run). This was one of the reasons I though the bottle was low and the high pressure side gauge might have been reading incorrectly. I now think think the high pressure gauge is OK. Also checked the puff fill pressure and it seemed reasonable at 2.4 e-5 for 1.5ms puff with the puff battery voltage at 223V (the input circuitry for the puffline says it needs 220V). Checked the egun and its working OK with 9A of filament current and 220V bias.

Later in the day I replaced the puff line battery with a new one (246V) and the breakdown became more consistent. I was able to get the puff time back down to 1ms and still had it breakdown. It looks like the battery was marginal and giving rise to most of the problems breaking down. Not much progress in setting up target plasmas as a result. More tomorrow.

Thursday February 19 05 55 Maurer 41533-41557 Discharge development

Spent the day trying to develop an Ip ramp down case with nice m=3 RWM fluctuations for feedback. Had no breakdown problems today. Discharges exhibited lots of tearing activity with minor or major disruptions before the q=3 surface went external. Some signs of RWMs but nothing as nice as the target plasma fluctuations of last weeks runs. The change in puff valve pulser battery has changed the machine fill pressure for a given puff valve pulse time. I am adjusting for this to effect to try an reproduce the previous runs RWMs with essentially the same bank settings. It hasn't quite worked out yet. I will continue tommorrow.

Friday February 20 05 17 Maurer

No shots taken today. Planning on Saturday feedback run.

Saturday February 21 12 58 Maurer none none

Came in this morning to run and found the base pressure at 5.5e-5Torr and the Ebara cryo head at 268K. The CTI cryo was still pumping, but the head temp was at 15K(a little warm, should less than 10K). After valving off the Ebara the pressure came down to 3e-7Torr and was composed mostly of water on the RGA readout. I spun up the turbo and switched over to it so I could regen the CTI cryo. The base pressure went up slightly to 5.3e-7Torr. I am currently pumping on the CTI head. I have also pumped out the Ebara cryo head. I turned on the bake to help get rid of the water.

The base pressure is now in the low 1e-6 range with the turbo only. I am not sure what happened to the Ebara cryo. After inspecting the compressor it appears to have blown one phase of its 220V, 3phase pwr. I found the cable plug had a blackened pwr pin on it. The circuit breaker appears not to have blown which I find confusing. Hopefully the compressor is not damaged. I will stay until the CTI head has been pumped on fully and turned back on to cool down. This takes 2 to 3 hours to reach final temperature. If someone who lives nearby wants to stop in at 5 or 6PM and switch over pumping to the CTI cryo once it is cold let me know. Otherwise I'll stop in Sunday and do it.

Sunday February 22 03 55 pedersen

I switched over from the turbo to the CTI cryo at ~3:30pm today (Sunday). Pressure was 2.7e-7 before the switchover and started dropping at a good rate when the gate valve was opened to the cryo. It is currently 9.7e-8 and continues to drop.

Monday February 23 03 53 Maurer None

Opened up the Hall probe array bellows to the vacuum chamber after the group meeting. The pressure went up to a few 1e-7 composed mostly of water. Kept the bake on until midafternoon. The base pressure is now in the low 1e-8 range. Nick is currently inspecting the compressor to make sure it has failed. We will try hooking it back up to pwr tomorrow when new plug ends arrive from McMaster. There was no obvious damage inside the compressor housing. Maybe it was just the plug ends that failed.

We have a quotation for an advanced exchange with a 2 day turn-around time (they have a matching compressor in stock) from Genesis Vac Tech if it turns out the compressor is bad. The price is $3950. I did not get a chance to take any cleanup shots today. I plan on starting tomorrow morning. After about ten shots or so if the machine still looks dirty I will glow in He. Also, Yuhong is ready to piggy-back and test the four Hall probe channels that she has installed.

Tuesday February 24 04 19 Liu 41560-41573 test Hall array

There were just two shots were caught in today's run and the rest of the day we experienced serial highway failure. Mikhail tried to bypass CTX and ICRH, but it didn't work out. The Hall array was working except they were more noisy than the old array. More shots need to run to analyze the reason.

Wednesday February 25 06 01 Maurer/Liu 41578-41639 Clean-up shots

Took clean-up shots today. Started to see SXRs and tiny sawteeth by the end of the day. The Ebara compressor is OK. The cable end only was damaged in the weekend failure. This is very good news. It has been re-installed and was opened during the last 1/3 of todays run. This helped quite a bit in pumping out stuff knocked off the walls. I had to raise the puff because I had trouble initially breaking down with both cryos pumping. It was after this that I was able to get some SXR signals on the fan array. Yuhong ran piggy back and worked on her Hall sensor signals.

The noise problem of the Hall sensors comes from the power of the amplifier. When I used a battery, one of which were used for Hall sensor current source, together with a car battery powered the amplifier, the annoying random noise was gone. Also, G=10/100 were tested for vacuum shots and both worked fine. I will monitor the Hall sensor's behavior when Dave runs tomorrow.

Thursday February 26 05 20 Maurer 41640-41682 Clean-up shots

Took more clean-up shots today. By the end of the day things were looking good enough that I was starting to develop a target plasma for feedback. Vloop was less than 10V at plasma start-up evolving down to 5 or 6V at discharge termination compared to greater than 12V yesterday during most of the run. I am switching to the shot type with a RWM generated by the initial qstar less than 3 and slowly evolving up to 3 (similar to shilov's case) that I used for RWM bias probe experiments this fall. This would typically give a burst of m=3 magnetic fluctuations between 2 to 3 ms. Yuhong also worked on her Hall probe sensor signals. There appears to be high frequency noise on the signals at discharge start-up and termination that might require filtering.

Friday February 27 05 16 Maurer 41683-41691

Only a few shots taken today. Tested out the Hall sensor array. The problems with noise are a result of layout and cabling problems. This is good news in that an input filter will not be needed. The actual layout of the 20 channel array will have to be done very carefully.

Monday March 1 04 12 Maurer 41692-41732 target plasma development

Spent the day trying to generate m/n=3/1 RWM fluctuations. Tried dialing in old shot number bank settings that Shilov used as a target shot. Plasmas generated this way were different from the database case, with typically higher loop voltage and different major radial and qstar evolution. The wall-limiter conditioning must still be not quite up to what we need to generate these plasmas. Spent the rest of the day trying to develop a case with the initial qstar at about 2.7 after discharge start-up evolving toward qstar of 3 by 2 to 3ms into the shot. This would traditionally have a RWM fluctuation from 2 to 2.5ms and a few Gauss amplitude in the past. Was somewhat successful at this but the plasmas did not have a good m=3 fluctuation to feedback on perhaps due to profile effects. I will continue tomorrow and turn the feedback on even if I don't have a mode just to check things out. I am out early today to go and get my allergy shots.

Tuesday March 2 07 15 Maurer 41733-41780 Setting up target

Took more shots today trying to develop target for RWM feedback experiments. I had one clear example of an m/n=3/1 RWM on shot 41744. This can be seen on the smps and ssi Br sensors just around 3ms. I took this shot after the machine sat for 1 hr, so the prior shots residual gas was well pumped away. I was unable to reproduce this plasma by lowering the puff (so far at least). Comments and suggestions welcome.

Also, the Ebara cryo head has recently started to make pinging noises most likely due to the seals starting to go bad in the head. It was rebuilt last June so I hope it is under warranty.

Wednesday March 3 05 17 Maurer 41781-41792 target development

Working on target plasmas (still). Trying to adjust puff time and bank settings to generate RWM fluctuation burst as qstart approaches 3. More work to be done tomorrow. Both beta and sherwood had to be restarted this afternoon. I hope this did not disturb peoples work too much. Nick has contacted the company who rebuilt the Ebara cryo head this summer about warranty coverage. Hopefully we'll know by tomorrow or the end of the week what the situation is. As it stands now the head is holding its 8 K temperature. There has been to signs of temperature cycling yet.

Thursday March 4 05 31 Maurer 41781-41826 target development

More of the same. Another day of trying to get a good RWM for feedback studies. I am contemplating going back to the ramp down case. Comments.

Sunday March 7 10 47 Maurer 41728-41778 target development

Took more shots Saturday. After trying to develop target discharges, I switched to a high Ohmic input plasma to generate SXRs and cleanup the limiters and walls.

Tuesday March 9 07 13 Maurer 41881-41891 clean-up

Had a short clean-up run only today. I will try the RWM case tomorrow. In other news the Ebara cryo head has stopped squeaking for at least the last two days (maybe Sunday too). Maybe it will last longer than was first thought after all. I will keep a close watch on it. The FPGA arrived yesterday. Alex is sorting out some software driver issues with NI. The leaking pipe in the bank room is sceduled to be fixed tomorrow morning. The run will commence after that is finished.

Wednesday March 10 05 29 Maurer 41892-41942 Clean Up/Target development

Took 30 clean up shots to start the day. Then switched to target development shots. Started to see a choppy burst of MHD fluctuations as q_star was approaching and going though 3 from below. The fluctuations were of lower frequency than the background MHD nosie. Reproducibility was better than during the last weeks, but still not too robust. These are the first encouraging results in about two weeks! Here's looking forward to tomorrow.

Thursday March 11 12 40 stillits 41953-41960 Testing RPFB

The RPFB is up and running again. Moreover, we have modified the control panel application.

Thursday March 11 04 36 stillits none Modification of Control panel application.

- the control panel application has been modified to use triangles instead of square-wave as input for RPFB.

Monday March 15 06 12 stillits 41962-42017 feedback

Feedback on major radius (with the new application) is now working. Feedback on q* seems to be working, but some more finetuning needs to be done.

Friday March 19 10 11 Maurer 42018-42027

Yesterdays run was cut short by a problem with the TF charging resistor. There was some arcing from the top of the resistor to the water column. The top electrode had broken off from the charging supply lead during the run. A new electode is being manufactured and will be installed asap.

Monday March 22 05 26 stillits 42069-42124 q*-feedback

We tried to perform feedback on q*. The algorithm does the right things, but there are some minor problems with the gains, which need to be finetuned. We'll work on that tomorrow.

Tuesday March 23 04 35 stillits 42125-42154 q*-feedback

We were able to get the feedback on q* up and running today. There are, however, some "small" deviations between q* in the scope and q* we calculate. The max. deviations in q* are approximately 0.2-0.3 and in some cases 0.4. The reason for this is the small value of the minor radius a squared (a^2 ~ q*), which will change q* by 5-10% when the major radius in the scope deviates by, for example, 4 mm compared to the major radius we calculate. When we take into account these deviations, we can still get very nice feedback values on q*:

Shot ^#42137-42146: We tried to put the plasma at q*=3.5. After having chosen the suitable parameters (which only took a few shots), we were able to get some nice shots ^#42141 and ^#42142 (and also ^#42143).

Shot^# 42147-42154: We tried to ramp up the plasma from q*=2.7 at 2.4 msec to q* = 3.1 at 4 msec. Again, it took only at few shots to tune the parameters. The nice shots are ^#42150, (^#42151), and ^#42154.

Monday March 29 04 12 stillits No shots. Application.

The Control Panel Application has been modified. Control of q* (including q*-ramps) has been included.

Tuesday March 30 07 23 James No Shots Maintenance

At about 9:45 last night I found a large amount of water on the floor in the basement. It was centered on the bias probe electronics and when I looked to find the source I noticed that it was the leak that Jim has been trying to get maintenance to repair. Over the weekend I guess it had filled the drip pan. It was a lot of water and it spread over to the dump, I vacuumed it up with the wet-vac, but before we run today that space (from the wall behind the bias probe to the dump) should be checked again to be sure I didnt miss anything before it is powered up.

Tuesday March 30 07 26 James No Shots Maintenance

At about 9:45 last night I found a large amount of water on the floor in the basement. It was centered on the bias probe electronics and when I looked to find the source I noticed that it was the leak that Jim has been trying to get maintenance to repair. Over the weekend I guess it had filled the drip pan. It was a lot of water and it spread over to the dump, I vacuumed it up with the wet-vac, but before we run today that space (from the wall behind the bias probe to the dump) should be checked again to be sure I didnt miss anything before it is powered up.

Wednesday March 31 05 10 Maurer RWM development

News from Tuesday: the leaking pipe in the bank room was finally fixed by facilities. Todays run was cut short by a minor arc on the TF coil set. A nut on one of the coils for the lug connection to the current feedthru had come loose. When the last shot of the day was taken part of the lug was vaporized giving rise to a big flash when the arc occured. The TF current trace looks just like the previous shot, so the full bank current still flowed through the arc to the coils. Nick and Jim have cleaned up the contacts and re-made the lug connection. All other TF connections have been checked also. I will check the resistiance and inductance of the coil and compare it to others before running tomorrow.

Thursday April 1 03 52 Maurer 42176-42185 checking out TF

We had another TF lug arc at the same spot as yesterday. Started the run today by bringing the TF up a kV at a time until we reached 6.1 kV to check out the new lug end. Everything appeared to be fine. Took one plasma shot with full TF, VF and OH banks. On the second plasma shot the lug arced at the TF current feed just like yesterday, blowing off the end of the lug. Apparently, the threads were damaged enough that the nut loosened up again an caused an another arc. We are discussing various possiblities about how to proceed. Input and any comments are welcome. New lugs have to be ordered and should be in tomorrow.

Monday April 5 05 05 Maurer 42185-42205 TF check-out

Checked out the new TF coil lug attachment today that was arcing last week. So far so good. I brought the TF up in kV steps initially checking the lug nuts between every shot. After reaching 6kV with the TF only I took about 5 shots with full banks making plasmas while still checking the lug nuts every shot. They were nice and tight every time. After that the remainder of the plasma shots were taken checking the lug nuts every 2 shots then 3 shots. They were OK the whole afternoon. I will continue to check them tomorrow gradually increasing the number of shots between inspections. Tomorrow I will be attempting to generate RWMs approaching qstar=3 from below using the Ro-qstar equilibrium feedback to help reproducibility.

Wednesday April 14 10 43 D. Maurer, Alex K. 42248-42270 FPGA tests

Yesterdays run summary...

Inserted FPGA processor into one poloidal set of sensors/control coils (bottom shell, upper set).

Digitized raw sensor signals, filtered signal that reach the FPGA, signals from the FPGA, and current in control coils. Will analyze data to see how it is consistent with mode control algorithm and filters.

We tried several gain settings on the analog boards to see at what point crunch amplifiers are overdriven.

More work needs to be done, including checking polarity of control coils and current shunts, (spatial) phase shifting of output signals.

Wednesday April 21 02 51 Alex Klein 42271-42315 FPGA testing

Yesterday, fixed polarity issues with current shunts. Also took many shots with sine wave driving things instead of sensors (no plasmas), to measure overall system transfer function.

Today, verified control coil polarities and then ran toroidal phase shift scan on bottom-midplane FPGA group with all signals looking very nice. Took shots at 0,45,90,135 and 180 degrees toroidal shift. Will analyze data.

Hope to wire in remaining FPGA groups by end of week.

Thursday April 29 09 53 A. Klein 42326-42434 wiring/repair/configure

The FPGA global mode control system is installed and functioning. All analog boards, FPGA wiring details, control coil current shunt polarities, crunch amplifiers, control coil polarities, sensor coil polarities, coil integrities, etc are fixed, configured, functioning, verified. Now ready to apply feedback to RWM plasmas at various spatial phase angles using 20 sensors/coils configured in 4 groups of 5.

Thursday April 29 05 07 D. Maurer, Alex K. 42438-42457 mode control testing

Applied DFT FPGA mode control with all available coils. Only covered 2 spatial phase angles for feedback: 0 and -85 degrees. Will require much analysis to see actual effects on the plasma, and much analysis to see what effects should be acting on the plasma.

Friday April 30 04 33 Maurer none

Bias probe reinstalled and bellows assembly leak checked this week. Opened gate valve to chamber this afternoon. Pressure rise was maximum of about 6e-7 Torr and composed mostly of H2O. The bake has been on for over an hour and the base pressure is already down to 2.8e-8 Torr. We will continue baking over the weekend.

Thursday May 6 04 54 stillits 42464-42519 clean-up shots.

Clean-up shots were taken with RPFB and biased probe turned on.

Friday May 7 04 53 stillits 42520-42591 clean-up shots.

Clean-up shots were taken with the biased probe and RPFB (R=94 cm and R=93 cm) turned on.

Monday May 10 03 43 stillits 42592-42605 H-modes (Biased probe + RPFB)

In our attempt to create H-modes, we ran the biased probe with RPFB. However, after a few shots, the biased probe arched. We measured a high-voltage short between the tip of the probe and ground. The probe has been pulled out, and the gate valve has been closed.

Tuesday May 11 05 39 Alex Klein 42607-42630 Global Mode Control

Initially, trouble with basement digitizer. Ch 2 on 8210^#2 in basement appears to be out, switched VF bank signal to Ch3. Then, developed discharges to yield q* < 3 at start, then slowly ramp through q*=3.
Ran several shots with FPGA mode control ON at t=2 msec, off at t=3.5 msec for 0 and 180 degree spatial phase shifts.
Did not run radial position control, will do so tomorrow.

Wednesday May 12 05 18 Alex K 42631-42668 FPGA mode control

Performed gain scan on FPGA system: It had been slightly overdriven previously (increasing gain caused no increase in crunch amp current). Reduced FPGA gain setting and found gain setting which causes max crunch amp output (5-7 amperes)for typical plasmas.

Ran with radial position control, with goal of keeping plasmas alive as long as possible while slowly ramping through edge q =3 from below. Found m =2 tearing modes to dominate everything at 4 msec, resulting in disruptions. While FPGA feedback was on at various times for many shots, these may not be too meaningful...

Plans now have changed: perhaps run through edge =3 quickly and from above, to look for RWM before tearing modes dominate ? Did some discharge development in this direction.

Friday May 14 10 15 Alex Klein 42669-42722 FPGA model control

Spent a while developing discharge with aggressive OH ramp, such that edge q runs through 3 from the outside early on in the shot. Got to the point where RWM consistently observed between t=2 and t= 3 msec as edge q dipped below 3.

With this type of shot, then ran with FPGA mode control. Control output gated on between t=1.5 and 3.5 msec. Took data at 45 degree spatial phase shift increments for FPGA output. Have at least two shots at each angle with RWM.

Will have to look at data carefully to be able to say reasonable things about mode control. The RC filters and digital integration, as well as control coil inductances are causing control coil output to be dominated by low frequencies, and phase between sensors and controls highly susceptible to frequency as well.

Puff valve: For unknown reason, puff valve required to be open for 2.2 msec to get comparable gas levels as existed earlier in the day with valve opening only for 1.2 msec. Battery on valve checked out fine (235V), gas lines were all open with adequate pressure.There may be something wrong with the puff valve.

Friday May 21 03 49 Alex Klein 42723-42747 Mode Control

Again used discharges where plasma current ramp = agressive, edge q comes from the outside and rapidly dips below '3', giving rise to RWM early on in the shot. Very clear and reproducible RWM's.

With FPGA mode control algorithm now using much reduced integration time (~15 micro-secs), determined what spatial angle to dial into DFT to have 90 degree spatial phase shift between N=1 sensor mode and N=1 control mode. Took many shots at this setting with and without FPGA running. Consistently had 90 degree phase relationship.

Saw evidence that at "+" 90 degrees, feedback system excites N=1 mode. Fairly reproducible increase in sensor mode amplitude and lower mode rotation rates when FPGA is on, vs. when it is off.

Then, adjusted DFT angle by 180 degrees, to attempt "-" 90 degree feedback, should be suppression of RWM. No clear results. Mode amplitude looks always similar, regardless of wether FPGA is on or not.

Why? -> Consistenly had NO fixed phase relationship between control and sensor coils, i.e. phase is all over the place. Mode rotation rates seem high.

Will now do more rigourous analysis, with some objective method for quantifying data.

Thursday June 3 04 46 Alex K. 42748-42792 FPGA mode control

Picked up where we left off 2 weeks ago. Plasma behaved exactly the same, without any tweaking required (edge q comes to 3 quickly, from the outside). Took many shots with spatial phase angle close to 0 degrees, and many shots with phase angle near 180. Cursory glance: RWM looks about the same for the two types. Will require more analysis to say anything meaningful.

Then drove control coils with function generator going into analog boards, to explore gain and saturation limits in crunch amplifiers.

Then with p[lasmas: began gain scan (at spatial phase angle ~90 degrees), will continue tomorrow and perform gain scan at phase angle ~ -90 degrees.

Friday June 4 04 43 Alex K. 42793-42817 FPGA mode control

Hose for ignitron cooling burst, flooded basement. Fixed and cleaned up in the morning.

Then continued gain scan from yesterday, at approx. +90 degree spatial phase shift and -90 degree target phase shift. Nearly finished, cua900 computer acting up... Gain scan should be complete after about 20 more shots (to be done next week).,

Monday June 7 02 34 Maurer/Rivera

The OHEL bank is back up and running...

The fpga is being programed for this weeks latency tests. The run will start tomorrow.

Wednesday June 9 03 56 Alex K. 42829-42871 FPGA mode control

Yesterday, ignitron cooling water hose had burst and flooded basement (again), therefor no runs.

Today: With the usual discharge (see past several run reports): Finished gain scan at spatial phase angle which causes mode suppression.

Then: Implemented FPGA program which has select-able latency in multiples of 10.2 microseconds, while keeping looprate/sample-rate the same. Took shots at 10, 50, 100, 200 microsecond latencies. It will require lengthy analysis to expose small effect, judging by quick look at the data

Monday June 21 05 29 stillits 42873-42894 clean-up shots + biased probe

In the first 13 shots the biased probe was not turned on. In these shots the plasma was cold, lasted between 3-4ms, and the plasma current reached 10 kA. The loop voltage was fairly constant ~ 12 in each of these shots. From shot^#42887 the biased probe was turned on. The biased probe voltage was increased from 0V to 350 V in 50V-steps. Saturation was reached at 300-350 V in which I_sat = 100-150 A. We saw a few more spikes in the loop voltage (~ 12), and from SXR the plasma was a bit hotter (but not much). However, there is a lot of noice in SXR, for example, so it is not easy to say anything about spikes in SXR.

The RPFB was not turned on in any of the shots because of problems with the hot decks on the LANL.

Tomorrow we will continue taking clean-up shots with the biased probe on.

Tuesday June 22 05 24 stillits 42895-42931 clean-up shots + biased probe

Clean-up shots were taken. The biased probe (BP) was slowly ramped up to 300V (it looks more like 200 V on the scope), and at 300 V the BP-current was 100 A. The plasmas created lasted for 3-4ms, and the plasma currents were ~10 kA. The signals are very noisy, so it is not very easy - by looking at the SXR-signals - to say anything about how hot the plasmas were, but the spikes in the loop-voltage seem to be absent in some of the last shots.

The RPFB was not on because of problems with the LANL, but Nick is working on it.

We will continue taking clean-up shots tomorrow.

Wednesday June 23 05 21 stillits 42966-43011 clean-up shots + biased probe

Clean-up shots were taken. The biased probe (BP) was slowly ramped up from 100 V to about 300V, and at 300V the BP-current was ~100 A. The plasmas lasted for ~ 4 ms and plasma current ~ 10kA. There are no spikes in the loop-voltage indicating the BP is clean. Furthermore, as soon as the BP is turned on (at 2ms), the loop-voltage remains constant (~ 10) or decreases indicating the plasma is getting heated (the plasma current ramp is 2.5 MA/s). SXR indicates a hot plasma but it is difficult to say because of noice.

The RPFB was not turned on, but Nick is working on the LANL.

Thursday June 24 05 00 Alex K. 43012-43076 FPGA open loop runs

Coerced the machine to produce the types of discharges I have been using for a while now, with agressive current ramp, edgq q approaching 3 from above early in the shot...

Spent the day running "open loop" experiments: All 20 control coils were energized with a set of signals that were out of phase with each other such as to drive a m=3, n=1 helical pertubation around the tokamak. At a certain time during the pulse, the polarity of the signals was reversed. (Phase Flip Experiments)

Initially (in the morning), I had the pertubation propagate downwards poloidally as the toroidal angle was advanced (along plasma current). Saw little or no mode locking at several frequencies (between 4 and 8 kHz). Then I switched helicity, and found mode locking (between sensor coils and control coils) for a range of frequencies. This means presumably that perturbations are tied to the electron fluid. Can now analyze data to measure plasma response to phase flip of magnetic perturbation, also looks like I can measure plasma amplification of resonant perturbations.

Friday June 25 04 21 Alex K. 43078-43107 open loop, closed loop

The transfer functions of a feedback system seem to be surprisingly important.

Tuesday June 29 04 02 stillits/maurer 4343108-43124 Biased probe

The biased probe (BP) was tested. With BP inserted 3 cm into the plasma, an IV-characteristic was obtained with I_sat = 45-50 A at V = 250 V (the BP voltage was increased to 350 V with no sign of increased BP current). There was no sign of impurities by looking at the loop-voltage (too much noice on SXR). Tomorrow we will perform an IV-scan by obtaining IV-characteristics at different positions of BP in the plasma.

Wednesday June 30 04 50 stillits/maurer 43153-43174 Biased probe

We had BP positioned at 1 cm and 5 cm from the edge of the plasma (it was positioned at 3 cm yesterday) to obtain IV-characteristics. The voltage was increased from 100V to 350 V in 50V-steps. For 1 cm, I_sat was 30 A, and for 5 cm it was 55 A. In the 5 cm case I_sat was reached at 100V-150 V, and in the 1 cm case I_sat was reached at 200V - 250V.

Thursday July 1 04 41 Alex K. 43175-43202 Mode control debugging

Was going to run latency experiments, spent the day chasing down bugs and problems instead. Output signals from the crunch amps on bottom poloidal group are not what they should be. Input signals as seen by digitizer are what they should be. Verified that FPGA module works as it should, and analogboard output stage performs properly as well.

Took some plasma shots as part of the bug-chasing work, no good data. Hope to fugure this out tomorrow.

Thursday July 8 05 05 Alex K. 43242-43249 Mode control debugging

The analog boards/rack seem to be fixed after Nick R. worked on it on the bench. All channels are working except for one output signal on one of the bottom control lines (Channel 3B C3). It is a very bizzarre problem that can't easily be diagnosed/explained. Just as I was getting to the bottom this, the CAMAC data aqcuisition failed, presumably because the basement rack is overheating - the air conditioner in the basement is down, Jim A. is on the job of getting it repaired.

Tuesday July 13 05 50 A. Klein 43262-43288 Mode control debugging

Still had problems with one group of output signals - but I think I finally figured out the problem. Will test tomorrow and hopefully everything will be working.

Developed discharges also: while the typical type of shot used previously for mode control studies appeared once in a while, it was not as reproducible and robust today as before. Will continue tomorrow, and hopefully run latency experiments.

Wednesday July 14 05 16 A. Klein 43288-43319 latency in mode control

All signals behave properly now. It took a while to get the target plasma, but then the old mode control shot (aggressive current ramp, edge q comes in to 3 and hovers there for a while, slow freq 3-1 oscillations visible) appeared. Ran feedback with 10, 60,110,160,210 microsecond latency in system. Data looks interesting. Every shot from 43302 to 43319 is relevant, no dud shots (which is nice for the operator).

Will now have to analyze the data more thoroughly.

Thursday July 15 05 01 stillits 43320-43349 biased probe + RPFB

The LANL and the triple probe are working again.

We inserted the biased probe (BP) 5 cm into the plasma, and slowly increased the BP-voltage from 100 V to 300 V. The max. BP-current was 50A-100A. We used RPFB to position the plasma at 94 cm (from 3 to 10 msec), which was fairly successfull.

In a couple of shots, for example shot^# 43346, we saw huge inboard edge densities up to 6-8*10^¹⁸ m^^-3. However, we have to smooth the signals before we come to any conclusion.

In shot^# 43332 we see some interesting stuff. At 4.1 msec to 5.1 msec we see a drop (from 60A to 10-20 A) in BP-current, an increase in SXR from 0 to about 0.1 (the loop-voltage decreases from 12 to 9), and a slowly increasing inboard edge temperature from 5-10eV to 20eV.

The inboard edge density seems to be fairly constant. All this could indicate an H-mode.

Friday July 16 08 35 Stillits 43356-43388 Bias Probe +RPFB

Today we ran the biased probe (BP) and RPFB. The BP-voltage was slowly increased from 100V to 250V, and BP was kept at 5cm from the outer limiter. We tried to position the plasma at 93cm and 94cm, but R=93cm seems to be the preferable choise because at R=94 cm we saw huge BP-currents (up to 150A), which could potentially destroy the probe. Beside the noice, we did not really see anything of interest. However, we are getting better at running BP; we were able to create long lasting (8-9 ms) and faily hot plasmas (SXR ~ 0.2-0.3) with BP inserted.

Monday July 19 04 48 stillits 43392-43406 No plasma shots.

We didn't take any shots today. There was water in the basement this morning, and after the basement was dry the OH kryon pulser broke down and some parts had to be replaced. However, everything is working now and as soon we get the D_alpha-meter installed tomorrow we will start taking shots.

Tuesday July 20 05 26 stillits 43407-43447 Bias probe + RPFB

We ran the bias probe (BP) with RPFB on. The plasma was kept at 94cm from 2.5 msec, and BP was inserted 5 cm into the plasma from the outer limiter and voltage was 200 V.

In shots^# 43411 and 43421 the RPFB and BP were turned off (however, BP was inserted 5 cm, but no voltage on it).

We had several very good shots (shot^# 43418,43420, 43426, 43428, 43430, 43433, 43438, 43439, 43441, and 43442) in which the BP-current almost dropped to 0 A, SXR increased, loop-voltage decreased, inboard temperature slowly increased, and inboard edge density decreased or stayed constant. Several of these shots (especially shot^# 43428, 43433 and 43441) experience an increase in SXR (and a drop in loop-voltage) so soon as the BP current drops to (almost) 0 A, and a decrease in SXR when the BP current increases. These shots could indicate an H-mode.

The D_alpha-meter needs to be installed before we can come to a final conclusion about creation of an H-mode. We had expected to install it today but because of APS abstracts we didn't have the time to do it.

Thursday July 22 08 09 James/Stillits 43450 43480 ICRF

Spent the morning putting the ICRF back together. There were some minor problems with filling the cooling system reservoir on Monday that were rectified but put us a little behind schedule. ICRF was revived almost without incident. We have a problem with the Jorway trigger and had to run a cable to the one in the west rack so we could run. We were able to take only a few shots bug free (mostly operator error).

Shots 43467-43480; our vacuum shot (base line) for comparison with the antenna q ratio for loading investigation was ^#43478. Most of the shots were not high density, but ^#43477 looks like a good loading candidate at first glance. The feedback system was not on for this shot (43477) but it was for the others. We hope to spend more time next week with in developing more ideal plasmas and using the bias probe.

Friday July 23 04 32 A. Klein 43481-43532 Mode control phase scan

Ran detailed phase angle scan for FPGA mode control using the 2nd order digital filter, lowest latency, moderate (fixed) gain. I adjusted the toroidal phase of the control field, covered 46 different angles. Found phase of maximum feedback, observed mode frequency increase (substantial) to the "left" of that, and frequency decrease to the "right", implying torques on the modes as expected. It was quite a good day: out of 51 shots, only one is not useful.

Wednesday July 28 08 20 James/Stillits 43533 43580 ICRF

Spent the morning developing target plasmas, and fired ICRF in the afternoon. Had to stop running due to difficulty the Jorway timing in the west rack, but fortunately, we were able to get some data and even the data that was time eschewed (baseline vacuum shots) may be salvageable. The bias probe did show increases in the density of close a factor of two, the radial positioning worked like a charm. Thursday we will start with testing the ICRH tree and developing H-mode like plasmas and position them near the ICRF antenna for loading and heating.

Friday July 30 06 14 James/Stillits 43616-43658 ICRF

We fixed the tree problem we were having with the trigger. We were able to basically dial up the plasmas that gave us possible H-modes in the past. We did see some good H-mode candidates during the run (43654-43656), but most were not. However, almost all the plasmas we did develop were ideal for loading studies. We observed loading not just in the initial launch of the ICRF, but also through the shot as well. See the comparison of antenna_q value for shots 43631, 32, 38, 39 (loaded plasma) and 43629, 33, 41 (vacuum) - the vacuum shots are in such good agreement that any of them can be used. I will try to have some print outs available for Monday.

More data analysis to come, but we dont expect to find much evidence of heating. I hope to take more shots like these with the bias probe / RPFB system combination. The densities were close to two orders of magnitude higher than those typically seen without them. So finally we have verified that our theories about positioning densities near the antenna for loading. Next we want to take more shots like these to determine the best density regimes and attempt to heat the plasma!

Tuesday August 3 05 28 stillits 43694-4375 ICRH + BP + RPFB

Today we had the ICRH, BP, and RPFB on. In most of the shots, the RPFB kept the plasma at 94 cm from 2.5msec to 10 msec. BP was kept at 200V in all the shots. With RPFB and BP on, we took shots with and without ICRH. We saw several plasmas that had an "H-mode"-structure (for example shot^# 43729). However, there is a problem with the ICRH measurements; the loading is not as high as previously thought, and there is not loading during the whole ICRH shot, as previously claimed.

Monday August 9 04 12 stillits/james 43778-43820 IRCH+RPFB+BP

The LANL is working again, and the D_alpha-meter is installed and working. ICRH: (We ran with Deuterium) The source of the pick-up was found and repaired and the triggering problem was fixed too. We tried all day long to create high inboard densities, but the plasma was pretty "stubborn" today; we never reached above 4*10^¹⁹.

"H-mode": The D_alpha-meter started working in shot ^#43815 (to 43820). In shot ^#43820, we see something that could resemble an "H-mode" while we see a (slow) decrease in the D_alpha signal. However, this should probably be taken with a grain of salt since we just got the meter to work.

Wednesday August 11 04 46 Maurer

summary for Tues and Wed...

Building power was shut down on Monday night to fix problems with a faulty phase. The cryos where valved off in case they lost power during the shutdown. The base pressure was 1e-4 Torr on Tues morning after not pumping over night and was composed of H20 and some air. After pumping the machine down the base prressure hung at 3e-8 Torr and was mostly H20. We baked in the afternoon and after cooling down the pressure is back down to 3e-9 Torr.

Today getting ready to run we had some CAMAC problems that took the afternoon to fix. Yuhong took some VF shots later in the day to test out the Hall amplifier box.

We have also been experiencing some problems charging the TF bank recently(last week or so). Nick has checked it out and suspects the charging resistor needs to be cleaned out of Cu debris that fills the bottom of the resistor as the top electrode corrodes. We will take care of this before tomorrows run in the morning.

Scheduled runs are still: Thurs ICRH and Fri Biased probe.

Thursday August 12 04 01 Maurer/James/Stillits

More problems today.

There was a small flood to start the day off in the bank room. The TF resistor was cleaned out and put back into service. In getting ready to take a shot the TF bank appeared not to charge. Upon going back down into the bank room we found the water leak had opened up again. After fixing this and cleaning up the floor we had more problems charging the TF. After inspecting the bank with Nick and charging it to low voltage locally we found no obvious problems. The problem appears to be intermittant. Several shots in a row at the end of the day were OK. Hopefully the power supply is OK. We are going to see if we can monitor the bank charging current meter on the camera system to actually see what happens when the bank does not reach full charge. No ICRH run as a result.

Friday August 13 05 18 stillits 43853-43908 "H-mode"

We had BP on = 200 V at 5 cm, RPFB = 94 cm from 2.5 msec, and we ran with 20% H_2, 80% D_2. We saw several shot that could indicate an H-mode: for example shot ^#43877, 43893, 43895, 43904, and 43906. In shot ^#43906, we see a very clear signal of an H-mode: BP-current goes to zero, a drop in the D_alpha-signal, an increase in the floating T-probe potential, and increasing SXR's. The major radius was kept at ~ 94 cm during the whole shot. The T-probe batteries were dead. After we changed the batteries we saw much larger inboard edge densities.That could probably be reason why were not able to make large inboard edge densities last week.

Tuesday August 17 04 52 stillits/james 43917-43972 "H-mode" + ICRH Shot ^#43917-43959:

We were able to lower the puff time from 2.0 msec to 0.9 - 1.0 msec and create some hot plasmas; in almost all the shots the loop-voltage started at ~ 9 and decreased to about 5-6 (we saw a lot of SXR's). With these settings we tried to create "H-modes"; in almost all the shots we saw weak and clear signs of an "H-mode" structure. Shot# 43960-43972: In these shots we ran with ICRH turned on. We increased the puff time from 1.0 msec to 2.0 msec to get high-density plasmas. The inboard density reached ~ 8*10^18 m^-3, and the ICRH was triggered at 25 kV - 30 kV. However, we can not say anything about loading; some more analysis is required.

Wednesday August 18 04 54 stillits 43973-44007 "H-mode"

We ran with RPFB (93.5 - 94 cm from 2.5msec to 10 msec) and BP (200 V, 5 cm) on. We saw several signs of an H-mode (especially shot ^#43977), but in most of the shots the D_alpha-meter was down because of high temperatures in the screenroom (83 degrees F). We tried to use a fan but it didn't help a lot. By the way, in some of the shots we were able to create very long lasting and hot plasmas: in shot ^#43985 the plasma lasted for 9.3 msec, and SXR reached almost 2.0 at 8 msec.

Friday August 20 05 39 stillits/pedersen 44017-44074 ICRH + "H-mode"

"H-mode": We ran with BP = 200 V at 5cm, and RPFB at 93.5 cm from 2.5 msec to 10 msec. We saw several cases with an H-mode structure and a few cases with a "Multi H-mode" structure (2 - 3 "H-modes" (of short duration) in same shot). We also tried to perform density-scans (push the plasma inboard towards the Triple probe during an "H-mode") but it was not a huge success. This was tried in shot ^#44050-44074 (the "interesting" shots are 44060, 44061, 44063, and 44065). ICRH: In some of the shots we saw some loading but it requires some more analysis.

Monday August 23 08 08 Y. Liu 44075-44094 VF only vacuum shot

These shots were used to figure out the noise problem of Hall probe array. As prof. Navratil predicted, it is amplifier cross-talk. After keeping on fixing problems, there are two worst channels left now - I can get clean signal by leaving them off. Further analysis needs to be done to have all channels work well. There was water leak in the basement. It was from the old wound of the tube close to TF banks. The ground under TF banks are wet. I mopped the open ground and left the fan on. Hope it will be fine by tomorrow.

Tuesday August 24 05 02 Stillits/Klein 44095-44131 "H-mode" + FPGA

We ran with 20% H_2 + 80% D_2. The RPFB was on and BP =200 V at 5 cm. First we tried to see if we could generate any H-mode structure. When that was achieved we tried to push the plasma inboard towards the triple probe in order to perform a density scan. It was not huge success, but in shot ^#44119, 44121, and 44122 we were able push the plasma inboard while the plasma was in some kind of an H-mode. However, we don't see any huge increase in inboard density as we would have expected We tried to do "feedback", i.e. turned on the mode control system to see what effect might be produced by control coils when plasma drops into/out of H-mode and very strong, high freq. MHD modes are present. FPGA algorithm is not at presently designed to deal with 30 kHz modes however, the data is difficult to interpret.

Wednesday August 25 10 25 klein 44132-44137 gain limit scan

Fired vertical field only and ran mode control feedback with various gain settings on analog board to explore limits. Had been running with G=10 for good feedback with plasmas. Now, found strong (saturated) self-oscillations at ~16 kHz for gain settings above 100. (G=100 still OK). Analysis shows oscillations are produced largely as n=1 mode, indicating feedback loop is self oscillating (as opposed to circuitry after FPGA).

Wednesday August 25 04 41 A. Klein 44138-44159 Mode control phase scan

Began filling in more data points for fine phase angle scan, will have data at 5 degree increments (71 shots) by tomorrow. Will then also try to nail down phase angle for positive feedback...

Thursday August 26 11 14 A. Klein 44160-44178 Mode control phase scan

Finished phase angle scan, have data at 5 degree increments for entire torus. Also took some shots around angle where negative feedback is maximized, to try to nail down 90 degree measured phase between control and sensor coil fields. TF charging supply quit for a while, messing with rheostat miraculously fixed the problem.

Friday August 27 02 56 Liu/Maurer 44179-44184

Just very few shots were taken. The amplifer box for Hall sensor has problems. I added filter caps for all the channels, but it didn't get things work better. Instead, some of the chips got hot when the power was turned on for a while. And couldn't see amp outputs even on bench test. Hope to continue to investigate the problem when Nick is back next Monday.

Tuesday August 31 05 08 Liu/Mauer/Nick 44179-44205 Test Hall probe

Both VF only and plasma shots were taken on Monday and Tuesday. We current have 19 out of total 20 channels working fine. There is one noisy channel left. We will continue work on it.

Wednesday September 1 04 38 Liu 44207-44241(84-117) Hall array

With 19 channels, we took both vacuum shots and plasma shots. VF only shots with TF sweep were taken to check TF effect; TF only shots with bank voltage from 170V to 320V were taken for calibration purpose. Shots 44234-44242 were plasma shots. All available hall sensor channels detect clear fluctuations. Hall sensor data is currently located on Langmuir with tree name 'my_tree'.

Friday September 10 04 12 Maurer

Due to building power being turned off to fix the main breaker for floors 1 and 2 of the Mudd Bldg the HBT cryos will be valved off and shutdown overnight and also cua900, beta, sherwood and langmuir will be shutdown overnight at 5PM today, Friday Sept. 10. cua900 and the linux machines will be back on tomorrow afternoon.

Tuesday September 14 05 08 A. Klein 44249-44302 Mode control phase scan

With new (and improved) digital filter, which should give flatter phase response for frequency band 1-20 kHz, began detailed phase angle scan... Will cover toroidal angle in 5 degree increments, expect to fibish tomorrow. Plasma type is as always, agressive current ramp bringing the edge q to 3 early in the shot. Good reproducibility, and data looks good at first glance.

Wednesday September 15 03 40 A. Klein 44315-44359 Mode control phase scan

Finished taking data for detailed phase angle scan using the new and improved digital filter. Had to replace puff valve battery midway through. CUA900 is full, Dave Maurer is working on backing up data to tapes, until he is finished no more runs are possible.

Tuesday September 21 05 35 Liu 44373-44412(149-185)

Plasma shots with SMP shell at both 0cm and 4cm out were taken. In each case, SS feedback was turned on to exciting and suppressing modes. So far, it doesn't look it makes a big difference with one pair of Al shells in(0cm) or out(4cm): phase shift to excite/suppress mode stays same; the regular plasma shots with same bank setup look same also. Hall sensor array was recalibrated by moving to different locations (R). Plan to have another run-day this week by moving Al shells in and out. P.S. ICRH crate seemed have some problem (fan made a big noise) when we tried to turn it on. Therefore, we had to bypass it.

Wednesday September 22 05 26 Liu 44413-44442

m=3 mode was investigated by moving Al shells with different combinations. There is no conclusion so far.

Thursday September 23 04 51 Andrello/Stillits installation

We installed and pumped out the new multiprobe. Tomorrow we will leak check it and hopefully open it to the chamber to bake over the weekend.

Friday September 24 04 40 Andrello/Maurer/Stillits installation The multiprobe vacuum installation is complete. The pressure jumped to 10^^-7 torr and quickly came back down to 10^^-8 torr. Bake is on for the weekend.

Tuesday September 28 04 38 Liu/Maurer Hall probe

Hall probe shroud developed a leak during todays run while it was inserted into the edge plasma bringing the vacuum chamber up to ~50 Torr. Cryos have been regened, the bake is on, and the base pressure is 2 e -6 Torr comprised predominantly of H20 pumping with the Ebara cryo only so far. Glow discharge cleaning planned for tomorrow and continued baking. The Hall probe tubing has been taken off the machine and melting at the tip is clearly seen. We are currently investigating ways to surround it with a refractory material like alumina.

Wednesday September 29 10 23 Liu CPCI digitizer

CPCI digitizer is now integrated into hbtep main tree. I already tested it - working well. If anybody finds any complain please let me know. Joshua Stillerman gave us a lot of help during the whole process.

Friday October 1 04 51 Maurer/Andrello

summary of cleanup details... Tues. up to 50 Torr of air in the machine. Wed. glowed in D2. Prior to glowing machine base pressure was dominated by water. Miniturbo would not turn on and will be sent back for repair quote. Glowed using a static fill of D2 and pumping out with horizontal turbo every half hour. Ebara cryo warmed up spontaneously during the afternoon(cause not understood might be due to incomplete regen on Tues or something more serious). Thurs glowed in He. Fixed roughing pump line accident on regen line for the ebara cryo and started regen. Machined new new alumina tip for Hall probe array and mounted on shaft. Fri finished cryo regen and installed Hall probe tube and bellows and pumped it out and baked it. Opened it to the chamber after the afternoon talk. Base pressure was 7 e-8 Torr composed mostly of He after yesterdays glow. It is not 1 e-7 Torr and mostly water after opening the Hall probe gate valve. The bake willl continue over the weekend.

Tuesday October 5 06 07 maurer/liu

Tried taking cleanup shots today but had problems with the tree and mdsplus when initializing the shot cycle that took the whole afternoon to figure out. Partial fix has been implemented so we should be able to run tomorrow. Not sure what exactly happened to the hbt tree to cause these problems.

Wednesday October 13 07 29 Liu 44523-44558 vacuum shots

Calibration shots for both old and new Hall arrays were taken. I accidently turned off chamber ion guage. Jimmy came back to the lab from home to check the vacuum system to make sure everything is all right.

Friday October 15 02 15 Maurer/Stillits 44559-44570 TP-array

We tested the triple probe array (TPA). The test was conducted by moving the TPA 1cm into the plasma after each shot. Thus, the TPA is now 4 cm into the plasma. All tips, except for one - the third tip on the third TP, are working. The amps need to be checked. Nick did some work on the TF-charger, but we are experiencing some problems with the TF-charger still.

Tuesday October 19 04 46 liu/stillits 44573-44623 clean-up shots

We took clean-up shots all day long. We started with a puff time at 2.2 msec, but when the puff time reached 1.6-1.7 mesec the plasma had problems breaking down. Beside that, we were able to produce up to 4-5 msec plasmas, loop voltage ~12, and major radius at 94 cm. We had problems with the TF-power supply. The power supply was monitored; when the TF bank got stocked at a certain level, the amps on the power supply dropped down to zero. Thus, there seems to be a problem with the power supply.

Wednesday October 20 04 46 liu/stillits 44624-44688 clean-up shots + TPAWe

Took clean-up shots all day long. The plasma stayed around 94 cm, loop voltage ~11-12, and lasted up to 6 msec. We tried to lower the puff time, but we had problems breaking down the plasma when the puff time was lower than 1.8 msec. We tested the triple probe array (TPA). Thanks to Nick, all 4 triple probes are now working. We will hook up the TPA to the CPCI digitizer tomorrow. The batteries for the triple probe were dead and were replaced. We will continue taking clean-up shots tomorrow.

Thursday October 21 04 23 liu/stillits 44689-44738 clean-up

We continued clean-up shots today. Although we had to increase pufftime instead of decreasing it to breakdown the plasma, the plasma got better comparing with the shots we took yesterday. With VFFB on, plasma typically lasts 5-6ms. Also we saw sawteeth from SXR signal.

. problems: (1) TF bank power supply is getting worse. It failed to charge for most of the afternoon. It's a very high priority for it to get fixed before we could productively run the machine. (2) The first channel of triple probe array still has a problem with the chip - it keeps getting blown - but Nick is looking into it.

Monday October 25 03 11 liu/stillits 44739-44756 clean-up shots

We were only able to take a few shots today before the TF went dead. However, we were able to produce some nice plasmas: loop voltage ~ 8-10, duration ~ 5-6 msec,SXR ~ 0.1 (sawtooth). All four TP's are now working, but we are having problems with blown chips on TP no. 1.

Tuesday October 26 05 03 Stillits/Liu 44757-44785 clean-up

We took clean-up shots all day long. We are beginning to see some very nice plasmas: loop voltage ~ 10, duration ~ 5-6 msec, SXR ~ 0.1-0.2, however, we were not able to lower the puff time below 2.0 msec. The RPFB was on: 94 cm from 2.5 msen till disruption. We checked the puff-battery and it showed 240 V, which should be ok. The triple probe array (TPA) was working well till we blew the board. The second board was installed, and we will continue to test TPA tomorrow. The Hall sensor array is working well and we are expecting to get some useful data if the plasma behaves repeatable tomorrow.

Wednesday October 27 03 54 liu/stillits 44786-44808 hall-probe + TPA

We are beginning to produce plasmas that are behaving very nice: duration more than 6 msec, loop voltage starting at 10-11 and decreasing to ~ 8, SXR ~ 0.1-0.2 and in some cases ~ 0.3-0.4. We ran the triple probe array (TPA) with the new board, and it seems to be working: we are getting very nice signal (one I_sat and two voltage measurements) for TP no. 1. Furthermore, we are not seeing any huge spikes in the signals. We will keep testing TPA tomorrow. The Hall probe array is working normally. We were trying to get some target shots for natural m=3 MHD mode study. The plasmas were quite repeatable. However, we couldn't take enough shots since TF power supply was working part-time. Nick has been working on it - in trying to pin down the problem, he has put light bulbs in parallel at different locations.

Monday November 1 07 14 Liu/Stillits 44809-44850 Mode structure

Good news: Nick has TF power supply fixed! So as we were able to run experiments smoothly today. Although the base pressure is a little bit high (~9*10^(-9)) comparing with the old days, plasmas are quite repeatable. Shots were taken by pulling all the Al shells out. Also, m=3 mode structures were investigated by partially turned on FPGA feedback control (Thanks Alex for adding new fuction on labview control pannel of this purpose). We are looking forward to collecting more data tomorrow.

Tuesday November 2 02 26 Liu/Stillits 44851-44877 Mode structure

We collected most needed data for mode study by investigating m=3 mode with different FPGA feedback scan. However, we had a hard time breaking down the plasma since late morning. PGA showed water and N2 are dominant effect to vacuum at current point. The base pressure is 1.7*10^(-8). TPA has been tested also. Data from the 1st channel is promising.

Monday November 22 02 59 stillits 44924-44939 clean-up shots

We had hoped to take some clean-up shots today, but we were not able to break down the plasma. The pressure was 1.1e-8 and we ran as high as 3.0 msec in puff time, 275 on OH elect, 6.5 on OH start. The e-gun is working and there is gas in the system (a signeficant jump in pressure during a shot). There seems to be a lot of water in the system, which probably stems from a leak.

Wednesday November 24 04 18 stillits 44940-44959 clean-up shots + TPA

The e-gun battery was dead. After having replaced the battery, we were able to break the plasma down. However, we have to run with at a very high puff time = 2.8 msec; if we try to lower the puff time below 2.8 msec, the plasma can't break down. Anyway, with a puff time of 2.8 msec, we were able to create some fairly hot and long-lasting plasmas: SXR ~ 0.3-0.4, duration of 7 msec, loop-voltage ~ 11 down to about 8. The start pressure was 1.2*10^-8. Furthermore, we were able to lower OH_start and OH_elect to 5.5 and 265, respectively. We can probably lower the OH values even further and still get nice plasmas. We will check the puff valve battery on Monday. After Nick fixed the problems we have had with the cable (from the TPA to the digitizer), the TPA signals seem to be working very well (we hooked the cables up to the function generator). On Monday we'll try to insert the TPA into the plasma and get some data. The RGA seems to be indicating that there's some H20/D20 and Nitrogen in the system: 2.5-3*10^-9 for both water and Nitrogen.

Wednesday December 1 07 12 Liu 44968-45040

First, CPCI communication problem has been fixed. The problem probably was caused by network frozen. Today's plasma is repeatable and controllable. With the Hall probe inserteded into plasma 2.5cm (1st channel of the probe at R=1.047), we took shots with current ramps from 4MA/s to 0A/s, et. al. Triple probe array is ready to capture data tomorrow.

Thursday December 2 04 19 stillits 45041-45058 Testing TPA (triple probe array)

We tested the TPA today. There semms to be some problems with the electronic; we are getting saturated signals on all except for one channel - even when we pull the TPA out of the plasma. We will look into the problems tomorrow.

Friday December 3 03 58 stillits 45059-45072 TPA-testing

We tested the TPA today. However, we are still experiencing problems with the TPA-electronic.

Thursday December 16 04 48 David Maurer 45147-45180 Discharge development, bias probe

We've been trying to develop the same shots as used for mode control experiments late in the summer. The puff valve is deteriorating, fresh battery still requires very long puff times to achieve reasonable gas fills. We also experienced a few computer problems, which have been corrected. The discharges now look almost like the target discharge, it may still be a matter of adjusting the gas puff in small increments. Current puff time is 8.5 msec (compared with 2 msec in August to achieve similar densities). We expect to have the traget plasmas by tomorrow morning and turn on the niased probe to affect rotation...

Tuesday December 21 05 26 Maurer/Klein 45198-45235 low kink rotation

By adjusting the fill pressure we were able to generate a plasma similar to those used for the recent feedback studies (see shot 45217 for example, the target shot was 43481). The puff valve response is changing on a daily basis. Current settings are Tpuff=1.15ms gives 3.1 e-5Torr fill pressure. Friday it was Tpuff=4ms for the same fill. In the future, people would be well advised to take static fills of the chamber with the pumps valved off to know their operating point. The biased electrode was inserted approximately 3cm into the edge plasma. Bias voltages of 50, 100, and 300V were applied to the probe tip. The higher settings (100 and 300V) were tried because it was difficult to see any rotational effects at 50V on the MHD. The reason for this is not clear. I plan on trying 5 cm insertion tomorrow and will keep the voltage high until I see an effect on the MHD such as reversal of the rotation direction and then back off the voltage and aim for only slowing the mode down to zero.

2005

Monday January 10 04 46 A. Klein 45292-45337 Gain scan feedback

Performed gain scan with full coverage feedback. Used only two toroidal phase angles (max and min supression). Found feedback effect visible in Rogowski coil signals even at gain 14 times lower than that used previously (although greatly reduced effect). Max gain used was 1.4 times greater than previous setting (distortion of control waveform out of crunch amps limits gain). Pulser for puff valve failed, switched to spare. Nick is repairing the unit. Will continue partial coverage feedback experiments throughout the week.

Tuesday January 11 05 23 A. Klein 45338-45426 gain and partial coverage mode control

Performed gain scan for partial coverage scenarios: midplane only coils activated, and midplane coils de-activated (top and bottom coil sets active). Did detailed gain scan for toroidal phase angle set to max ecxitation and max suppression. Data looks very good (3 shots at each setting) and interesting. Maximum gain (1.4 time greater than previously used) does NOT result in maximum suppression due to distortion of waveforms, high freq. modes excited. Best supression seen when gain is a bit lower than setting used for phase angle scan this summer. Extreme low gains of course diminish feedback effectiveness, and at gains 14 times lower than previously used the kink modes seem unaffected. Puff valve behavior drifted and might require that I re-do the data set for mid-plane only coil sets active, because base shot was a bit different and more varied than usual. The data set for midplane coils turned off case looks very nice though, after I adjusted gas puff time. Problems with the new digitizer (for triple and hall probe arrays) resulted in errors and restarts of cua900, after some time YuHong was able to make the system work. Will continue these types of experiments in the next few days.

Wednesday January 12 03 54 A. Klein 45431-45497 partial coverage mode control gain scan

Todays partial coverage scenarios: Toroidal positions de-activated, did gain scan with shells 1,3,4 on but 2 and 5 off, then did gain scan with shells 1,3 on but 2,4,5 off. Cursory glance indicates that such feedback does not work very well. Minor differences between feedback on and no feedback shots, even at high gain, both for exciting and suppressing phase angle setting. Scan was detailed (three shots each setting, 6 gains, both exciting and supressing phase angle). Puff valve drifted again compared with yesterday but periodic checking of puff time on fill pressure made this not an issue. Will continue these types of experiments in the next few days.

Thursday January 13 04 38 A. Klein 45501-45599 gain scan partial coverage mode control

Ran more gain scans for various partial coverage scenarios: 1) One toroidal location disabled (pretty successful feedback) 2) Redid midplane only coils active (plasma is much better and cleaner data) 3) Every other coil disabled (fairly successful) As before, 2 phase angles (max suppression and excitation) were used at 6 gain settings, three shots each. In general, plasmas are very reproduceable. The machine has cleaned up considerably since Monday. Exception: When the triple probe array was inserted 6 cm into the plasma, very different discharges were observed. These were very quiescent except for intense sawtoothing on the x-ray, very long lived, very hot, edge q went down to between 1 and 2 and stayed there. All other parameters were the same (intense current ramp). These shots were reproducible (5 times in a row), but then probe was withdrawn and I continued with gain scans on the "old" shot types.

Tuesday January 18 05 31 A. Klein 600-627 gain scan partial coverage mode control

Finished taking data for partial coil coverage gain scan. Had some problems with ignitron cooling flowmeter interlocks, but they were resolved. Puff valve was very unpredicteable for the first few shots (50% variability) but then stabilized. Will now analyze the ~400 good shots taken last week.

Wednesday January 19 07 49 Yuhong Liu 45628-45699

It's a long run-day. Current ramp scan was done with all Al shells pulled back.The plasma shot length was not affected by withdrawing Al shells. The puff valve was wavy at the beginning of the run, but it was adjusted from time to time and stayed stable for most of the shots we took taday. Machine behaves well.

Thursday January 20 04 35 stillits 45700-45732 TPA testing

We tested the TPA today. We tried to rotate the TPA 180 degrees (at different positions: 2, 3.5, 5, and 6.5 cm) to see if the results (temperature and density) would be the same. There does not seem to be any change in the values. However, we began to experience some problems with the signals from the TPA after we had rotated it, which seem to be related to some of the connectors. We'll look into that tomorrow.

Friday January 21 03 38 stillits 45733

Friday January 21 03 40 stillits 45733-45744 TPA testing

We ran the TPA today. However, we are experiencing problems with the electronic (connectors) still. We'll look into that next week.

Monday January 24 04 38 Maurer ICRH

Fixed CTS and traverser problems with the ICRH crate Jorway. The trigger is now working fine(several shots at least!). FYI the L8210 does give NOX storage errors in the action monitor and serial crate controller (SSC) errors if the crate cooling is not on and the door is not well closed. The ICRH was turned on and a couple shots taken at 15kV and 25kV to check triggering, etc. The bank fired but the plate currents do not appear in the data base properly yet. I assume this is a cabling and rectifier hookup problem since the digitzers come up with no signals and appear to be storing properly in the data base. I will try to piggy-back and check this out tomorrow or Wed.

Tuesday January 25 04 07 Stillits/Maurer triple probe

Started to diagnosis problems with the triple probe array. Did not get too far along due to CAMAC serial highway sync problems and communication problems between cua900 and the CPCI crate that appears to freeze MDSplus during initialization of the serial highway. It was discovered that the basement u-port adaptor is causing the no-sync problem presumably due to over-heating. CTX has been temporarily jumpered out of the highway during trouble-shooting of the system. Will continue tomorrow.

Wednesday January 26 03 23 A. Klein 45760-45802 partial coverage mode control gain scan

Had intended to run two more partial coverage scenarios with mode control. Instead, experienced and worked on problems all day: U-port adapter in basement failed, needs repair or replacement. Took adapter from CTX experiment and installed in basement. Fiber optic link to ICRF rack appears to have failed, intermittant link took a while to diagnose. Had to bypass that crate. Experienced CAMAC and initialization problems, needed to back up, but back up disc was full. Yuhang deleted shots, then archived data on cua900. Finally, puff valve was extremely irratic, esp. at first. Sometimes no gas at all was admitted to chamber (perhaps CAMAC failure to trigger), when it did open, the amount of gas varied wildly. This appeared to have stabilized towards the end of the day. However, I was unable to achieve target shot, seems as though OH start is much lower than previously despite match in bank settings. Upped OH start considerably, which improved things, then they became worse and worse again (low initial plasma current). Will try again at later date.

Tuesday February 1 04 46 A. Klein 45815-45860 diagnose problems

Had intended to finish partial coverage gain scan for mode control feedback. Instead, dealt with problems. OH Start bank was not providing the voltage one would expect, one of the Klystrons had failed and was replaced by Nick. After that, the OH bank was still not up to par, and produced inconsistent results even for fixed settings. Then determined the OH Start bank is leaking off charge fairly rapidly. Took much of the day to realize this. Played with voltage settings, "refreshing" the charge, adjusting the charge rate so that bank finishes close to when the TF bank is finished charging. This may be moderately succesfull. There must be a capacitor in the bank that is bad. Higher voltage settings result in lower start voltages, it is a mystery and is not explained by leaking capacitors (is it?). Also, had trouble breaking down to form plasma. This may be related to the OH start bank problems, or perhaps not. Checked out e-gun, seems to be working fine. Gas puff also works fine (although long pufftimes are required to get target fil pressure). Formed plasma only 3 times today, which is a bit inexpliceable. The North Rack contiunues to have problems with digitizers for the SMP's.

Wednesday February 2 04 02 Maurer/Rivera

Fixed problems with the Ohmic heating circiut today. The OHST bank dump had failed. The ST bank dump oil was almost black in color. The reduced standoff voltage was enough that the bank was continually discharging through it leading to yesterdays problems. The dump oil was changed and some tracking on the dump structure was cleaned up. The dump was then reinstalled. The bank was the checked and is working properly now. FYI: the CTX crate is back in the camac loop and will have to be turned on before running tomorrow. In addition, the ICRH crate is also back and will need to be turned on. Alex will resume Tuesdays incomplete feedback run starting tomorrow. Lets hope the breakdown problems from earlier this week do not occur.

Thursday February 3 04 09 A. Klein 45867-45951 partial coverage mode control gain scan

More partial coverage gain scans: Today, had no problems with the banks. Plasmas were good (very first shot was target plasma). SMP's still not being digitized properly. Puff valve failry stable, it drifted a little during the day but I compensated with the puff time. I ran 2 scenarios: top two poloidal groups only, and uppermost (single) poloidal group only. The data is very surprising. Recall that the previous data suggested midplane coils only were not as effective for feedback as top and bottom only, and that perhaps this had to do with not having enough poloidal coverage to grab onto the modes. Well, todays data says:Top group is very effective at suppressing/exciting 3-1 mode, with top-midplane group or even without ! At medium gain, the one lone top poloidal group (which is furthest from the plasma surface since the plasma is shifted down somewhat) is as able to suppress the kink as the entire system (all groups on) is. All the data so far suggests that a) midplane coilsets do not couple as effectively to the external kink mode, or b) what we think is the midpalne is the top, and vice versa. I spoke with M. Shilov, he seemed to be sure of what is what, confirming that what I think is the top/midplane is in fact the top/midplane. Perhaps he has it wrong? Thus far I have the coils numbered 1 assigned to the midplane, and the ones numbered 3 assigned to the top (or bottom). Does anyone have documentation to confirm/refute this? Will analyze the data more formally, we can then decide what it might mean or what other experiments should be run.

Monday February 7 04 43 A. Klein 45973-46015 partial coverage mode control gain scan

Investigated partial coverage gain scan for one poloidal group (bottom midplane) only active scenario. Doesn't look like it performed too well. Will keep running these types of experiments tomorrow. ad some trouble due to CPCI data acquisition system, ended up restarting cua900 and disabling the CPCI stuff.

Tuesday February 8 02 30 A. Klein 46015-46100 partial coverage mode control gain scan

Completed partial coverage gain scans for top poloidal group only active and top-midplane poloidal group active only. The data conforms to previous data and trends. The puffvalve now requires 35 msec pulse times to give required fill pressure... I think this may conclude the partial coverage gain scan experiments (I hope). Will analyze the data in the coming days...

Monday February 21 05 07 Maurer 46241-46260 Ran initial tests with new vertical poistion bucking coil.

The coil was run at 50, 100, 200, 400 and 600A of current. No breakdown problems were experienced even when running at the highest 600A level. Plasmas typically had about 350 to 450A of plasma current in the OH transformer bias phase indicating a medium gas fill level. Initial observations showed the global plasma parameters were not changed significantly until 200 to 400A were run in the new coil. Plasmas were actually pushed in major radius at the 400A level. Perhaps indicating that we are pushing it too high. There is preliminary evidence that the poloidal sensor arrays top-bottom asymmetry is less at the 200 to 400A level. No major jumps in SXRs, loop voltage or other parameters were noticed during the expts. We are still having camac problems with the shell mounted coils. I found what appears to be sin(theta) coil cabling in the west rack. Have not yet been able to check out if it is ok yet due to some afternoon camac-vax problems. Nicolai continued to check out ion saturation current measurements on the triple probe array and will also piggy back tomorrow.

Wednesday February 23 04 31 Maurer 46257-46307 Vertical position

Moved top and bottom limiters out yesterday 2cm to allow some wiggle room for the plasma as the vertical position bucking coil experiments continue. Today was a day of problems... shorting stick left on the OHST bank caused some localized arcing and damage to the stainless steel resistor but nothing serious. This was cleaned up in the morning, the bank is OK, and a new shorting stick built and installed. The CPCI crate was taken out of the loop due to a fan failure on the digitizer side of the rack which was extremely noisy and discovered during the mornings run. The company has been contacted about a replacement and the problem fan was removed temporarially and the crate re-installed for tests tomorrow. The polarity of the leads for the vertical bucking coil were reversed (to move the plasma down) as a check on Mondays initial results before proceeding with a more detailed position-current scan. I tried to set up a target plasma but had great difficulty with the puff valve and intial fill pressures. They were widely varying the whole day. So much so that the plasmas were ranging in lifetime from 1.5ms to 6ms. The initial bias phase currents went from about 0 to 1000A. The puff valve pulser was found to be putting out about 100V instead of the 235V battery value when measured on a scope. The input trigger appears solid. The backup pulser was installed but did not give out a pulse when triggered. Nick is checking out the two units. Perhaps the transistors are failing because of the factor of ten to twenty larger puff times that we now require to attain the same fill pressures used in the past (due to puff valve aging?). I would suggest that we take time tomorrow to recalibrate the SMPs, flux loops, etc which has still not been done until the puff pulser is fixed. There is only Mondays data to discuss for Thurs meeting.

Thursday February 24 05 06 maurer

Figured out north rack problem with the smps and triple probe diagnostics in the afternoon. The local serial ribbon cable input on the fiber optic uport adapter is failing. This causes information on the store part of the shot cycle in crates 7 and 9 to be lost (or mostly lost). This appears to be the cause of the recent smp and triple probe problems in the database. The ctx uport adapter was borrowed to trouble shoot the problem.

Tuesday March 1 04 32 A. Klein 46369-46433 vertical position shift

Collected ample data on vertical shift coils and their effect on one type of discharge. Will analyze in detail. Only with extreme currents in the correcting coils does the plasma behave grossly differently/badly. Hottest plasmas near 200 Amps driving plasma upwards. Tried to digitize Sin-theta Rogowski coil, the coil is not functioning (probable short circuit, 0 Volts).

Wednesday March 2 04 51 Maurer/Liu Calibrations

Started calibration checks today on the flux loops, smps, bank currents, etc... work still in progress. Tomorrows run is with the biased probe and triple probes (Stillits). FYI: Dmitry and I will be at PPPL going over VALEN again and in particular learning how to do the partial coverage runs to help in understanding Alex's recent data presented at the last physics meeting.

Sunday March 6 07 11 Maurer 46524-46551 Calibrations

Sunday run... tested HFS triple probe. Even with the north rack camac problems fixed the isat measurement is still not working on the triple probe. The digitizer checks out OK with a sine wave input. Pulled the amp out of the north rack and checked the battery is OK. Must be the opamp or other circuit problem. Will have Nick check it tomorrow. Spent rest of the evening running sine and square waves through the shell mounted probe electronics to check for any AC response differences that might give rise to measurement errors. All the working channels looked OK with a 1kHz square wave input. Also checked the flux loop in a similar manner on Wed with Yuhong. All the flux loop channels were OK also. Still have to go over the bank current rogowskis. Did not get the bucking vertical position coil on today due to the calibration checks.

Wednesday March 9 02 46 Jmaes/Stillits 46560 -46603 Beta Enhancement

Spent the morning checking equipment, ICRH, Triple Probe, RPFS, & Bias Probe. We were able to get ICRH signals in the tree on Monday. Replaced an op-amp in the triple probe, but signals still seemed strange in the tree; still debugging. In the meantime, we will use a t-probe tip to make an Isat probe for a direct relationship to the inboard density. This will allow us to continue the ICRF studies this Thursday and Friday. We were able to develop plasmas that resemble h-modes with the ICRF and bias probe on and the plasma positioned at ~94 cm. We had some problems with the RPFS, Nicolai and Nick are investigating. Dave Mauer has offered us assistance in rerouting the screen room out of the serial highway for tomorrow and Friday. I am not sure if that affects the RPFS or not but we are confident that we can work it out. We needed the extra u-port adapter to replace the one removed form ICRF a few weeks ago. We borrowed CTXs U-port, but returned it yesterday afternoon so they can run this week.

Thursday March 10 10 54 Maurer 46607-46638 vertical position shift

Yesterdays run... HFS triple probe isat measurement is now working. Found bad opamp chip. Hooked up shell probes and had flux loops, hall array, LFS triple probe, poloidal feedback sensors on for todays vertical shift expt. Ran the vertical shift coil in 50A increments to scan position. Target plasma was outboard limited with Ro = 94 to 95 cm. Took three shoots at each current up to 350A. Coarsened scan to 1 shoot per 50A increment from 350 up to 600A due to limited space on cua900 at the end of the day. Data analysis underway. CTX uport returned. Screen rm jumpered out of camac loop to allow for ICRH run today.

Thursday March 10 06 56 James/Stillits 46643 - 46695 ICRH

We bypassed the screen room first thing this morning. We used the bias probe, to increase inboard plasma densities all day, but each time we tried to move the plasma inboard with the RPFS, the densities seemed to fall. We weed to investigate this further, but today, we were just trying to get target ICRH plasmas by whichever tools were available. We left the plasma edge at ~.94 m and found that when the radial current is at 200 A, the edge densities increased noticeably and relatively remained so throughout the plasma lifetime. We then fired the ICRF at lower energy at first (~18 kV) and then up to 28 kV for the final 5 or so shots. I am analyzing the data now, and plans to begin a more systematic sweep tomorrow of edge densities to lock in the ideal parameters for FMSW Coupling.

Tuesday March 15 05 54 Maurer 46765-46829 HFS density optimization

Ran plasmas today to otimize the HFS density measurement on the triple probe. Was able to develop target plasma with edge density of around 4 e-12 per cc as measured on the HFS triple probe. This was done by a combination gas puff, vertical shift coil current, and OH bank adjustments. I was able to make a relatively long lived plasma with Ro at 92cm and another target with Ro at 91cm for a 2.5 ms period. This was done so far without the biased probe or radial position control. Late in the day I turned on the baised probe at 250V and expected the density to increase even further, but this was not experimentally seen. The probe current and voltage signals were very noisy. The major radial evolution changes slightly with the probe on. As as a reference case we usually get a factor of two or so increase in edge density with the biased probe (this gives values of about 5 e-12 per cc). So we are getting close to that edge density without the probe just by todays discharge programming with increase gas puffing, but it is not clear why the probe is not enhancing the density even more. There were some Lanl problems during mondays run that hopefully will not impede tomorrows work. I plan to run from the start with the probe in the morning to see if I can increase the density even further. If this shows no promise I'll switch on the ICRH and use the target I have currently going in the afternoon. In todays run I was using 200 to 250A in the vertical shifting coil.

Wednesday March 16 08 32 James 46738 - 46763 Radial Field Scans

This from Mondays runs: We were unable to use the Lanal, seems to be a repeat of a problem experienced last year see Nick, Dave or Nicolai for more information. The bias probe was on for all shots. Nicolai, help me to set the plasma between 93 and 92 cm with lifetimes from 3-5 ms. Once we were able to have a suitably long lifetime, I began a sweep of the vertical positions starting at 175 A radial current and ending at 225 A, at 25 A increments with about five shots for each step. Looking at the 2.3ms-4.1ms window density averages, the largest were shots 46752 (4.18E+18) and 46760 (4.01E+18). From the sweep, it looks like the 225 A current produced the highest densities in that window. I also turned on the ICRH for shots 46761 63 and 46764 (vacuum shot) to continue to look for some loading and any other signs of coupling while I everything is still working. I ran out of time before I could try 250 A radial field currents, so it may produce an even higher density measurement.

Thursday March 17 10 32 Maurer 46830-46880 High density plasmas with ICRH

Forgot yesterdays run report... I continued making high density plasmas and using the vertical shift coil at 225A. The two target plasmas developed yesterday for ICRH studies are: (1) Ro= 92 cm plasma with edge density of 4 e-12 per cc and (2) Ro= 91 cm inboard limited (on the RF antenna limiter) plasma also with edge density of approximately 4 e-12 per cc. Ran ICRH pulses at 28kV into both of these target plasmas. The SXR fan array has enoungh pickup that it is not useful in diagnosing any temperature rise. I will try to play with cabling to see if it can be reduced. No easily seen signature of loading or heating seen just looking at the data initially during the run. Looking at the Q measurements for subtle loading effects today.

Thursday March 17 04 54 Stillits/Maurer 46881-46948 H-mode + TPA

We had problems with the LANL, which we have had for some time now; each time we took a shot the banks got dumped. We were, however, able to create long-lasting (~ 6-7 msec) plasmas without the RPFB. Furthermore, we couldn't get the D_alpha meter to work; we'll look into that tomorrow. The BP (Biased probe) was on at 250 V and 5 cm into the plasma, and the TPA (triple probe array) was inserted 3.5 cm into the plasma. We saw several cases of "H-modes."

Monday March 21 05 08 stillits 46982-47063 H-mode + TPA

We ran with BP on at 250 V and inserted 5 cm. The TPA was inserted 3 cm into the plasma. We saw some cases of H-modes, some more interesting than others. Among the interesting shots are shot# 47003, 47026, 47043, and 47044. We'll continue with these kinds of shots tomorrow and perform the BP-TPA test. By the way, the high-field TP is not working.

Tuesday March 22 04 21 stillits 47064-47100 H-mode + TPA

We ran with TPA inserted 3.5 cm into the plasma, and BP on at 250 V and inserted 5 cm into the plasma. We saw several cases of H-modes, but the run day was cut short when the TPA stopped working.

Tuesday March 29 04 16 A. Klein 47103-47126 Mode control coverage

Ran to tie up loose ends with partial coverage mode control experiments. Previously, Dr. Mauel had suggested two control coils were always highly correlated with feedback effect (magic coils). There is something to this, judging by todays experiments. Toroidal position ^#3 apparently is close to Rogowski coil used for analysis, and activating coils at that location strongly affects feedback analysis when based on Rogowski coil measurements. Will have to re-evaluate lots of data, using 2nd Rogowski coil, poloidal sensors, and shell mounted probes.

Wednesday March 30 04 38 A. Klein 47126-47182 discharge development

Developed discharge with R ~ 93 cm, current ramp, external kink when edge q goes below 3 while vertical shift coil was powered with 260 Amperes. Shot 47182 is the target, I think. Will use such shots to do some mode control feedback experiments.

Friday April 1 04 52 stillits 47184-47201 TPA

Thanks to Nick, the TPA is working again. We tried to test how close to the BP we could move the TPA before experiencing any major problems. The BP was on at 150 V and 5 cm into the plasma, and the initial position of the TPA was 3 cm. After having experienced problems with the CPCI digitizer this morning (we couldn't initialize), we got as far as inserting the TPA 4.2 cm into the plasma - 8 mm from the BP - before we again started having problems with the CPCI. Up to that point we didn't have any problems with the TPA getting to close to the BP. Next week we will try to address the problem with the CPCI. On the positive side, the high-field TP is working again and the digitizer for the D_alpha-meter is working again (we will run with the D_alpha-meter next week).

Wednesday April 13 05 15 stillits 47221-47308 H-mode

This is run report for Monday, Tuesday, and Wednesday this week. We experienced problems with the CPCI this Monday; we were not able to initialize the TPA channels. However, Yuhong seems to have been able to fix this problem by creating a completely new publicly - for the HBTEP group - accessible CPCI-tree, which will be tested tomorrow. Furthermore, we are having troubles with the D_alpha-meter. The digitizer is working, and we have tested all the cables (which are also working), but we are not able to get any signals from the spectrometer. We will try to fix this problem tomorrow. Since we have had problems with the CPCI, we have no been able to perform the BP-TPA scan. Instead, we tried to create "H-modes" in which we had the BP inserted 2 cm into the plasma with 300 V. It seems that we are able to create "H-mode" structures with this "BP- setting" too; for example shots ^#47279 and ^#47305 are interesting.

Monday April 18 05 06 James 47320 - 47355 ICRF

I used the scope to measure the period of the ICRF oscillations T=2.28x10^-6 s so f= 4.38 MHz (should be 4.5 MHz). I put this into Prof. Mauls program, but we are still unsure of the toroidal B-field, so it is not clear how significant this change is. We were able to develop some fairly high densities (centered around ~5x10^18 [+- 1.5x10^18]). ICRH was fired at full strength, ~31 kV, and a few shots showed significant loading. After looking at some previous shoots, it seems that the inboard density can be double what I got when the biased probe is inserted 5 cm. Combine this with clearing the toroidal B-field uncertainty so we can adjust the frequency, and we should be able to determine weather or not more drastic measures should be taken.

Monday April 18 02 33 Liu CPCI digitizer status - works well

Want to clear that both hardware and software of the cpci digitizer are working well. I spent time last week created a new mdsplus tree on langmuire /home/hbtep, which is accessible for the whole group. The tree's name is 'subtree_cpci'. Anybody who wants to retrieve data from the tree, please just use the standard mdsplus/idl commands/programs. If you want to get data from terminals that are not langmuir, please use "mdsconnect, 'langmuir.ap.columbia.edu' ". The commands "mdsopen, 'subtree_cpci', shot#", "a=mdsvalue('top...')", etc still work the same way as you get data from 'hbtep' tree. Please be sure the correct node name to get data - you can check it using traverser. One thing needs to be mentioned: the problem that cpci digitizer sometimes stops working is quite possibly because the CPU usage on langmuir is more than 99%. If anybody has the idea of limiting user's usage of CPU, it will solve the problem.

Friday April 22 05 18 stillits 47435-47450 BP-TPA scan

Thanks to Yuhong - we were able to run the CPCI/TPA today. The BP was position 5 cm into the plasma, and the voltage was 150 V. The initial position of the TPA was 4.2 cm (= the center of the deepest inserted TP was 4.2 cm into the plasma). we then slowly moved - in step of 2 mm - the TPA into the plasma. We stopped taking shots when the deepest inserted TP was inserted 5.8 cm into the plasma. We did not see any spikes - neither on the I_sat or voltage (float or +) for each of the TP - at all at any position.

Monday May 2 04 00 Alex Klein 46504-47538 mode control compass scan

Began detailed compass-scan of mode control system and tokamak configured exactly as during August 2004, but with toroidal position ^#3 control coils de-activated. This is the position near an m = 3 Rogowski coil which otherwise is subject to direct pick up. The plasmas look good, data looks good at first glance, will finish tomorrow (am trying to cover 72 different angles).

Tuesday May 3 02 41 Alex Klein 47539-47589 Mode control compass scan

Finished compass scan with mode control system configured as in August 2004, but toroidal position 3 control coils deactivated. Will now analyze data to see how rogowski is affected when nearest control coils are far away.

Tuesday May 10 05 15 stillits 47605-47653 TPA + BP

Thanks to Yuhong, we are now able to take data with the TPA. After we had kicked Bialek of langmuir and changed OS to the old version, we were able to take shots without any major problems, except for a few shots in which the CPCI was not able to store data. we ran with BP at 2 cm & +300V and TPA at 3.8 cm (position of the deepest inserted TP). We saw some cases of spikes in SXR while the BP-current was ~ 0A; at the same time we saw drop in I_sat/density on the deepest inserted TP's, so the "barrier" seems to appear deeper inside the plasma than 3.8cm. But some more analysis is required. Next time we'll try to insert the TPA deeper than 3.8 cm.

Wednesday May 11 04 02 Alex Klein 47657-47699 mode control compass scan

Performed full coverage mode control control coverage scan, to confirm that the feedback system actually still works as before (there had been a question, as plasmas appear to have slightly faster natural rotation rates compared to previously). At first glance: Mode control works, dramatic exitation/suppression on Rogowski coils as before. Tomorrow, will do phase angle scan with toroidal position ^#7 (far from either Rogowski) control coils muted.

Thursday May 12 03 05 Alex Klein 47699-47778 partial coverage mode control gain scan

Ran detailed compass scan study with mode control coils at toroidal position ^#1 muted. This is the set of coils near Rogowski m eq 3 coil number two, and also the set mounted on the shells which are partially withdrawn due to Hall probe access. At first glance, it looks like feedback is working, as the Rogowski coil ^#2 mimicks Rogowski coil ^#1. Will do analysis to quantify feedback effect.

Friday May 13 02 34 Liu/stillits 47779-47822 Hall probe calibration

Yuhong tried to calibrate the Hall probe today.

However, during these shots we had to reboot the CPCI and langmuir several times. We don't know what makes the system so unstable - but we have emailed Josh and Peter about the problems.

Friday May 20 02 08 stillits 47889-47900 TPA calibration

12 shots were taken today (three for each TP). Data was stored on floppy disk, and further analysis will be done in IDL.

Monday May 23 09 56 James 47901 - 47921 B-dot Probe

Tested the B-dot probe today, ensured the cable was indeed attached and checked it for shorts. Initially took an ICRH only shot to compare with B-dot readings in the past (see shot ^#36547), with the rectified probe raw signal in the raw bottom antenna current digitized channel. The signal seemed to just be pickup in the wire (see shot ^#47902). Then bench tested the Ge rectifier, but sig. gen. only went to 1MHz and at this frequency it was hard to tell if the signal was truly rectified. Went back to the ICRH crate and used the o-scope while firing the ICRH to see both the un-rectified and rectified signals over the lifetime of the ICRH pulse. Seems like the rectified signal oscillates with the carrier as it drops, pointing to a possible capacitor failure, but I ran out of time before I could replace them will do so next time and start the toroidal B-field scans with ICRH.

Tuesday June 14 03 50 stillits/liu 47952-47982 cpci

The CPCI seems to be working now. We took about 30 shots, and in each shot all the data was stored; this even happened while Bialek was running (using more than 99% of CPU).

Thursday June 23 02 46 Stillits_Nick 47987 OH test

Tested the OH bank to see if there was any damage from the water leak. It worked fine. Will test triple probe tomorrow.

Friday July 8 05 25 stillits 48192-48218 TPA

The TPA is now working. After Nick had cleaned the pins, allocated new channels on the CPCI (we are now using ch. 49-54 for TPA1 and TPA2) and after we had recalibrated - by means of the effective areas - we were able to get the temperature profile we had anticipated. The temperatures were underestimated by: TPA1 (outermost): 3%, TPA2: 11%, TPA3: 13%, TPA4 (innermost): 5% However, we have not taken into account the ~25% - related to the electronic - we are overestimating the temperature on each TP, but it will not change the profile structure. All this will be added in the tree. Furthermore, after we put a fan in front of the CPCI, the CPCI has become more reliable. There are, however, instances in which data is not stored, but the solution to that problem seems to be to turn the CPCI off and on - without rebooting langmuir. We have also changed the sampling rate on the CPCI from 200K to 250K and (almost) gotten rid of the time delay ~200e-6 sec, but there are still some a few minor problems that need to be solved. I'll try to look into these problems and put it all on the tree this weekend.

Tuesday August 2 04 43 Maurer Fixing broken quartz section

Summary of Monday 8/1 and Tuesday 8/2. Found the HBT vacuum chamber up to air prior to Mondays group meeting. The quartz piece between the Hall probe and SMP chamber pieces had cracked around the circumference sometime over the weekend. The exact cause of the break is unknown. The failure mode is however similar to the quartz piece that cracked next to the ICRH section about 5 years ago when a cinder block wall was taken down in another part of the lab. It was thought then that floor vibrations due to taking the wall down caused enough of an impulse to start a crack propagating in the quarz. In checking with facilites, there was no scheduled heavy construction on our or ajoining floors so I am not sure what caused it (if that is the cause).

The quartz piece was taken out this morning. It took a bit of time to pull the bolts, move the Hall probe stand, pop off its flange and dismantle the connections to the control and sensor coils mounted off of those stainless shells so they could be moved over inside the chamber to allow the quartz piece to be removed (since the gap coils extend over underneath the quartz piece). I broke one wire of the smart shell coils in dismantling them that will have to fixed. Upon inspecting the broken quartz we found a part of the crack that does go over to the edge of the piece and might have been caused by contact with the next door chamber flange. If anyone wants to do more post mortem analysis the pieces are on the table by the door to the control room. We currently intend to install an old single-convolution stainless steel piece in place of the broken quartz. This is being done for ease and quickness of installation. Also, there were no diaganostics at that location that were in use. There is a m/n=2/1 saddle coil that we are planning on not re-installing. The quartz was also jumpered out and not acting as an insulating break either.

Using the convolution piece we have a good chance of installing it tomorrow and perhaps even doing an initial pump down and with baking we might be able to run at the end of next week perhaps. If people think this is not the way to go and they want one of the new thick, machined quartz pieces installed (Mike or Thomas?) please email Jerry and I and we will discuss matters since everyone is gone. Right now we are cleaning the spool piece and orings, making new mylar covers to hold the orings in place, and preparing to install the section tomorrow. We have checked the gap between the chambers and the convolution spool piece should fit with its orings and mylar cover, There is a gap of a little less than 0.1 in. flange to flange. We also checked to make sure the bolt circles match alright. In addition to the above, we pulled off the top microwave feedthru that has been broken for some time so that it can be fixed in anticipation of having the interferometer up and running in the future. I will either blank it off or if i can find it install a window and shutter than S. Paul loaned to us in the past. This could perhaps be used for the new Dalpha array temporarily since the microwaves will not be workijng for sometime. I have tried to inspect the SXR Be filter to see if it was damaged by the pressure wave when the quartz cracked by looking through the bottom window opposite the array. I frankly cannot see it well enough to say if it has minor damage or not. It does not appear that there is major cracks or flaking of the Be.

If there are other up-to-air tasks that people would like to see done now is the time!

Wednesday August 17 05 32 Hanson/Stillits 48559-48632 clean-up

Yesterday there were problems relating to the TF-CB circuit. We chose a new trigger channel and amplifier, and got the circuit to work. This mornining, we still had problems with the TF bank firing for low charging voltages ( < 6.1 kV ). It would seem that the one of the TF ingitrons is no longer good. After achieving solid TF pulses, a clinking noise was heard from inside the machine area. Two loose bolts were found resting on the concrete pad underneath the hbtep near where the quartz piece had been replaced, and were removed. Subsequently, clinking noise was no longer heard. We plan to check again for loose tools and hardware before running tomorrow. Intitial shots (48569-48578) showed little or no ionization. Gas puff time was gradually increased to 3.0 ms, and OHST was gradually increased to 6.5kV. Breakdown was achieved at shot 48580. For the rest of the day, pulse length gradually improved, and was just shy of 3 ms by the end of the day.

Thursday August 18 05 54 Hanson/Stillits 48633-48678 Clean Up

Started the day with a second missing tool search as promised in yesterday's run report. Jim Andrello found a small section of hex stock (that was used in losening the socket head cap screws holding the broken quartz piece) on the machine area floor beneath the concrete base pad. Jim still believes that an allen wrench is missing, but no clinking sounds were heard when firing the TF bank. Runs today showed spotty improvements in discharge time. Typical shots lasted 3-4ms, but we had a lovely 8ms shot towards the end of the day (48677). In order to get breakdown, we needed to keep the gas puff time around 3.0ms. In the afternoon, it was increased to 3.5 ms. We plan to continue making clean-up plasmas tomorrow with the hope of making 8.0+ ms discharges consistently. Ideally, we would test ICRH early next week.

Friday August 19 01 30 Hanson/Stillits 48680-48701 Clean Up

Before running this morning, we installed a new puff valve pusler box. We did at one point swap the old box back in again, but it did not seem to fire. Then we promptly reinstalled the new box. All shots taken today failed to break down. We tried to explore various settings: increased OHB to 03, increased OHEC and OHST to 280 and 6.5, respectively. VFEC was already at 40 from yesterday. We verified visually that the E-gun lights up, and that gas puffs consistently bring the system to order 10^-4 pressures. We ended up increasing the gas puff time to 20.0 ms. Also tried to adjust the tree setting TOP:CAMAC:TIMING:PUFF_VALVE2 a little bit, but I am not sure about the units, and whether it is the length of time or start time for the puff. We've been using nothing but D2 gas all week. Perhaps the missing quartz piece really does make a difference?? On several occassions yesterday, I noticed strings of shots that had progressively longer discharges. However after building to a 6-8 ms discharge, the following shot would fail to break down at all. I'd say this happened about 3 times yesterday.

Monday August 22 05 34 Hanson/Stillits 48704-48708 Clean Up

After poor plasma performance last Friday, the idea for today's shots was to asses the effect of fill pressure on our ability to achieve breakdown. To this end, we scanned gas puff times from 1.0 ms to 10.0 ms with little success. Around mid afternoon, Nic remembered the following: a) there was a small leak in the vacuum chamber which now seems to be fixed. Thus breakdown problems might be due to the chamber being too clean. b) the e-gun has a bias power supply that runs on a battery. We checked the battery on the e-gun bias supply and found that it had about 200 V out of 250. A fresh battery was then inserted. Subsequently, breakdown was achieved when the puff time was set at 20.0 ms. After achieving breakdown and some ~8 ms discharges, the puff time was slowly decreased to 16 ms with no negative result (see shots 48740-48748). Plan tomorrow to keep attempting to lower the puff time and make reproducibly good plasmas.

Tuesday August 23 05 35 Hanson/Stillits/Maurer 48749-48805 Clean-up

Smooth progress today. There were some questions about whether the gass lines were clear of He this morning, so they were pumped out. SXR emissions are slowly creeping up, and loop voltage is decreasing. Discharge times are in the neighborhood of 6ms.

Wednesday August 24 05 05 stillits/hanson/maurer 48806-48859 clean-up

The plasma is slowly getting cleaner. At the end ot the day we were able to create plasmas with puff time = 1.3-1.4 msec, loop-voltage starting at 12-14 and dropped to ~ 6. The duration of plasmas was on average 8 msec, initial plasma current 5 kA and max. current 7-8 kA (OH elect. = 260 and Oh start = 5.8). By the way, there were signs of increasing SXR's.

Thursday August 25 02 54 maurer none baking

After running the past few days with clean up discharges base pressure remains at 2 e -8 Torr. The dominant partial pressure is H20 on the RGA. The ratio of O2 to N2 as about 5 so if there is a leak it is small. I think the easiest way to get rid of the H20 is by baking rather than making plasmas, which is not very efficent at getting rid of the water. The bake is back on and I will see what it pumps down to on Fri morning. I think the fact that some bake controllers are not working appears to have limited our ability to get rid of the H20. The pressure with the bake on is 7 e-8 Torr.

Monday August 29 05 11 Maurer/Hanson 48860-48883 clean-up

Dave turned off and removed the bake system this morning, and we started running at about 1:45. Discharge times today seemed to fall pretty consistently in the 6.0-8.0 ms range. Soft x-ray signals are still small or non-existent. Loop voltage seemed to average about 11-13V. Most notably, the major radius seemed pretty bouncy on a 1-2ms time scale.

Wednesday August 31 04 34 A._Klein 48884-48929 Clean Up

Took many clean up shots. Plasma still prone to multiple minor disruptions, ended up running shots similar to past several days: large R and repeated "crashes" into outboard side, with hopes of cleaning machine that way. Did NOT adjust TF settings as reproducibility is still sorely lacking. Vertical Field only shot revealed poloidal sensor T1 position had its polarity reversed during the recent repairs.

Tuesday September 6 05 27 Hanson/Stillits 48931-48965 clean-up

Took more clean up shots today. Loop voltages averaged 10-15 V; Discharge times, 4-8 ms. Soft x-ray signals were very small. Major radius seemed quite chaotic. Planning to continue clean-up shots tomorrow, and maybe scan the TF a bit in the afternoon.

Wednesday September 7 05 31 Hanson/Maurer 48966-49022 cleanup/tf scan

<p>Spent the morning & early afternoon doing cleanup shots, and switched to coarse TF scan later in the afternoon.</p> <p> Regarding cleanup, discharge length seemed to average about 8 ms, loop voltage~13, no soft x-rays to speak of, major radius still quite chaotic. Nearly all of the shots shared some interesting characteristics: plasma current seemed to ramp up for about a 3rd of the flat-top, hit 7-8 kA, and then ramp back down to ~5 kA by the end of the discharge. During this ramp down period, loop voltage decreased to less than 10 and conductivity temp increased to maybe 10-20 eV. See shot 49001, for example.</p> <p> The results of the tf scan (49006-49022) seem encouraging. Scanned the tf voltage from 4.9 kV to 6.3 kV, in steps of 0.2 kV, with a 3 ms puff time. Was able to reproducibly achieve breakdown even at 4.9 kV, but discharge times here were very short since the vf was not adjusted to compensate. Shots taken at between 5.7-6.3 kV all looked quite similar.

Tuesday September 13 04 55 maurer

Glowed in D2 on Monday. Ratio of N2 to O2 was about 10 with the bake on this morning. Did a initial D2 glow in the morning and switched to He in the afternoon. Current base pressure is 5.0 e-7 Torr after turning off the glow. Mostly He and H20 products (DHO, D2O). Will take shots starting tomorrow morning.

Thursday September 15 04 57 Maurer

Only a few shots taken today. The OH coil started making a louder than usual noise when I was making plasmas. The sound resembles a dull thud. It does not appear to be an arc. It happened quite distinctly on one shot. We are in the process of trying to find out what is loose, moving around and making the noise. It is probably the bottom OH winding that has caused problems in the past. After inspection it does not look like the clamps have broken anywhere. So it is at least not a large scale clamp failure. It is possible that the bolts holding them together are a little loose from pulsing the machine. The clamps we have been able to check so far we finger tight. We will check the rest of the bolts before testing the circiut again. It is somewhat harder than before to access the bolt clamps due to the bottom-middle radial position control coil that is now installed.

Friday September 16 02 12 maurer

Friday September 16 02 21 Maurer

We were able to tighten the clamps on the bottom OH coil enough to reduce any noise back to similar levels as before yesterday. In the process of localizing where the coil was loose an arc flash was observed for a few OH only bank shots. This arc is intermitant in occurance. It was not observed with shots taken later in the day. The chamber sections where we believe the arc was localized were inspected as much as possible given the limited access to examine the clamps and cabling. There was no visible damage to either the clamps or cable that i could see or feel with my hand. Currently the VF winding, turnbuckles and other items that where taken off the machine to be able to inspect those clamps are being reinstalled.

Tuesday September 20 09 56 Alex_Klein

Dear All, I'm doing a dry run for a talk I am scheduled to give at MIT next week. The talk is an overview of the mode control experiments on HBT in the past 2 years + present status. You're invited. If interested, it will be in the control room on Thursday 9/22/05, 9:30 am. Thanks Alex K.

Wednesday September 21 05 13 stillits 49058-49095 clean-up shots

After having installed the Hall probe array this morning, we started to take clean-up shots. The plasma is slowly getting better with small signs of SXR's. The flashes we have been seeing lately were due to a trigger cable from an unused puff valve that was arcing where its connecter was touching the pump stand. It has since been insulated and the probelm has been fixed.

Friday September 23 04 39 stillits 49152-49190 clean-up shots

We spent the day taking clean-up shots. Small improvements of the plasma are seen but we probably have to spend at least one more day taking cleaup shots. We had problems charging the OH bias bank; fortunately it was only a water-resistor that needed to be refilled.

Monday September 26 04 29 stillits 49191-49247 clean-up shots

No major improvements of the plasma. We ran with OH-start ~ 6.9 and OH-elect ~ 290 but the plasma did not last more than ~ 4-5 msec, loop-voltage ~ 12 and only small values of SXR ~ 0.05.

Tuesday September 27 05 18 hanson 49248-49289 clean-up shots

Clean up shots today looked more or less like they did last week. Little or no SXRs, plasma current peaking at 10 kA, loop voltage ~10-15 V, discharge times 4-7 ms. Major radius still seems quite chaotic.

Wednesday September 28 03 14 A._Klein 49291-49359 Clean up/phase flip

Tried to reproduce Shilov-type shot with edge q evolving from below 3. Played around in bank settings but as unable to achieve the shot. Generally, plasmas appear to be fairly clean, some discharges had long duration (9 msec) with some soft x-rays and relatively smooth equilibrium evolution. Only, Shilov target shot is difficult to achieve....

Thursday September 29 03 41 stillits 49360-49417 clean up

No major progress today. The plasma is still fairly lousy. We ran with puff time ~ 1.7 msec, OH_elect ~280, OH_start ~ 6.8, VF_elect ~ 50-100, and VF_start ~ 2.9-3.8. The major radius is very volatile and loop-voltage ~12. Duration of plasma was ~3-4 msec, and plasma current stayed below 10 KA (~8-9 kA).

Friday September 30 04 39 stillits 49418-49454 Clean-up + Biased Probe

No major progress today. We are still faced with a lousy plasma. Max. duration ~ 4-5 msec, no SXR's, loop-voltage ~ 12 even with OH_elect. ~ 280, OH-start ~ 7.0. We tried to insert the biased probe to 3.5 cm with +200 V (biased current ~ 75-80 A) to see if we could see any changes in the plasma. But that was not a success either.

Monday October 3 04 43 stillits 49458-49522 clean up

After Dave had removed the straps/jumpers, we were able to get higher initial plasma current (~7 kA) at a lower OH_start: it is now OH_start = 5.2 compared with OH_start = 7.2 earlier. However, the plasma does not really seem to heat up: SXR's very small < 0.05, loop-voltage ~12-13, duration 4-5 msec, and max. plasma current ~ 10 kA. We will (probably) keep taking clean-up shots tomorrow.

Thursday October 6 12 13 A._Klein 49618-49625 test

Unable to break down plasma. Checked electron source, filament and bias. Confirmed gas fill. Confirmed bank currents. Unable to breakdown, will try again later after consulting w D. Maurer

Tuesday October 11 05 05 stillits/hanson 49626-49810 clean-up

Plasmas seemed very unwilling to breakdown. Nicolai increased the puff time throughout the afternoon to get results. When I took over, puff time was at 15.0 ms, corresponding to a fill pressure of ~2.3E-4 Torr. Shots with 10 ms puff times failed to breakdown. Dave noticed that even with the high fill pressures, plasma current during the bias phase of the shot is more or less zero. Bank traces, however, looked normal.

Monday October 17 03 23 Maurer none

Checked out e-gun circuitry. Bias voltage and heating current tested OK. Initially had a hard time measuring the emission current. After testing various parts of the circuit we were able to measure about 30mA of emission current (lots!). Not sure if we bumped a bad or intermittent conection or not that led to the earlier zero emsision current results. Decided not to pull the e-gun and inspect it since it appears OK. Opened up Hall probe gate value to main chamber and turned on bake. Will insert alumia coated probe shaft tomorrow and take shots. Plans in progress for up-to-air installation of quartz piece.

Tuesday October 18 04 16 Maurer

Took shots today to make sure breakdown is working well enough to make plasmas for testing the Hall probe shaft coating. Will test shaft tomorrow morning witharound 50 shots.

Friday October 28 03 55 Jim_Andrello Quartz installation

The quartz section is installed and sealed. The seals roughed down to 10^-3 Torr. Monday I will leak test it with Helium. The ICRF 10" flange has also been taken off for access to the Faraday shield.

Thursday November 3 05 58 Jeremy_Hanson ICRF hardware work

I completed removal of the Faraday screen from the ICRF antenna as planned on Monday. I then replaced the antenna limiters in their original positions. It turns out that it's possible to remove the Faraday shield/antenna limiter assembly without perturbing the antenna straps themselves. Once this assembly is removed, the Faraday screen straps are easily unbolted. I haven't had a chance to really look closely at the B-dot probe yet. It's covered in a little Faraday screen of its own that will be difficult to remove. My current understanding is that the probe circuit is open. I'll check that tomorrow and then see what else can be done.

Wednesday November 16 02 02 Hanson 49716-49732 ICRF testing

Yesterday, Nick and I removed all but one of the stages from the ICRF delay line in order to be able to test the ICRF with a short pulse. Today, I took vacuum ICRF shots with increasing voltages. I also added one more stage to the delay line and took more vacuum shots. The current traces in the antenna straps seem to track well with the top tank current (bottom tank ws not digitized). All traces exhibit an exponential RC decay, so I don't suspect any arcing. However, a snapping sound is heard when the antenna fires at higher voltages. The delay line was left in its truncated state because I intend to repeat this experiment with plasmas whenever they are available.

Wednesday November 16 02 52 Pedersen none FYI

Just a reminder of the HBT-EP physics meeting tomorrow morning at 9.30 am. I will be talking about our research on bias induced edge transport barriers.

Thursday November 17 01 45 Hanson 49733-49734 icrf testing

Today, Nick and I took a couple more ICRF shots to see if we could determine the source of the snapping sound heard yesterday. However, the sound could not be reproduced today.

Tuesday December 13 04 14 hanson/klein 49735 feedback test

Alex and I did a vacuum test of the feedback system. We put a sine wave on the control coils using the fpgas and then checked to make sure the sensors picked it up. They did. The feedback system seems to be in working order.

Thursday December 15 05 57 hanson/maurer 49738-49794 clean-up

First day of plasma shots since the re-installation of the quartz piece. Plasma currents were ~5 kA, pulse lengths 1-2 ms. Breakdown and the character of the shots seemed quite consistent; much better than the intial clean-up shots taken towards the end of last summer.

Friday December 16 01 03 hanson 49795-49812 clean-up

Friday December 16 01 14 hanson 49795-49812 clean-up

Plasma shots today looked more or less like yesterday. Low plasma current: ~6 kA, and short lifetime ~ 2ms. Discharges are quite consistent and there haven't been any big problems breaking down. The vacuum base pressure has been somewhat problematic; I've been waiting 10-15 min for the pressure to return to the 10^-8 range between shots. Fill pressures are ~3E-5 Torr. Around 11:30 this morning, the base pressure began to climb a little while after a shot. The temperature on the cryo head was about 10K, and a look at the RGA showed Dt and H20 to be dominant; ratio of N2 to 02 was about 10. After climbing slowly to 4.5E-7 Torr, the base pressure began to recede. It's now at 1.0E-7, a little more than an hour after the incident.

2006

Monday January 9 04 22 hanson 49813-49822 icrf test

Made the first tests of the ICRF system with plasma since the removal of the Faraday shield. The ICRF delay line was trimmed to two sections, as with the previous vacuum tests, making the ICRF pulse about .5 ms. Due to the long pump-down time between shots (30 min - 1 hr), I was not able to complete a scan of ICRF voltage settings. Preliminary results don't show any signs of arcing or misbehavior. The plasmas themselves look much the same as they did two weeks ago: plasma current peaking around 7 kA, ~2 ms pulse lengths. Turning on the ICRF doesn't seem to impact plasma parameters. I plan to resume the voltage scan on Wednesday, and eventually add more sections to the delay line.

Wednesday January 11 04 42 hanson 49823-49862 icrf test

Today, I resumed scanning the ICRF bank voltages and added two sections to the delay line. For the most part, the ICRF traces looked normal, and I did not see any big changes in plasma behavior. Evidence of an arc was observed in the ICRF traces for the highest bank setting (20) and 4 sections in the delay line, in the presence of a plasma (49861). No arcing was seen for a vacuum test of the ICRF with an identical configuration. I plan to explore the arcing issue further, with Nick's help.

Tuesday January 17 05 18 hanson 49863-49888 clean-up/icrf test

Took clean-up shots and continued testing the ICRF system. I kept the ICRF bank voltage lower (~20) and did not see any signs of arcing. On one shot (49883) the having the ICRF on seemed to correlate with slightly higher conductivity temperature and plasma current. Most shots showed little or no effect. It turned out that the cause of the bottom tank circuit not being digitized was that the digitizer was not connected to the coil. I'll get that fixed up tomorrow.

Wednesday January 18 04 55 hanson 49889-49917 icrf test/clean up

Today, I digitized light from the ICRF tests using a photo diode pointed through the quartz window towards the antenna straps. Firing the ICRF introduced noise into the photo diode signal, but didn't cause any DC jumps. There was another arc in the ICRF system (49907), when the charging voltage was set at 30 kV. After this, I went downstairs and noticed that the ICRF cooling water purity light had turned red -- indicating impure water. The light did not turn green afterwards, so the ICRF was not fired again.

Thursday January 19 05 11 hanson 49918-49946 clean-up

Took clean-up shots today, without firing the ICRF.

Friday January 20 05 00 hanson 49947-49976 clean-up

More clean up shots today. Plasmas seem to slowly be improving. Average loop voltage has been about 11 V; soft x-rays still negligible; pulse lengths are averaging 5 ms. Replacement ICRF cooling water pump has arrived; plan to replace it on Monday morning.

Monday January 23 05 06 stillits 49977-50044 clean-up shots

Clean-up shots were taken. Outboard limited plasmas were created R~94-98cm. Loop-voltage ~ 14-6, plasma current ~ 5-8 kA, duration ~ 8 msec, and only weak SXR signals. Puff time ~ 2.0-2.5 msec. Will continue to take clean-up shots tomorrow.

Tuesday January 24 04 57 stillits 50045-50131 clean-up shots

Again, clean-up shots were taken. Outboard limited plasmas were created R~94-99cm. Loop-voltage ~ 15-6, plasma current ~ 5-7 kA, duration ~ 7-8 msec, and only weak - but improving - SXR signals. Puff time ~ 2.1-2.5 msec; no break down of plasma when puff time < 2.1 msec.

Wednesday January 25 03 22 maurer/hanson Took about 20 shots and tried to diagnosis why there is no plasma current in the bias phase of the OH transformer. Typical fill pressures to have the plasma breakdown are 1e-4 Torr. Thats about 3 to maybe 5 times what we normally would require. After chasing down a few possibilites iI think its the egun again. Nick and I will check it out.

Friday January 27 05 27 hanson 50153-50222 clean-up/icrf test

Took clean-up shots and attempted further testing of the ICRF system. The good news is that we now seem to be seeing some plasma current in the bias phase of the shot (so the e-gun seems to be functioning after yesterday's repairs). However, large fill pressures (~1E-4 Torr) are still required for breakdown. The plasmas seem to be cleaning up slowly but surely: SXRs are becoming more noticable, loop voltage is decreasing, etc. The ICRF system is still suffering water purity problems in spite of the new pump. The indicator light seems to go red after an ICRF shot, and then turn green again later. Firing 3 ICRF shots in close succession resulted in an arc.

Monday January 30 03 44 hanson 50223-50250 clean-up

More clean-up shots taken today. Attempted to do current-ramp shots as discussed at this morning's meeting. I was able to ramp up to 9.5 kA plasma current, maximum.

Tuesday January 31 05 05 hanson 50251-50313 clean-up

More clean-up shots taken today. I'm now able to get plasma currents up to 10 kA using an agressive ramp. Plan to test run the feedback system tomorrow or Thurs.

Thursday February 2 05 45 hanson 50354-50445 clean-up/feedback tests

Made clean-up discharges and fired up the feedback system today, using an FGPA routine written by Alex. The feedback system seems to work inasmuch as nontrivial voltages appear across the control coils. It's hard to say if any modes were suppressed. I did a scan over the toroidal phase-angle input to the FPGA routine.

Friday February 3 05 02 hanson 50446-50502 clean-up/feedback tests

Took more clean-up and feedback phase-scan shots today.

Tuesday February 7 05 44 hanson 50503-50586 clean-up/feedback tests

Took clean-up shots and attempted to generate magnetic perturbations with the FPGAs in open loop. I had some difficulties getting the FPGAs to generate waveforms on the coils; plan to try again tomorrow.

Wednesday February 8 05 43 hanson 50593-50613 clean-up/phase-flip

Took more clean up shots today, and managed to put some 3:1 modes on the control coils. Also, I had a look at the loop voltage voltage divider. It appears to be in working condition, and the gain settings in the tree give accurate results to within 2-4%.

Thursday February 9 05 39 hanson 50614-50696 clean-up/phase-flip

Took more clean-up shots and did phase-flip experiments.

Wednesday February 15 05 59 hanson 50726-50777 clean-up

Took more clean-up shots. After clearing out the puff-valve lines an switching to pure deuterium, plasma bias current seems to have improved.

Monday February 20 05 12 stillits 50830-50855 New Biased Probe

We tested the new biased probe (BP) today. We took a few regular shots in which we had BP inserted 1cm, 2cm, and 4 cm into the the plasma and no voltage. Thereafter, we inserted BP 1 cm into the plasma and at +50 V we measured BP-current ~50 A, +100 V => BP-current ~100-120 A, and +150 V => BP-current ~ 150 A. Tomorrow we'll try to insert BP 2cm - 5cm and V_BP = 50 - 150 V (it will depend on the current drawn). Clean-up shots were taken too. We are able to create plasma with puff time ~1.2-1.3 msec.

Tuesday February 21 05 05 stillits 50856-50896 Biased Probe

We finished getting I-V characteristics for the new biased probe (BP). We had BP inserted 1cm to 4-4.25cm (max. insertion) into the plasma and slowly ramped the voltage up from +50V to +150V at each position. The BP current is ~1 A/V and fairly independent of the position. Furthermore, we also took clean-up shots. Small improvements are seen: duration of plasma ~ 6msec, puff time ~ 1.3 msec, max. plasma current ~ 10 kA, weak SXR < 0.1, q* ~ 1.5.

Thursday February 23 05 29 hanson 50897-50938 clean-up

Took clean-up shots and experimented with moving the aluminum shells. Putting in the Al shells does not seem to impact breakdown greatly, but does lead to more quiescent discharges. Also, the Ebara cryo-pump has been misbehaving again, oscillating in temperature and causing temporary increases in base pressure.

Friday February 24 05 00 hanson 50939-50981 phase-flip

Attempted to drive 3:1 modes rotating at 5 kHz using the control coils in an open-loop configuration on discharges with little ambient 3:1 activity. I didn't see much correlation on the SMPs. Plan to analyze runs and do a frequency scan next week.

Tuesday February 28 03 58 stillits 50982-50988 TPA + BP

Most of the day was spent on getting the CPCI to work. Afterwards, we took a few plasma shots with the old TPA, but the cpci is still having some problems storing data. On Thursday we will try to get a few plasma shots with the new BP and old TPA to see what floating potentials and electric fields we are generating with the new BP.

Wednesday March 1 05 09 hanson 50989-50139 clean-up/open loop fpga

Took clean-up shots and conducted open-loop fpga tests. I attempted to impose a rotating 3:1 mode on plasmas that had some ambient 3:1 activity. Rotation frequencies varied from 1-8 kHz. Plan to continue frequency scan on Friday. Also, the Ebara cryo pump continues to misbehave, warming up and discharging into the chamber every hour or two.

Friday March 3 05 03 hanson 51069-51110 open-loop fpga

Completed a frequency scan of rotating 3:1 magnetic perturbations using the fpgas. Plasma discharges continue to improve, but the Ebara cryo pump still has temperature cycling issues. Bake is on.

Tuesday March 7 04 41 hanson 51111-51156 open-loop fpga

Applied more rotating 3:1 perturbations with control coils, at various frequencies. Hopefully, these results will guide phase-flip experiments to be done tomorrow and later in the week. Also, the Ebara cyro pump continues to misbehave.

Wednesday March 8 01 43 hanson 51156-51176 discharge shaping

Attempted to create "Shilov style" discharges in which the edqe safety factor slowly crosses 3 from below. These shots seem to be more quiescent than shots with agressive current ramps and may be more suitable for open-loop phase-flip experiments.

Wednesday August 30 01 54 maurer/hanson 51528-51545 systems test

Test fired all banks and the puff valves. Monitored the TF coils with the new flux loop sets, and the traces looked fine. We achieved plasma breakdown on shots 51544 and 51545. Plasmas showed low plasma current (max 5 kA). Pumping has been a bit slow and the RGA still shows lots of water. Plan to bake, glow in Dt this afternoon, and maybe regen a cryo or two.

Wednesday October 11 04 27 hanson/maurer/james 51555-51590 clean-up/neutrals

This was the first real HBT-EP run day since spring. Plasma currents peaked consistently around 10kA, loop voltages were around 10-13 V, pulse lengths grew as the day went on, reaching about 3 ms. Plasma behavior seemed quite consistent from shot to shot with most discharges experiencing a noticable disruption about halfway through the shot. Used the caemra housing, small filter, and single channels on the PDA to test the amp. Signals without the filter were about 1V on average peaking at about 2V and negative (as expected). the H-alpha filter signals were somewhat more reduced, so it was difficut to seperate them from the noise. This could be a result of the plasmas being cold so we will continue tomorrow. If results are unchaged tomorrow, I will revisit the amp.

Thursday October 12 08 59 James_Hansen 51590 - 51627 Cleanup & Neutrals

Cleanup shots continued today, we reached longer lifetimes (about 5.5 -5.75 milliseconds) than yesterday, but we were around 7 kA Ip for most of the day close to yesterdays. We did notice that we had to raise the fill pressure throughout the day after long breaks between shots as expected after a long pump down. We were able to continue testing the PDA to day as well. Filtered signals with a single channel from the array were still too small to distinguish from the noise. All these shots can be viewed in the tree with the PDA.jscp in the HBTEP directory; shot ^#51579 is a good example to see this. We then replaced the array with the single diode and same filter. The single diode with an area 4 times that of the arrays single channel showed a clear and smooth twin peaked signal as predicted. Next we put two and finally three adjacent elements from the array in parallel to increase the surface area and finally got a noisy but clearly discernable signal, shot ^#51627. We will be able to test more channels in this fashion once the larger filter comes in on Monday.

Tuesday November 14 05 42 hanson 51636-51668 clean-up

Made clean-up plasmas. Plasma parameters are slowly improving, loop voltage averages around 14 V, plasma current around 10 kA, pulse-length 3-4ms. No soft x-rays yet. Some initial tests were done of the Hall probe hardware and D-alpha amplifier.

Friday November 17 02 56 hanson 51669-51678 clean-up

Friday November 17 02 58 hanson 51669-51678 clean-up

More clean-up shots today. Plasma parameters continue to improve. Most notably, loop voltage now stays below 10V for a msec or two on most shots. Pulse lengths are in the neighborhood of 3-4ms.

Monday November 20 04 23 hanson 51690-51716 clean-up

More clean-up shots today. Loop voltage averages around 10V, and SXRs are still quite small.

Tuesday November 21 06 04 hanson 51717-51757 clean-up

More clean-up shots today. Plasmas are behaving fairly consistently, with loop-voltages a little below 10 V (on the good shots), but no soft x-rays yet.

Wednesday November 22 06 20 _hanson 51758-51796 clean-up

More clean-up shots today. Loop voltage seems to be coming down. Plan to bake over the long weekend and resume clean-up shots on Monday.

Tuesday November 28 06 04 hanson 51797-51828 clean-up

More clean-up shots today. Attempted to make hot plasmas with lots of ohmic heating input. This was somewhat successful: core soft x-rays in the neighborhood of .15 V were observed, and some shots even had some small saw-teeth.

Wednesday November 29 05 14 James/Hansen 51833 - 51873 Clean-Up/D-alpha

More of the same today, clean-up shots mostly, with a mixed with Hall probe and D-alpha testing. The Hall probe seems to be functioning well, calibrations to follow. The D-alpha detector single channel is operating nicely with little gain (about a factor of 2 M-ohms), so the Thompson Scattering port is a good place for collecting light from the plasma. We are still working out bugs in the array amplifier.

Thursday November 30 04 21 hanson 51894-51897 clean-up

More clean-up shots today. We continue to see small improvements in the soft x-rays.

Thursday November 30 04 25 hanson 51894-51917 clean-up

More clean-up shots today. We continue to see small improvements in the soft x-rays.

Tuesday December 5 06 44 hanson 51919-51983 clean-up

More clean-up shots today. Plasmas look a lot like they did last week. Also, the sensor coil digitizers were tested using a function generator, and found to all work. Additionally, OH only and VF only shots (51955,51956) were taken to try to assess whether these effects are corrected for properly in the major radius calculation routine.

Wednesday December 6 05 14 James,_Hansen 51983 - 52030 Clean-Up

More clean-up shots today. The cooling oil fixes seem to be holding up well.

Thursday December 7 04 32 hanson 52031-52070 clean-up

More clean-up shots today. Plasmas continue to improve, but do not behave consistently from shot to shot.

Monday December 11 05 03 hanson 52071-52109 clean-up

More clean-up shots today. The battery in the puff-valve circuit was slightly low, so we replaced it this morning. I did repeated static fills throughout the day to determine if the puff-valve was working consistently -- it is. The minimum fill pressure required for plasma breakdown was about 1.0E-4 T. Plasmas behaved similarly to how they did last week, but did seem a bit more consistent shot-to-shot.

Tuesday December 12 04 04 hanson/james 52115-52123 clean-up

A few clean-up shots were taken this afternoon; during the morning, additional hardware related to the D-alpha experiment was installed on HBT-EP. Also, did some troubleshooting on one of the m=2 rogoski coils (which hadn't worked in recent memory). The coil was installed and in good shape; the problem turned out to be a misconfigured AM502 amplifier.

Wednesday December 13 05 09 hanson 52123-52160 clean-up

More clean-up shots today. Some testing of the D-alpha hardware (with a tangential view of the plasma) was done as well.

Thursday December 14 05 52 hanson 52161-52205 clean-up

More clean-up shots today. There have been problems getting the CPCI to initialize after a restart of langmuir.ap a couple days ago. I've requested help from Yuhong and am waiting to hear back.

2007

Friday January 26 04 26 hanson/maurer 52218-52250 testing

Tested all of the banks, looked for tools dropped near the machine, and sorted out some small kinks that had worked their way into the system since the last run (back in Dec). The TF coils were tested with care, and traces from the diamagnetic loops mounted on the coils looked normal. Also, one plasma shot was taken. Plan to take clean-up shots all next week, as everything now seems to be in good working order.

Tuesday January 30 05 17 hanson 52251-52284 clean-up

Took clean-up shots today. Plasmas are behaving fairly consistently shot to shot, with loop voltages around 8-10V. The sensor coils were tested and found to work, with the exception of two whose analog boards have not been fixed yet. We looked into getting the optical rotation diagnostic connected to a digitizer, but some of the configuration settings for the 6810 were not transparent. Steve Paul has been contacted for advice. Plan to take clean up shots for the rest of the week and work on getting data from the hall and rotation diagnostics.

Thursday February 1 04 33 hanson/james/maurer clean-up

Took more clean up shots. By the end of the day we had nice sawteeth and loop voltages down to about 6 Volts by the end of the shot. The longest lived plasmas lasted about 8 msec. Had problems with the vacuum controller interlocks again (similar problems occured with the main ion gauge last week and caused the vacuum system controller to trip over the MLK day weekend at the start of the semester). It appears that many of the connections on the backside of the gauge controller are starting to fail due to age and fatigue. There might be some problems with the gauge controller itself as well.

Friday February 2 12 07 hanson 52382-52399 clean-up

More clean-up shots today. Saw some nice plasmas with lots of soft x-ray/mhd activity and loop voltages descending to around 6V. We're ending early today to fix oil leaks, bake relays, the ion-gauge connection, and the remaining analog feedback boards.

Wednesday February 7 05 19 hanson 52400-52425 clean-up

Made more clean-up plasmas today. Plasmas continued to look good, but the RGA showed high water levels after the run. The crunch amplifiers and control coils were tested and seem to be working. The new Kalman filtering routine was run and it seems to produce coil signals similar to those made by Alex's routines. Plan to continue with the clean-up shots with some focus on creating discharges with (m,n)=(3,1) activity.

Thursday February 8 05 06 hanson 52425-52490 clean-up

More clean-up shots today. Plasmas continue to look good. Pulses typically last longer than 6 ms and have loop voltages that descend from 10-5 V over the shot.

Monday February 12 04 27 hanson 52491-52528 clean-up

More clean-up shots today. Plasmas continue to look good with loop voltages that approach 5V towards the end of the shot and plenty of MHD activity.

Tuesday February 13 02 58 hanson 52528-52556 clean-up

More clean-up plasmas today. Plasmas continue to look good. As there seems to now be plenty of mhd activity, the plan for tomorrow is to do some initial testing of the Kalman filter feedback algorithm and a compass scan.

Thursday February 15 04 50 hanson 52557-52622 feedback system testing

Did some testing of filter gain and feedback phase settings using both new (Kalman filter with lead-lag compensator) and old (lead-lag compensator only) FPGA routines. Also, more Hall probe calibration data was taken. Plan to do some further FPGA tests tomorrow and think about the results over the weekend. I should be in shape to do some more formal phase scan experiments next week.

Wednesday February 21 05 59 hanson 52654-52672 feedback testing

Did some more tests of the feedback system with a slightly re-designed Kalman filter. (An oversight in the program was discovered earlier this week.) Some D-alpha hardware tests were done and some more Hall probe calibration data was taken.

Friday February 23 02 43 hanson 52691-52710 Rotation Diagnostic/Feedback

We recorded data for the first time using the rotation diagnostic. 90% He, 10% D2 plasmas were made and signals were seen on both channels. Unfortunately, the gas received for these experiments was mixed incorrectly. It should have been 10% He, 90% D2. The gas supplier has been contacted. Throughout the week, I've been trying to pin down the source of some large DC offsets seen in the control coil traces when the feedback system is on. Tests done today seem indicate that at least some of these are due to hardware problems, not bugs in the FPGA routines (as originally suspected).

Tuesday February 27 04 15 hanson/james 52711-52729 feedback/d-alpha

Did more tests of the feedback system today. Some potential problems in the FPGA algorithm were eliminated. I hope to have the rest of the bugs worked out by the end of the week. D-alpha array is working (minus the slit). It turns out that the negative lead to the PDA was borken and has been repaired. I will test with the slit tomorrrow.

Wednesday February 28 04 04 James/Hansen 52730 - 52740 D-alpha Tests

I did initial tests on the Photodiode array today. The tests show that we did remove most of the noise from the system and that there is plenty of white light available. We were unable to get specific levels of D-alpha for a single element of the arrray due to problems with the VF-Start control panel. It is unclear if it is just a problem in the Control Panel or in the basement bank system. Nick is investigating. We plan to complete ths D-alpha tests tomorrow whern VF problem is corected. No feedback work was done due to problems with the analog boards. They don't seem to be getting the correct voltages from the power supply anymore. Work to fix this issue is underway.

Friday March 2 04 46 hanson 52772-52796 feedback testing

After bench-testing all of the analog boards yesterday (all were in good, working order), the feedback system was reassembled for online testing. (It was suspected at one point that one or more boards might have shorts or bad op-amps.) Replacing the +/- 12 V DC power supply for the analog boards with two large 12 V batteries fixed some of the problems we were seeing during the previous two weeks. These were: DC offsets appearing inconsistently in the control coil signals and incorrect power supply voltages appearing across the boards. In any case, eliminating the power supply seemed to do the trick. A replacement has been ordered.

Wednesday March 7 02 40 James 52827 - 52840 D-alpha

Installed a moving mirror setup for the D-alpha array. I used a laser to confirm the angle range to sweep the solid angle from major radius about R=107 inward. The Levels of white light increased as the view was moved in to the center of the plasma, but they were substantially lower than the levels seen with out the mirror, compare shot number 52831 with mirror and 52724 no mirror. When the filter was added the levels drastically dropped to almost zero without the slit even being added yet, compare 52838 with mirror and 52729 no mirror and both no slit. I will print these out for Monday morning meeting. To see the data mentioned look in ICRH_L8210_1:Input_2 in the tree and the signal is negative so it must be multiplied by -1.

Thursday March 8 04 39 hanson 52854-52891 feedback testing

Today, a working Kalman filter was tested on plasmas for the first time. However, the filter did not work as expected on all of the FPGAs. One FPGA/poloidal group seemd to work flawlessly, but the others performed strangely. I'm still pondering this. (Other programs have either worked correctly or not on all of the FPGAs.) Also, plasmas looked quite good; I was able to create discharges with lots of m=3 activity.

Friday March 9 02 33 hanson 52892-52905 feedback testing

I got the Kalman filter working on all four FPGAs today. (There had been a bug in the software preventing matrix data from being loaded correctly onto some of the FPGAs.) I did a few shots with the filter turned on at a constant feedback phase angle and a few reference shots. The plasmas showed lots of m=3 activity, but it was difficult to tell from the raw signals whether I was driving or suppressing anything. I plan to compare the filter output with my model over the weekend and undertake more elaborate experiments next week.

Tuesday March 13 05 46 hanson 52906-52950 feedback

More Kalman filter tests today. I played around with some different proportional gain and phase angle settings. By looking at the time delay in gating on the feedback, I got an estimate for the FPGAs' loop rate. Updating the matrices with this parameter produced control coil currents that were in good agreement with my model for what should happen. Setting the phase angle at 180 degrees (for positive feedback on the kink mode) causes the Kalman filter to blow up. I can even make the FPGAs ring on line noise (with no plasma) at this setting. I Plan to attempt a fine phase scan tomorrow, hardware permitting.

Wednesday March 14 05 50 hanson 52951-53045 feedback phase scan

Did the greater part of a feedback phase angle scan today. I took two feedback on and two control (feedback off) shots for each 10 degree step. Plan to finish up the scan tomorrow and start analyzing the retsults.

Thursday March 15 05 45 hanson 53047-53092 feedback phase scan

The Kalman filter feedback phase angle scan continued today and is nearly finished. The run day was cut short by a failure in the charging power supply for the OH electrolytic bank. The failure point was a splice connector at the 3-phase power input to the supply. The connector had become loose and was arcing and melting. A substitute connector was made in the shop this afternoon, so we should be ready to run again in the morning.

Friday March 16 12 47 hanson 53093-53127 feedback phase scan

Completed the Kalman filter phase angle scan today. The OHEC charging supply was reassembled this morning and appears to be working fine.

Thursday March 22 05 05 hanson/debono 53150-53163 rotation diagnostic tests

We tested the Doppler rotation diagnotic using 10% He 90% D2 plasmas. Signals coming from the optics looked like they were in qualitative agreement with earlier results.

Tuesday March 27 05 13 hanson 53176-53132 feedback

More feedback tests today. After a numerical error was found in the alorithm last week, the first plasma tests of the corrected filter were done today. The control coil currents seem to track quite well with expectations, but it is difficult to tell yet what the impact of the feedback is on kink modes. Further testing and analysis to come.

Wednesday March 28 04 44 hanson 53233-53298 feedback

More feedback testing today. Some time was spent developing a reference discharge with a strong current ramp and lots of m=3 activity. I have not seen any very obvious evidence of suppression yet, but a frequency analysis of the shots has yet to be done. Also, one poloidal group seems to be performing poorly; I hope to be able to address this issue tomorrow.

Friday March 30 04 42 hanson 53299-53310 feedback

More feedback testing. Frequency analysis from shots taken earlier this week showed that the filter algorithm drives the control coils at frequenies that are somewhat smaller than the mode. The filter matrices were recalculated for slightly higher frequencies and tested again today. Also some more data was taken using the rotation diagnostic.

Tuesday April 3 06 11 James 53317 - 53325 D-alpha

Success! I was able to repeat last Friday's tests in Deuterium (vs the 10 percent He mix). The white light and D-alpha output concurrently increased. I then masked the single diode to 1mm and added the filter and lens and was able to sweep the mirror to find the highest D-alpha levels. Outputs were around the 2.5V level through the shot, after the start-up spike. I also documented the rise and fall of the signal around this peak mirror angle. After these initial tests, I replaced the masked single diode with the array and after some quick repair work on the input connector, D-alpha signals between .7V and 1.3V were achieved! Tomorrow I will vary the PDA elements and mirror angle to try and get a framework to begin mapping the chords.

Wednesday April 4 05 06 hanson 53327-53376 feedback testing

I made some changes to the way that the Lead-Lag filter is implemented in the feedback algorithm and tested them piecewise this morning. Initial analysis shows (both in theory and practice) that the lag filter does a good job of flattening the transfer function going into the FPGAs, so that the signal seen by the Kalman filter stage is proportional to B (and not, say, dB/dt). In the afternoon, the algorithm was tested as a whole. Anecdotal evidence of some m=3 suppression and excitation was seen for different feedback phase angles.

Thursday April 5 04 38 hanson 53377-53430 feedback

I did a coarse feedback phase angle scan using the algorithm as described yesterday. Analysis is pending.

Friday April 6 01 13 hanson/debono 53431-53442 rotation diagnostic testing

Took some data using Steve Paul's rotation diagnostic that will hopefully lead to a better calibration.

Tuesday April 24 04 59 hanson 53468-53501 feedback

Two sensor coils that exhibitted signal saturation problems were disconnected from the feedback loop. The analog boards had been eliminated as possible sources of the saturation; there is some suspicion that the shielding on the coils themselves may be damaged. (Recall that the poloidal sensors are mounted on the plasma facing side of the stainless steel shells.) Eliminiting these inputs left the xT_S1 group with only two inputs (a third sensor coil in this set is open-loop). Control coil signals for the xT_S1 group were subsequently much more well behaved, and looked similar to those on other groups. However, I am not yet sure if the spatial DFT coefficients for this group should be modified to compensate for the unevenly sampled data. Plasmas were repeatable, with one or two shots showing signs of possible feedback suppresssion of m=3 modes. I plan to think about the DFT coefficients tonight and continue with feedback experiements tomorrow.

Thursday April 26 05 09 hanson 53524-53566 feedback testing

I attempted to make plasmas in the style of M. Shilov for feedback studies. These plasmas break down between edge-q=2 and 3 on the outboard edge and move inwards, pulling in the q=3 surface as they expand. This type of shot seems to be m=3 unstable without perturbing the two "bad" sensor coils as much. I was able to keep these coils plugged in and have the Kalman filter work reasonably well on the xT_S1 set. I found it more difficult to reproduce the Shilov plasmas exactly, shot to shot. However, I believe that I've seen some anecdotal evidence of m=3 suppression. I plan to continue in this vein tomorrow.

Friday April 27 05 01 hanson 53569-53640 feedback

More feedback shots today, using the "Shilov shot" described in the run report yesterday. I adjusted the banks a little and was able to achieve better repeatablility. I covered a few feedback phase angles; the rest will be done next week.

Tuesday May 1 05 30 hanson 53672-53750 discharge shaping

I've been trying to cultivate a shot with large, low frequency (~5 kHz) m=3 activity. As a model, I'm using shot 43102, from Alex Klein's work. Some important characteristics of this shot are subtle and difficult to reproduce, but I'm making progress. I hope to get back to doing feedback studies tomorrow.

Thursday May 3 04 59 hanson 53823-53875 feedback

I've been able to develop a plasma with a more slowly oscillating m=3 mode--in the neighborhood of 5-7 kHz. Initial experiments with the Kalman filter appear to indicate that I can both suppress and drive it.

Friday May 11 04 41 hanson 54163-54213 feedback

I spent the week developing a discharge with a robust, current driven m=3 instability. I am not able to exactly reproduce the discharge used by Alex Klein, but I have found one that behaves similarly. I seem to be able to both supress and excite the mode using the feedback system.

Tuesday June 5 05 04 James 54326 - 54385 D-alpha

Finally tested the D-alpha prototype amplifier with three signals. The designed board worked perfectly. One of the output connector channels between the CPCI connector and the amp is a little flaky but the amp itself works great. I was able to take a white light shot and get signals that were clipped into the 5 volt region. D-alpha light ranged between .2 V for an inboard view to 1 V in the edge region. These views are not exact but in the general vicinity. The noise was in the .0012 V region and frequency response seems good. Nick and I are working up drawings to submit for a quote on a multi-channel amplifier board made on the economy. Tomorrow morning I will take a few shots with the power supply to check noise levels and hopefully a repaired connector to get rid of the one flaky channel between the CPCI and the amp.

Tuesday June 19 03 01 James 54412 - 54417 D-alpha

I was able to conclusively show that the amplifier boards work. I will verify their freq. response on later this week. I was also able to take a picture of the plasma that I will send out or post after I put it in a better format. I will also attempt to create better pictures of the plasma and hopefully use the d-alpha filter in a few shots!

Monday June 25 04 32 James 54437 - 54440 D-alpha

Theses were short lived plasmas, ~3.5 ms. The pictures were taken through the ICRH port to get a fairly unobstructed view. Able to see D-alpha Light - the center of the plasma does not seem to be emitting much, but the edge of the plasma and not the chamber wall seems to be the source as would be expected when not looking at a limiter. I have e-mailed out a pdf with pictures in it.

Friday July 6 05 02 Hanson/DeBono 54517-54532 Rotation Diagnostic

Tested rotation diagnostic using 10%He, 90%De mix.

Tuesday July 10 06 06 James 54563-54587 D-alpha

Did some calculations with Dave Maurer and determined that the transfer fxn roll-off was mostly due to the 15pF cap in the amplifier circuit. I replaced the caps in each of the channels with a 1pF, 3.6pF, and 6.6pF capacitors and did some signal to noise tests on the bench. I then put the PDA on the machine and tested the cabling and CPCI digitizer. I spent the rest of the afternoon attempting to align the lens & PDA with the mirror - with no success. I will continue tomorrow with this process and then determine noise levels and frequency response for each channel.

Thursday July 12 09 57 James 54588 - 54624 D-alpha

I was able to calibrate the detector and get D-alpha signals into the 1V range through the shoot lifetime. There are still noise problems from the PDA connector, but they will be rectified in constructing the final unit. There was noticeable smoothing in the 6.6pF and minor smoothing on the 3.6pF channels. The 1pF had some extra wiggles but it was on a channel with an obscured view, so it was hard to compare side by side. Yet the freq response test confirm a 31KHz - 35KHz corner freq for the 1pF channel with 30.6KHz calculated. I didn't test the others yet, but the 3.6pF has a calculated corner freq. of about 10.5KHz which seems a bit low. For now we are going with the 1pF and I will do the freq. response for the other two channels to make a decision on Monday for the final design to be processed.

Thursday July 12 03 37 hanson 54633-54647 kalman filter testing

Tested a newly written version of the Kalman filter algorithm today, with actual plasmas. The new algorithm has been optimized for a significantly improved latency: 10-15 microsec, down from 30-40 microsec. Qualitatively, the algorithm performed more or less as expected. Control coil currents seemed to be a bit attenuated on some toroidal sections--this may be due to a numerical error (possibly an underflow) in the algorithm. I plan to check the new algorithm carefully and to work on making plasmas with robust m=3 modes.

Friday July 13 04 23 hanson 54648-54697 discharge development

Spent the day trying to fine tune m=3 kink-mode discharges. I plan to pick up where I left off on Monday and spend the rest of next week doing Kalman filter experiments.

Monday July 16 03 19 hanson 54698-54729 kalman filter

More discharge development and Kalman filter testing today. I've been able to create a fairly repeatable kink mode. There's a small bug in the Kalman filter algorithm that seems to track with the adjustable phase angle. I think I know what's wrong and will fix it tonight and recompile. I hope to be ready to start compass scan experiments tomorrow.

Tuesday July 17 05 41 hanson 54730-54790 kalman filter

This morning, I found and fixed the bug in the Kalman filter (there was some incorrect memory addressing in the inverse DFT step) and was able to start experiments. I completed about a quarter of a feedback phase angle scan. This afternoon, there was some trouble with the TF bank charging electrode: a piece had broken off. With Nick's help, I filled the resistor high enough to submerge what was left of the electrode. Eventually, it will need some attention.

Wednesday July 18 05 39 hanson 54791-54850 kalman filter

The feedback phase angle scan continued today. It's nearly halfway finished.

Thursday July 19 07 35 hanson 54851-54930 kalman filter

More of the Kalman filter feedback phase angle scan today. I am nearly done; plan to finish up tomorrow morning.

Friday July 20 03 51 hanson 54931-54965 kalman filter

The Kalman filter feedback phase scan was completed today. I plan to do initial analysis of the results over the weekend, so we can talk about what to do next on Monday. Also, I managed to overheat the TF charging resistor again: the electrode is now almost gone and needs attention. I'll repair it with guidance from Nick and Dave on Monday.

Tuesday July 24 05 01 hanson 54966-55005 discharge development

This morning, I inserted the aluminum shells fully and re-created the shot used for feedback experiments last week. The Al shells were able to suppress the mode about as well as optimally phased feedback (last week), but this suppression was not spectacular. Perhaps the mode I used behaved too ideally. I am now working on developing a more resistive mode.

Thursday July 26 03 39 hanson 55006-55106 discharge development

I've spent the last couple days fine tuning my kink mode discharge. I hope to get back to taking feedback data tomorrow.

Friday July 27 04 29 hanson 55107-55151 kalman filter

Took some feedback phase scan data today. Much more of the same planned for next week.

Monday July 30 05 05 hanson 55152-55195 discharge shaping

Did some work trying to make a better, stronger external kink mode. Plan to continue in this vein tomorrow.

Tuesday July 31 05 12 hanson 55196-55258 discharge development

Work on making a better external kink discharge continues. I have a candidate that I'll test with feedback tomorrow.

Wednesday August 1 06 10 hanson 55260-55304 discharge development

More discharge development today.

Thursday August 2 05 22 hanson 55308-55379 kalman filter

I started a feedback phase scan with a new, stronger kink mode today.

Friday August 3 06 05 hanson 55381-55447 kalman filter

More feedback phase scan shots today. Initial results show some promise, but the going is slow.

Monday August 6 04 28 hanson 55448-55506 kalman filter

More feedback phase scan data taken today.

Tuesday August 7 04 54 hanson 55507-55579 kalman filter

More feedback phase scan shots today. I'm now done with over half of the compass points.

Wednesday August 8 06 09 hanson 55580-55659 kalman filter

More feedback phase scan data taken today.

Thursday August 9 05 49 hanson/shiraki 55660-55738 kalman filter/hall probe

Finished up the feedback phase scan today. Initial analysis of the data points to some suppression over a decently broad range of phase angles. I'll have some plots for discussion on Monday. aisuke got some hall probe data as well.

Friday August 10 05 00 shiraki/hanson 55739-55781 Hall probe

Took some hall probe data, attempting to recreate Yuhong's shots.

Monday August 13 05 09 hanson/shiraki 55782-55834

In the morning, we pulled the Al shells all the way out to try to replicate a shot from Yuhong Liu's thesis. In the afternoon, the shells were fully inserted to see what effect they had on the mode used for feedback experiments last week. Some suppression was seen, but I'd like to take a few more of these shots.

Tuesday August 14 04 37 hanson 55835-55861 discharge development

First, the mode from last week's phase scan was re-created with the aluminum shells pushed all the way in. Little suppression was seen. Next, the Al shells were pulled back 4 cm from the plasma surface and an attempt at making a shot similar to the one used in Liu's thesis (45645) was made. The behavior in this shot seems to be a little easier to reproduce than seen in Klein's discharges. I plan to fine tune my reproduction tomorrow.

Wednesday August 15 05 47 hanson 55862-55915 discharge development

Attempts to make a better kink mode continue.

Friday August 17 04 50 hanson 55916-55960 discharge development

The search for a better kink/RWM continues. Today, I spent a some time attempting to make shots similar to those used by M Shilov (e.g. 38318). One thing that's nice about this shot is that the mode appears much later in the discharge compared to the shots that Klein used--usually around 5 ms. By this time, the plasma has had a chance to heat up more and there is sometimes a good, solid SXR signal. In any case, I feel like I've been having difficulty achieving good soft x-rays, especially early on in a discharge. Loop voltages are usually below 10 V for "usual" plasma currents of 10-20 kA, but I've been finding that I usually need to grow the plasma in major radius to see much of anything in the SXRs. I've played with the fill pressure a little bit, but perhaps some more systematic studies are warranted.

Monday August 20 05 08 hanson 55961-55997 discharge development

Friday and today, there were some problems getting the banks to fire. The issue has been intermittent, so I haven't been able to pinpoint the cause yet. Also, the screenroom digitizer did not want to power on this morning. It's bypassed for now. I did have time to make a few plasmas. The idea now is to see if I can get them to heat up better by breaking down a little further out (R~=95cm) and then pulling the plasma in. This does seem to be helpful. I will not run tomorrow in order to allow for Thompson scattering and D-alpha hardware work to be done in the machine area. I'll pick back up again on Wednesday.

Wednesday August 22 05 57 hanson 56003-56069 discharge development

Today I made some progress in creating external kink modes in scenarios where edge-q is ramped down through q=3 and where it is ramped up. Mike suggested that I adjust the start time of the VFST bank to coincide with the start time of the OHST bank. This seemed to create a more smoothly evolving discharge in the ramp-up case. I plan to continue tuning both kinds of shots and reproduce them with the Al shells in. Perhaps feedback experiments can begin next week.

Thursday August 23 05 09 hanson 56070-56120 discharge development

I have completed fine-tuning my q ramp up and ramp down discharges. I am now trying to reproduce them with the Al shells fully inserted to see how the modes behave. I hope to have this done by the end of tomorrow.

Friday August 24 04 21 hanson 56122-56165 discharge development

I attempted to re-create my ramp-down and ramp-up modes with the Al shells all the way in, today. The ramp-down evolution was a little easier to get right and showed more mode suppression.

Tuesday August 28 05 02 DeBono/Hanson 56166-56169 rotation

Ran 10%He & 90% Deuterium for rotation diagnostic testing. Had problems with new screenroom digitizer rack initializing.

Wednesday August 29 06 17 B.DeBono/_J.Hanson 56170-56198 Velocitometer

Ran shots 56170-56198 for rotation diagnostic calibration.

Thursday August 30 06 03 DeBono/Hansen 56199-56223 Rotation Diagnostic

Ran shots 56199-56223 for rotation diagnostic calibration

Wednesday September 5 05 12 hanson 56224-56269 discharge development

I studied the effect of the start time of the vertical field on the dynamics of my external kink modes. I made some interesting plasmas, but did not see any modes stronger than what I've made already.

Friday September 14 03 41 DeBono/Maurer 56399-56431 Velocitometer

Ran Cleanup/Rotation calibration shots today with 90%Du/10%He mix.

Wednesday September 19 06 22 DeBono/Hansen/Maurer 56432-56500 Rotation Diagnostic

HBTEP was run with 10%He, 90%Du for rotation diagnostic calibration: Tests were conducted with velocitometer detector mounted to view the plasma perpendicular to its flow (on the ICRH port).

Thursday September 20 06 12 hanson/shiraki 56501-56582 clean-up

In the morning, I tried to re-create the modes discussed at the physics meeting last week. I was able to get very close, but I feel like there might be a little cleaning up that needs to be done yet. However, the plasmas look pretty good. In the afternoon, Daisuke got some hall probe data for edge-q ramp-up style shots.

Monday September 24 04 28 hanson 56583-56606 kalman filter

I was able to get a few shots in this afternoon. Previously, the FPGA driving the top-middle control coil set was using inputs from the bottom-middle sensor coils. After discovering that the phase-shift between top-middle and bottom-middle in the sensor coil data is almost exactly 180 deg (for the m=3 mode), I added a correction to the FPGA algorithm. Data taken today confirmed that the correction works as it should. I plan now to redo my Kalman fitler phase scan, since the correction was quite drastic.

Tuesday September 25 04 57 hanson 56607-56677 kalman filter

I began my Kalman filter phase scan today. Plasmas seem to be a little less reproducible; perhaps we are still cleaning up a little. I had some trouble this morning with a control coil not energizing. After swapping analog boards around and testing one on the bench, I concluded that there was simply a loose connection somewhere. Getting the analog board plugged in "just right" by trial and error seemed to fix the problem.

Wednesday September 26 02 58 hanson 56678-56724 kalman filter

More Kalman filter phase scan data taken today. Plasmas looked just fine.

Thursday September 27 05 50 hanson 56746-56813 kalman filter

More Kalman filter phase scan data taken today. Shot to shot reproducibility seems to be somewhat poorer than in past months; this experiment will likely continue through most of next week.

Friday September 28 02 26 hanson 56814-56834 kalman filter

More Kalman filter phase scan data was taken today. We ended the run day early to get the bake turned on for the weekend. Brian and I had hoped to get some optical rotation data this afternoon, but a shortage of liquid nitrogen prevented us from being able to pump out the puff line to switch gases.

Tuesday October 2 05 37 Hanson/DeBono ?-56927 Kalman Filter/Rotation Diagnostic

Shots were taken by Hanson and by DeBono Shots with 90%Du, 10% He were taken for shots 56905-56927 with the detector looking perpendicular to the direction of plasma flow. The purpose of this run was to assume the rotation velocity of the plasma was 0 (as viewed from the detector) with the data used for detector re-calibration.

Wednesday October 3 05 56 hanson 56928-56990 kalman filter

More Kalman filter feedback shots taken today. Reproducibility is still somewhat poor, although there's no other evidence of the vacuum vessel being unclean. (Base pressure this morning was 1.0E-8 T.)

Thursday October 4 05 10 hanson 56992-57039 kalman filter

More Kalman fitler phase scan shots taken today.

Friday October 5 05 08 hanson 57040-57093 kalman filter

More feedback phase scan data taken today. I'm nearing the end of this experiment; hopefully another day or two will be all I need to finish.

Monday October 8 06 31 hanson/debono 57094- kalman filter/rotation diagnostic

The morning was spent fixing a problem in the cos 1 theta Rogowski diagnostic. The L8210 used for this signal appears to have gone bad, so the signal was moved to a different L8210. In particular, the cos1 signal was moved from crate 5, slot 13, input 1 to crate 5, slot 21, input 4, and the tree was updated accordingly. In afternoon, rotation diagnostic shots 57101-57118 were taken. The detector was moved from its perpendicular viewpoint to a view looking along the torus (i.e, the detector was moved to see plasma rotation). The signals taken were relatively weak and characterized by nonexistant termination shock profiles. I will attempt to correct for this by increasing the light level output from the plasma by increasing the current and/or the pressure fill density.

Tuesday October 9 07 16 hanson 57120-57172 kalman filter

More Kalman filter phase scan data taken today. Yesterday's fix for the cos1 digitizer appears to be working fine.

Wednesday October 10 06 01 hanson 57173-57228 kalman filter

More Kalman filter feedback phase scan data taken today.

Thursday October 11 06 19 hanson/debono 57229-57321 kalman filter/rotation diagnostic

In the morning, Hanson continued the Kalman filter feedback phase scan; in the afternoon, DeBono got some rotation data from 10% He plasmas. The phase scan is nearly finished and preliminary analysis of the m=3 Rogowski signal shows a trend that looks a lot like that of Klein's experiments, but with no excitation above 5 kHz. 3 more shots are needed to complete the dataset; these will be taken tomorrow. Rotation diagnostic was taken with the detector setup to observe plamsa rotation. The tokamak was operated with high OHST and OHEL values in order. These parameters were chosen to gain as much signal as possible from the plasma. Data showed rotation velocities within the acceptable error (+-1km/sec) and with low velocity values (0.3 km/s < velocity < 3km/sec).

Friday October 12 05 10 hanson 57322-57358 feedback

A variety of things was done today. The last data points in the Kalman filter phase scan were recorded; Hall probe data was taken for feedback phasings of clear suppression and clear excitation; an open-loop, static (ie non-rotating), (m,n)=(3,1) field phase-flip FGPA algorithm was tested on the both edge-q ramp-down and ramp-up plasmas. There were problems communicating with the North rack throughout the day. Since this didn't directly impact any data taking, the issue was not fully resolved. Plan to let the crates cool off and see how things are next week.

Monday October 15 06 25 DeBono/Hanson 57360-57397 Rotation Diagnostic

Today I revisited the perpendicular 0-velocity shot after obtaining a wide spread in rotation velocities for a given non-perpendicular shot last week. The detector was moved to a perpendicular view; results showed inconsistent velocities obtained by the diagnostic. More shots need to be obtained for this.

Tuesday October 16 10 23 hanson 57398-57453 mode rigidity study

I'm working on doing Kalman filter feedback runs with one toroidal group of control coils muted. Rogowski coil and Hall probe data will then be analyzed as a function of the muted coil set. The scan is nearly finished; I plan to have it done with by the end of the day tomorrow.

Wednesday October 17 01 15 hanson 57454-57488 mode rigidity investigation

A preliminary mode rigidity experiment was completed today. The Kalman filter feedback algorithm was used with a phasing of 60 deg (for suppression) and a single toroidal group of control coils muted. Four shots were taken for each toroidal location.

Wednesday October 17 07 39 DeBono/Hanson 57489-57559 Rotation Diagnostic

Rotation Diagnostic data was taken with the detector oriented perpendicular to the plasma flow. Tokamak was run with 10%He,90%Du. Shots 57489-57559 were taken. Data seems encouraging with several excellent shots taken.

Thursday October 18 05 19 Jeff_Levesque 57566-57579 Thomson Scattering

Testing the Thomson scattering triggering and data acquisition. Observed scattered light in all channels of the polychromator.

Friday October 19 05 50 hanson 57581-57622 partial coverage

More partial coverage Kalman filter feedback data was taken today. Feedback was performed with different toroidal sections muted, one section at a time. Additionally, vacuum shots were taken to investigate the pick-up from the feedback coils in various diagnostics. These shots were done with the TF, VF, and OH fields present as normal, and a single toroidal section of feedback coils ringing at ~5 kHz.

Tuesday October 23 05 47 Levesque 57633-57691 Thomson Scattering

More tests of the Thomson scattering system. Attempted to increase scattered light collected by adjusting the collection lens. Signal levels stayed approximately the same over a small range of adjustment. Shots taken with laser firing at 3 ms and 5 ms into the plasmas. Verified that only channels 1 and 2 of the polychromator receive stray light.

Wednesday October 24 06 02 hanson 57696-57736 feedback testing

After analyzing SMP data from the last phase scan and not seeing much of a trend, I decided to move the SMPs up close to the plasma and do some feedback excitation and suppression shots. I also spent a little time doing feedback with Alex's old routine to get some data for comparisons between the algorithms. Additionally, Royce plugged his photodiode array into the CPCI and saw signals due to (unfiltered) light on most of the channels.

Thursday October 25 03 03 hanson 57737-57767 feedback

I finished up my run of feedback shots with the SMPs fully inserted today. Several phasings of Alex's old filter were tested in addition the Kalman filter.

Wednesday October 31 09 44 James 57697 - 87854 D-Alpha

Piggy-backed Jeremy's feedback runs. I was able to get all 20 channels working properly, showing both unfiltered and d-alpha. Unfiltered light was in the .15 V range, and the d-alpha .075 V. I was able to work out the channel errors I had. It turns out that I had the channel order on the board backwards. I was able to confirm this with the filament when I was aligning the PDA and mount. The chordal signals track the plasma nicely showing more visible light in the center channels where the most plasma is intercepted and less on the edges. Conversely d-alpha signals are highest on the (outboard) edge and fairly constant in the center channels. Data analysis is the name of the game for the next few days to get ready for the poster.

Friday November 9 11 02 Jeff_Levesque TS Calibration

Performed Rayleigh and Raman calibrations with nitrogen (on November 7 & 8) in order to calibrate Thomson scattering density measurements from 10/31/07.

Friday November 9 12 04 J._Andrello,_N._Rivera

We pumped the chamber back down. The pressure quickly dropped to high 10^-8 Torr. The bake is on and the pressure is presently 2.4x10^^-7 Torr

Thursday December 6 05 40 DeBono 58006-58065 Rotation

Shots were taken at the perpendicular viewing port to test anti-aliasing amplifiers on rotation diagnostic.

Thursday December 13 03 35 hanson 58065-58094 clean-up

I am attempting to reproduce the (m,n)=(3,1) kink mode used for Kalman filter feedback experiments in the summer and fall. It is quite possible to match the evolutions of the plasma current, major radius, and edge-q

Thursday December 13 03 48 hanson 58065-58094 clean-up

I am attempting to reproduce the (m,n)=(3,1) kink mode used for Kalman filter feedback experiments in the summer and fall. It is quite possible to match the evolutions of the plasma current, major radius, and edge-q with previous shots, but the mode does not always appear. Possibly, impurity levels are a bit higher (after the Thompson scattering calibration), making it harder to sustain a current density gradient at the edge. The mode did seem to get stronger as the day went on. I plan to turn on the RGA tonight and continue with these shots as time allows, eventually starting phase-flip experiments when plasmas improve.

2008

Monday January 14 05 14 hanson 58235-58264 clean-up

Clean-up shots were taken today, with the eventual goal of producing stable resistive wall modes for use in phase-flip experiments. The plasmas currently exhibit large disruptions before modes have a chance to grow. Loop voltages are around 8V for plasma currents in the neighborhood of 10 kA. Plan to continue running these kinds of shots all week.

Tuesday January 15 05 03 hanson 58265-58312 clean-up

More clean-up shots taken today with the intention of producing a good model discharge for phase-flip experiments. Plasmas are still plagued by large disruptions in mid-shot, but these seem to be getting smaller. Plan to continue this campaign tomorrow.

Wednesday January 16 07 13 Levesque_+_Hanson 58313-58396 Clean up

Clean up shots. Testing automatic recording of Thomson Scattering data. Raw data from TS oscilloscope traces is being added to the HBTEP tree. Royce was able to see a difference in D-alpha emission with changing fill pressure.

Thursday January 17 05 24 Hansen/DeBono 58415-58444 Cleanup shots

Cleanup shots taken today. Plasma seems to be improving.

Friday January 18 03 47 hanson 58446-58470 clean-up

More clean-up shots taken today, with the goal of developing a good model shot for phase-flip experiments. This discharge has been improving over the week, with discharge lifetimes increasing slowly, loop-voltage decreasing to < 8V, and soft x-ray signals starting to come up. We ended early today to insert the filament probe for d-alpha array spatial calibration. Plan to continue these shots next week.

Friday February 8 05 43 hanson 58610-58626 clean-up

We were finally able to get back to running HBT-EP this afternoon. (There were delays due to pumping on the bias probe bellows and air-conditioner maintenance.) There is definitely some more clean-up work to be done, but the plasmas do not look that bad, all things considered. Loop voltages were reaching 8 V towards the end of some shots, for plasma currents around 12-15 kA. On the side, there was some progress made in getting the CPCI to function more consistently. A firmware upgrade to the box fixed some internal triggering problems, so the CPCI now appears to record data at the correct time for every shot. There have been some new troubles getting the time values for the data, but this is less serious.

Saturday February 9 02 02 hanson/maurer 58631-58681 clean-up

More clean up shots taken today. Plasmas continue to improve, with loop voltages that drop below 8 V towards the end of the discharge. Plan to start mode development early next week.

Monday February 11 04 55 hanson 58682-58721 mode development

I am now trying to develop plasmas with external kink modes in which the edqe safety factor slowly ascends from about 2.7 to above 3.0. Currently, plasmas made in this style suffer large disruptions, indicating that there is still some clean-up work to be done. Past experience has been that the disruptions will become less frequent as we clean up.

Tuesday February 12 04 36 hanson/debono 58722-58762 discharge development

Continued working on creating a good q-ramp-up discharge to day. Plasmas continue to improve, but large disruptions are still observed occasionally. Some testing of the optical rotation diagnostic was done as well.

Wednesday February 13 03 41 hanson 58774-58810 discharge development

Attempts to create a Shilov-style kink mode continue. The modes seem to be improving slowly, but these shots are difficult to make. Shot 58808 looked good, but it was not reproduced.

Thursday February 14 04 45 hanson 58821-58862 discharge development

The q* ramp up discharges I've been working on this week are slowly improving in that they last longer and the rise through q*=3 is less dramatic. We are seeing large external modes on some shots, and lots of internal modes, too.

Friday February 15 06 11 DeBono/Hanson 58899-58949 Velocitometer Scan

Shots were taken with 90% De and 10% He this afternoon. Problems involving the Ch1 connector box appear to have been repaired. A shot with a constant major radius profile @ ~94cm for ~4msec was used as the observed chord was scanned across the plasma.

Monday February 25 07 18 DeBono 59023-59063 velocitometer

Shots taken on 59023-59063 with 90%De & 10% He to obtain the HeII emission profile of HBTEP plasmas. After 59063, a loud noise was heard in the basement after the start panel resets were punched. Banks were not charged. No initial sign of damage in the capacitor room.

Tuesday February 26 09 52 DeBono 59064-59094 velocitometer

Ran shot numbers 59064-59094 after refilling OHEL water resistors with CuSO4 solution. No problems, no abnormal occurances.

Thursday February 28 10 53 DeBono 59096-59140 velocitometer

Shots 59096-59140 were taken with 10%He, 90%De gas to finish velocitometer chord sweep of unaccelerated (i.e. no bias probe) plasmas.

Wednesday March 5 06 32 R._James 59164 - 59224 D-alpha

Attempted to understand discrepancies in calibration measurements made from the filament and data taken from the plasmas. Made several variations of inboard limited and outboard limited plasmas to try and determine if I was indeed only seeing part of the plasma as the calibration measurements suggested or the entire plasma which would point to a mishap in the calibration results. Unfortunately, the results are inconclusive. As I lock down constraints (i.e. camera angle, mirror angle, and camera tangential position), things are beginning to get clearer. To do this, I have been switching back and forth between plasma shots and filament scans to investigate the outer and inner channels respectively. I did discover what seems to be either a bad diode channel or a broken connection inside the 'camera' before the amp box in the lower channels (^#33) or in the input to the amplifier, it would need to be repaired before the calibration was complete as well. I will need a day or two more to systematically confirm the chord spacing and chordal view next week.

Tuesday March 11 03 02 James 59225 - 59251 D-Alpha

Calibration complete! Channel 33, repaired, Bias probe to be installed this afternoon. Details on calibration to come as white paper.

Friday March 14 05 21 hanson 59252-59273 clean-up

The bias-probe bellows was opened to the vacuum chamber yesterday night, and the first clean-up shots were taken this afternoon. In order to get an idea of what the vacuum conditions were like, I tried to reproduce my kink mode model shot from last fall's experiments: 56079. I was able to get close on a couple shots, indicating perhaps that there is not much clean-up work to do. The bake will be turned on over the weekend and clean-up efforts will continue on Monday.

Monday March 17 05 06 hanson 59281-59301 clean-up

More clean-up shots taken this afternoon. Strangely, shots today did not look quite as good as on Friday. Perhaps the weekend bake re-deposited impurities onto the limiting surfaces. Also, I'm shaking out a couple bugs in the feedback system, but it is more or less working.

Tuesday March 18 03 58 hanson 59302-59353 clean-up

More clean-up shots taken today. We are starting to see more mhd activity, but plasmas are not yet behaving very consistently.

Also, the feedback system is now in working condition. (A D-connector in the south rack was unplugged, disconnecting half of the control coil circuits).

Wednesday March 19 07 21 James/Hanson 59354 -59429 clean-up

Lots of clean-up shots today under the guise of mode development. Shiraki took more triple probe data, Hanson concentrated on Kalman filtering (59369 - 59386), while I did Shilov style RWM development (59387 - 59429). Hanson was able to recreate modes from last fall, yet it seems like the vacuum is still to dirty for RWM's. I plugged in the RGA so we can compare it to Mondays readings. More of the same tomorow.

Thursday March 20 05 27 hanson/james 59432-59493 clean-up

More clean-up shots were taken today. Royce has been attempting to create a reliable q* ramp-up shot for phase-flip experiments, but this has not yet been successful. The q* ramp-down shots are looking pretty good.

Friday March 21 05 03 hanson 59494-59556 clean-up More clean-up shots taken today.

Monday March 24 05 03 hanson 59557-59583 kalman filter

Experiments in adding adjustable amounts of noise to closed-loop feedback algorithms continued today.

Tuesday March 25 10 17 James 59584 - 59659 d-alpha

More clean-up and RWM mode development. The loop voltage has come down to about 8.5 V from 12V last week and Shilov's shots were at about 6V so still a fair amount to do. More of the same tomorrow.

Wednesday March 26 05 14 James 59641 - 59700 Clean-up/ RWM

More clean-up today. A bit of a late start after some CPCI com. issues that Jeremy and I resolved. Base pressure began to creep between shots. From 1.8 e-8 to 4.1 e-9 by the end of the day - I put on the RGA. Did a couple of phase flips, just to get familiar with the system, shot#'s:59700 (3.3 msec) and 59681 (2.5 msec).

Monday March 31 08 24 debono 59733-59790 Velocitometer

Rotation Diagnostic data was taken with bias probe inserted into plasma. Bias probe voltage was varied from 0V to 275V at 25V increments. Data shows sharply reduced plasma lifetime at higher voltages, as well a tendency of the plasma to crash into the inner wall at higher bias probe voltages.

Wednesday April 2 05 58 hanson 59834-59860 kalman filter

The newly reworked Kalman filter algorithm was tested on plasmas today after modeling in IDL indicated that it would behave reasonably. (The accuracy of the IDL model was checked against the actual FPGA algorithm at 10 frequency points and found to be in very good agreement.) Several phase angles and gain settings were tested today. A more formal phase scan will begin tomorrow.

Thursday April 3 06 01 hanson 59861-59921 kalman filter phase scan

I scanned a good number of Kalman filter feedback phase angles today. I plan to continue this experiment into next week.

Monday April 7 07 21 R._James 59993 - 60022 RWM Development

Initiated fill pressure scan today. Plan to complete tomorrow and start phase flip experiments.

Tuesday April 8 04 55 james/hanson 60023-60067 d-alpha/kalman filter

Royce did some fill pressure scan experiments this morning, and I did a few more points in my phase angle scan this afternoon.

Wednesday April 9 05 59 hanson 60068-60102 kalman filter

The first set of phase angles in the Kalman filter phase scan was completed today. I will probably try to do a couple more sets for averaging. The initial results are about the same as those from last fall, but with shifted regions of suppression and excitation. Now that the optimum phase angle for supressive feedback is known, I'm taking a few shots to see if the proportional gain setting can be further optimized. This should be finished by tomorrow.

Thursday April 10 07 40 James/Hanson 60103 - 60141 RWM - Feedback

Hanson continued yesterdays tests following up on the phase scan completed yesterday. More pressure scans, pressure seems to cap at 5.5e-5 no matter how long the puff valve is open and bottom out at ~ 4.4e-5 to still achieve breakdown. More analysis over the weekend will help to determine if is a large enough window to appreciably alter D-alpha emissions and eventually RWM stability. It would be nice to get a factor of 10 and at least a factor of 2 difference. We will investigate the puff electronics and pressure to see if we can improve it next week.

Thursday April 10 09 46 James/Hanson 60103 - 60141 RWM - Feedback

Hanson continued yesterdays tests following up on the phase scan completed yesterday. More pressure scans, pressure seems to cap at 5.5e-5 no matter how long the puff valve is open and bottom out at ~ 4.4e-5 to still achieve breakdown. More analysis over the weekend will help to determine if is a large enough window to appreciably alter D-alpha emissions and eventually RWM stability. It would be nice to get a factor of 10 and at least a factor of 2 difference. We will investigate the puff electronics and pressure to see if we can improve it next week.

Friday April 11 04 18 hanson 60142-60185 kalman filter

More Kalman filter gain scan shots were taken today. I plan to analyze the results of this and the phase scan over the weekend, and we can discuss what experiments to do next on Monday.

Wednesday April 16 05 19 hanson 60191-60255 discharge development

I attempted to make a stronger kink mode discharge for feedback studies. Some progress was made, but I will probably need another day or so to continue developing the shot. In other news, Royce and I added a second puff of gas during the middle of some discharges. This had a dramatic effect, temporarily damping mode activity, and causing some very large disruptions later on. We imagine that higher fill pressures could be obtained by moving this second puff to a time before the plasma breaks down, but this has not yet been tested.

Monday April 21 10 44 James 60343 - 60379 RWM / D=Alpha

Investigated variations in puffed gas during shots to see how the plasma reacts. Used 3.75 ms as a base puff time and initiated a second puff of 1/8, 1/12, and 1/36 of that time at 2.5 ms. It generally takes approx. 1.5 ms after the puff is initiated to see an increase in D-alpha. Emissions were about equal to breakdown peak, half breakdown peak, and 1/5 breakdown peak respectively. The later was still about twice the emissions before and after its flow and ebb - except when theres a disruption immediately following. As expected, 1/36 puff time seems to be the least disruptive to the plasma. The lifetime of the plasma is shorter with a major disruption usually just after the ebb of the d-alpha increase and sometimes during that ebb. Plasma current ramps always leveled off with 1/36 puff having a significant effect just like the 1/8 and 1/12 puff times. However the q seemed to behave much in the same manner as usual as long as the major radius fell off fast enough. I also attempted to increase the pressure before the shot with multiple puffs since I have had better luck with developing the mode for plasma lifetimes between 5 ms to 7 ms. With an increased puff time of 5 ms, I used 3 pre-breakdown puffs, at -95 ms, -35 ms, and -.5 ms to get a slightly higher fill pressure than with 2 puffs, up to approx. 7.6 e -5 torr from 7.0 e-5 torr seen with only 2 puffs. Later I added the 1/36 puff time at 2.5 ms as before totaling 4 puffs. See shot ^#60379 use scope hbtep_da_review.jscp in my home directory Jscope configurations under. D-alpha emissions are higher at the higher pressures yet the neutrals appear to burn out as the shot continues and the increase between breakdown and final disruption is subtle.

Tuesday April 22 05 25 hanson 60380-60409 discharge development

By examining the effects of small changes to the bank settings previously used for feedback studies, I have been successful in making a mode that is both stronger and more reproducible than the old one. With the new settings, the current ramp is kept about the same, but the edge-q is kept closer to (but just below) 3. This seems to result in a robust 5 kHz mode. The mode also seems to respond to feedback: I've been able to reduce amplitudes in the sensor coils by 40-50% in a couple shots. One shot done at a positive feedback setting showed excitation of the mode. I also did a shot in which the FPGAs were not gated on until 2.5 msec (in the middle of the mode). This appeared to have a near-immediate on the mode amplitude, but more shots are needed to confirm this result. I am planning next to do a scan of the growth rate parameter in the Kalman filter model, as run time allows.

Wednesday April 23 06 00 hanson 60410-60467 kalman filter

I began a scan of the Kalman filter's real growth rate parameter today, using the mode developed yesterday. I'm about halfway finished with the scan; it should take another day to finish up.

Friday April 25 06 10 DeBono 60565-60586 rotation diagnostic

Shots 60565-60578 taken with 90% Du 10% He, and shots 60579-60586 taken with 100% He. Bias bank appears to be working well; probe voltage holds well with bias probe driven currents well exceeding 50 amperes in plasma. Very low signal levels observed with the rotation diagnostic while using 90% Du 10% He mix, obeserved signals became usable with use of 100% He plasmas.

Tuesday April 29 06 11 hanson 60617-60640 diagnostics testing

The RC filtering for the SSI sensor coils was added to the signal path this morning and tested with plasmas. The plasmas themselves were cold and did not display much MHD activity, but the filtered sensor signals were observed to have a quasi-dc component that I believe should come from the Ohmic and vertical fields. One filter was also bench tested to make sure the frequency roll-off was as expected. With regards to the SMPs, the L8100 amplifiers have a gain and a multiplier setting. Originally the these were set to G=100 and M=.5, giving an effective voltage step of dV=.0002 V in the signal. The gain cannot be set above 100, but the multiplier was turned up to 1.0, for an effective voltage step of dV=.0001 V, improving resolution by a factor of 2. In order to achieve plasma break-down, I had to use the double puffing technique to inject extra gas. I was then able to reduce the amount of gas puffed in over the afternoon, but I could not get breakdowns with normal puff times. The plasmas had loop-voltages above 12 V and showed chaotic behaviour and little MHD activity. It also seemed to take a long time to pump out after a shot. An inspection on the RGA showed levels of He/D2 at the same order of magnitude as the base pressure. I've been told that He pumps out slowly, so I've turned on the bake to see if this will accelerate things. I will consult with Jim and Dave about regening the cryos in the morning.

Wednesday April 30 06 03 hanson 60642-60705 kalman filter

More Kalman filter gain scan was done today. The hope is that the modifications to the magnetic diagnostics and the new SVD analysis will reveal a clearer pattern in the scan. After this, I plan to start a phase scan at the optimal gain. (The previous phase scan was done at a rather low gain.) Plasmas looked good today. Poor plasma performance yesterday was likely due to some confusion over what gas was in the puff lines from last week; I may have been puffing in a large amount of He.

Thursday May 1 05 45 hanson 60706-60785 kalman filter

The Kalman filter proportional gain scan begun yesterday was finished today, giving a fairly clear indication of what the optimal gain is for negative feedback. A phase scan was then begun at this gain and is a little over half finished. I plan to finish at least one run of the phase scan tomorrow and possibly move on to scanning the real growth rate parameter.

Friday May 2 04 22 hanson 60786-60838 kalman filter

A first round of the Kalman filter phase scan was completed today.

Monday May 5 05 33 hanson 60862-60899 phase scan

More Kalman filter phase scan shots were taken today. Progress was slowed by repeated CAMAC problems. We were getting NoX errors on large numbers of digitizers during the afternoon. After turning off the Jorway serial highway driver and crates for approximately half an hour, we were able to get 4 shots in before the errors came back. This makes me think the root causes of the problems was thermal -- perhaps related to our air conditioning outage in the server room.

Tuesday May 6 10 24 hanson 60905-90987 kalman fitler

More Kalman filter phase-scan data taken today. There were CAMAC problems again due to the AC outage in the control room. However, CU facilities came by with a temporary unit that is now cooling the server room, and we were able to further eliminate CAMAC errors by setting the screenroom crate's u-port to bypass. The hypothesis is that the screenroom crate was overheating, triggering NoX errors in everything else downstream. In any case, the second round of the phase scan should be complete by tomorrow morning, if there are no further hardware problems.

Wednesday May 7 06 18 Hanson 60995-61063

Jeremy continued with his phase scan, and some Mach probe tests were also done.

Thursday May 8 05 54 hanson 61064-61118 kalman filter

More Kalman filter phase scan data taken today. I am just a couple shots shy of being done with the 3rd round of the scan. The trends in the data are becoming clearer with more points to average over, so I think I will try to do some added noise experiments tomorrow.

Friday May 9 05 38 hanson 61119-61165 kalman filter

More Kalman filter shots today. I finished the third round of the phase scan and started investigating the effects of added noise.

Monday May 12 04 32 hanson 61166-61192 kalman filter

More noise scan data was taken today.

Tuesday May 13 04 32 Hanson/Shiraki 61193-61210 kalman filter/mach probe

The Kalman filter algorithm was tested with external noise inputs today. The Mach probe was tested also.

Wednesday May 14 05 39 hanson 61211-61269 kalman filter

More Kalman filter experiments with added noise today. Initial analysis (on just a few shots) revealed differences between the efficacy of the Kalman filter and proportional gain only filter in both exciting and suppressing the kink mode. I hope to finish getting data by tomorrow and to start writing things up on Friday.

Thursday May 15 05 50 hanson 61270-61329 kalman filter

More added noise experiments today with the Kalman filter and proportional gain only filter. Discharge repeatability has been poorer these past couple days, slowing progress. I'll probably try to run most of tomorrow, and maybe will regen a cryo on Saturday.

Friday May 16 05 39 hanson 61330-61389 kalman filter

More added noise experiments today. The krytron for the VFEC bank failed and we replaced it with a spare. However, this was the last spare; we have no more spare krytrons. Aside from that interruption, the runs went well.

Tuesday May 20 05 05 James 61391 - 61414 RWM Stability

Did some mode development shots today. We decided to regen the other cryo so it was slow going. Was able to get a decent mode relatively quickly plus test the Hall probe channels I will be able to use at the same time as D-alpha. Phase flips will start tomorrow.

Wednesday May 21 10 06 James 61415 - 61458 RWM Stability

Opened up the 2nd cryo today after a few shot and it seems to be doing well. I did a series of static phase flip experiments and a few dynamic ones at the end (61456-58),at 2.5 and 2 ms respectively (before the q*=3 surface was broken. The hall probe, triple probe D-alpha can all be run simultaneously. We should be ale to continue testing tomorrow and include gas puffs.

Thursday May 22 08 34 R._James 61459 - 61521 RWM Stability

Perfected low fill pressure stable RWM and performed static phase flip tests at 2 ms, 61492 - 495. Developed a 3 puff pre-breakdown stable mode which gave an increase from 4.4e-5 to 5.9e-5 and repeated the static flips at 2 ms, 61511 - 61519. These plasmas were a little harder to reproduce from shot to shot, but I was able to get some stable RWM's to experiment with. I will do one more static phase flip experiments tomorrow with a puff early in the shot to complete the static phase flip method. If time allows I will initiate the dynamic phase flip experiments as well.

Friday May 23 03 02 R._James 61523 - 61555 RWM Stability

I completed the Phase flip experiments today. If time allows I will do the resonant response test around the end of next week. Preliminary results are promising, it looks like the realignment times after the flip for a stable mode with normal fill pressures around 4.4e-5 torr are around the same time as those Shilov reported about .25 ms. I increased the fill pressure in two ways. With 3 pre-fill puffs 61510 - 519 and with 2 pre-fill and one puff early in the shot 61541-555. The realignment times for these were subsequently longer indicating a probable decrease in stability with more neutrals!

Tuesday May 27 06 21 DeBono 61556-61590 Velocitometer Scan

Shots were taken with 10%He 90%Du. Measured velocities are inconsistent shot-to-shot with the bias probe inserted (but not with the probe removed), perhaps additional shots are needed to remove impurities from the surface of the bias probe.

Wednesday May 28 06 26 DeBono 61592-616-24 Velocitometer w/ Bias Probe Shots

Plasmas were run from 10% He, 90% Du. The bias probe was energized to voltages of 50V, 150V, 300V to see measured affect on plasma rotation (Probes outer edge inserted to 105.5cm with the velocitometer looking at 99.53cm). The diagnostic detects increasing toroidal rotation as the bias bank is fired, and it appears that increasing bias probe voltages result in higher rotation speeds. Unfortunately, the bias bank pre-fired ~50% of the shots taken (possibly due to craytron noise).

Thursday May 29 06 08 DeBono 61625-61657 Velocitometer scan

Plasmas were run with 10% He 90 % Du. Bias bank pretriggered for nearly every shot taken, upon examination it appears that this is likely due to craytron noise from the VFST and VFEL banks, and possibly also the OHST and OHEL banks.

Friday May 30 06 14 Shiraki 61658-61688

Took hall probe and triple probe data. Results to be analyzed over the weekend.

Wednesday June 4 10 46 R._James 61733 - 61779 RWM

Shot development today, experiments to follow tomorrow.

Thursday June 5 01 47 DeBono 61780-61792 Bias Bank Prefiring Tests

Tests were run in 100% He to test an attempted fix to the bias bank prefiring problem. The tests proved to be successful mostly: the bank did prefire for most shots taken at high voltage (>300 volts) but no pretriggering occured for shots taken at 50 or 150 volts. The fix to the problem is described below.

bias bank power supply was rerouted to a plug outside the capacitor room, as earlier it was plugged into the same power outlet as the crytrons. The power supply connects to an isolation transformer, which has cheaters plugged into both its power outlets to provide power to the avionics trigger and to the power supply for the bias bank as well as potentially avoiding any ground loops. The lid to the bias bank was closed (previously it was left open) as well to minimize EM interference with the triggering circuit.

Friday June 6 04 45 debono 61793-61817 velocitometer scan

Plasma were run with 90% Du 10% He. The repairs to the bias probe triggering appear to work well for voltages <=200V, however, at higher voltages the bank still suffers from pretriggering. A sweep was conducted between 50V and 250V to attempt to characterize the increase in He ion velocity with temperature. Data to be analyzed over weekend.

A loud noise in the capacitor room was heard during the last shot of the day; one of the relays in the TF bank appears to have been the culprit.

Wednesday June 11 12 28 R.W._James 61819 - 61843 RWM

Regend Cryo again today the air conditioner in the room with the Tokamak was repaired just after lunch so we got a late start. When I left, the temp was pegged at 20K and had been for the last 2 hours. It was slow running with the single cyro and mode development was all that I was really able to accomplish. Phase-Flip experiments in the morning. There is leaking in Mo's area from the wall above next to the reactor. The engineering students downstairs came up to let us know and they also called facilities. The water is localized to the floor just under the wall and damage looks minimal to Mo's space, but there is about 3-5 gallons of water on the floor.

Thursday June 12 02 33 James 61819 - 61968 RWM Stability

I was able to get a plasma response from the Phase Flips after a lot of wrestling with the tokamak. I did phase flips at 2ms, 2.5ms, and 3ms which also corresponded to different D-alpha emissions. Plasmas were not very reproducible so it took a long time. I will attempt tocomplete this experiment on Monday after the meeting.

Thursday June 12 06 01 debono 61977-61992 Bias Probe

Shot development with bias probe inserted and 10%He 90% Du plasmas. Slow going due to only one cryopump operational. DeBono

Friday June 13 06 23 debono 61993-62023 Bias Probe

Shots taken in 90% Du 10% He. Shots were taken with bias probe was run at voltages from 0-200V in 25V increments. Pace was slowed considerably due to lack of a second cryopump.

Tuesday June 17 03 04 hanson 62028-62043 discharge development

I've started creating a mode for experiments that will scan the Kalman filter parameters. This will be very close to the mode used in previous experiments, but some small adjustments need to be made, probably due to differences in the vacuum conditions.

Thursday June 19 05 06 hanson 62056-62110 discharge development

More discharge development for Kalman filter studies today. I've had some good shots, but reproducibility has been poor. Hopefully this will improve as I continue to run.

Friday June 20 05 39 hanson 62111-62163 kalman filter growth rate studies

I started an experiment today to scan the real growth rate parameter in the Kalman filter equations for negative feedback. Plan to continue and hopefully finish this scan by mid next week.

Monday June 23 04 42 hanson 62164-62192 kalman filter

The scan of the Kalman filter's real growth rate parameter continued today. Unfortunately, there were problems with the in the output on several control coils: either the output was null or had a large dc offset. It appears that the analog boards may be part of the problem, but investigations are still underway.

Wednesday June 25 02 40 James 62193 - 62231 RWM stability

Completed aditional stability / phase-flip experiments at 2.0, 2.5, and 3.0 ms. Shot to shot variability was better than in the past, but still not the best. these shots will give more statistical might to the results I got two weeks ago.

Thursday June 26 05 31 hanson 62232-62296 kalman filter

More Kalman fitler growth rate scan today. The analog boards seemed to be working normally again. I hope to be finished with this scan tomorrow.

Friday June 27 04 52 hanson 62297-62335 kalman filter

I finished the scan of the real growth rate parameter in the Kalman filter today. Plan to begin a scan of the imaginary part next week.

Monday July 7 07 00 James 62341 -62372 RWM More shot development with additional puff. More tomorrow with Phase-Flip experiments too.

Wednesday July 9 12 16 James 62441 = 62405 RWM

Was able to create shots with 2nd puff during the shot. Plasma parameters seem unaffected for a few ms after the affects of the puff are seen in d-alpha. This is promising. However, our back of the envelope calculation puts the perturbation delivered by the new control coils onto the plasma at about 1/6 of the old coils. Tomorrow I will attempt to get a more stable mode so the response will have some hope of being measured.

Wednesday July 9 11 41 James 62407 - 62501 RWM

Attempted to squelch the MHD noise in the magnetic probes by altering the VFS timing and subsequent mode development. This was not too successful. At times, the MHD noise was much worse that shots I took two weeks ago.

Tuesday July 15 08 35 James,_Levesque 62518 - 62539 Density/Temp Scan

Used Thompson Scattering to investigate density and temperature levels before, during, and after affects of the 2nd puff are seen in d-alpha. Initial inspection shows temp and density to be fairly inversely proportional. I will compare TS data to triple probe data and hall probe data to see how the increase in neutrals changes the plasma. The triple probe had all three tips inserted and one hall probe from the array was inside the plasma.

Wednesday July 16 05 27 hanson 62537-62569 Kalman filter

I started a scan of the Kalman filter's rotation rate parameter today. I'm pleased to report that I was able to reproduce my target plasma relatively quickly. Hopefully the scan will be nearing completion by the end of this week.

Thursday July 17 04 15 hanson 62570-62591 kalman filter

More Kalman filter phase scan data taken today. The run was cut short by a dump relay failure on the TF bank. Work is currently underway to repair the damage.

Tuesday July 22 03 44 shiraki 62602-62604 TF Test

Tested TF bank, up to the usual 6.1 kV. Normal running will resume tomorrow.

Wednesday July 23 04 47 hanson 62605-62634 kalman filter

I attempted to continue my scan of the Kalman filter's rotation rate parameter today, but there was a small vacuum mishap. Sometime between yesterday and this morning, the gate valve for the horizontal turbo pump was opened. Fortunately, the mini turbo was turned on, backing the larger one. However, the RGA showed higher partial pressures of N2 and O2 after this problem was noticed. The behavior of plasmas after the incident was somewhat different from last week, but strong external kink modes were still achievable. We plan to continue running and will start regenning the cryo pump on the large pump stand tomorrow. The base pressure at the moment is down to 1.3E-8 Torr (with one cryo pump valved off).

Thursday July 24 06 34 hanson 62634-62685 discharge development

I attempted to recreate external kink unstable discharges used in earlier Kalman filtering experiments. Plasma behavior seemed to change throughout the day, and I was getting close to my old shots by late afternoon. Base pressure this morning was 6.6E-9. The cyro on the large pump stand was allowed to warm up last night, and the cold head was pumped on today. Pressure in the line was about 10 mTorr at the end of the day, so the regenning will continue tomorrow.

Friday July 25 04 51 hanson 62688-62746 kalman filter

A Kalman filter rotation rate parameter scan was started today. Plasmas and modes looked good. Regen of the cryo pump is still ongoing.

Monday July 28 04 23 hanson 62747-62775 kalman filter

More Kalman filter rotation rate scan today. Plasma behavior was good. Regenning of the cryo at the large pump stand continued.

Tuesday July 29 05 19 hanson 62776-62818 kalman filter

More Kalman filter rotation rate scan today. Regen of the cryo continued as well.

Wednesday July 30 02 15 hanson 62819-62826 kalman filter

The scan of the Kalman filter's rotation rate parameter was finished up today. Regening of the cryo pump is still ongoing.

Thursday July 31 05 07 Levesque 62827-62838 SXR tomography

Tested several amplifier boards in soft x-ray tomography system. Banks off, no plasma -- just amplified test signals.

Friday August 1 06 16 debono 62841-62859 Bias Probe Radial Electric Field Scan

The objective of todays run was to characterize the radial electric field generated by the bias probe using the triple probe array, and to simultaneously measure rotation using the doppler shift diagnostic. Plasmas were run in 10% He 90% Du plasmas. Plasmas appeared cold (<30eV core temp as measured by thomson scattering) but good radial profiles were easily obtained. Bias probe voltages were scanned from 0-150V in 50V increments. Triple probe data seems anomalous, more to follow on this.

Monday August 4 06 33 DeBono 62861-62882 Triple Probe/ Mode Development

More work was done today to characterize the radial electric field generated by the inserted bias probe. The error in the triple probe was confirmed today; the first channel and third channel of the triple probe failed to produce the same signal when inserted into the exact same position in the plasma (done by inserting/withdrawing the probe by 1cm). For some reason the signal from ch1 seems to be inverted compared to the signals obtained from channels 2 and 3, not sure why. This inconsistency problem in the triple probe does not seem to depend on the bias probe being inserted into the plasma, although inserting the bias probe did significantly alter the signal we obtained from the triple probe.

On a positive note, mode development proved highly productive today. Thomson scattering shows core temperatures ~100eV (compared to ~30ev on friday) with a peak plasma current of 17kA with a very strong current ramp, plasma lifetime~7msec, and a very clean major radial profile. Rotation diagnostic signals are very clean, although perhaps some additional signal would be desirable. Plasmas were run with 10%Du 90% He.

Tuesday August 5 05 09 debono 62883-62906 Mode Development

Additional mode development was done today, following experiments from yesterday. Shots developed on 8/4/2008 appear to be favorable for rotation diagnostic studies, however, the presentation of a very low edge q (1.5<q<2.5) results in high MHD activity as well as making it more difficult to potentially use Jeremy's feedback stabilization. Accordingly, efforts today were dedicated to obtaining shots of similar profile to those on 8/4/2008 but with a higher edge q and hopefully also lower MHD activity. Shots were conducted 10% He 90% Du.

Thursday August 7 05 19 hanson 62924-62970 kalman filter

I wanted to get a few more points in the scan of the Kalman filter's rotation rate parameter today, because the optimal phasing for the different parameter settings is not clear from looking at the transfer functions. This work will continue tomorrow.

Saturday August 9 11 38 Hanson/Levesque 62971-63022 Kalman filter + TS

Jeremy continued Kalman filter scan. Took Thomson scattering and SXR tomography data during the scan. TS data show a significant drop in central density at 3ms for shots lasting >~4ms. Temperature shows a linear rise throughout shot, though major radius is also changing. Drops in SXR emissivity around 3ms corroborate TS data.

Tuesday August 12 05 14 hanson 63023-63039 kalman filter

More Kalman filter rotation rate scan data taken today.

Much time was spent troubleshooting a CAMAC problem. The FO line connecting the north and basement racks had become pinched in the door of the basement rack and failed. Jeff and I removed the line and installed a new one. Various racks then had problems initializing properly. Powering off the crates, letting them cool for an hour, and restarting matterhorn fixed this problem. So things are back in working order.

Wednesday August 13 06 13 hanson 63040-63084 Kalman filter

More Kalman filter rotation rate scan data taken today.

Thursday August 14 08 28 debono 63090-63117 Radial Eelctric field

Work was conducted today to observe the radial electric field generated by the bias probe.

Wednesday August 20 05 46 hanson 63143-63183 discharge development

I had intended to begin a scan of the Kalman filter's rotation rate parameter at the different value of the q0 parameter, but I had difficulties reproducing the modes I was seeing last week. It seems the plasmas might be heating up a little more and rotationally stabilizing the mode. I intend to continue my efforts to bring the mode back tomorrow.

Thursday August 21 05 43 hanson 63185-63232 kalman filter

Some Kalman filter data taken today, for a scan of the rotation rate parameter.

Monday August 25 04 40 hanson 63282-63299 clean-up

Clean up shots taken this afternoon following a weekend bake and glow discharge experiments on Friday. With some small adjustments to the banks settings, I am able to come close to reproducing shots from last week, but mode amplitudes are lower. Plasmas improved over the afternoon.

Monday August 25 05 38 Shiraki/Levesque 63233-63281 He glow discharge

(Run report for Friday August 22)

Ran helium glow discharge to attempt a relative calibration of probe areas in the triple probe array using a uniform density plasma. Tried to measure emission using the rotation diagnostic for a zero-rotation calibration. The resulting plasma density was too low to measure with the existing triple probe circuitry, and there was not enough emission appropriate for the rotation diagnostic due to low ion temperature and density.

After the glow, took digitizers-only shots (63241-63281) as part of testing the SXR tomography system.

Tuesday August 26 05 07 hanson 63302-63348 clean-up

More clean-up shots taken today. The kink mode activity continues to improve, getting closer to what was observed last week.

Wednesday August 27 06 17 hanson 63349-63372 kalman filter

External kink modes improved today and some Kalman filter parameter scan data was taken. We had trouble this afternoon with the dump relay on the OH Bias bank. The relay did not fail completely, but was arcing enough that we could smell it in the control room. A couple hours were spent finding, replacing, and testing the relay.

Thursday August 28 04 50 hanson 63373-63420 kalman filter

More Kalman filter rotation parameter scan data taken today.

Friday August 29 05 50 hanson 63420-63483 kalman filter

More Kalman filter rotation rate parameter scan data taken today.

Tuesday September 2 04 41 hanson 63484=63510 kalman filter

More Kalman filter rotation rate parameter scan data taken this afternoon.

Wednesday September 3 11 23 hanson 63511-63587 kalman filter

More Kalman filter rotation rate parameter scan data taken today.

This evening there were problems with low-amplitude signals on a couple of the control coils. Some inspection of the crunch amp cabinet revealed some poorly made connections in the power bus for the amps. The connections were either improperly soldered or not soldered at all, and covered with electrical tape. A dozen or so of these were fixed, but some problems with the amps remain. Debugging of the system will continue in the morning. <

Thursday September 4 05 39 hanson 63589-63608 feedback system debugging

Further attempts were made to get to the bottom of the feedback signal problems mentioned yesterday. Connections at the inputs to the crunch amps were examined and redone, and some testing of the analog boards was done. The problems still have not been resolved, so work on this front will continue tomorrow.

Friday September 5 04 55 hanson 63609-63641 feedback system test

More debugging of the feedback system today. The crunch amps and analog boards have been eliminated as causes of the smaller signal amplitudes observed on several coils. Connections in and out of the crunch amps are now being investigated.

Monday September 8 05 13 shiraki 63645-63649

A glow was created this afternoon for calibration of the triple/mach probe, but was unsuccessful because the transimpedance amplifiers were not designed for dc currents. The bake was turned on at the end of the day.

Friday September 12 07 02 Levesque+Shiraki 63664-63704

Cleanup Cleanup shots following Monday's deuterium glow. Broken krytron in the VF electrolytic bank was replaced using one from the ICRH bank. Had problems seeing light with the Thomson scattering system, but found that the problem occurred because the TS viewport shutter was still closed from the glow (ooops).

Monday September 15 06 11 Shiraki/Levesque/Hanson 63705-63746 Cleanup Cleanup shots.

Jeremy was able to get the feedback system working again by swapping analog boards.

Tuesday September 16 05 26 hanson 63747-63801 clean up

More clean up shots taken today. MHD activity is approaching reasonable levels.

Wednesday September 17 05 52 hanson 63803-63836 clean up

More clean up shots taken today. Hopefully, ELM noise experiments can begin tomorrow.

Thursday September 18 05 31 hanson 63837-63902 Kalman filter

I attempted to take some data for Kalman filter experiments with added ELM-like noise. However, MHD activity in the plasmas was very inconsistent from shot to shot, indicating that perhaps the vacuum is not quite clean yet.

Friday September 19 07 00 DeBono/Hansen 63919-63966 Feedback & Bias Probe

Shots 63919-63966 were taken with the bias probe inserted into the plasma. The objective of the experiment is to apply bias voltage in order to slow down the rotation velocity of the n=1 mode to 0hz (or as close to 0 as possible) and subsequently to apply Jeremy's feedback algorithm. Mode development was conducted today, more results next week.

Monday September 22 07 35 hanson 63967-64019 feedback

Feedback algorithms were tested in vacuum with both elm-like and white-noise inputs. While the algorithm without the Kalman filter appears to be more or less transparent to both types of noise, the Kalman filter reacts to both noise varieties at frequencies near the programmed mode rotation frequency. The vacuum response of the Kalman filter was greater than I expected for both types of noise, given the results of previous experiments with plasma.

Plasma-mode development continued, but there was not much time left for it this evening.

Tuesday September 23 04 45 hanson 64020-64057 kalman filter

Experiments with adding ELM-like noise to feedback algorithms started today, but the run was cut short by problems firing the VF crowbar in mid afternoon. Subsequent investigation pin-pointed the trigger box for the ignitron, probably the krytron tube. Nick is still working on the circuit to eliminate other possibilities. We are now out of krytron tubes.

Wednesday September 24 06 16 hanson 64060-64102 kalman filter

Nick was able to substitute a circuit with a thyrotron for the one with the broken krytron this morning, and runs resumed in the afternoon. More elm-noise experimental data was taken.

Thursday September 25 04 56 hanson 64103-64158 kalman filter

More elm-like noise experiments today. I have enough data to start analyzing, and I will decide tonight if I need to get more tomorrow or if I want to pick up where I left off with the rotation-rate parameter scan.

Friday September 26 2008 7:31 pm hanson/shiraki 64159-64247

More ELM-noise experiments today. Also, a radial scan of basic plasma parameters was taken with the triple probe.

Monday September 29 2008 4:46 pm hanson 64248-64269 Kalman filter

More Kalman filter rotation rate parameter scan data taken today. jmh

Tuesday September 30 2008 4:59 pm hanson 64270-64320 kalman filter

More Kalman filter rotation rate parameter scan data taken today. jmh

Wednesday October 1 2008 5:49 pm Hanson 64321-64358 Kalman filter

More Kalman filter rotation rate parameter scan data taken by Jeremy today. -JPL

Thursday October 2 2008 8:22 pm hansen/debono 64388-64399 Feedback & Bias Probe

Plasmas were run with 100% Du. Bias probe was inserted and effects on plasma rotation were measured. The effect of the edge biasing is to reverse direction on the n=1 mode, however, unfortunatly the amplitude of the observed m=2 modes are rather weak. More work to be done tommorow.

Friday October 3 2008 6:40 pm DeBono/Hansen 64400-64444 Bias Probe/Kalman Feedback

More work was done today on utilizing the bias edge probe to slow down the rotation frequency of the n=1 mode. Mode development work done today, feedback studies to be done next week. DeBono

Monday October 6 2008 8:21 pm debono 64446-64494 bias edge probe

Bias probe inserted from 104.1cm-106cm in 100% Du plasmas. More work was done today on using the probe to generate a shot with ~0velocity for the n=1 mode, with the goal of then conducting feedback studies. It proved difficult to obtain a desired mode unfortunatly, with the two main difficulties being 1. obtaining sufficient signal from the SSI's and 2. flattening out the n=1 mode velocity profile around 0 km/sec. DeBono

Tuesday October 7 2008 4:57 pm hanson 64495-64529 kalman filter

More scan of the Kalman filter's rotation rate parameter today. (The biased probe was turned off and retracted outside the plasma.) jmh

Thursday October 9 2008 1:05 am Hanson/Levesque/Shiraki 64530-64579 Kalman filter

More scanning of the Kalman filter's rotation rate parameter today. Brief testing of the mach/triple probe array and SXR tomography.

One of the two power supplies (PS) in the SXR tomography amplifier box was found to be the source of an overwheming amount of noise in data from this week. Powering all amplifier boards from the other PS alone reduced the noise significantly. Problems with the PS or its connections will be investigated further tomorrow. -JPL

Thursday October 9 2008 8:32 pm debono 64602-64638 bias probe induced n=1 mode braking

More work done today on the n=1 mode breaking done today. Bias bank pretriggering ended up costing about half the runday, however this cleared up evenutally allowing a few hours of productive plasmas. Will try a modification to the bias bank soon in order to try to eliminate the pretriggering.

Friday October 10 2008 2:26 pm hanson 64648-64682 kalman filter

More Kalman filter rotation rate parameter scan data taken today. Scan of the plant noise covariance parameter to begin next week.

Monday October 13 2008 10:13 pm DeBono 64683-64715 Bias Probe

More mode development work done today with the bias probe inserted. The edge q was kept slightly below 3.0 throught the shot duration in attempt to provoke an m=3 mode from the plasma. Biorthogonal decomposition revealed this to lead to a short (~0.75 msec) and relatively weak m=3 mode followed by a much stronger (~4x) m=2 tearing mode which terminates the plasma. More work to follow tommorow. DeBono

Tuesday October 14 2008 11:10 pm debono/shiraki 64716-64791

More mode development today.

Wednesday October 15 2008 11:09 am debono 64722-64782 Bias Probe

More work done today regarding mode development with the bias probe inserted into the plasma. The goal of the days run was to develop a strong RWM with the bias probe inserted in the plamsma for future feedback studies. Biorthogonal decomposition showed that the shots we developed today had an acceptable (but not incredibly strong) n=1 mode between 2-3 msec followed by a tearing mode which eventually disrupts the plasma. After developing the mode the bias probe was initialized at 2msec at a voltages ranging from 0-125 volts to spin the n=1 mode. As expected, the n=1 mode reversed direction, and did so more rapidly as the voltage on the probe was increased.

Wednesday October 15 2008 9:52 pm hanson/levesque 64792-64847 Kalman filter

Jeremey continued Kalman filter studies. Observed internal m=2 mode activity in recent SXR tomography data.

Thursday October 16 2008 10:37 pm hanson/shiraki 64848-64939

More Kalman filter data taken. More shots were taken for the triple probe profile scan.

Friday October 17 2008 4:38 pm hanson 64940-64988 More Kalman filter and Triple probe data taken today.

Tuesday October 21 2008 10:26 am DeBono 65004-65044 Bias Probe & External Mode Feedback

The purpose of these experiments is to spin a sufficiently large n=1 mode using the bias probe, and then to use the feedback system to reduce the size of this mode. We obtained a n=1 external mode with sufficiently usable amplitude last week and showed that the bias probe can be used to drive its rotation frequency negative. Todays experiments was to determine the repeatability of this n=1 mode created so that we can accuratly determine the effect that feedback will have on the mode.

Half the runday was hampered by pathological prefiring of the bias bank, however, the problem seems to have been fixed by moving the bias' banks digital delay generator outside the cap room.

10 shots were taken with the bias probe at a set voltage (voltage dialed on the power supply, not the actual voltage of the probe in the plasma) of ~1.5V, and 10 shots with the probe at a set voltage of 75V. Care was taken to make sure that the plasma parameters of each of the shots was similar. More analysis is needed to determine the shot-to-shot deviation of the n=1 mode amplitude, however, by eye it seems to vary considerably. More work to be done tommorow.

Tuesday October 21 2008 10:04 pm hanson/shiraki 65054-65119

More Kalman filter and Triple Probe data taken. Also, quick test showed that several tomography chords are blocked by the existing stainless shells in the hall probe section.

Wednesday October 22 2008 7:56 pm debono 65145-65206 Feedback & Bias Probe

Feedback experiments conducted on shots with the n=1 mode spun negative by the bias probe. Feedback shots were taken with the Kalman filter both on and off. The start panel was behaving a little erratically towards the end of the runday. Analaysis and more work to follow.

Thursday October 23 2008 5:06 pm hanson 65207-65263 kalman filter

More Kalman filter scan data taken today. There was a small problem with the OHB charging resistor. It had run out of water and was arcing. It was cleaned out and refilled, and then it functioned ok.

Monday October 27 2008 3:18 pm Shiraki/DeBono 65264-65337 Bias Probe / Triple Probe

Tests were done with the triple probe while the bias probe was inserted into the plasma edge (104.1 - 106 cm) and energized.

Tuesday October 28 2008 6:27 pm hanson 65340-65393 Kalman filter

A phase scan with an optimized Kalman filter was started today, following a quick gain scan to determine a satisfactory gain for suppressive feedback. There was some trouble with the serial highway driver. This was fixed by restarting both the serial highway driver and matterhorn.

Wednesday October 29 2008 6:04 pm hanson 65394-65462 kalman filter

More Kalman filter phase scan data taken today.

Thursday October 30 2008 9:36 pm hanson 65463-65543

More Kalman filter data taken today.

Friday October 31 2008 6:05 pm hanson 65544-65605 kalman filter More Kalman filter phase scan data taken today.

Saturday November 1 2008 6:07 pm hanson 65606-65680 kalman filter

A third round of the Kalman filter phase scan was completed today. This will probably be a sufficient amount of data for comparison with earlier results.

Tuesday December 2 2008 3:56 pm shiraki 65749-65753

Tested the OH bias bank for the first time since the pipe burst. Everything seems to be working fine.

Tuesday December 9 2008 6:28 pm DeBono 65754-65781

I worked to reproduce shilov-style shots today as an attempt to increase the length of time the bias probe is able to perturb the plasma rotation (from the current pulse time of ~1msec). Most of the runday was spent doing cleanup shots from the last glow discharge.

Wednesday December 10 2008 7:17 pm DeBono 65782-65823 Bias Probe

More work done today on extending the length of time the bias probe is able to drive n=1 toroidal mode rotation.

Friday December 12 2008 5:59 pm DeBono 65824-65854 Bias Probe

More work was done today on improving the length of time that the bias probe can influence the n=1 plasma mode rotation. It was found that the depth of bias probe insertion into the plasma has a significant effect on this time length. By inserting the bias probe into the plasma to the maximum permitted by the bellows setup (outer edge 104.4, inner edge 102.5 cm) it was seen that the bias probe was able to influence the n=1 mode rotation for ~ 3msec. This is an improvement over previous runs, where the bias probe was only able to influence the plasma for ~1msec. Bryan DeBono s

Tuesday December 16 2008 6:02 pm debono 65862-65881 Bias Probe

More bias probe work done today. A crate error occured at the end of the runday while taking a shot; this prevented the firing of the banks, which were then dumped. DeBono

Wednesday December 17 2008 6:56 pm DeBono 65883-65924 Bias Probe

More bias probe work done. The CTX Uport was giving problems, but we were able to run using the Uport's bypass switch. It proved difficult to obtain repeatable shots at the beginning of the runday, but this cleared up somewhat by the afternoon. n=1 mode amplitude was lower today than in previous runs. DeBono

Thursday December 18 2008 8:39 pm shiraki 65925-65956

The Hall probe was inserted 2.5cm into the plasma. After 17 shots, the probe appears fine. This data has not yet been analyzed. Instead, the majority of the day was spent repairing the SMP's... Channels 13, 15, and 16 were repaired by swapping waveform analyzers and 8100 modules in the North rack. However, Channel 3 went out at the end of the day; the reason is still unknown

Friday December 19 2008 5:49 pm Levesque 65957-65974 Repairing SMPs

Repaired SMP channels 2 and 3 by swapping amplifiers, waveform digitizers, and memory modules in the North rack. Now it appears that the only non-working SMP is channel 14.

2009

Thursday January 8 2009 7:12 pm DeBono 65975-66020 Bias Probe

More Bias Probe work done today. Probe was inserted to 102.5-104.4cm and activated to voltages between 75 volts and 200, in 25V increments, in order to characterize the effect of biased voltage on the probe.

Friday January 23 2009 5:58 pm DeBono 66049-66074 Bias Probe

More work was done with the intention of creating a strong (3,1) external kink with the bias probe fully inserted into the plasma.

Plasmas were created with a current rise of ~6 MA/sec by using a combination of strong OH and a very short (1.35 msec) Deuterium puff time. This plasmas proved to be quite unstable, subject to sudden minor disruptions.

Monday January 26 2009 5:42 pm DeBono 66075-66094 Bias Probe

More mode development work done today.

Wednesday January 28 2009 7:13 pm DeBono 66095-66131 Bias probe

Plasmas were biased at 100V and 200V with the probe fully inserted (inner edge 102.5cm, outer edge 104.4cm). More mode development work done.

Thursday January 29 2009 8:28 pm Debono 66132-66159 Bias Probe

I had been advised by Daisuke and Jeff that the SMPs had been withdrawn in the bias probe shots taken 2/28/2009. Therefore more biased shots were taken today with the SMPs re-inserted for use in mode analysis.

Thursday February 5 2009 3:17 pm DeBono 66160-66190 Bias Probe

CAMAC and matterhorn proved rather testy today, which slowed down the runday. Bias Probe shots taken at 250 Volts.

Upon consultation with Jim we will begin regening the cryos tommorow. Minimum base pressure is ~1.0E-08, a factor of 10 higher than what was obtainable a couple of months ago.

Wednesday February 11 2009 7:04 pm DeBono 66191-66221 Bias Probe

The goal of the run was to examine the current profile of the strong, bias-probe induced (3,1) external modes we have been obtaining. The Hall probe was inserted into the plasma edge (innermost channel @104.5cm). The bias probe was at (102.5,104.4cm) for all shots. Shots were taken with the bias probe @0V and @250V.

Friday February 13 2009 1:02 am DeBono 62222

One shot was taken today to prepare the tokamak for a deuterium light show for some truant NY times reporter.

Upon consultation with Daisuke yesterdays run will have to be repeated due to the Hall Probe not being calibrated. The second cryo should also be regened, will try to accomplish this tommorow. More crate problems today...

Wednesday February 18 2009 12:16 am shiraki 66243-66259

Hall Probe was recalibrated. Shots were taken before and after to confirm results of calibration.

Wednesday February 18 2009 5:29 pm DeBono 66265-66273 Bias Probe / Hall Probe

Shots taken with Hall probe inserted into outer edge of plasma.

Wednesday February 18 2009 5:29 pm DeBono 66265-66273 Bias Probe / Hall Probe

Shots taken with Hall probe inserted into outer edge of plasma. More to follow.

Friday February 20 2009 5:57 pm DeBono 66287-66318 Bias Probe/ Hall Probe/ Mach Probe

Biased edge shots taken with the hall probe and the mach probe inserted to 104.5cm (bias probe at 102.5-104.4). Their was a problem with the D gas puff towards the end of the runday, perhaps the D bottle is empty.

Monday February 23 2009 6:17 pm Levesque/DeBono 66319-66354 Puff Valve

We fixed the problem with the gas puff valve today. The problem was a loose wire.

Wednesday February 25 2009 7:52 pm shiraki 66356-66372

The puff valve circuit failed again today. The problem turned out to be a combination of a blown transistor and a loose ground. The circuit was repaired and seems to be working again.

Friday February 27 2009 2:58 pm shiraki 66373-66391

Tried to find a good shot for n=2 feedback experiments, similar to the ones from the Kalman filter experiments. Also, a problem with the analog boards was fixed, by replacing a broken op-amp dating back to December '08.

Tuesday March 3 2009 6:26 pm shiraki 66465-66538

Continued the n=2 phase scan. Results so far don't show any significant trends, so the plan now is to do more tests with the FPGA emulator.

Wednesday March 4 2009 5:08 pm DeBono 66538-66572 Bias Probe

The days run was severly hampered by crate errors from the west rack. At first I thought the error was due to a defective J222 channel, however the J222 channel tests out fine and the problem ended up resolving itself by late afternoon.

Tuesday March 10 2009 10:45 pm DeBono 66611-66638 Bias Probe

Q scan done, with 3 <= edge q <= 4. Shots were taken with both the bias probe activated to 250V and to 0V.

Significant (tearing?) mode activity was observed with edge q>3, limiting plasma duration.

Wednesday March 11 2009 6:28 pm shiraki 66639-66645

Verified that the corrected n=2 feedback algorithm works properly. The Kalman filter with increased rotation rate has yet to be implemented.

Monday March 30 2009 7:04 pm shiraki 66646-66684

At the beginning of the runday, the oil on the machine was extremely low. There was one leak at the entrance to one of the TF coils, which was tightened and cleaned up... Experiments to measure the effect of mismatched sensor coil sensitivities were performed. VF only shots at the end of the day suggest there may be significant miscalibration or other source of error.

Thursday April 2 2009 5:48 pm DeBono 666688-66710 Bias Probe

More bias probe q scan work done today.

Tuesday April 7 2009 6:31 pm shiraki 66722-66762

Made plasmas with 3/1 mode activity with and without static perturbations. At first I tried to create a base case with the bias probe inserted but turned off, but was unable to see a good mode. I then took shots with the probe retracted, but the mode doesn't appear to differ in an obvious way between having or not having the static perturbation. The sample size is small though (less than 10 shots each) so more data will be added tomorrow, included biased shots.

Wednesday April 8 2009 8:01 pm shiraki 66762-66802

Took more shots with naturally rotating modes, with and without the static perturbations. While yesterday's pessimistic analysis was based mostly on the rogowski signals, analysis of the partially integrated SMP's (B-dot) looked more promising. SMP signals when the perturbation was present seemed to have more "spiky" delta function like fluctuations. At the end of the run day, I blindly sorted through all of the available data (38 shots) and correctly predicted whether the static perturbation was there or not in 26/38 cases (68%). On shots I felt fairly confident on, I was 5/6.

Friday April 10 2009 4:43 pm shiraki 66802-66814

Applied some static perturbations to biased plasmas. Not too many shots taken today, but I'll continue next week.

Monday April 13 2009 5:24 pm shiraki 66815-66829

Applied static perturbations to more biased plasmas at full amplitude (before clipping). We now have data from about 20 biased plasmas (with and without perturbations), and analysis is underway.

2010

Thursday October 7 2010 4:45 pm shiraki 67411-67436 Copper Plasma

Completed the last of the copper plasma experiments, with the last of the Rogowski data taken. Disassembly will begin tomorrow. -DS

Tuesday November 30 2010 5:14 pm shiraki 67604-67610 TF test

We fired the TF today, beginning at 1kV and increasing 1kV at a time up to 6kV. The banks and magnets seem to be working well. A few small (~1mm) air bubbles were seen coming out of the tops of the magnets above 4kV, and a few turnbuckles worked themselves loose from the vibrations. These will be re-tightened before taking any further shots.

Wednesday December 1 2010 6:46 pm shiraki 67611-67654 OH test shots

A good portion of the day was spent resolving a problem with the OH coil, which turned out to be due to overheating of an amplifier in the basement. The remainder of the day was spent firing the OH to find tools below the machine. A few still remain, but we should be able to finish this tomorrow.

Thursday December 2 2010 4:26 pm shiraki 67655-67684 First Plasma!

We achieved our first plasma since the upgrade today. The plasma broke down with fill pressures of between 2.3 and 4.2E-5 Torr. Plasma currents were between 4 and 8kA, lasting up to 5ms. Loop voltages were between 10 and 15V. Currently we have no major radial information, but this data will be post-processed.

Friday December 3 2010 7:04 pm shiraki 67685-67717

We now have major radius data based on copper plasma and vacuum shot calibrations. The plasma usually breaks down around 93-94cm on these shots. Ip is typically 8-10kA, lasting for about 4ms. Loop voltages have come down to 8-9V. The plan next week is to continue taking shots while continuing to add more diagnostics to the machine.

Monday December 6 2010 8:33 pm shiraki 67718-67762

Plasma currents as high as 15kA were reached today. The major radial force balance seems to respond well to the bank settings, and is fairly stable. The first set of feedback sensors (4 poloidal and 4 radial at one toroidal location) were digitized today, and the signals look good despite temporary cabling and improper shielding. Some of the soft X-ray tomography channels appear to be working as well. Additional Fourier Rogowskis will be connected tomorrow, followed by optical diagnostics (fast camera and spectrometer).

Tuesday December 7 2010 7:06 pm Levesque 67763-67827 Shot development

Attempted to make long-lived plasmas with edge q between 2 and 3. Reached plasma currents as high as 18kA. Spectrometer has been implemented, looking at D_alpha light. Lots of D_alpha light is seen during the breakdown and disruption, with the disruption light peak being up to 4 times as intense as the breakdown. The rises in D_alpha light during disruptions correlate well with drops in SXR emission. Mode activity is seen in the feedback sensors and sin3T Rogowski coils when edge q is near rational numbers.

Wednesday December 8 2010 7:33 pm shiraki 67828-67882

Today's shots consistently reached 18kA of plasma current, with loop voltages coming down to 5 or 6V as the plasma heats up. Half (40 total) of the feedback sensors were digitized, including two complete toroidal groups. Biorthogonal decomposition of this data shows a rotating n=2 mode with about 15% of the mode energy. However some grounding problems still need to be worked out for the feedback sensor preamplifiers.

Thursday December 9 2010 7:45 pm Levesque/Angelini/Shiraki 67885-67920 Fast camera testing

Tested fast camera by viewing light through a midplane window port. High light levels were seen during breakdown and disruption. The highest tested frame rate was 24000 frames per second. No obvious current filament was observed during breakdown, but a higher frame rate may help.

Friday December 10 2010 7:45 pm Levesque 67921-67973 Shot development

Attempted to make long lived plasmas with higher edge q values. No significant improvements in plasma duration -- good plasmas still last ~5-6ms at the longest. Having the edge q between 2 and 3 has reproducibly given the longest lasting plasmas so far.

Monday December 13 2010 8:08 pm Levesque 67974-68024 Long shot, spectrometer tests

Attempted to make long-lived, low-current plasmas. Longest plasma durations were 11.5ms, though the plasma currents and major radii were very erratic. Currents for long plasmas peaked at 7.5kA and dropped to ~2kA during the shots. Spikes in d_alpha light were seen during minor disruptions for these long shots. Several spectrometer wavelengths were tested in order to verify the d_alpha wavelength setting and to look for an oxygen line.

Tuesday December 14 2010 7:23 pm Levesque 68025-68065

Long-lived, low-current plasmas. Record duration so far is 15ms with a peak current of 8kA. Long-duration shots achieved so far have produced little or no x-ray emission.

Wednesday December 15 2010 8:31 pm shiraki 68066-68135

More shot development. The longest pulse today was 20ms, with a slow ramp-down of Ip after a peak of 17kA. Bank timing was adjusted slightly, but has not yet produced any obvious effects.

Monday December 20 2010 10:52 pm Levesque 68136-68165 Bank timimg tests

Made large changes to timimg of OH and VF bank firing to investigate different operating regimes. Was able to make several plasmas without firing the OH start or OH electrolytic banks, where the plasma was formed using the OH bias bank and maintained for 5ms using the two VF banks. Was further able to drive this type of plasma by firing the OH start and OH electrolytic banks at 1.5ms into the plasma shot to raise the plasma current to ~13kA. Interesting mode activity was seen throughout these shots.

2011

Wednesday January 26 2011 5:15 pm shiraki 68177-68186

The full set of feedback sensors was tested with the new breakout board. There are still some cabling/grounding issues, which we will begin to resolve tomorrow.

Tuesday February 1 2011 5:45 pm shiraki 68187-68200

Continued debugging of the x10 feedback preamps. Grounding the boards to the rack helps, but the problems remain when more than 3 boxes are connected.

Wednesday February 2 2011 5:28 pm shiraki 68201-68214

All 10 feedback preamp boxes are now working properly, after moving the power supply to minimize the loop voltage on the DC power lines.

Friday February 18 2011 4:58 pm shiraki 68217-68248

The first half of the high density arrays was tested successfully. A few amplifier channels need repair, but signals look good on both poloidal and toroidal arrays. Biorthogonal decomposition of poloidal arrays show mode numbers ranging from m=2 to 7.

Monday February 21 2011 5:25 pm shiraki 68249-68280

Continued testing of the high density array. One issue so far is low signal levels on high-field side sensors on outboard limited plasmas, particularly for high m numbers.

Thursday February 24 2011 7:07 pm Levesque 68309-68331 Feedback sensor tests

Developed low breakdown current shots to prevent feedback sensor clipping during breakdown. Low initial current shots can be used to test equilibrium reconstructions before FB sensor amp boards are modified to prevent clipping during normal shots. Three FB amp boards have been modified so far, and do not clip during the breakdown or disruption for typical shots. The remaining 7 FB boards can be modified in the next few days.

Monday March 14 2011 5:41 pm Levesque 68488-68498 Testing high-resolution sensors

Tested second high-resolution sensor amplifier set. All poloidal sensors and amplifiers are currently working, except for 1 in the toroidal array. Radial sensor data was taken, but has not yet been checked.

Wednesday March 16 2011 6:45 pm Levesque 68521-68535 Fast camera

Tested fast camera mounted 2-3 feet above the Thomson scattering viewport. The camera was able to withstand the high fields without movement or noticeable video distortion. Frame rates were ~47000 fps. Videos generally show if the plasma is far outboard or inboard limited, but there is no obvious emission centroid or clearly defined edge. Fluctuations in light can be seen, but have not yet been compared to magnetic fluctuations. A D_alpha filter was mounted in front of the camera lens for a few shots. Upon first inspection, little difference was seen between the visible light and D_alpha-only videos, other than a large reduction in light intensity due to the small diameter of the filter.

Several magnetic sensor amps are currently not working; trouble-shooting is in progress.

Monday April 4 2011 6:35 pm Levesque/Shiraki 68536-68549 Vacuum field pickup shots

Ran vacuum shots to measure pickup on toroidal array sensors. Various combinations of VF and OH bank settings were used. There are discrepancies (toroidal asymmetries) in vacuum fields as seen by toroidal array sensors. We haven't yet come up with an appropriate explanation for this, but are working on it.

Tuesday April 5 2011 7:07 pm shiraki 68550-68576 Control coil force test

We energized a set of control coils in the presence of the toroidal field, while viewing the coil with the fast camera through the Thomson scattering collection port. Currents were gradually increased from 2A to 36A. Unfortunately, the fast camera view is in the radial direction, as is the jxB motion of the control coil wires. However, motion of the wires is easy to detect when the TF is on, while being almost undetectable when there is no TF, even at full current. There may also be slight flexing of the control coil side of the shell, which may be because of reduced rigidity of the shell due to the Thomson cutout.

Repeating these tests with the fast camera at the bifurcating port should show radial motion of the wires more clearly.

Thursday April 7 2011 11:43 pm Levesque 68600-68612 Vacuum shots

Ran vacuum shots to measure pickup on toroidal array sensors. Tried to isolate possible problems with amplifier boards or sensor mapping. No problems have been found so far.

Friday April 8 2011 4:52 pm Rath 68613-68621 Equilibrium Test

Made some plasma shots with a new tree model that includes magnetic sensor position information. Attempted to do automatic equilibrium reconstruction, but weren't able to do so because of missing polarity information.

Also had trouble getting plasma breakdown at normal puff times, but longer puff times didn't show any problems. We may have problems with the puff circuit or the gas may be running out.

Monday April 11 2011 5:30 pm Levesque 68624-68630 Vacuum shots

Ran vacuum OH-Bias-only shots to measure pickup on toroidal array sensors. Discrepancies in vacuum fields as seen by the high-resolution sensors appear to be mostly caused by pickup along the ribbon cables in the copper bellows between the tokamak and the amplifier boxes. Individual sensor polarities together with the ribbon cable groupings are consistent with anomalies in all toroidal array data taken so far. The fields and pickup areas along the ribbon cables are comparable to the fields and areas of the high-resolution sensors.

Tested this idea by twisting one set of 3 ribbon cables within the bellows (Bottom West set) without disconnecting the bellows. This changed the vacuum pickup by a significant amount. Pickup was reproducible in subsequent shots. Then twisted the same ribbon cables within the bellows further, as much as reasonable without disconnecting the bellows. Again pickup changed, then was reproducible in subsequent shots.

The resolution for this problem will be to replace these standard ribbon cables with twisted-pair ribbon cables.

Monday April 18 2011 6:42 pm Levesque 68673-68686 Vacuum shots

Ran OH and VF vacuum shots to measure pickup on toroidal array sensors. Took shots before and after replacing one of the ribbon cables running from the tokamak to the amplifier box with a twisted pair cable. Twisted pair bundle cable was run directly from the conflat feedthrough to the amplifier, through the front door of the amplifier box (not through the bellows). Non-sensor pickup was dramatically reduced. All sensors in that bundle now have the same apparent polarity for vacuum shots, versus having 2 distinct sets of pickup. Spreads in nominally axisymmetric poloidal field signals are now ~40% while firing the OH coil alone, and ~10% while firing the VF coil alone. Both of these are down from spreads of over 100%. More testing should be done with improved shielding.

Monday April 25 2011 10:08 pm shiraki 68750-68805 Static phase flips

Today we began static phase flip experiments using the FPGA. The plasmas were standard current ramps with edge q starting at around 4 and falling to slightly below 3. The applied control coil currents had an amplitude of 20A, with resonant 3/1 helicity. Timing of the phase flip was chosen to occur after the q fell below 3 and initial fluctuations were diminished. Initial biorthogonal decomposition analysis of the full magnetics data did not show conclusive results, but further analysis and experimenting will continue tomorrow.

Tuesday April 26 2011 5:48 pm shiraki 68826-68835 phase flip experiments

More data was collected attempting to compare similar current ramp discharges with and without static phase flips in order to extract the plasma response to the external field. Pertubations with both 3/1 and -3/1 helicities were applied. Comparisons so far have been limited by plasma reproducibility and small sample size. This dataset will be expanded in the coming days.

Tuesday April 26 2011 11:59 pm Levesque 68806-68825 Vacuum pickup shots

Ran OH and VF vacuum shots to measure pickup along toroidal array sensor cables. Tested two cables between the tokamak and the amplifiers: a twisted pair bundle cable and a twisted pair ribbon cable. Pickup along the cable was isolated by connecting each to a shorting d-sub connector with no effective NA, instead of the sensor feedthrough. Also tested pickup along the output ribbon cable between the amplifiers and the west rack.

The bundled pair cable and the twisted ribbon cable had similar pickup for vacuum shots. Pickup along these cables does not appear large enough to account for differences in vacuum shots with the sensors connected, suggesting that there is more pickup at the feedthrough or elsewhere within the chamber. Pickup along the output ribbon cable was measureable, but was not enough to account for differences in signals.

Wednesday April 27 2011 7:51 pm shiraki 68836-68858 phase flips

A systematic error was found in the wiring of the power amps, meaning that previously applied fields were not actually resonant perturbations. This was corrected, and the new wiring was confirmed by using the control coils as sensors and detecting the helicity of rotating plasma fluctuations. Static 3/1 phase flips were then applied to plasmas with edge q just under 3, and initial measurements suggest there may be a measurable plasma response, but further analysis is necessary.

Thursday April 28 2011 6:23 pm shiraki 68859-68877

The timing and length of the applied phase flip perturbations were adjusted today, varying from 1ms (0.5ms up and 0.5ms down) up to 2ms (1ms up and 1ms down). In addition to a possible response to the phase flip, there may be some interaction with the pre-flip RMP, i.e. a static response. However, the primary limitation to extracting the plasma response has been discharge reproducibility, so an effort was made at the end of the day to develop a more reproducible marginally stable plasma.

Friday April 29 2011 4:03 pm shiraki 68880-68898

More data was collected with 3/1 static phase flips on similar plasmas as earlier this week. An attempt was also made to apply phase flips on discharges with more of a multimode spectrum. The FPGA and power amp system appear to be working well, but the challenge is still in properly deducing the plasma response.

Tuesday May 3 2011 8:29 pm Levesque 68934-68957 Phase flips

Applied static 3/1 phase flips while attempting an edge q scan. Plasmas were more erratic and non-reproducible than those on April 28; it was difficult to apply the phase flip with the edge q near 3. Also tried applying fields at half the amplitude as part of an RMP amplitude scan. Analysis still needs to be done for the usable plasmas, however lack of reproducibilty should be considered when analyzing these plasmas.

Wednesday May 4 2011 10:25 pm Levesque/Shiraki 68960-68992 Retracted shells, phase flips

Retracted shells 2cm from their fully-inserted positions to try to observe different mode characteristics. Observed 4/1 modes that were more prominent than usual when the edge q was above 3.

Plasmas with large initial currents (~10kA) had difficultly getting below edge q=3. Plasmas with lower initial currents (~9.3kA) and stronger current ramps usually made it through edge q=3 with no problem, including during phase flips. Phase flips were done with helicity -3/1. More shot development is needed to get plasmas with steady q values near 3.

Also did 2 separate phase-flip periods for 3 plasmas (shells still retracted): on for 0.5ms, flipped for 0.5ms, off for 1ms, on for 0.5ms, flipped for 0.5ms. The 3/1 mode typically locked during the first CC on+flip stage, then went away or started rotating when the control coils were off. One plasma disrupted during the second phase flip.

Friday May 6 2011 9:14 am Levesque 68997-69006 Shot development, multimode

Yesterday's run focused on getting a reproducible shot with edge q near 3 for normal operating conditions. This shot type will be used for a q scan at different applied RMP helicities. The shells were fully inserted prior to the run. Plasmas formed around 92.8cm at the end of the start phase, versus 93.3cm for the shells retracted 2cm with the same start settings.

Evidence for multimode activity was seen, especially for shot 68998. BD analysis shows a 6/2 and a 7/2 mode in addition to the 3/1 and 4/1 mode as the edge q crosses 3 from above. This and neighboring shots were similar to shot 68498, which had been our best multimode shot to date.

FPGA code for applying a rotating perturbation was in development, so RMPs could not be applied to these plasmas.

Friday May 6 2011 6:41 pm levesque/shiraki 69007-69014

Some shots were taken today to try to look at naturally present multimode fluctuations, without RMP's. Initial attempts to apply rotating RMP's suffered from lack of synchronization between the five FPGA's, but this issue is being addressed. Plasma experiments with rotating perturbations will begin as soon as possible, possibly this weekend.

Monday May 9 2011 5:59 pm levesque/shriaki 69027-69057 Static phase flips

Due to continuing problems with implementing the rotating RMP's, more static phase flip experiments were conducted today. In particular, a detailed amplitude scan was performed. For control coil currents less than ~12A, the 3/1 static response scales linearly with the perturbation, with a slope of about 2Gauss/Ampere. Between ~12 and 22A, the response appears to saturate at about 3.5G. Above ~22A the plasma seems to always suffer a minor disruption just after the perturbation is turned off. There was some shot-to-shot variability of major radius and edge q, but this was not considered in the analysis so far.

Finally, several vacuum shots were taken to ensure that direct sensor-coil pickup does not affect the response calculation.

Tuesday May 10 2011 7:46 pm Levesque/Shiraki 69058-69100 -6/2 RMP amplitude scan

Applied -6/2 static phase flips to attempt an RMP amplitude scan. At first we saw a linear trend in the n=2 plasma response as measured by the feedback sensors, but later vacuum shots showed that this was mostly/entirely due to direct pickup. Direct pickup for the 6/2 helicity is much larger than for the 3/1 helicity (by an order of magnitude), most likely due to the relative phasing of control coils near each sensor. Toroidal array data were taken for these shots, but have not yet been extensively analyzed.

There was no observed trend in the 3/1 response with changing 6/2 applied amplitude. After stopping the 6/2 scan due to the direct pickup, we continued the -3/1 phase flip experiments and started a q scan. Analysis for the q scan has yet to be done, but only a few shots have been taken so far. We also applied one -4/1 and one -2/1 phase flip (separate shots) for the same equilibrium characteristics. The plasma suffered a minor disruption after the -2/1 perturbation was turned off, though the edge q was close to 2.9. Again, analysis is backlogged for different applied helicities.

Wednesday May 11 2011 11:30 pm Levesque/Shiraki 69101-69127 RMP q scan

Applied -3/1 static RMP phase flips during a q scan of plasmas similar to those from this week's amplitude scans. Phase flips were always applied at 2.5ms to try to minimize the variation in equilibrium plasma properties. So far we have seen the RFA increase as the "mean-edge-q" increases from ~2.6 to ~3, where mean-edge-q is defined as the average q over the period of control coil activation (2ms-3ms). More shots are needed with mean-edge-q above 3 in order to hopefully observe a peak in RFA near edge q=3.

Much of the q variation done so far has been due to ~5mm variation in major and consequently minor radii, which may influence coupling to the control coils and feedback sensors. Systematic reduction of plasma current to raise the edge q with the same major radius was in progress at the end of the run day, and will continue in the morning.

Thursday May 12 2011 6:20 pm Levesque 69128-69153 RMP q scan

Continued edge q scan during -3/1 static RMP phase flips. Took shots with mean-edge-q around and above 3 to extend the range of q's in the scan. The results of the q scan show that the RFA amplitude is largest when the mean-edge-q is between 2.9 and 3. The range of mean-edge-q's in the scan was 2.65 to 3.25. In each shot, the edge q decreased significantly during the 1ms window of control coil activity (e.g. from 3.2 at 2ms to 2.8 at 3ms), which probably spreads out the peak of calculated RFA amplitude. A future q scan should be done during a time window where the edge q is steady for various plasma currents in order to get better RFA peaking.

Friday May 13 2011 6:50 pm DeBono 69163-69173 Bias Probe

Bias probe was run @75Volts into electron saturation current.

The CPCI's suffered a digitization lag of ~150ms. The A14's were not affected. Problem unresolved, will repeat efforts on Monday. Bryan

Sunday May 15 2011 11:55 pm DeBono 69174-69189 Bias Probe

The purpose of this run was to determine the correct polarity of the bias probe voltage and bias probe current signals. It appears the bias probe V,I signals are inverted from their correct polarity by the South Rack A14. Crates only, no bank shots.

Monday May 16 2011 6:44 pm DeBono 69198-69214 Bias Probe

Shots were plagued by same ~150ms CPCI latency as last Friday. However, a couple somewhat useful shots were taken.

Bias probe triggered sucessfully at desired time/voltage. It was necessary to swap the 1V/10A Rogowski for a 1V/100A Rogowski due to high currents collected by probe (For Vprobe = +100V, Iprobe was > 50 Amps).

More data to be taken once CPCI latency problem is solved.

Tuesday May 17 2011 9:39 pm DeBono 69231-69262 Bias Probe

The day's run was made more difficult by the frequent appearance of Store_Action NoX errors in the A14's, which occured primarily in the North Rack but also occured in the Basement, South, Screenroom and West Racks. I suspect that these NoX errors are related to thermal effects in the basement crate: opening both basement crate doors and blowing the high-power fan on the rack seemed to help (although NoX errors continued to occur throughout the runday).

Shots were taken @+25V,+40V, +50,+100,+150 with the bias probe fully inserted. As was seen before, large positive biasing resulted in the rotation frequency of the m=2, m=3 modes to reverse direction from the natural, positive rotation frequency.

Thursday May 19 2011 12:16 pm DeBono 69266-69292 Bias Probe

More work done on using bias probe to slow down mode rotation near 0, as in shot ^#69265.

Probe voltage was set to floating, 0, +30, +35V for these shots. More shot development needs to be done here to achieve repeatable, large, near 0 rotation modes.

Also bias probe was set to high V values to spin the plasma. Probe voltages set to 200, 250, 300V.

Friday May 20 2011 7:26 pm DeBono 69293-69314 Bias Probe

The goal of the run was to use the bias probe to slow down the natural plasma mode rotation and achieve a repeatable, non-rotating mode. First, the radial position of the probe was experimented with. The probe was biased to 30V and moved so that the probe tip was placed [1.234, 2.234, 3.234]cm inside the plasma. The probe placed at 3.234cm (maximum probe penetration allowed by the bias probe bellows) seemed to have the best results. At the end of the runday, we adjusted the shot to straddle the q=3 surface in order to achieve a stronger mode. Increasing the VFEL to push the plasma inwards raised the edge q closer to 3 and also increased the plasma current ramp. This resulted in a significantly stronger mode and some encouraging shots towards the end of the runday.

Friday May 20 2011 9:47 pm DeBono 69315-69331 Bias Probe

The purpose of the run today was to use the bias probe to slow down the natural mode rotation. This run built upon the promising shots taken at the end of the runday yesterday.

The B.D. applied to the shots taken today with a floating, fully inserted bias probe showed the natural mode rotation around 8-9kHz. This is consistent with natural mode rotations observed earlier (see shot ^#69142 as an example). The probe was used to slow down mode rotation below 4kHz; possibly as low as 2kHz for a period of 1.5msec after probe activation. Shot-to-shot major r, q, plasma current traces show good repeatability. Shot-to-shot repeatability of the MHD mode rotation is questionable, although all shots show reduced mode rotation compared to the natural mode rotation.

Wednesday May 25 2011 7:34 pm shiraki 69333-69363 static phase flips on biased plasmas

The static response to a m/n=-3/1 static phase flip was compared for plasmas with and without biasing. The probe was inserted inside the plasma in both cases. When turned on, the probe was biased to ~50V, drawing 35-40A. This results in about a factor of three increase in the measured plasma response over non-biased shots. In tomorrow's run we will continue to scan over probe voltage/current.

Thursday May 26 2011 6:21 pm shiraki 69364-69411

The phase flip response as a function of bias probe current was scanned, with bias voltages between 0 and 80V. The plasmas seemed to disrupt more often when the probe was turned on, even when no RMP was present. In cases where the plasma does not disrupt, the plasma response was often delayed relative to the perturbation, making the previously used correlation analysis difficult, even when the raw signals show a clear response. This scan will be continued tomorrow.

Tuesday May 31 2011 3:01 pm DeBono,_Shiraki 69412-69452 Shot Development

The purpose of the run today was to do shot development to optimize the shot for RMP response experiments with edge-biased plasmas. We attempted to optimize the shot for an edge q=3 and also to remove shot-to-shot irrepeatability.

Shots taken throughout the runday suffered from shot-to-shot irrepeatability. In addition, the bias probe suffered from pretriggering problems; instead of firing at 2msec as was programmed, the bias probe would pretrigger around 1.0-1.2msec. This problem appeared pre-HBT upgrade, but has not been a problem post-upgrade until this run.

Tuesday June 7 2011 11:38 pm DeBono 69453-69522 Bias Probe

This run report covers 3 days of running.

The bias probe pre-triggering problem appears to be fixed. It was found that the OHST bank caused the 111A pulse generator to prefire and activate the bias bank SCR prematurely. The problem was fixed by cheating the 111A's power supply.

The rundays were plagued by store_action no-x errors; >9 no-x errors appeared every shot, and sometimes as many as 12 errors per shot. The basement A14_07 was swapped with the west rack A14_15 in an attempt to fix this problem, but did not succeed although perhaps reduced the frequency of basement no-x errors. The lack of a functioning cap room AC could be a culprit: Facilities work order ^#703420 was placed to fix the AC.

Friday June 10 2011 7:03 pm Levesque 69523-69536 Vacuum pickup tests

Testing vacuum pickup along a new twisted pair ribbon cable which will be used to replace the existing high-resolution sensor cable runs between the tokamak and amplifier boxes. Like the run on April 26, Pickup along the cable was isolated by running the cable from the amp to a shorting d-sub connector with no effective NA, instead of to the sensor feedthrough. Two of the three flat sections along the cable (40" twist, 3" flat, repeat) were manually twisted using the technique we decided upon. When the non-twisted flat section was near the amplifier box, pickup along that flat was small but still measureable (~0.3 Gauss with typical bank settings) though it was 4 feet away from the basepad. This suggests that we should manually twist all flat sections of the cable, including the sections farthest from the tokamak.

Pickups along the rest of the cable and shorted Dsub were ~1-2 Gauss for both VF and OH coils. This is down from ~60 Gauss pickup along the existing flat ribbon cables and feedthrough. When the twisted cable was reversed in order to place the flat part ~5 inches away from the outer VF coil, pickup was ~10 Gauss from the VF. These quoted field strengths are the spurious fields that would be inferred in the poloidal field sensor measurements. More tests may be done next week if deemed necessary.

Thursday June 16 2011 6:28 pm Byrne 69539-69544 Vacuum Pickup Test

A twisted pair ribbon cable (the same as Jeff is using, but unmodified) was hung as it will be when the GPU is connected to the amplifier rack and the VF and OH banks, separately and together, were fired.

The ribbon was shorted at one end by twisting to eliminate NA, and the other end was routed into the North Rack A-14's. Recorded pickup was very low, and comparable to that seen in other channels.

Finally, the other end was unshorted and connected to the Bias Probe trigger, which as a 5V signal is comparable to the signal to be output by the GPU. It was unappreciably affected by the pickup. This confirmed that the small signal was indeed the pickup in the cable. Al shielding is expected to reduce this effect further. Interference on GPU->Crown Amp signals is not expected to be an issue.

Brian Debono checked the basement rack temperature midway through the run, found outside temp to be 27 C, and 19 C inside.

Thursday June 16 2011 6:52 pm Levesque/Byrne 69539-69542 Vacuum pickup shots

Ribbon cables from the feedthrough to the amplifiers for 1/4 of the high-resolution sensors were replaced with twisted pair ribbon cables. Piggy-backed on Pat's vacuum pickup shots to measure fields from the VF and OH in the TA sensors. Updated TA sensors showed much better agreement with each other, versus the wild disagreement when using the flat ribbon cable. More-detailed comparisons will be done after all flat ribbon cables have been replaced.

Monday June 27 2011 7:41 pm Levesque 69545-69551 Vacuum pickup shots

Ribbon cables from the feedthrough to the amplifiers for all of the high-resolution sensors have now been replaced with twisted pair ribbon cables. Ran vacuum pickup shots to measure fields from the VF, OH, and TF banks separately. All high-res sensors now show much better agreement with each other, except for 2 of the PA sensors and 1 of the TA sensors. Discrepancies with these 3 sensors may be due to amplifier problems; this will be investigated.

Two plasma shots were taken (69550 and 69551). Although these two shots were not good-quality plasmas, there should be enough agreement among the high-res sensors for equilibrium reconstructions. Differences in equilibrium fields seen by the axisymmetric TA sensor sets are comparable to the differences in vacuum shots, so subtracting the vacuum fields should help in reconstructions. Paul and Jim will start adding flux loops to the machine tomorrow.

Thursday June 30 2011 7:41 pm shiraki 69552-69587

Two of the new flux loops have been mounted on the machine and were tested today. Raw signal levels were around 1.5V at breakdown and about 200mV during the rest of the discharge. These signals were digitized for all of today's plasma shots. A few shots with a 3/1 phase-flip were taken today, but the plasmas may have been somewhat different from those in previous experiments such as the amplitude scan.

Friday July 1 2011 9:03 pm shiraki 69589-69644 phase-flip amplitude scan

With appropriate bank settings, today's plasmas were close to the target plasma from the last 3/1 phase-flip amplitude scan. However, the "saturation regime" from ~15-30A of control coil current was NOT observed. Instead, plasma response increased linearly with coil current (with the same slope as before), up until about 25A, where the plasma usually disrupted after the perturbation was turned off. In the 20-30A range, there is a noticeable separation between the two datasets (larger than the spread of each).

The shots that resulted in disruptions have not been analyzed yet.

Tuesday July 5 2011 7:47 pm shiraki 69645-69686 phase-flip amplitude scan

More shots were taken as part of the phase-flip amplitude scan, with an effort to consider the variation in edge q from shot to shot. The spread in the plasma response measurement was larger today, for still unknown reasons. The cumulative data so far may be consistent with the theory that the plasma response saturation only occurs for edge q near 3 (~2.85-3.0), while only a linear regime is observed for plasmas with edge q of 2.6-2.85. More data will be taken in the coming days to test this theory.

Wednesday July 6 2011 9:03 pm Levesque/shiraki 69687-69739 Phase-flip amplitude scan

Continued RMP amplitude scan using -3/1 phase flips. Additional data from today is consistent with observations from the previous runs. Only a linear plasma response is observed for plasmas with 2.7<q_*<2.85. Linear plus saturated regimes are observed for plasmas with 2.85<q_*<3.0. The magnitude of the 3/1 plasma response for linear regime shots is consistent with earlier q-scan results.

Thursday July 7 2011 7:33 pm Levesque/Shiraki 69740-69789 Phase-flip helicity scan

Performed a phase-flip helicity scan with n=1 applied fields. The applied m numbers varied between -5.0 and +5.5. The maximum control coil current in each case was ~10A, which is in the linear regime of the recent RMP amplitude scans. Clear peaking in 3/1 static response was measured for m numbers around -2.5 and +4.0. Note that previous RMP amplitude and q scans were done using m=-3. In addition to the m-scan, Bryan took a couple shots at the end of the run day to test the bias probe with a 16-ohm resistor added to its circuit to prevent pre-trigger bias currents.

Monday July 11 2011 8:21 pm shiraki 69790-69829 n=2 phase-flip helicity scan

n=2 phase-flips were applied using 'm' numbers ranging from -7 to +6. All phase-flips had the same amplitude of ~40A. The measured -6/2 response peaks near m=-6 and m=+1, as expected from the corresponding peak in the -6/2 component of the applied field. In addition there may be another unexpected peak at m=+5. However, unlike the n=1 phase-flips, the measured response is not always well characterized by fitting to the passively measured "natural (unstable) mode" structure. Also, while the peaks in the response are higher than the direct vacuum pickup, the pickup may still represent a significant portion of the signal and may require more careful subtraction.

Wednesday July 13 2011 6:47 pm DeBono 69830-69841 Bias Probe

The bias probe was attached to the Crown amp biasing setup today: the setup can place an 85V, 1kHz sine wave on the probe. Plasmas will be taken tomorrow; flux loop installation prevented machine running today.

Thursday July 14 2011 9:59 pm debono 69842-69855 bias probe

The bias probe voltage divider was rebuilt today following an arc. The negative bias supply was re-tested: a 109V, 1Khz sine wave pulse (1 full wavelength) was successfully placed on the probe.

Friday July 15 2011 8:13 pm debono 69856-69873 bias probe

The crown bias voltage supply was tested for the first time today. Shots were taken at 20,40,60,80V sine wave pulses @1kHz. Large pickup voltages were present on the shots with plasmas; the crown amp will be moved either into the south rack or outside the machine room to prevent these pickup voltages.

Monday July 25 2011 5:24 pm shiraki/levesque 69874-69920 Clean-up shots

The bake was turned off yesterday, and clean-up shots were taken today. The discharges improved over the course of the day, lasting up to 5ms with 15kA of plasma current. More clean-up shots will be taken tomorrow if necessary, followed by the high-current (~25A) phase-flip q scan.

Tuesday July 26 2011 10:53 pm shiraki/levesque 69921-70000 high current phase-flip q scan

More clean-up shots were taken in the morning until plasma reproducibility greatly improved. Then the phase-flip q scan was begun with ~25A of control coil current, scanning between (time-averaged) q of 2.6 and 3.1 during the phase-flip.

The initial results are again consistent with a linear response at lower q (q<2.8), and a saturation for plasmas near q=3. It is notable that the installation of the Hall probe and following bake may have changed vacuum and plasma conditions in the time between this and the low (~10A) current q scan.

Wednesday August 3 2011 9:23 am Byrne 70010-10011 Hall Probe Calibration

Two shots were taken with the TF bank set to 500V. Banks fired at this low voltage, but CPCI 12, to which the probe was attached, did not acquire any samples.

Probe was moved to CPCI 10, but then changes to the setup of the probe (probe standoff/leveling cantilever) were noticed, and the run was scrubbed until Thursday so that the changes could be investigated and understood.

Thursday August 4 2011 6:10 pm Paul_Hughes 70026-70027 Flux Loop Calibration

Tested performance of prototype flux loop preprocessing circuit. At full OH, signal saturated. Dividing factor will be increased, and the circuit will be re-tested.

Friday August 5 2011 9:44 am Byrne 70023-70037 TF Only Hall Probe Calibration

Probe was connected and shots were taken with TF @ 500V. Preliminary results look as expected, although data hasn't been fully analyzed. One probe was dead, after invesitgation was found to be a broken lead wire. Wire was repaired but probe now shows an offset of ~70% of signal. Will be investigated 8/5.

TF probe was also scanned along angle to find maximum (Phi direction) and minimum (Theta direction) of signal. Mor work remains to be done however.

Friday August 5 2011 2:30 pm Paul_Hughes 70038-70043 Flux Loop Calibration

Shot 70038 lost to powered-off basement rack. Run actually starts at 70039.

Ran max-setting OH and VF vacuum shots to check for saturation with new circuit prototype. Spike was below 9V with VF at 150, 050 and OH at 295, 105, 02. Success! Prototype will be kicked up to implementation.

Friday August 5 2011 5:42 pm Byrne 70044-58 Hall Probe Angle Calibration

Good data was gathered, but the TF probe shows variation shot to shot, which reflects strongly in hall probe data. Consequently signal is not only a function of angle. multiple shots have been taken for each angle in the region of max signal, TF strength will be normalized out, and the maximum angle extrapolated from the data.

Three bad probes have been located. One is extremely noisy, seems to be a bad amp channel, one seems to be dead, has not been diagnosed, and one (the bad probe from last run) seems to have been blown on the board (Hall current is zero, but signals are output. Very confusing)

Remaining work to be done: Find probe sensitivity (V/T), characterize transverse field effects, and trouble shoot bad probes/amp channels.

Tuesday August 9 2011 5:56 pm DeBono 70059-70092 bias probe

Today brought the first successful tests using the crown amp setup to power the bias probe during a plasma shot. The following waveforms were tested:

Sine waves @500hz, 20V, 100V

Square pulses @-100V, 1msec, 2msec duration

The sine&square waveforms were clearly visible in the voltage/current measurements of the probe. Negative biasing caused positive acceleration (in the natural direction) of the (3,1) saturated external kink, while positive biasing resulted in counter-acceleration.

I-V probe characteristics showed a near-linear plot, with an V/I plasma resistance of 3ohms. -100V was not enough voltage to reach ion saturation current, the old LANL transformer will be hooked up to the crown setup to allow higher voltages to the probe.

Tuesday August 9 2011 7:45 pm Byrne 70093-70112 Hall Probe angular Calibration

Data today was much better than on Friday. Found a cosine-like dependence of signal on angle with respect to toridal field. Peak of plot shows a plateau from ~ -2.5 degrees to +10 degrees away form an arbitrary zero.

Plan going forward is to possibly take more shots (probably no more than one per position) in the plateau region to see if signal really is insensitive to angle.

The shots taken at the decided-upon optimum angle can be immediately used in the next test, which will determine the probe sensitivity (nominally 7V/T). If time permits, the VF banks will be used to quantify signal attenuation due to tansverse field.

Wednesday August 10 2011 7:07 pm Byrne 70113-70134 Hall probe calibration

Run was initially delayed by lack of repeatability from yesterday. A positioning error was found, but signals all displayed a systematically reduced normalized response of ~99% of yesterday's response. Probe currents were the same as yesterday, but the battery voltage was slightly lower (~99.5%). This was the only difference that could be found, and will be tracked in the future.

TF only shots were taken to test probe sensitivity. Data has not been fully analyzed yet, but working probes do show a linear response. VF/TF shots were then taken to determine signal attenuation. During this test it was noticed that the probe amp noise seems to be larger than it was pre-upgrade. This will be looked into moving forward.

Expect to have data analysed and hall probe provisionally ready for use by end of day tomorrow. Systematic day-to-day drift, the malfunctioning probes mentioned in today's meeting and the increased amp noise problem will continue to be investigated and hopefully fixed sometime soon.

Tuesday August 16 2011 6:36 pm DeBono/Levesque 70136-70159 Bias Probe

The runday was shortened today due to a CAMAC software error which took up the morning and early afternoon. Shot development work was done today with the bias probe inserted but floating (i.e not connected to crown voltage source). Shots improved throught the runday, suggesting impurity reduction. More shot development to follow: shots with duration >6msec were acheived today but the edge q profile still needs refinement.

Wednesday August 17 2011 6:31 pm Levesque/DeBono 70165-70190 Bias Probe

Bias probe run today with the following settings

1. Probe floating, no RMP's & static RMP's from 2-3msec

2. Probe biased to approx. floating potential (-45V), no RMP's & RMP's from 2-3msec.

3. Probe biased to (-90V,+45V), with RMP's.

The shots at +45V resulted in slowed plasmas which disrupted during the RMP. Other shots did not disrupt during the RMP.

Analysis pending.

Thursday August 18 2011 3:06 pm Byrne 17191-17210 Hall Probe re-calibration

The probe was recalibrated as before, only this time with the probe gain reduced. Signal did not fall as expected ~30-50%, three previously non functioning probes began working, two previously dead probes began passing 60Hz noise, and noise was reduced across all channels. This suggests that there is something wrong with the 10X setting on the PGA202 amps. An apples-to-apples comparison has been added to the wiki.

During the run it was realized that a toroidal shot measured in the poloidal direction provided a much stronger signal of proper poloidal alignment, as the signal had a zero cross at poloidal alignment, rather than a flat plateau of signal, as was seen with toroidal alignment. Probe was found to be ~5 degrees off axis. Poloidal shots were taken to ensure min n*Bphi corresponds to max n*Btheta. No difference could be seen from one angle to another, meaning poloidal signal is not reduced, while TF pickup is minimized.

Probe reference was redone to allow proper alignment, then measurements of TF field at a variety of voltages (300-700 by 50V increments), and VF field measured with and without a full TF field.

Thursday August 18 2011 6:01 pm Hughes/Levesque 70233-70238 Flux Loop Calibration

We took vacuum shots with full regular settings for VF only, OH only, OH+VF only, and all banks. Also took vacuum shots emulating the bank settings of recent plasma shot 70232 for comparison. * NOTE: we examined reproducibility of VF current, and found VF current is lower for same settings on shots with TF. We suspect this may be due to the VF charge degrading while the TF charges up.

Friday August 19 2011 6:07 pm debono/levesque 70211-70232 bias probe

Bias probe was run with the following settings:
1. Probe biased to (-8,+22,+45V) from 1.5-4.5msec, with static (-3,1) RMP's from 2-3 msec.
+45V reduced the mode rotation velocity of the modes from 6-9kHz to 2-3kHz. Half of the +45V biased shots disrupted after the RMP pulse terminated.
The -8,+22V shots did not disrupt after the RMP.

Friday August 19 2011 6:58 pm Levesque 70239-70266 Top shells retracted

Made plasmas with all top shells retracted 4cm in to try to observe changes in poloidal mode structure. Four shots were done with half of the top shells retracted (sections 1-5). Bottom shells were fully inserted for all discharges. Shots with retracted shells broke down ~5mm further out radially for the same bank settings, assuming the major radius calibration is still valid for the different shell configurations.

Several shots appear to have multimode activity. Mode activity seemed qualitatively similar to modes with fully inserted shells. These have not yet been quantitatively compared to previous shots. Further analysis will be done next week.

Thursday August 25 2011 5:58 pm Byrne 17269-17313 Hall Probe Troubleshooting

Shots 17269-17271 showed very little agreement with expected behavior based on previous rundays. Probe 1 has begun behaving like probe 7, that is, no current is drawn, but signal is output as would be expected.

Furthermore signal levels were down significantly from previous runs with the exact same bank settings/probe position.

Decision was made to once and for all attempt to characterize Hall Probe system's behavior and understand its various pathologies.

Shots from 17272-17313 have been crates only shots, piggy backed on B. Debono's run. Sig gens input sinewaves into the amplifiers, removing probes from the equation for now. Variables controlled for were signal inputs (sinewave, open, shorted) amp circuit (there are two, each servicing 10 probes) pickup to magnetics, and amp gain.

Circuit/Circuit pickup varies from amp to amp, and in some cases can be significant ~6% of signal, however this is only a problem if the amp inputs are left open. Shorted, or with signal, pickup is reduced to roughly the level of the backgroud noise (but still identifiable)

Gain seems to work as expected, which contradicts earlier work whereby signal levels were only slightly changed (not the factors of 10 in the amp spec) and certain probe channels which did/did not function based on gain setting appeared insensitive to the setting this time.

Noise on all amps seems small ~10mV, except for probe channel 12, which outputs slightly more than .5V. Large level signals of 2 V or greater are readable, and filtering may help. Short of replacement I do not know how to repair this amp channel.

Today's work suggests the majority of the problems associated with the hall probe are localised in the probes themselves, or maybe in the cables used to carry the probe power/signal to the amps. This will be looked into tomorrow. And of course the work will have to be checked tomorrow, due to day-to-day variability of probe behavior.

Friday August 26 2011 9:54 am debono 70272-70314 bias probe

Beginning of RMP amplitude scan done today.

Bias probe set @45V from 1.5-4.5 ms, (-3,1) RMP from 2-3 ms with phase flip @ 2.5 ms.

Shots taken @ RMP currents of 1.3, 13.3, 26.6 Amperes. The 1.3 A shots did not disrupt, the 13.3A shots disrupted after the RMP (~3ms) while the 26.6A shots disrupted during the phase flip (~2.5ms).

More shots to be taken tomorrow.

Friday August 26 2011 6:23 pm DeBono 70272-70363 Bias Probe

CC current scan continued today. (-3,1) RMP's were applied to bias probe slowed plasmas from 2-3 ms, with the phase flip at 2.5ms. Shots were taken with the CC current @ currents ranging from 1.33A to 33.3A. The highest CC currents tended to disrupt the plasma during the phase flip@2ms: intermediate currents tended to disrupt the plasma AFTER the RMP @3ms, while currents ~<10A did not disrupt the plasma.

Tuesday September 6 2011 8:12 pm Byrne 70376-70387 Crates Only Hall Probe Troubleshoot

Probe cabling has been refurbished, south rack has input bulkhead installed, signals are properly shielded, Hall/Triple/Dalpha probe breakoutboard is installed. Two broken traces have been repaired, which correspond to non-functioning probe channels. It is expected that no more physical work is necessary for the operation of the Hall Probe

Zero-level signals of the probes have been looked at with crates only shots. Three probe channels display problems.

Probe on channel 7 (numbers are in order from closest to furthest from Plasma Center) displays a .6V offset. Noise on signal no worse than average (4-6 mV), should be easily averaged out by presampling with existing code. Offset has been localized to the probe by checking D-sub output directly with a voltmeter. Amp works fine

Amp on channel 12 displays noise of roughly 100 mV amplitude. Depending on signal levels, this may be problematic.

-Both of these problems are well known from previous work.

Amp on channel 17 seems broken. With gain set at 10X *or* 1X, probe will saturate at -10V output with probe zero-level input (1-3 mV). Probe displays zero-level signal no higher than average when checked with Voltmeter. At 10X gain, probe output shows 1kHz fluctuation with amplitude equal to noise, which is ~3-4X (20mV) average. When shorted, amp outputs zero signal regardless of gain, and noise is reduced to average. Currently unsure how to address this, short of replacement.

A runplan involving calibration-style VF and/or TF vacuum shots is forthcoming to double check that the remaining probes work as previously, and that the repaired probes have been fully brought back into operation.

Thursday September 8 2011 10:27 am Byrne 70388-70402 Hall testing

So it's all in the run report, but basically, I didn't realize the legs of the chips were so delicate, and broke them on Amp chip 14 before I realized what I was doing. This was a channel that had been working fine.

Moving on from there, and keeping myself only to the "busted" chips, I found that:

The noise on chip 12 seems to be due to the circuit, not the chip. With Chip 17 in place, the same noise level is output, and 12 doesn't put out that noise when put in 14's place.

The sensitivity of 17 seems to be *partly* due to the circuit. When 12 is placed in its socket, the DC offset goes to -5V from -10V but that is still WAY higher than any signal I've seen from a magnetic measurement. Oh and also, when 17 and 12 were switched, the noise frequencies of ALL probes, good and bad dropped by ~8X from 40k to 5k. So I guess the circuits aren't independent either?

Does this mean any changes to one circuit, like say, replacing chip 14 means that all the other chips need to be recalibrated as well?

Ideas: I was only working with the busted chips, *maybe* a good chip placed in either 12 or 17 won't behave badly. But I'd rather have someone who knows what they're doing around electronics i.e. you do the switching. Also 17 has a resistor across the terminals, and 12 does not. Maybe the resistor is why swapping them changed the frequency. Although why all channels had the same frequency before they were switched, and then all had a different frequency after, rather than varying board to board, I don't know.

Thursday September 8 2011 3:44 pm Byrne 70403-70410 Hall Probe repair

Amplifier chips had a high frequency noise filter cap attached across two leads. On Chip 12, the one outputting large noise levels, this had come loose. This is not the best method of attachment, and it is possible if not likely it will happen again. A note will be added to the wiki, as this cap is not called out in the circuit diagram and is not visible unless the amp box is disassembled.

Probe 17 had an power input wire broken, but still in physical contact with the dsub pin. Why this would have led to the behavior noticed, which was a steadily declining output voltage, (which may reset itself after a certain point, certain measurements suggest it but inconclusively) is unclear. However, stripping the wire and soldering it properly has removed the behavior.

All probes seem to have been returned, or returned themselves, to operation. Nevertheless they have not actually been used to measure a magnetic field since the repairs.

While it should be fine to use for J. Levesque's upcoming run, 4 of 20 channels were inoperative at the time of the last calibration. Furthermore, while it is likely just due to the delicate nature of the probe, there has been significant run-to-run and shot-to-shot variation observed, as well as strange noise detected when the (nominally constant) transverse TF field was present.

Therefore a separate runplan is forthcoming, in order to fully recalibrate the probe, observe, understand and if possible, repair any time-dependent behavior and allow time for any on the spot repairs.

Friday September 9 2011 7:20 pm Levesque 70411-70439 Toroidal shell asymmetry

Ran with top and bottom shells in sections 2, 3, and 4 retracted 3.9cm from their fully inserted positions. The intention was to break the toroidal shell symmetry to attempt to observe changes in the natural toroidal mode structure. No obvious mode structure differences have been seen yet when comparing to shots with shells fully inserted, but analysis is very preliminary.

Also, Hall probe data were taken with the innermost shield tip at 107cm. This puts the innermost sensor ~1.3cm outside of the limited plasma edge. Fluctuations in Hall probe data correlate with fluctuations seen in the other magnetic diagnostics. Clean-up shots were required at the beginning of the run day, presumably due to the introduction of the Hall probe shield into the scrape-off layer.

Control coil amps were powered on while the FPGA (amp input) was off, which should allow the large control coils to be treated as flux loops when looking at the measured CC currents for all shots from today.

Some interesting shots are:

70420: High frequency counter-rotating mode activity appears around the time that the plasma becomes high-field-side limited (~3.5-3.6ms). This is seen most clearly in the HFS sensors. I have seen this for several shots in the past, but have not cataloged it. The fast mode rotation has been both co- and counter- to the usual mode rotation.

Other interesting shots: 70425, 70428, 70430, and 70433/4.

Monday September 12 2011 7:20 pm Levesque 70440-70472 Toroidal shell asymmetry

Ran with top and bottom shells in sections 2, 3, and 4 retracted 3.9cm from their fully inserted positions. Took 7 shots with shells in sections 1-5 retracted. Tried to reproduce the discharge evolution from shot 69142 in order to directly compare mode activity, but changes to the radial equilibrium with retracted shells made this difficult. Was able to roughly match desired major radius, plasma current, and edge-q evolution for the first ~3.5ms in the retracted-shell shots after changing the VF bank timing by 40-60us.

Hall probe and control coil "flux loop" data were also taken in the same manner as on 9/9/11.

Noteworthy shots:

70449: Large drop in D_alpha light during 2.3-2.5ms. Subsequent spikes in D_alpha are seen during 3-3.2ms, though there are no apparent minor dusruptions (which is when we typically see similar D_alpha spikes).

Other interesting shots: 70447, 70464, and 70468.

Tuesday September 13 2011 5:02 pm Byrne 70473-70484 Hall Probe Calibration

Problems were discovered with the angle finder. There are a few degrees of slop which make repeatable measurements of the angle difficult. This was not discovered as the cause of measurement problems until the end of the run day.

The issue is being corrected by affixing the anglefinder to the probe shaft with two drops of epoxy. This will allow the anglefinder to rotate rigidly with the probe, yet still be removed in the event that proves necessary in the future.

One issue that is not explained by the slop issue:

The point of the angular measurement was to aim the probe further away from the toroidal direction while measuring the toroidal field, in order to pinpoint the angle at which the signal had a zero cross and became negative.

This occurred at different angles for different probes. Despite not being able to say with certainty at what angle the probes are directed, it is certain that the probes are aimed in different directions. Without repeatable measurements, it is difficult to say for sure, but it seems to be of the order of 5 degrees of misalignment.

This will be addressed by redoing the angular measurements once they have been made repeatable, applying a linear fit to the data, and then attempting to find a "best poloidal" by least squares fitting the 20 zero cross angles.

Probe sensitivity and transverse field signal attenuation tests were not done because it is unclear how strongly they are affected by angular uncertainty, as well as time constraints.

Tuesday September 20 2011 3:54 pm Byrne 70492-70518 Hall Probe Calibration

The runplan ran as planned.

The hall probe was subjected to a more rigorous angular calibration, but the suggested change to the nominal poloidal angle was only .03 degrees.

Probe sensitivities were gathered, and in all cases agree well with the results of the previous calibration, in 2007. In addition, two previously inoperative probes are now working, meaning the entire 20 element array is in proper working order. Results will be posted on the wiki this evening

VF/TF tests were run, and the probe seems to be behaving very similarly to the previous calibration, based on eyeball analysis of the plots. More mathematical analysis will be performed, and will be available at tomorrow's meeting.

It is expected that the repair and calibration work on the hall probe is now finished, and that once the analysis and code updates have been made, the data can be entered directly into the tree for use in others' code. Until then, code used in the diagnosis and calibration of the hall probe are being commented and uploaded to the wiki for use. Ask me if you have any questions about their operation.

Byrne

Tuesday September 27 2011 2:43 pm debono 70519-70529 Bias Probe

The bias probe is connected to the cap bank and is performing normally. However, a BiRa CAMAC power supply in the north rack failed after a couple shots and ended the runday. The affected modules have since been swapped into the (previously) unused north rack crate. DeBono

Friday September 30 2011 6:42 pm DeBono 70597-70625 Cleanup Shots

Cleanup shots done today. Plasmas got longer as runday went on, but base pressure pumps down slowly. Starting base pressure before runday was ~5.6e-9.

Wednesday October 5 2011 6:59 pm DeBono 70664-70735 Bias Probe

(-3,1) phase flips were applied to rapidly rotating biased plasmas. The plasma response appears to show saturated and linear regimes with an amplitude less than that of naturally rotating or slowed modes. DeBono

Thursday October 6 2011 6:16 pm shiraki 70736-70794 RMP timing scan

Plasmas similar to shot 70000 was chosen as a model discharge for the RMP timing experiment. This resulted in minor disruptions early in the discharge (~0.5ms in) which affected reproducibility. The plasmas used instead (see 70781 for example) had a higher initial q and did not cross q=3 until about 3ms (rather than 2ms, in the case of shot 70000). Phase-flips were applied to these shots starting with 3ms up to 5ms, in 0.5ms increments. Only 6 shots were taken in this scan, and there is no discernable trend in the RFA outside the spread of the measurements. Some effort will be made tomorrow to achieve plasmas closer to the target discharge, where the edge q stays constant at ~2.6 for over 2.5ms.

Friday October 7 2011 4:59 pm shiraki 70795-70825

More attempts were made to recreate the desired target plasma for the RMP timing experiments. Bank settings and puff times were adjusted, but reproducibility remains a problem. No RMP's were applied and the Hall probe was not inserted today.

Monday October 10 2011 7:23 pm shiraki/levesque 70826-70882 cleanup shots

The focus of today's run was to clean up the system by creating outboarded limited plasmas to clean the limiting surfaces. This was interspersed with attempts to create the target discharge for the RMP timing experiment. Reproducibility of these plasmas appears to have improved over the course of the day. If this can be confirmed again in the morning then the RFA experiments could be started again.

The runday was ended when the OH electrolytic bank began to have problems fully discharging. Peak currents dropped from about 20kA to 19kA for the same bank settings, with a subsequent shot (without charging) resulting in about 1kA being discharged. This issue will need to be resolved before tomorrow's shots.

Tuesday October 11 2011 6:35 pm shiraki 70883-70931

The issue with the OH bank was not fully resolved, and partially persisted throughout today's run. However, with appropriate adjustments in bank settings this did not seem to affect plasma performance.

Plasma reproducibility has improved since the clean-up shots, and the RMP timing scan was begun. The number of shots is still small, but initial results suggest that the plasma response may be larger earlier in the shot rather than later. This may simply be due to the time-evolution of the safety factor, as the earlier RMP's occur closer to the q=3 crossing which occurs about 1ms after breakdown. The scan will be continued tomorrow.

Wednesday October 12 2011 4:44 pm shiraki 70932-70965

We continued to adjust the timing of the 3/1 phase-flip and measured the plasma response as a function of the timing. Applied RMP's had an amplitude of 10A for all shots, with 500usec in one phase and 500usec flipped. The beginning of the RMP was varied from 2ms to 4ms in 0.25ms increments. The results now seem to indicate that any changes in the measured response are due to the time evolution of q, rather than increased beta over time.

Wednesday October 19 2011 7:40 pm debono/levesque 71010-71070 bias probe

The bias probe was biased to ~-170V,-12A (the target voltage was -300V, loading reduces this to -170V). The mode is torqued in the direction of natural rotation, from ~8-9kHz to ~13-14kHz.A (0<2.5)amplitude scan was done for(-3,1) RMP's applied from 2-3ms, with the phase flip at 2.5ms. Extra measurements were taken at A=0, 0.5 to observe consistency of measured n=1 plasma response phase. DeBono

Friday October 21 2011 9:50 am Debono 71071-71114 Bias Probe

We started a rotation scan for the (-3,1) standard phase flip. RMP amplitude was set @10A. The bias probe was used to accelerate the modes to (+4-5, 14-15) khZ by applying (+13.5,-11.5)A of current from the bias probe. These are the maximum currents the probe can output (in either direction) given its current configuration.

Friday October 21 2011 9:52 am Byrne 71073 and 71114 Hall Probe

2 shots taken: 71073 and 71114

Conditions were: inner hall probe at 103 cm, TF bank at 750V.

70103 was the first shot of the day at 11:50, the Hall probe had just been turned on, and was at as close to room temp as was feasible.

71114 was the last shot of the day at 6:21, and the Hall probe had been left on the entire time.

Good news: The hall probe's temperature closed 92% of the distance between initial and steady state within 19 minutes. All calibrations were done long after a longer delay between turn-on and turn-off.

The signal dropped as expected, though the data has not been analyzed yet. The probe's De-calibration was not affected from morning to evening, however. All shots seemed to be reduced by the same percentage, but the over all shape of the plot (perfect calibration would be a straight line) was not affected.

The thermal issues can be helped by bleeding in coolant gas, or made irrelevant by allowing the probe to arrive at its steady state temperature (About 35C, or 6.5 kOhm as measured across the thermistor) before attempting to take measurements with the probe.

The de-calibration over time issue remains unclosed, however

Sunday October 23 2011 4:58 pm Levesque/DeBono 71115-71155 TF issues

The following summary pertains to the machine problems and resolutions that occurred during Friday's run. Physics goals and results of the run will appear in a separate run summary by Bryan.

-Jeff

While high-potting, Jim and I found a small continuous arc below chamber section 10 between a TF magnet foot and a bolt in the G10 spacer that the TF magnet stands on. We discussed the arc details with Nick and followed his recommendation for correcting it. The bolt in question has no direct connection to ground -- the arc seems to have been a capacitive effect similar to when the I-beams around the base pad were arcing to the upright support columns a few months ago. There was no apparent immediate danger to the machine, but the problem was resolved with mylar to be cautious. This was done by lifting the magnet slightly to take the weight off the outer foot, and inserting a small mylar sheet in the arc region. The TF magnet shifted slightly while inserting the insulation, but the change should be negligible. A subsequent high-pot test did not show any problems. For the first shot of the day (full vacuum shot), I watched the magnet and arc location to make sure there was no arcing or movement of the magnet. Everything appeared as usual.

In a separate occurence, a very loud noise came from the basement during shot 71121. Bryan, Nick, Jim, and I checked the basement for what caused the sound. Nick found problems with 2 of the TF dump relays. They had clearly arced, and were now stuck in the open position. This was the only abnormality we were able to find. Nick's best guess is that the arc occurred by tracking along insulating components of the relay to its ground. The damage appeared to be only superficial. Nick and Bryan cleaned the 2 relays and surrounding plywood. Nick also added vacuum grease to the those and other TF dump relays to inhibit surface tracking, and gave the okay to continue the run after lunch. We started back up with a half-power vacuum shot, then a full vacuum shot. Bank current traces appeared normal, and we continued the run.

Monday October 24 2011 3:47 pm DeBono 71115-71155 Bias Probe

Machine Report (written by J.P.L); While high-potting, a small continuous arc was seen at chamber section 10 between a TF magnet foot and a bolt in the G10 spacer that the TF magnet stands on. The bolt in question has no direct connection to ground -- the arc seems to have been a capacitive effect similar to when the I-beams around the base pad were arcing to the upright support columns a few months ago. There was no apparent immediate danger to the machine, but the problem was resolved with mylar to be cautious. This was done by lifting the magnet slightly to take the weight off the outer foot, and inserting a small mylar sheet in the arc region. The TF magnet shifted slightly while inserting the insulation, but the change should be negligible.

Shot 71121 was interrupted with arcs in two of the TF bank dump relays. The damage appeared to be only superficial, so the relays were cleaned and reinstalled. Vacuum grease was added to several components of the TF dump relays to inhibit surface tracking. The cause of the arcs was unknown.

Physics Report (written by B.A.D): Rotation scan done with 10amp standard (-3,1) RMP phase flips. Shots were taken with naturally rotating modes, and with modes accelerated in the direction opposite to natural rotation. To achieve the latter, the 1:1 Hammond transformer was reconnected to the bias probe circuit to achieve greater Ibias than was possible with the 4:1 LANL transformer.

Monday October 24 2011 9:44 pm shiraki 71156-71207 Hall probe internal measuremetns

Another arc was found between a TF magnet foot and a bolt in the G10 ring in the basepad. The problem was resolved in an identical manner as last Friday, with the insertion of a mylar sheet between the two conductors, and the run continued as normal.

The Hall probe was recalibrated at the beginning of the day, after powering the electronics for about half an hour to allow thermal equilibrium. This calibration was rechecked periodically through the day, and several channels were found to have drifted by a few percent. It is currently unclear how to compensate for this drift in future runs.

The probe was inserted until the back of the Boron Nitride cap was at the back of the shell, which puts the first channel at 105.5cm, with four channels inside the plasma. At this position, the magnitude of the first two channels are seen to drop relative to the trend of the channels sitting in vacuum. It is hard to quantify this reduction (particularly with the drifting signals), but it is clearly noticeable and repeatable.

Finally, vacuum pickup measurements were taken between the Hall probe and the control coils. With the previously used m/n=-3/1 phase-flips, maximum pickup was about 0.5G/A. However, by choosing an appropriate toroidal phase, the null of the external field can be placed at the location of the Hall probe, reducing pickup on all channels to below .01G/A, which is below noise levels for coil currents as high as 30A.

Tuesday October 25 2011 7:07 pm shiraki/maurer 71208-71259

There was an accident today with the TF bank, although there was no major damage to the machine. The accident was likely caused when the dump relays were reset while the banks were charged. Rather than a full dump which keeps the relays closed, the reset would momentarily close then reopen the relays while the bank was still discharging. Much of the wiring in the dump circuit was damaged, but only one relay was destroyed. This has been replaced with a spare, along with all of the damaged wiring except for the box controlling the power for the relays. As soon as this is repaired, we can begin to test the banks.

Prior to the accident, internal poloidal field measurements were taken with the Hall probe at full insertion, while applying static phase-flips. This was preceded with another calibration of the probe at the beginning of the day. Applied coil currents ranged from 0-30A, which is high enough to induce disruptions.

it was observed today that during the applied RMP, there is an increase in the D-alpha light seen by the spectrometer. This increase over background levels is only observed for coil currents above ~15A, which corresponds to where we begin to see a saturation in the plasma response. At about 25A of coil current, the light is increased by about 50% over the non-RMP shots. This will be studied more closely once the machine is back running.

In addition, the Mach probe was inserted 3cm into the plasma and testing of the probe was continued. I_sat signals appear to now be working, with typical signals of ~100mA.

Wednesday October 26 2011 4:24 pm shiraki 71262-71268 TF bank testq

The TF bank was repaired and tested today. The bank is working fine and two plasma shots were taken. Experiments can resume again tomorrow.

Thursday October 27 2011 9:32 pm Levesque/Shiraki/Maurer 71269-71313 RMP q-scan with Hall & mach probes inserted

Applied m/n=-3/1 phase flips to plasmas with the innermost Hall probe sensor at R0=105.5cm. A q-scan was done with an RMP amplitude of ~20A. Average q_* during RMPs was between 2.75 and 3.05, which is in the saturation regime observed in our previous RMP studies. More shots with lower q values will likely be done in the continuation of the run. Increases in D_alpha light were seen during the RMP periods.

Cleanup and development shots were necessary at the start of the run day. The Hall probe was calibrated at the beginning of the run. The mach/V_float probe was inserted at 104cm. Several shots were taken with the D_alpha-array covered in order to measure non-light pickup.

The TF charge panel has had a minor problem since it was repaired yesterday: The "full charge" indicator remains stuck on after each shot, though the bank fully fires. The resolution is to cycle the power to the TF/VF panel prior to each shot. This does not seem to influence the run, but should be repaired when we have a chance.

Friday October 28 2011 5:43 pm shiraki 71314

There was another arc with the TF bank today, although again there was no damage to the machine. This time the arc occurred when the computer failed to fire the banks, which then timed-out and tried to auto-dump. Six of the dump relays were damaged, and all of these will be replaced with spare 25kV relays from the old LANL system. Several sections of the mylar insulation were also damaged and are being repaired. In addition, the wiring will be reverted to the original configuration with the relays on the low-voltage side of the water resistor, while properly insulating the base of the resistors to prevent tracking along the plywood base.

A J221 in the basement rack was found to be broken later in the day, although it is not known whether this was the cause of the banks failing to fire, or if it is a consequence of the accident. The repairs to the bank will continue on Monday.

Wednesday November 2 2011 12:42 am Levesque 71315-71318 Testing banks

Low power vacuum shots to test bank triggering and currents. All systems appear normal, though a full-power TF shot has not yet been done. The root cause of the banks failing to trigger on 10/28/11 is still unknown. Standard plasma shots will commence Wednesday morning.

All TF bank dump relays have been replaced with clean spares from the old radial positioning system, which have a 25kV rating (instead of the previous TF dump relay ratings of 12.5kV). Dump relays were moved to the low voltage side of the water resistors, which should reduce current in the event of future arcing. The dump resistors have also been raised on PVC pedestals to reduce the possibility of tracking. The TF charge panel issue mentioned in the 10/27 run report has also been fixed.

Wednesday November 2 2011 7:08 pm Levesque/shiraki 71319-71360 Clean up shots

Attempted to continue RMP q-scan with lower edge q values and the Hall probe inserted. Most of the run day was spent doing cleanup shots and trying to achieve the target discharge. Reasonable discharges were reached at the end of the day. Three shots had applied RMPs, but these did not have edge-q's as low as desired. The run will continue Thursday morning.

Thursday November 3 2011 7:06 pm shiraki/levesque 71361-71416 Internal hall probe measurements

More data was taken for the q-scan with internal Bp measurements using the Hall probe. Bank settings were adjusted over the runday to produce plasmas with lower edge q's during the time of the phase-flip. Analysis of Hall probe data from this run is ongoing.

Wednesday November 30 2011 10:34 pm byrne/shiraki 71417-71489 RMP phase scan

The plasmas at the beginning of the day were erratic and reproducibility was poor. The majority of the day was spent running outboard limited clean-up shots. However by the end of the day vacuum conditions were back to normal, and some useful shots were taken. m/n=-3/1 phase-flips with various toroidal phases were applied, all at an amplitude of 15A. There were a total of five such shots that were close enough to the target plasmas, and within these shots, the kink locking-phase tracks one-to-one with the RMP phase. The amplitude of the plasma response had more variation, but is probably within the expected scatter based on previous experiments.

Friday December 2 2011 6:14 pm Levesque/Stoafer/Peng 71490-71536 Hard x-ray noise tests, Phase-flips with fewer CCs

Ran vacuum shots to test for noise and pickup in the hard x-ray system due to the banks. Tests were done with and without a radioactive source near the detector. Analysis is pending.

Phase-flips were done on plasma shots with control coils active in only sections 4 and 5. All large coils were used in the 2 sections, totaling 8 coils versus the normal 40. The applied field had a toroidally localized m=-3 structure. Coil currents were ~22amps peak. A 3/1 global response is seen for most shots. Multiple-shot averaging made the global response very clear. An exemplary is 71530.

Five usable shots were also taken with control coils on in section 4 only. The response is less clear, but may be measurable.

Monday December 5 2011 4:26 pm shiraki 71537-71560 Longer phase-flip RMP's

Some shots were taken with RMP's lasting up to 1ms in each phase. This is in contrast to the 500us in each phase which has been used up to now. Only a few shots were taken, so the experiments will continue tomorrow.

Tuesday December 6 2011 5:15 pm shiraki 71561-71567

Today's run was interrupted early when Facilities arrived to repair the air-conditioning in the experimental hall. The run will resume tomorrow if the A/C repair is completed.

Wednesday December 7 2011 6:22 pm shiraki 71570-71603

Phase-flips lasting 1ms in each phase were applied today at several amplitudes. Several (but not all) shots with 20A in the control coils disrupted towards the end of the phase-flip. This is in contrast to the ~30A which was necessary to induce disruptions when the RMP lasted 0.5ms in each phase.

It is also observed that in many cases, the plasma response tends to be larger in the second half of the phase-flip. This can be seen in shot 71584, for example. It is also common to see a "phase-flip like response" (meaning an n=1 response which reverses phase along with the coil current) with a large rotating mode on top of this. Shot 71591 is a good example of this.

Tuesday December 13 2011 8:20 pm debono/shiraki 71648-71711 Low amplitude phase-flips

Low amplitude (Ic<5A) static phsae-flips were applied today. Analysis of today's and previous shots suggest the possibility of a locked 3/1 mode and a rotating 3/1 mode simultaneously existing. As the perturbation amplitude is increased, the amplitude of the locked mode increases as well.

A few shots were taken at the end of the run with the bias probe inserted and powered with the Crown amp with the 1:1 Hammond transformer. No clean-up was necessary and vacuum conditions remained good. Application of small amplitude phase-flips to these biased plasmas will continue on Thursday.

Thursday December 15 2011 7:56 pm shiraki/debono 71712-71760 low-amp RMP on rotating modes

The bias probe was used at a low positive voltage (+22V) to slow modes down to 5khz (vs naturally rotating 8-9khz), and then (-3,1) 3amp RMP phase flips were applied. Most shots taken today had weak mode amplitudes; however mode locking was observed during the RMP for some shots. Star shots: 71755, 71757, 71758.

Friday December 16 2011 5:42 pm debono/shiraki 71761-71790 bias probe

Continuation of yesterdays experiments to apply Low-amplitude RMP's to bias probe slowed modes. Today A=0.1 and A=0.2 RMP's were applied to slowed plasmas rotating at ~5kHz from 8-9 kHz. Bias Probe set to +22V. Mode locking was not observed in todays best shots (shots that best matched the q,major r profiles of reference shot ^#70000). Star shots 71766,71744,71784.

Monday December 19 2011 5:40 pm Levesque/Peng 71790-71812 Puff line problems

Attempted to run Qian's experiment with uniform control coil currents for all large coils on 2 neighboring shells, but the run was hindered by problems getting reproducible breakdowns. It looks like there is a problem with some part of the gas puff system. The OH-bias-phase reversed plasma current was inconsistent shot to shot, and was absent for several shots. This suggests that the valve was not puffing in enough gas, or is puffing at the wrong time. The gas bottle pressure, manual valves, and puff valve battery appear fine. Puffs with closed pumping gate valves and no banks showed reproducible static fill pressure at 3.3e-5 torr. The problem will be troubleshot more tomorrow if it persists.

Tuesday December 20 2011 7:18 pm Byrne 71813-71846 Hall Probe Re-commisioning

Shaping experiments were not undertaken today, due to the postponement of the hall probe recommisioning yesterday, and unexpected problems with the hall probe.

While trimming th potentiometers, two of the hall probe chains (there are now 4 chains of 5, rather than 20 individual probes drawing current) had fluctuating currents, from 1-7 mA. This represents the entire range of currents allowable in this configuration, but the potentiometers were barely touched. This was not seen when the probe was tested yesterday. The hall probe was disassembled and inspected, and intermittent losses of continuity were discovered on some probe chips, but the actual location of the loose wires were never found. The probe eventually fixed itself, and has been running well all day.

The run began in the afternoon, and time was lost trying to troubleshoot channels that seemed to be malfunctioning. Two channels turned out to have their gains set incorrectly, while two others, which also output signals much larger than expected, simply fixed themselves. Again, the problem did not recur, and the cause was not discovered.

The full range of calibration tests were performed, and the hall probe does seem to be in good working order.

However, the air pump which was installed to force air and facilitate cooling does not seem to be very useful. The throughput is rather small and the line to the probes is extensive and narrow. There is likely a large pressure drop from the pump to the probes.

The measured thermistor resistance of 6.7 kOhm corresponds to a temperature of ~35° C. 10kOhm represents a 25° C room temp. At the end of the day, the pump was turned off and the resistance was found to drop to 6.29 kOhm. This suggests the pump is cooling the probes by 1 degree C. The pump MAY be useful in holding the temperature constant, and preventing fluctuations, but that will have to be tested over the course of a run day.

Thursday December 22 2011 6:00 pm Byrne/Peng/Shiraki/Angelini 71847-71890 axisymmetric shaping/non-axisymmetric perturbations

The first few shots of the day failed to breakdown, which was resolved by replacing the e-gun biasing battery. The old battery had dropped to 200V, and the new battery measured ~225V. There were no problems with the puff or breakdown for the rest of the day.

Local perturbation shots with no poloidal structure were taken. Unlike the previous partial coverage experiments with m=3 structure, no obvious plasma response was seen. Control coil currents were increased to 40A with still no response.

Shaping experiments were run from shots 71886 to 71890

Due to time constraints, shots were only taken with coils firing at high amplitudes (±35A) shots 886, 887, and 890 were "triangular" shots, while 888 and 889 were "square" shots

In both cases, very coherent, non-resonant modes were created which the plasma survived for its usual lifetime. Triangular shaping seems to draw the plasma's Major radius outwards, while square shaping pushes it inwards. It is yet to be determined if this is plasma coupling or simply the control coils coupling to the MR rogowski. Small effects were also noticed in edge q, but it's still too soon to say if it's an important effect (or a real one).

The modes seen in the triangular shots had a primarily 3/1 component, but the outboard midplane "lobe" was much larger than usual. There was also a coherent n=2 mode, but the m number was harder to distinguish. The square modes were similar, with an 3/1, 5/2 and an n = 3 mode with an unclear m number. The shape of the primary m = 1 mode was similar to that of the triangular mode, but with a phase shift of 180 degrees though the outboard midplane node was much less exaggerated. In both cases the primary modes seen clearly resemble those modeled with Tokamak.

Square shaped plasmas had a longer lifespan than triangular shaped plasmas (comparison to unshaped has not been done yet) Both shaped plasmas had an impurity spike event during their early evolution (~1.8ms) which may have had an effect.

2012

Tuesday January 24 2012 9:34 am Niko 71915-71958 Feedback

First shot did not break down. E-gun bias voltage was 220 V at battery and 200 V on machine. Raising e-gun current from 6.2 A to ~7 A fixed the problem.

Initial plasma shots were short lived and several cleanup shots necessary.

First feedback experiments resulted in unexpected behavior of the control outputs. The problem turned out to be that the data buffers send by the ACQ196 are not ordered by channel number but in some (more or less) random permutation.

With the permutation taken into account, effects of turning on feedback were clearly visible in the mode structure. Modes with active feedback had rotation frequencies down to 0.5 kHz. However, no relationship between mode activity and feedback phase could be observed yet. Reproducibility of plasmas was low.

The feedback system occasionally failed to trigger, or deadlocked while running. Debugging of the problem is not possible with the current hardware configuration and would require to either disable one of the ACQ196s or waiting for D-TACQ to deliver the RTM-T support

The run had to be interrupted once for about an hour due to network problems.

Tuesday January 24 2012 5:44 pm Niko/Daisuke 71960-72011 Feedback

Plasma reproducibility was better today, about 3 out 4 plasmas were usable. A feedback phase scan in 45 degree steps was completed, and a finer scan with 36 degree steps started. Phases around zero resulted in much larger amplitudes and rotation frequencies around 1 kHz. Phases around 100-200 degrees resulted in smaller amplitudes and rotation frequencies of 5 - 8 kHz. The transition, however, seems discontinuous and there is not enough data to tell if the differences are significant or within experimental error.

The digitizer deadlocked two times, so this problem does not appear to be too frequent.

The feedback system did not finish control computations in the available cycles times for roughly 12 samples in every shot.

The run was stopped because plasmas did not break down any longer. The e-gun bias voltage has dropped from 220 V yesterday to 200 V today, which is the most likely reason for the problem. There are no more spares, and this type of battery is not manufactured anymore.

Wednesday February 1 2012 4:50 pm Shiraki/Angelini 72012-72032 Cleanup / D-Alpha

The shots for today were primarily to test the new e-gun and run cleanups. The e-gun worked without any difficulty.

We also had the opportunity to test what the D-Alpha saw. We were able to confirm when a plasma was outboard or inboard limited because of the high emissions at the edge during contact. The lens was covered for a shot to confirm that the measured fluctuations were due to the light and not simply pickup. The light emissions appear to be hollow, but that could simply be due to large emissions from the limiters.

Thursday February 2 2012 6:43 pm Levesque 72012-72032 Feedback

Tested new CPU feedback algorithm which uses fully integrated B_p signals with a smoothed signal subtracted. Gain was set high enough that some control coils reached 30A transiently during feedback. Phasing of the control coil currents tracked the mode phase well, at least for modes at ~5-10kHz. Feedback phase was tested at 4 values separated by 90 degrees. One of the CC current A14s had problems digitizing throughout the run day, making the timebase wrong for 6 of the CC current measurements; presumably these control coils were activated correctly during the shot though.

Lack of plasmas reproducibility hindered the run. Minor disruptions often occurred just after startup and would change the edge q significantly. Non-reproducibility was unrelated to feedback, as shot behavior would diverge before feedback was activated. Shot-to-shot comparison for feedback efficacy at different phase settings might not be possible due to large plasma parameter differences. The time at which edge-q crossed 3 varied by ~1ms among the shots, making it difficult to choose the best time to activate feedback. Within individual shots, feedback did not conclusively enhance or suppress modes while activated -- changes in mode amplitudes were within the standard range of variability.

Friday February 3 2012 7:08 pm Levesque 72070-72098 Feedback

Continued feedback phase angle scan from yesterday. Plasmas were more reproducible than yesterday, but still not enough for good shot-to-shot comparisons. Large -3/1 mode amplification was seen for 2 shots with the feedback phase set such that the measured difference between detected Bp and applied Br was 90 degrees. These 2 shots (72079 and 72080) also had the edge q closest to the target, slightly below 3. For the feedback phase set 180 degrees away from this (72078), no clear suppression was seen. The 180 degree difference in FB phase only changed the angle between detected Bp vs applied Br by 90 degrees, instead of 180.

Monday February 6 2012 7:18 pm Levesque 72099-72140 Feedback with slowed rotation

Applied feedback to shots with the bias probe slowing down mode rotation. Plasmas were fairly reproducible. Modes were non-existent or very low amplitude due to the presence of the bias probe. Feedback did not seem to have much/any effect on the biased shots. Feedback currents peaked at around 10A. Later attempts using a higher gain failed due to a problem with the CPCI outputs; this problem has been resolved and can be avoided in future feedback tests.

Rotating RMPs were also applied to several biased plasmas. With RMP rotation at 2.5kHz, a co-rotating response was seen in the Bp sensors with a phase difference of 90degrees. This was the same phase difference seen in the shots from Friday that had feedback amplification of the mode. This is consistent with a feedback phase instability, where the feedback result was similar to that of the rotating RMP. At RMP rotations of 1.3 and 8kHz, modes did not lock to the perturbation.

The e-gun power supply was found to have a problem in the morning. An indicator light on the power supply was not working, and E-gun bias voltage was measured as ~400V, even though it was set at 250V last week. Nick was able to decrease this voltage to 338V, but it won't go lower. We decided to run with this as is since it's within the design spec of the e-gun. A resistor was also replaced in the power supply.

Tuesday February 7 2012 10:24 pm Levesque/Peng 72141-72183 Feedback phase scan, shot development

Continued feedback phase scan from Friday without the bias probe. Shots were taken at 45-degree increments. Mode frequencies varied shot-to-shot, including natural modes that existed before feedback was applied. During feedback, mode frequencies were 4-10kHz. Results were similar to Friday, with no clear suppression. The closest-matched shots (with programmed FB phase angle) to compare for the 2 run days are: 72078 (90deg), 72080 (270deg), 72146 (0deg), and 72147 (180deg). Shot equilibria had much more scatter for other phase angle settings.

Also worked on shot development at the end of the day to target plasmas that formed with edge q below 3. Delayed VF bank timing by up to 220us to try to get plasma break down with a larger major radius. Breakdown major radius ranged from 92.5cm to 95.0cm. We could not get plasmas with an initial edge q below 3 due to 3/1 tearing modes during and just after the startup.

Wednesday February 15 2012 5:25 pm Hughes/Shiraki 72184-72188 Calibration of The Big Rogowski

Took shots with TBR wrapped around the upper VF coil 1 time, 10 times, and 15 times as noted in shot descriptions. The results are still somewhat puzzling: the VF current curve integrated from the Rogowski signal (accounting for the partial integration of the noise-suppressing RC circuit) is off by what looks like a roughly constant factor of two, but more analysis is needed, and the reason for this is not yet clear. With a little luck, the calibration can be completed with data analysis, and won't require any more run time.

The Big Rogowski is now undergoing an upgrade to improve its insulation for use with the TF magnets. Data from today's shots is recorded in the tree as input 71 from the west rack CPCI; if you use something that references that spot in the tree, it might be worth making note that the above shots are recording something else entirely.

Thursday February 16 2012 7:31 pm Shiraki/Peng/Byrne 72189-72234 Local perturbation + some vacuum test

Get trained on the start up procedure. It takes a long time to get a nice reproducible shot but we finally are able to work on the local perturbation.

The Plasma do give a roughly m=3 response from n=1 local perturbation (only the bottom inboard coils are excited). Control coils are excited when q_edge cross 3. Further analysis is needed.

Another thing is that the background poloidal field fluctuations a bigger when q is around 3. and for all the shots taken, q stays around 3 ( above or blow a little bit). So the response to perturbation is not obviously seen for a single shot, but becomes more clear after average over shots.

A few vacuum shots are also taken in the end.

Monday February 20 2012 6:46 pm Hughes/Levesque 72236-72243 Calibration and Trial of The Big Rogowski

Ran two zero-net-turn calibration shots which are yet to be fully analyzed. Ran several shots with The Big Rogowski around TF ^#14 (by Jeff's reckoning), starting with only half-normal TF bank charge and working up to a full regular setting vacuum shot with all banks. Next episode: how the B.R. signal compares with the TF Probe!

NB: The Big Rogowski is still on the TF magnet, so watch out for that if you plan to run. Please don't step on the cable in the basement hanging down in front of the basement rack; it really pulls the coil together.

Thursday February 23 2012 9:08 pm shiraki 72247-72291 Long duration phase-flips

Phase-flip RMP's lasting 1ms in each phase were applied to our standard discharges crossing q=3 from above. The FPGA was used to drive the control coils. Vacuum shots show that it takes ~1ms for the eddy currents to decay and for the radial field to soak through the walls.

The most dramatic observation is the occurrence of disruptions at much lower coil currents, usually around 20A but on one occasion as low as 15A. This is in comparison to the approximately 40A of coil current required to induce disruptions when the phase-flips lasted 500us per phase. An example of such a disruption can be seen in shot 72279, which disrupted shortly before the currents were flipped at 3ms.

Applied currents in today's shots were in the range of 7-20A. More shots will be taken tomorrow.

Friday February 24 2012 4:33 pm shiraki 72292-72333 Long duration phase-flips

Yesterday's run was continued to collect more data with long phase-flips (1ms up, 1ms down). Control coil currents were limited to 10A to avoid disrupting the plasmas. Plasmas today were noticeably more reproducible today than they were yesterday.

The bake was turned on at the end of the run day in preparation for next week's shot development.

Monday February 27 2012 11:37 pm Levesque 72334-72353 Cleanup shots

Cleanup shots following the bake. The bake was turned off at 8:40am, and plasma shots started at 4:00pm. The chamber was not fully cool yet, but the base pressure was at 1.1nTorr according to the pressure indicator in the control room (nearly bottomed out) and still falling slowly. Dominant lines on the RGA are deuterium and nitrogen.

Shots started out terrible, presumably due to redistribution of contaminants onto limiting surfaces. Plasmas got longer throughout the run, lasting up to ~4ms. Achieved breakdowns with puff times as low as 100us, while our standard operating puff times have been 1000-1250us recently with the same regulator pressure. D_alpha light in the spectrometer was higher than normal for all shots. Cleanup and shot development will continue tomorrow.

Tuesday February 28 2012 7:24 pm Levesque 72354-72394 Low breakdown q*

Developed discharges where the edge-q starts below 3 then approaches/crosses 3. Was able to do this by delaying the VF bank timing by 100us versus its normal timing, and having higher breakdown current. Successful plasmas finished the "start" phase with 2.5 < q* < 2.7, and 94.0 < R0 < 94.5cm. After settling on a discharge style, between 1/3 and 1/2 of the discharges remained at low q with a smoothly changing major radius and no major disruptions. Shots so far have approached q=3 quickly, or have had q changing slowly but possibly too far below 3. Shot development is still needed to have a better edge-q evolution.

Good shots are: 72374, 72388, 72390, 72392, 72394

Wednesday February 29 2012 9:00 pm Levesque 72396-72434 Low breakdown q*

Continued development of discharges with q* starting below 3. 3/1 and 6/2 modes appear when q* is below 3, thus do not begin as internal modes. More analysis is needed to determine if clear growth rates can be inferred -- the changing major radius complicates interpretation of mode amplitudes. Shot development will continue tomorrow.

Shots that might be good to analyze while q* is rising are: 72398, 72400&1, 72404&5, 72408, 72413, 72415, 72416, 72423, 72433

Thursday March 1 2012 10:57 pm Levesque 72436-72482 Low breakdown q*

Continued discharges that start with q* between 2.5 and 2.8. About half of the plasmas evolve in the style intended, with q* rising toward 3 for 4ms as Ip rises and R0 slowly decreases. The other half of discharges typically fall inward too fast immediately after the startup, and q* rises above 3 within 0.5ms. There are very few shots between these 2 extremes. This bifurcation may be caused by mode activity during startup. The discharges begin with 3/1 activity that usually decays away within 1ms.

In the intended shot style, 3/1 and 6/2 modes grow as q* rises toward 3 later in the discharge. Several shots show clear 3/1 growth that may be fit with a growth rate. The shot set from this week should be extensive enough for a study of natural 3/1 and 6/2 mode activity with q* exclusively below 3. Lots of analysis remains to be done.

Some additional notes from the run:

- cos(theta) rogowski pickup from the VF was recalculated for the delayed VF shots. This had a small but noticeable effect on the calculated major radius and edge q. Future shots with different bank timings may need different compensations for pickup.

- CC amps were on to allow control coils to act as pickup coils for all shots. North rack A14_14 failed to initialize about 20% of the time. I didn't check if its data (6 of the CC currents) were good for any of the shots since the control coils were not energized, but there may be problems if anyone wants to look at that data.

Several good shots are: 72461-3 (steadiest q*), 72464, 72469, 72477

Thursday March 8 2012 5:42 pm debono/shiraki 72483-72524 Bias Probe

The new bias probe 2:1 transformer combo is working well: highest current drawn from probe was ~40amps. More shot development tomorrow is necessary, as we need our plasma parameters to closely match the target shot ^#70000 from ~1.5-4.5ms for the RMP analysis to be accurate. Although we were able to produce shots that closely matched the target shot from 1.5ms-3.2ms, it proved difficult to obtain shots that closely matched shot ^#70000 from 1.5-4.5ms.

Friday March 9 2012 6:32 pm debono/shiraki 72525-72552 Bias Probe

Machine report: Although there was no serious damage to the machine, there was a bank failure today. This was most likely due to a water spill in the hallway above the TF bank, which made its way through a metal door in the floor and on to the caps below. The arc occurred while the bank was charging, and the dump relays do not appear to be involved in the accident. Damage is limited to several TF fuses and some mylar insulation. Repairs will begin on Monday.

Physics report: Pre-bank failure, more work was done on the bias probe run from yesterday. Shots proved to be erratic with the probe inserted and activated, limiting the amount of useful data obtained. Star shot ^#72540.

Thursday March 15 2012 2:44 pm shiraki 72553-72559 Bank test

The TF was tested today, working up in voltage to 6.1kV. A full vacuum shot was taken as well and everything is working normally.

Friday March 16 2012 6:00 pm DeBono 72560-72601 Bias Probe

The new battery-powered egun bias voltage supply has been installed and is working well. More "slowed rotation" bias probe shots taken today. The 2/1 stepdown transformer is working nicely; the bias probe typically drew 33.44 amps during shots today. Plasma's suffered the same irrepeatibility experienced during the run last friday, however a few good shots that matched the standard shot#70000 were taken. Star shots #72598, 72592, 72590.

Monday March 19 2012 7:50 pm Byrne/Shiraki 72602-72646 Control Coil Shaping

Most of the day was spent replicating the model shots (71000 and 70000). Around 3pm, a good equilibrium was arrived at and shots began to be taken.

While running Athena (the eebox computer running the smaller of the two monitors) experienced a glitch. The Ubuntu environment crashed, and the FPGA became frozen and unresponsive. About an hour was lost, but the banks were not charging when it occurred, so no harm was done. Two of the ignitron water outlet cables had been knocked loose, possibly during the repairs last week, and had flooded the area behind the OH power supply to a depth of about 1/8 inch. We shop-vacced up the water and replaced the hoses in the drain channel.

72634-72646 were the shaped shots. Shots 72634 and 35, shots 72641 and 42 are good examples of "negatively" shaped shots, While, 72637 and 38, as well as 72643 and 46 are good examples of "positive" shaping. Due to the fact that it is unknown which direction B_r points for a given coil current, I can not say for sure which represents triangularity or squareness.

However, there does seem to be an effect on the plasma edge q and major radius. Depending on the sign of the control coil currents, small jumps in the values in one direction or the other are seen. I have to double check to make sure my code is eliminating pickup from the coils properly, but if that were the case, it's likely we can correlate the jumps to a shaping factor, and from there constrain the direction b_r points.

The edge q and Mr jumps are the only obvious effect though. Sensor arrays (stripey plots) do not show any obvious changes to mode structure, rotation or strength compared to unshaped shots. It is hoped that the run tomorrow will be able to start shaping sooner, and that more data will be available as a result.

Wednesday March 21 2012 9:25 am byrne/shiraki 72647-72697 Control Coil shaping scan

Was able to return to yesterday's target shot by noon, and many more useable shots were taken as a result. 4 or so shots were taken with each polarity before being switched to ensure that drift in the characteristics of the plasma were adequately controlled for. Additionally, unshaped shots were taken from time to time to compare the shaping currents' effects.

Internal 3-1 modes were strongly apparent as the plasma's edge q fell from ~3.3 to 3. at the crossing, the mode disappeared, and while an external 3-1 established itself some microseconds later in the unshaped shots, the shaped mode spectrum was very "smeary" with no clear mode dominating.

"Positively" shaped shots tended to fall more slowly so that the tearing mode had more time to establish itself. "negatively" shaped shots seemed to kick the plasma out in major radius, and drove q down more quickly, so the internal mode was harder to see. firing the control coils earlier did allow the plasma to be negatively shaped at a higher starting edge q, and the mode behavior did look similar. No B-D analysis has been done yet to determine the actual content of the mode spectra at both these points.

While code has been created to eliminate the pickup in the major radius and IP coils due to the shaping pulses, it is unclear whether it has been correctly integrated into the analysis code, so radial "kickout" can not be said to be a real effect yet, although its magnitude suggests that it is not solely due to pickup effects.

Monday March 26 2012 8:39 pm debono/shiraki 72699-72767 bias probe

RMP phase flips were applied to bias slowed RWM's today. Overall the run day was pretty good: the plasmas tended to disrupt for shots with RMP currents >12A; this is significantly less than that for naturally rotating modes. Shot irrepeatability was still an issue but at least 11 shots were taken that had good agreement with the target shot. Star shot examples: 72734,72764 DeBono

Monday April 2 2012 6:14 pm debono 72769-72809 bias probe

Bias probe run with counter-rotating shots today Maximum bias probe electron saturation voltages/currents are +80V/50-60Amperes in the current setup. Overall the system works well when driven into the electron saturation regime. Unfortunately, well defined contour rotating shots were not seen for our best shots of the day. However, a few shots with q>3 showed defined counter-rotating modes moving at ~-6kHz. Star shots 72789,72804,72806,72805,72807 DeBono

Tuesday April 3 2012 10:55 pm Levesque 72810-78542 Asymmetric walls while approaching q*=3 from below

Began low-edge-q experiments with asymmetric walls. Most low-q shots with walls in the normal fully inserted position were similar to the target shot (72464). After retracting 8 of the top shells 4cm each, achieving the target shot was difficult. Only a couple of shots were similar enough to the target for direct comparison.

Most shots with retracted shells suffered from a strong, quickly rotating 3/1 mode during the startup. This led to minor disruptions which steered shots away from the target discharge. These shots are unusable later in the discharge, but analysis of the startup may be useful. I will continue trying to match the target with the poloidal asymmetry before moving on to the toroidal asymmetry.

Wednesday April 4 2012 8:34 pm Levesque 72855-72905 Low q, poloidal shell asymmetry

Continued poloidal wall asymmetry experiments. The amplifier for the plasma current Rogowski was not working at the beginning of the run day; vacuum and crates-only shots showed a strange oscillating I_p signal. Trouble-shooting showed that the problem was in the AM502 amplifier module itself, not in the rest of the signal path. This problem was circumvented by using the AM502 from the cos(3*theta) Rogowski coil for the plasma current instead. The digitized plasma current should be correct with the replaced module, as long as there are no minor gain differences. However, we're no longer digitizing the cos(3*theta) signal -- we don't have enough working AM502's.

Plasmas were much better than yesterday. Using a longer puff time than yesterday (1000us versus 500us) gave more D_alpha and SXR emission, but did not prevent the strong 3/1 mode from appearing during startup. Shots got better throughout the run. Mode analysis shows presence of 3/1 and 6/2 modes during the target period with q* staying below 3. Tomorrow morning I'll continue with the poloidal asymmetry, then will switch to the toroidal shell asymmetry in the afternoon.

Star shots: 72877, 72881, 72885, 72894 (compare with 72469 with normal shells), 72902, 72905.

Thursday April 5 2012 6:05 pm Levesque 72907-72954 Low q, shell asymmetry

Finished poloidal wall asymmetry experiments. Took more shots with 8 of the top shells retracted, then took a few shots with only 4 adjacent top shells retracted. I will thoroughly analyze the poloidal asymmetry shots next week.

Started toroidally asymmetric shell experiments. Top and Bottom shells in sections 1, 2, 9, and 10 are retracted 4cm. I was able to get close to the target shot by the end of the day. Toroidal asymmetry shots will continue tomorrow.

Star shots: 72913, 72951

Friday April 6 2012 11:57 pm Levesque 72955-73015 Low q, toroidal shell asymmetry

Finished low-q shell asymmetry experiments with top and bottom shells retracted in 4 sections. Most shots had shells in four adjacent sections (sec 9, 10, 1, and 2)retracted, giving an odd toroidal configuration. Also ran several shots with diametrically opposed sections (sec 9, 10, 4, and 5) retracted, giving an even configuration. Equilibrium evolution was slightly different for each of the shell arrangements, which may influence mode behavior. The data set from this week should be extensive enough for comparing mode activity as q* is below and approaching 3 with non-symmetric walls. Next week will be spent analyzing this data set in detail, comparing with shots from March 1st with the standard wall configuration.

3/1 mode activity during startup appeared with all wall arrangements. Characteristics of these startup modes will be compared, in addition to the original intent of studying mode activity that occurs later in the discharge as q* rises toward 3. Qualitatively the modes appear similar in each shell configuration, but detailed analysis has yet to be done.

Star shots: 72961, 72965, 72966, 72968, 72981, 73012

Monday April 9 2012 5:42 pm DeBono 73017-73035 Bias Probe

The goal of the shot today was to do RMP's on rapidly counter-rotating modes. Typical bias parameters were ~80V in-plasma 40-50A. This had the mode rotating weakly (~-4kHz) in the direction opposite to natural rotation, but was not enough to force strong counter-rotation of the mode. RMPs were not done due to the similarity of the shots to "slowly rotating" modes. DeBono

Wednesday April 18 2012 5:11 pm Hughes/Shiraki 73044-73048 Big Rogowski mkII Calibration

Ran preliminary calibration of Big Rogowski against VF current monitor. This data requires some analysis, but suggests a gain factor of roughly 1.65, and if this is accurate, it suggests that the TF probe values are low by about 10%. However, this is still only a preliminary result, and more calibration shots might be needed to more precisely characterize the Big Rogowski's behavior due to some disagreements between the Rogowski and other sensors in the time domain. More on this at the meeting tomorrow (4/19/2012), and results will go on the wiki.

Wednesday April 18 2012 5:54 pm Byrne 73049-73053 Amp wiring check

The amps were fired, first in 0/0 configuration to find reversed polarities (one reversed, many gains moved, a few bad connections), then in 3/1 mode to ensure the correct amps are connected to the correct transformers.

Signals look better, and all connections seem correct. Will re-set all gains tomorrow to ensure that the amplitude spread is due to factors outside the new work

Monday April 30 2012 4:42 pm Hughes/Levesque 73055-73069 VF Shunt Calibration

Hughes: Calibrated VF Pearson against 50mV/600A shunt. Semi-conclusive results, but disagreement seems to be under 2%.

Levesque: Ran VF only and OH only shots to check the calibration of the cos(theta) Rogowski coil pickup. Shots were taken with the unused bank delayed by 50ms in order to remove mutually induced currents. This especially affected the OH-only shots, which previously had significant current induced in the VF coil.

Tuesday May 1 2012 4:35 pm Hughes 73070-73083 TF Calibration Runs

Many hours spent debugging invalid timing setup on basement A14, but once that drama was over, got a few good shots recording data for the Big Rogowski, the TF Probe, and the 50mV/600A Shunt reading the TF current. Some script debugging is still needed to analyze results. However, one immediately apparent quirk was that the Shunt apparently sees the TF current peak at roughly 45ms before t=0, suggesting ~50ms soak-through time between the driven current and the toroidal field seen by the magnetic sensors from outside the TF casings. Further VF calibrations using both Big Rogowski and Shunt will be pursued.

Wednesday May 2 2012 7:17 pm DeBono 73084-73108 Bias Probe

The new Crown amplifier bias setup is capable of putting a ~170V, 3.5ms square pulse on the probe using the current 2:1 stepdown transformer. The VF bank was reconnected today, without a shunt in place. The fast camera has been installed, and a bright lamp has placed in the machine room for illumination. Plasma currents were very low today, approx 5-6kA. The cause of this has not yet been determined. More work to be done tomorrow. Bryan

Thursday May 3 2012 9:14 pm debono 73110-73115 bias probe

It was discovered this morning that one of the three cables (the green cable) connecting the egun to its power supply was wrongly disconnected. The egun is now working normally (~-230V bias). Plasmas taken today appeared to be low current 4.5-6.5kA, the same as yesterday. However, the Bp magnetic signal amplitudes for shots taken today are very comparable to previous "good" shots, suggesting that the Ip measurement is faulty (example: shots 73115,73038). I suspect that the Ip AM502 is broken. Neither CTX or HBT have spare AM502's to swap it out with, and indeed we are already an AM502 short (see April 4th's run report).

Wednesday May 9 2012 7:17 pm Niko 73214-73247 Feedback

Daisuke developed a code for calculating the plasma current from the poloidal arrays. The results were in agreement with the Ip rogowski measurements in earlier shots. However, when plasma shots were taken, the Ip rogowski was found to be working again, so the failure must have been intermittent.
The different control coil current measurements for constant amplifier input were crosschecked with a pearson rogowski for two channels by Pat. It was found that A14 measurements using the shunt resistor agreed with the rogowski measurements. Therefore, the variation is most likely due to varying gains in the amplifiers and can be calibrated for when generated amplifier input.

A short was found in a ribbon cable connecting the FB sensor breakout board to ACQ_388 and ACQ_350. This may have influenced FB sensor data in earlier shots as well. The cable was replaced. A systematic check of all used ribbon cables is probably advisable.

A 0.15 V fluctuation was found in the *input* of the GPU breakout board (i.e., measured in the connector that normally goes to the analog output board). The source of this noise is still unclear. The noise appears even if the downstream amplifiers are turned off.

Plasma shots were taken and a reasonable shot with some mode activity and acceptable reproducibility was found. With the feedback system active, the measured control coil currents, however, were all zero. For further investigation, the output of the FB system was rerouted to ACQ-12. This digitizer saw significant output, so something on the path from AO32 to control coils must have broken again. This was not yet investigated.

A second problem was found in the AO32 output itself. The output did not show any correlation with the control input. Further testing showed that a replay of the sensor data from the MDSplus tree generates correct output, but when the data comes directly from the digitizer, a different output is generated. The problem is being investigated.

Thursday May 10 2012 6:51 pm Niko 73248-73369 Feedback

Plasma shot reproducibility was very good. After the first two shots, all shots were within the target range. A good baseline for mode activity without feedback has been established, but unfortunately no shots with active feedback control were taken today.

The feedback system was further debugged by re-routing control output to another digitizer. The two major problems identified yesterday were fixed. The apparent lack of any control coil currents was due to disconnected A14s. The apparently incoherent control output was because the digitizer does not deliver the channels in order. The correct mapping has already been established previously, but it was assumed that when operating over PCI Express, the channels would be delivered in order. This problem has been fixed.

The state of the control system at the end of the day was unknown: different test scripts generate inconsistent results -- some indicating correct operation and some indicating further problems. I will try to resolve the discrepancies between the scripts first. Depending on the result, either more control system debugging will be required, or plasma shots with active control may be possible tomorrow.

Friday May 11 2012 6:47 pm Niko 73270-73283 Feedback

* The oil pump was found running when the run began at noon on Friday. It was turned off after the last run, so it must have been turned on again sometime between Thu evening and Friday noon. A puddle of spilled oil was found in the basement, most likely because the oil valves in the basement were closed and the piping not designed for the resulting increased pressure.

* At first, the VF bank was not discharging. The problem was fixed by removing the shunt that was apparently not installed correctly.

* The discharge bar for the bank to the left of the basement rack looks broken to me. This should be looked at by someone more competent than me.

* The inconsistencies in the control system data that were observed yesterday have been traced to broken feedback sensors. After these sensors were excluded from all computations, all data is now consistent. Most likely the high-frequency oscillations in the broken sensors resulted in significant differences after partial integration when sampling with 10us (feedback system) vs 2us (normal data acquisition).

* Instead of tracking the n=1 mode in the four toroidal arrays, the control system now tracks the (-3,1) mode and the (-6,2) mode. Motivation for this change was that this is probably a better example for multi-mode control, and that the integration of more sensors would result in a less fluctuation rotation frequency.

* After successful testing on ACQ-12, the control output was wired to the amplifiers and plasma shots with active feedback were taken.

* The control system seems to be working now. Amplitude and phase of the field generated by the control coils (calculated from the digitized control coil currents) track amplitude and phase of the fields seen by the feedback sensors (calculated from the integrated signals in the MDSplus tree).

* The effect of FB control on the plasma has not yet been analyzed, and the number of available shots is still very low.

* The dynamic adjustment of control phasing based on the mode rotation frequency still needs some work. In the tested configuration the phase was adjusted every 1ms. However, since frequency estimates jumped wildly between +15 kHz and -20 kHz, it is doubtful that the adjustment is really beneficial.

Tuesday May 15 2012 10:31 am Niko 73284-73353 Feedback

The feedback parameters were adjusted to compensate for the transfer function of the amplifiers and plasma shots with active feedback were taken.

A total of 20 shots was taken with different base feedback phases (0, 90, 180 and 270). So far, no significant differences in mode activity have been identified (see attachments).

Reproducing the target discharge from last week turned out to be difficult. The shots from today should therefore only be compared to each other, but not with the shots without feedback from last weeks.

Additional problems with the A14s used to digitize control coil currents were observed. In addition to the familiar problems (some signals appear with incorrect time axis or very strong noise), some channels now seemed to record entirely wrong voltages.

Tuesday May 15 2012 6:05 pm Niko 73354-73404 Feedback

Plasma shots were taken with feedback phases of 0, 90, 180 and 270 degrees. In contrast to yesterday's shots, q=3 was approached from below between 4 and 5ms. About 1 out of 3 shots was sufficiently close to the target discharge, other shots tended to fall in rapidly, reaching q=3 around 2.5ms and disrupting at 5ms. Feedback with 0 and 270 phasing was found to result in slightly smaller mean mode amplitude, but standard deviations of all phase sets as still largely overlapping, so the effect may also be coincidental. At the end of the day, plasmas stopped to break down for unknown reasons.

Monday May 21 2012 9:10 pm Debono 73512-73531 bias probe

The new egun filament is installed and is working well.

IMPORTANT NOTE ON IP: The Ip rogowski did not work correctly from the beginning of the runday today, and the problem degraded through the runday. I am now depending on the magnetic sensor measurement of Ip rather than the direct rogowski measurement. Please note that significant differences between the two measurements existed at the end of Niko's last run (ex:shot 73401).

After a number of cleanup shots, shots were taken with excellent plasma parameter agreement with the target shot (#70000), with the bias probe inserted. Biased plasma data to be taken tomorrow.

Wednesday May 23 2012 6:50 pm Debono 73585-73614 bias probe

More accelerated-mode bias shots taken today. (-3,1) phase-flips applied to the plasma with currents from 0-32A. Plasmas looked pretty good; reproducible, and good match to the target shot. Star shot 73613 (30 ampere RMP shot).

Tuesday May 29 2012 6:02 pm Debono 73620-73655 bias probe

CC shots were taken on bias-probe slowed plasmas after the previous run applying them to rapidly rotating plasmas. The plasma response was similar to slowed-rotation shot data taken in the past, and helps to validate the rapidly-rotating plasma response results.

Wednesday May 30 2012 7:05 pm Debono 73677-73700 bias probe

The goal of the run today was to debug the Ip Rogowski. However, the Ip Rogowski problem did not surface today after 10 plasma shots taken over the space of several hours.

We now have two different AM502's amplifying the Ip rogowski voltage into two different West-rack CPCI channels; the prexisting AM502 is connected to CPCI channel 70, and the teaching lab AM502 is connected to CPCI channel 71.

The 2nd AM502 will be left in place in the event that the Ip rogowski problem resurfaces.

Friday June 1 2012 7:55 pm Debono 73725-73750 bias probe

RMP scan taken today with bias probe inserted into plasma and floating (i.e. probe disconnected).

Crate problems in the middle of the afternoon interfered with the run; we were however to get data in the early and late afternoon. More running to be done on Monday.

Monday June 4 2012 9:41 pm Debono 73725-73750 bias probe

The Ip rogowski is working well with the teaching lab AM502 but not with the original hbtep AM502: the hbtep AM502 outputs a significant amount of noise. Today was a model day for excellent plasma reproducability & plasma parameter agreement with the target shot. A (3,-1) RMP phase-flip amplitude sweet was done today with the probe inserted into the plasma and floating. The plasma response to the RMP was very weak, however the mode response was not weak due to the physical presence of the probe.

Tuesday June 5 2012 9:41 pm Qian 73795-73811FeedBack

This is a copy of yesterday's run report. Using the wiki page

I tried to develop the tearing mod shot for torque analysis today. There was some issue with the battery for biasing the Egun in the middle of the day and the plasma didn't break. After re-charging the battery, it works again. I tried to make both the Ip current and Major Radius flat and it kind of works but the duration is short and some fine adjustment of the parameters are need. One can look at 73810 and 73811 for details of the shot information. More time is needed for shot development. .

Thursday June 7 2012 7:55 pm DeBono 73813-73843 Bias Probe

The plasmas today agreed well with the target shot but seemed unresponsive to ~30A (m,n)=(3,1) phase flips, similar to Mondays run. I am wondering if the control coils are actually applying a (3,1) RMP to the plasma. Comparing control-coil only shots 72729 (Slowed-mode run from March 26) & 73845 (today run) shows identical control coil currents, but significantly different FBp sensor pickup. More work to be done... Bryan

Monday June 11 2012 6:13 pm Qian 73889~73919 normal

Control coils are functioning well with the switching. Rotating perturbation is done on normal plasma shots ( not the tearing modes which has not yet developed)

I forget to multiply the frequency with 2pi and the perturbation rotation direction was in reverse to the mode rotation for the first few shots. But positive torque is observed... Then the bug is fixed and I tried some rotation perturbation with 4kHz ~ 13kHz and fixed 12kHz, 10kHz, 8kHz. The torque is always negative.

So I got back to the case where positive torque is observed and did some scan with -1kHz ~ -5kHz, fixed -5kHz, -3kHz~0kHz. -2kHz~1kHz, 0.5~2.5kHz. No matter the rotation is from positive to negative, or from negative to positive. The torque code show that the torque is positive around -1.5~1.7 kHz. ( multiply it by 2pi one gets around 8kHz)

Although I am still not 100% of the validity of my torque analysis code. It definitely shows that something is going on with -1.5kHz frequency. ( The natural mode we usually see is rotating at 7~9 kHz )

Should say that the bug with 2pi factor and direction helped me notice this region accidentally.

The interesting shot with positive torque is 73915~73919

Tuesday June 12 2012 7:44 pm Angelini 73922-73960 Inboard Limited/Fast Camera

The goal of today's run was to make inboard-limited plasmas which may make fast camera analysis easier. After some shot development, I was able to create shots whose major radius didn't vary much. In these shots, the q still changed drastically from around 4 to 2.5 as the plasma current increased. The majority of the shots were taken with the D-Alpha filter and the camera's framerate was 88,000 fps.

Notable shots:

73942 - The plasma is inboard limited most of the shot, and the MR varies the least

73953 - This shot ran for a while between the limiters

73955 - The longest shot of the day (disruption at 7.8 ms)

73958 - Highest current of the day (19.3 kA before the start of the disruption)

During the day I confirmed that the camera was still recording light even if it couldn't be seen in the video. It seems that 88,000 fps was sufficient for all the shots.

~Sarah

Friday June 15 2012 4:46 pm Hughes/Byrne 73966-73979 Big Rogowski mkII Calibration

Shots were taken with the Big Rogowski mkII wrapped 9 times around the upper VF, centered with a used cardboard cable spool. Several shots were dedicated to testing the 500microOhm shunt for safety and minimizing noise (settling on firing only the VF crowbar at 300V as the best approach), followed by shots testing the effects of coil end joint proximity to the VF coils, and proximity of the two coil ends to one another. Analysis still has to be done, but the new shunt has a great SNR, and fills the digitizer range perfectly at 300V (~4.5V peak signal out of +/-5V range). With any luck, no boxcar will be needed, so we'll have ideal time resolution! Note on changes made for this run: The Big Rogowski is still wrapped around the VF, but this shouldn't interfere in anyone's running. The shunt is removed from the VF, the VF return cable is reconnected (although someone misplaced the nut that was used to connect it!), and the associated BNC cable is coiled around the pillar between OHST and VF for safety. The basement A14 is back to TF-esque timing. All krytrons are back on. Everything should be normal, but whoever runs next might want to double-check those affected areas.

Tuesday June 19 2012 6:03 pm cziegler 73981-73998 clean-up

We ran clean-up shots after opening the probe vacuum to the main vessel.

Plasmas were short lived, typically disrupting around 3.5-4ms.

The 5-tip probe's electronics was on until (including) 73989 when it was realized that the voltage sensing circuits are railed. Starting from shot 73990, the probe circuit was off, in order to protect it and the digitizer. On shot 73993, the probe was inserted into the plasma in order to start cleaning the probe tip itself as well as the vessel. At a nominal position of R=107.5cm no effect was detected on the plasma. In the following two shots, the probe inserted a further 1cm each time. The longest and best behaved shot was 73998 (5.5ms). More clean-up shots are required before "real" running.

Wednesday June 20 2012 2:43 pm Hughes 73999-74001 Digitizer test

Three crates-only shots were taken using a 1.606V battery measured before and after with a DVM as a constant-voltage source to check that Basement Rack A14 inputs 5 and 6, and North Rack CPCI Input 69, were recording proper voltages. The good news is that all three recorded 1.606V +/- 1% (the most inaccurate, the North Rack CPCI, appears to have recorded ~1.59V). The bad news is that this doesn't help to explain why the Pearson transformer registers a VF current 25% higher than that recorded by the shunt. I'll take a look at the data again to make sure I didn't miss something by poking at it with only jScope, but it seems pretty cut-and-dry.

Wednesday June 20 2012 6:55 pm Levesque 74002-74020 Cleanup shots

Ran outboard-limited cleanup shots. The bias probe and new 5-tip probe were inserted into the plasmas to clean their surfaces. Both probes were inactive. Plasmas had returned to their normal duration and behavior by the end of the day. External modes were weaker than normal, but this is usually seen when the bias probe is inserted and inactive.

Thursday June 21 2012 5:30 pm Cziegler 74024-74034 5-probe test

We tried to put the 5 tip probe to use today, after the problems noted on Tuesday were resolved - i.e. the probe read-out circuits were re-wired. The machine behaved reasonably well, we ran shot that lasted about 5ms. Unfortunately, the probe recorded negative ion-saturation currents, and at a nominal depth of R=105cm even the floating potential signals bottomed out.

Vacuum shots confirmed that this is not due to pickup. It will only happen with plasma shots. The large negative Isats are even more confusing because

a) the read-out circuit cannot generate negative voltages

b) the problem persists even if the probe is behind the gate valve, BUT (once again)

c) ONLY with the plasma present in the machine.

We will need to re-evaluate the behavior of the probe and the components connecting it to the digitizer.

Saturday June 23 2012 7:51 am Angelini 74035-74053 Fast Camera

Today's half-day run was a continuation of the fast camera run from last week. Using the same settings, I was able to reproduce the inboard-limited plasmas, many of which stayed at a constant major radius for most of the shot. Unfortunately, for the later shots, about halfway through the shot the D-Alpha light intensity decreased drastically. I was able to observe this at different exposure limits, so it may be a physics phenomenon rather than a technical one. I'll see if this drop in intensity corresponds to a particular q or current. One difference in today's run was that the base pressure was lower than last week's.

Good shots for the day are 74044, 74049 and 74053.

Monday June 25 2012 7:15 pm DeBono 74054-74102 Bias probe

Ic amplitude scan taken today for (-3,1) RMP phase flips; the bias probe was inserted and floating in the plasma. Qian's software fix to the CC miswiring appears to be working; data taken today is similar to Daisuke's RMP phase-flip data taken without the probe in the plasma. More running will be done tomorrow. Note: New shorting sticks have been attached to the TF, OHEL, VFEL, and the Ohbias bank. (Note that the originals are still attached at the moment). Bryan

Tuesday June 26 2012 6:37 pm DeBono 74103-74148 Bias probe

More data taken today for the experiment started yesterday (6/25/2012). Plasmas were somewhat irreproducabile today. Overall,the RMP data taken with a floating bias probe is very similar to the data with no probe in the plasma. The plasma(s) disrupted for shots with control coil currents above 25A. No machine problems, although caliban was tempermental in the early afternoon. This slowed down the run somewhat. Bryan

Wednesday June 27 2012 7:20 pm DeBono 74149-74169 Bias Probe

The floating bias probe Ic scan was completed today. Overall the floating probe plasma response is very similar to the plasma response with no bias probe inserted into the plasma. Plasmas readily disrupted at control coil currents above 25A. No machine issues, although the florescent lights above the machine (towards the south rack) need to be replaced. Bryan

Thursday June 28 2012 7:54 pm DeBono 74191-74238 Bias Probe

An easy way to detect broken/missing fuses in the CC rack was found today. 1. Power on the CROWN amps 2. The RED "Clip" LED will light up for ~2s if the fuse is ok 3. The GREEN "Signal" LED will light up for ~2s if the fuse is broken/missing. We replaced several broken fuses today. RMP shots on accelerated modes done today. Bias probe voltage -150V, current -4A. The mode response again was again seen to be slightly less than that of "naturally" rotating modes. DeBono

Monday July 2 2012 6:27 pm Niko 74259-74355 CC Test and Feedback

To ensure that last Friday's rewiring fixed the wiring problems and did not introduce any new ones, every single large control coil was fired individually and checked against radial feedback sensor measurements. The wiring was verified to be correct for every coil.

The amplifier gains for two control coils were found to be set to zero, which may explain problems with these coils in previous runs.

In 3 out of 43 shots, the amplifiers apparently received no signal. This problem was not reproducible and disappeared when taking a second shot (without having touched any system). It is suspected that this may be a result of the RTM-T problems that result in computer crashes at higher sampling frequencies. For now, it is advisable to briefly check control coil currents after every shot.

Reproducing the target shot (approaching q=3 from below) for feedback experiment consumed the rest of the day. With bank timing and voltage settings similar to the reference shot, plasmas behaved drastically different. Towards the end of the day, a new bank voltage configuration was found that resulted in plasmas resembling the reference shot in about 1 of 3 cases. The feedback system was turned on for a few shots and confirmed to be behaving as expected, but no useful feedback data has been gathered yet.

Tuesday July 3 2012 7:06 pm Niko 74356-74450 Feedback

Producing usable shots continued to be difficult in the morning. Plasmas disrupted very early (3-4ms). Therefore, a different target shot evolution (with q crossing 3 at 2 and 5ms) was selected at noon. This target was more reproducible and used for feedback experiments. For all runs, the bias probe was inserted and driven with +60 V from 1.5ms to 5ms, and in later shots 1.5ms to 8ms. Modes were typically found to be rotating with 4 kHz.

CC currents reached peak values of 60 A without breaking any fuses. During one such shot, the voltage across FB02_C2 suddenly dropped to zero and remained there during several test shots. However, this problem disappeared during debugging before any faulty components could be identified.

"Blackouts" (with no signals coming into the CC amplifiers) in either the first 32 or last 8 control coil channels (corresponding to the outputs of the first and second D-TACQ AO32 analog output board) were observed several times.

In feedback shots, feedback was active from 1.5ms to 6ms on the 3/-1 mode. Feedback phasing was adjusted in steps of 45 degrees, and a total of about 3 usable shots was taken for each phase.

Shots with active feedback generally differed from shots without feedback. In shots with feedback, the q evolution stayed much closer to q=3, and often plasmas disrupted early. This did not occur in shots without feedback. A phase dependency was not observed, but an elaborate data analysis has not yet been performed.

Friday July 13 2012 10:56 pm Rath 74452-74495 Pickup calibration

Today's shots were taken to improve the subtraction of pickup currents from OH, TF and VF in the cos 1 rogowski. 4 TF-only shots were taken. Then the VF was disconnected at on top of the center stack and OH only shots were taken. To get a more diverse set of waveforms, some shots were taken with subsets of the available OH banks. After that, the VF was reconnected and the OH disconnected at the same spot. VF only shots were taken with and without VF start bank.

Control coils were then fired individually to determine the physical location of the cos1 rogowski. It was found between FB09 and FB10. No significant pickup from FB08_C1 and FB01_c1 currents was found. Calibration shots were therefore taken with FB09_C1 to FB09_C4 and FB10_C1 to FB10_C4.

Finally, a few plasma shots were taken to evaluate the possibility of reducing the control system cycle time from 10us to 6us. No problems showed up, so next week's runs will most likely be done with 6us.

The interlock at door 2 was found broken and had to be repaired before the beginning of the run. Inspection of the wire strongly indicated the involvement of a postdoc in this issue.

Monday July 16 2012 6:44 pm Niko 74496 - 74555 Feedback

Goal of today's run was to demonstrate feedback effects on a target shot that approaches q=3 from below, while avoiding feedback while q is decreasing at the beginning of the shot.

Initial shots with active bias probe resulted in almost completely suppressed mode activity, so the bias probe was pulled out for the rest of the day.

The control system was configured to track modes from t=1.5ms to t=6.5ms. Control output (and thus feedback) was active from t=4ms to t=6.5ms. Cycle system cycle time was 6us.

Most of the day was spend on shot development. However, at the end of the day good shot reproductivity was established and a total of 14 usable shots were taken in sequence.

Without feedback, the target shot crossed q=3 around t=2.2ms, bottomed out at q=2.4 around t=4ms and reached q=3 again at t=5.5ms before disrupting at t=7.5ms.

When feedback was active with a 180 degree phase, shots resembled the no feedback reference shots until feedback began at t=4ms. Shots then reproducibly disrupted at t=4.5ms, with the major radius jumping from 93cm to 91cm within less than 100us.

When feedback was active with a 0 degree phase, shots resembled the no feedback reference shots until feedback began at t=4ms. After that, the major radius decreased slightly faster than without feedback, resulting in the shot terminating ~0.5ms earlier. Mode activity compared to no feedback shots was generally higher.

Please refer to separate email for illustrating plots.

Tuesday July 17 2012 5:19 pm Niko 75556-75605 Feedback

The scan of feedback phase started yesterday was continued today. Data was acquired for 90 and 270 degree phasing, as well as some additional no-feedback cases. Shots with 270 degree phasing behaved similarly to 180 degree shots and caused a disruption within 1 ms of activating the feedback system. Shots with 90 degree phase behaved similarly to 0 degree shots, having a slightly reduced shot length compared to the no feedback case.

Comparison of perturbation amplitudes between no FB, 0 degree and 90 degree FB is complicated by the large shot-to-shot variability and so far no significant differences have been extracted.

About 1 in 4 shots suffered from failure in either of the RTM-T modules, resulting in only 32 or 8 feedback coils being used.

The amplifier for feedback sensors in section 10 was replaced with a spare. Although this changed the digitized signals, they can still not be considered correcty. Some problem must thus exist in either the sensors itself, wiring or digitizer channels.

Thursday July 19 2012 6:39 pm Niko 74613 - 74670 RWM Shot Development

Goal of today's run was to find bank settings that result in a plasma with distinct RWM activity.

The original plan was to develop a shot with a breakdown at the low field side with the plasma then moving inboard. This turned out to be impossible, as the plasma breaks down on the high field side before any VF bank has fired.

Instead, a shot was developed that started with the major radius increasing to a maximum of 96 cm at 1.3 ms, while edge q fell from +infinity to 2.5 at 1.3 ms. This q value was steady until about 4ms, after which it rose up to 4.5 at 6.5 ms. While significant 3/1 RWM activity was expected for this kind of shot, mode amplitudes at the q=3 crossing did not exceed 4 Gauss, while normal fluctuations were around 2 Gauss for the entire shot. Example: 74658

Attempts to reduce the fluctuation level by staying further away from q=2 generally resulted in very different shots with the plasma immediately falling inward.

Significant mode amplitudes in the order of 8 G were found only in 2/1 and 4/1 tearing modes. Growing 2/1 modes were observed with q=2.3 and r=96cm almost constant for 1ms. This growth was always followed by a minor disruption and the plasma beginning to move inwards. Example: 74632

It was theorized that the lack of any apparent RWMs might be due to the plasma being stable to external kinks. However, an attempt to make the plasma less stable by increasing the puff time and decreasing the toroidal field did not result in any significant changes. However, it was observed that prolonging the puff time from 1.25ms to 2ms results in plasma currents of ~3 kA already during the OH biasing. Example: 74656

It was also found that the edge q calculation was using a hardcoded toroidal field strength of 0.33 Tesla instead of using measurements. The shotplot.py script has been updated to use the measurements of the TF probe, scaled to the current major radius. The resulting corrections are in the order of 10% for a TF voltage of 6.1 kV.

Shots were taken with TF voltage ranging from 5.5 kV to 6.3 kV, thus exceeding the standard 6.1 kV.

In summary, it seems that adjustments of bank timings and voltages in the tested ranges do not result in plasmas with RWM activity. While the parameter space still has large unscanned regions, there does not seem to be any particularly promising direction. However, it seems that in order to demonstrate feedback a new type of shot that exhibits RWM activity is required, so I'm not sure how to best proceed at this point.

Monday July 23 2012 6:26 pm Niko 74671 - 74738 RWM Shot Development

Efforts to find a reproducible shot with RWM activity continued today. Today's attempts were focused on keeping the q evolution from shots like 74658 while increasing the plasma current ramp. So far this has not been achieved, shots with suitable plasma current evolution typically fall too far towards q=2, resulting in significant tearing mode activity, or bounce immediately back above q=3. Shots taken with the same settings fell in either of these categories. It seems that some bifurcation prevents the plasma from taking an intermediate path.

In a second approach, we tried to utilize the control coils in a static m=1, n=0 configuration to move the major radius at breakdown further out. Initial results confirm that we can reprocibly affect the major radius, but so far no shot parameters were found that allowed a return to q=3.

Tuesday July 24 2012 5:45 pm Niko 74738 - 74786 RWM Shot Development

Efforts to find a reproducible shot with RWM activity continued today. The control coils were used for major radius control during startup, but even with this additional know the desired shot evolution could not be achieved. A new parameter set with a large current ramp and a q evolution that drops below q=3 only around 2ms was explored. With these parameters, resistive wall modes with amplitudes up to 12 Gauss were observed in roughly one out of three shots. However, when averaging over multiple modes, this is washed out to 6 Gauss (compared to background fluctuations of 3 Gauss) which still seems too low for use as a base case for comparison with feedback based mode suppression.

Thursday July 26 2012 3:17 pm Hughes 74787-74794 Big Rogowski mkII Calibration, cont'd

In order to verify the calibration of the Big Rogowski, we independently counted wrappings around the VF and fired only the VFEL. All shots of this type produced gain factors (between BR and VF Pearson Transformer) deviating less than 1% from a VF Pearson gain factor from the previous calibration campaign, regardless of geometry of the BR relative to the VF coil, and regardless of the number of wrappings (9 vs 10).

As a second check, we also attempted to calibrate against the OH (wrapping around the upper outer winding), using only the OHEL. This produced a consistent 6%-7% disagreement with the 'characteristic' VFEL shot from the previous campaign. At present, it is not known if this is due to the OH's high dI/dt enhancing some neglected term in the calculation, an error in the calibration of the OH current monitor (a home-made Rogowski, rather than a Pearson), or some other unidentified source of error.

More refinement will follow, in particular, including the Big Rogowski's R and L in the calculation (previously disregarded as insignificant due to the L/R time being around 2us) and calculation of digitization error effects due to the Big Rogowski's relatively small signals. At some point, shots may be taken using the VFST to test the code's ability to keep up with a high dI/dt.

Special thanks to Pat for overseeing, lending a hand, offering some good ideas, and climbing around on the machine a bunch. He really tied the experiment together.

Tuesday July 31 2012 6:53 pm Levesque/Peng 74826-74873 Troubleshooting HD amps

Ran crates-only shots along with several control-coils-only shots to troubleshoot problems with the high-resolution sensors in sections 1-5. One amplifier board which was associated with 3 bad sensor channels was found to have all 12 amplifier channels working; however, 2 of the inputs had bad/loose connections at the ribbon connection for the amplifier inputs. One of the channels (TA03_S3P) was fixed by tightening the connector further in a vice, which is our standard method for these connectors. The other 2 sensors (TA03_S2R and PA1_S17P) are currently working for control-coil test shots, though nothing was done for these other than unplugging and replugging the input and output cables. TA03_S2R may be intermittent -- further testing can be done if the sensor goes bad again.

We also started troubleshooting a second amplifier board which is associated with 2 bad sensors (PA1_S15R and PA1_S16P). This board produces extra unexplained noise, and will be swapped with a working bench-tested board for further tests.

The strong 12-20kHz noise that appeared in all high-resolution sensors in sections 1-5 (refer to crates-only shot 72857) is now gone. This may have been caused by the power source for entire amplifier box, or by some bad connection on the amplifier board for the "TOP 1" signal group -- both were changed before the noise was recorded. The power source (a heavily-damaged extension cord daisy-chained through 2 outlet strips and another extension cord) was the likely culprit, as it was also producing lots of noise when powering an oscilloscope. The box is now powered through a direct source.

Wednesday August 1 2012 6:03 pm Jeff/Qian 74897-74914

*FB10_S3P had a bad connection to the breakout board and is fixed.

*Troubleshooting the FB sensors in section 10 is still in progress. Backup boards were put in and changed the bad sensors on section 10. But there are still bad ones..

*More test to be done tomorrow.

Monday August 6 2012 7:51 pm Levesque/Cziegler/Peng/Rath 74943-74976 Cleanup shots and troubleshooting

Ran cleanup shots. The high frequency noise in the high-resolution sensors in sections 1-5 returned at the beginning of the day. The problem was eliminated by powering the amplifier box through an extension cord to the control coil power rack, but this is only a temporary solution. We should assess the power sources/sinks in the area to look for what is causing the problem, and find a permanent solution. Circuitry for the new mach probe was tested on the machine during cleanup and shot development. Problems in the test setup have not yet been resolved. While attempting to reproduce the target shot for asymmetric wall experiments, the TF and OH Start banks stopped triggering. Troubleshooting of the problem has been inconclusive so far, and will continue tomorrow.

Wednesday August 8 2012 8:01 pm Levesque/Peng 74992-75048 Troubleshooting HD amp problem

For the first shot of the day, only the TF bank was charged to low voltage, and it did not trigger properly and was dumped. Subsequent troubleshooting without charging the banks showed no problems with the krytron firing, implying that the J221 outputs were pulsing. The TF bank was then charged again, and fired successfully even though no connections or software had been changed. All banks fired correctly for the rest of the plasma shots, though only 3 plasma shots were taken.

Most of the day was spent troubleshooting the high frequency noise in HD amp box 1 (for sections 1-5), which returned yet again. It seems that the noise appears under the following conditions, which is consistent with testing from the past week:

-The noise goes away when power to the amp box is cycled.

-The noise does not reappear as long as the high voltage panel is off (and thus krytrons do not fire, though J221 pulses are sent).

-The noise appears when the high voltage is on as soon as the VF Start krytron fires. This occurs even when the banks are not charged.

-After the noise appears, it remains for subsequent shots until the amp box power is cycled. This is true regardless of whether the high voltage is on or off.

Previous troubleshooting was done without considering the state of the HV switch, thus disappearance of noise may have just been due to the HV switch being off after changing power cables. For now, it looks like the problem cannot be avoided for plasma shots as long as all sensors are connected. The problem remains even when the amp box is powered by a UPS on battery.

The noise problem seems to be tied to the "TOP" amplifiers, specifically TOP1 & 3. When all amp outputs except TOP3 are connected to the CPCI, the noise does not appear. When TOP3 is individually connected, it does not produce enough noise to account for when all outputs are connected -- there seems to be a synergic effect between the three TOP outputs. The noise also propagates through the CPCI into the SXR tomography measurement, though the noise is much less pronounced. Troubleshooting will continue tomorrow.

Monday August 13 2012 9:05 am Rath 75082-75103 debugging

Various shots were taken to test and debug the high frequency noise, the automatic triggering and the control system.

The automatic triggering on completed TF bank charging is working fine.

The high frequency noise is still being debugged.

An unexplained phase difference between input and digitized control output was found. The phase difference changes over time, and does not seem to correspond to a constant time lag between the signals. Investigation is ongoing.

Monday August 13 2012 5:27 pm Niko 75112 - 75182 Feedback

The previously observed phase differences between control system input and output were further investigated and an explanation was found. In short, the precision with which the feedback system can compensate for latency depends on how much the average mode rotation frequency agrees with the instantanous frequency, i.e. the phase difference from one sample to the next. Varying sensor gains can cause the measured sin and cos components of a mode to differ in amplitude, which results in a modulation of the apparent frequency, i.e. instantanous and average frequency appear different even for a perfectly rotating mode.

Most of today's shots were "warm-up" shots and not yet reproducible. The target shot has not yet been reached again. At the end of the day, the shot parameters of the target shot seemed to give rise to a mostly reproducible shot with different evolution and less mode activity. Debugging of high frequency noise in high density sensors continued throughout the day as well.

Tuesday August 14 2012 9:59 pm Niko 75183 - 75256 Feedback

Plasmas were reasonably well behaved today. Plasma lifetime increased continuously throughout the day, so that good shots were achieved in the middle part of the run. Early and late plasmas differed too much from the reference shot.

Initial feedback shots were aborted after an unusually high number of broken fuses (about 37 fuses were lost per shot), even though recorded coil currents never exceeded 2 A. The cause turned out to be a race condition at the end of the shot: it was possible that the GPU powered down before the analog output module, resulting in the output module outputting garbage data for a small period of time. Since this happened at the very end of the shot, the digitizers were no longer recording and thus never showed the currents. The problem was fixed by explicitly turning off the output module before the GPU, and no fuses were broken for the rest of the day.

After dealing with the fuse problem, coil FB02_C2 continued showing zero current. The problem was eventually tracked down to a short at the A14 input, where unisolated ends of the ribbon cable touched a metallic component. The ribbon cable was moved to fix the problem, but a proper fix (insulating the ends of the ribbon cable) still needs to be implemented.

Feedback experiments were run on four independent n=1 modes (one for each toroidal FB sensor array). Control cycle time was 4 us, and control latency 16us. The former exceeds our original design goal of 5 us, the later is 4 us larger than the expected value (12us). Shots were taken with 4 different feedback phases spaced 90 degrees apart. Typical control coil currents were up to 2 A. For analysis, shots were aligned by the peak q value. Mode activity generally began to increase when crossing q=2.6.

The following effects have been observed:

• Compared to no fb, fb at -10 degrees slows down the positive frequency components.
• Compared to no fb, fb at -100 degrees introduces additional 1 kHz rotations.
• Compared to shots without feedback, feedback at -190 degrees slowed mode frequency from -10 kHz to -4 kHz at similar amplitude.
• Compared to no fb, fb at -280 degrees slowed frequency to about 7 kHz and increased mode amplitude.
• When comparing feedback at -100 and -280 degrees, -100 degree feedback results in slower modes of lower amplitude.
• When comparing feedback at -100 and -280 degrees, -100 degree feedback results in slower modes with lower amplitude.
• When comparing feedback at -10 and -190 degrees, -190 degree feedback results in slower modes.

Passive observations using the FB coils as sensors indicate that the theoretical phase for mode suppression should around -100 degrees. (assuming an adequate amplifier response model).

Fourier spectra showed distinct frequency peaks at -10 kHz for all modes. However, modes 1 and 3 also showed additional peaks at +10 kHz kHz (zero based indexing). This is consistent with a systematic n=1 error in sensor gain and/or alignment in two arrays.

Comparison of mode phase and control coil phase showed significant variation when feedback was active. I believe this may indicate that the control system latency is too high, so that the mode slips instead of being suppressed, resulting in the changed rotation frequency. Another explanation would be an inaccurate amplifier response model.

The heat lamp for the OH ignitron was found broken, but this seemed not to affect bank firing. Presumably, due to the broken AC the ambient room temperature is warm enough to provide a sufficient temperature grading to the water-cooled components.

From Jeff Levesque:
The source of high-frequency noise in HD sensors in sections 1-5 was isolated to one amplifier circuit (amp 6 in the "TOP3" assembly). The specific problematic component or cable was not found, but the high-frequency noise was eliminated in all sensors by removing the op-amp from this one circuit. Note that the op-amp itself works fine, we just disconnected it because that was the easiest way to disconnect the amplifier circuit. Although the noise only appeared under certain output connection combinations, this change has removed the noise for all further shots, including plasma shots. The end result is that the high-frequency noise is gone (or is as low as other system noise), and sensor PA1_S29R is disconnected.

Thursday August 16 2012 8:24 pm Niko 75275 - 75318 Amplifier Response

Most of today was spent on investigating the variation in control coil current mode amplitude with mode frequency.

Three issues were identified quickly:

• The formula for the frequency dependent gain correction was missing a constant term.
• At low frequencies, the filter meant to subtract equilibrium fields was filtering out part of the signal in the test setup.
• The constants for amplitude and phase correction were derived based on measurements of the plasma response rather than on the control coil currents. Measurements of the control coil currents therefore do not show the additional effects of eddy currents in the shells, which modify the field felt by the plasma.

Issue 1 was rectified right away, and dealing with issue 3 has been postponed. Correcting issue 2 (in order to verify that the amplitude variation was indeed reduced) turned out to be unexpectedly time consuming.

Elimination of the equilibrium filter introduced a constant drift in all amplitudes. This was first blamed on the function generator running in continous mode, so that the offset subtraction would actually subtract an average over a potentially fractional number of wavelengths.

The straigthforward way to resolve this by triggering the function generator turned out to be less straight, because the low input resistance of the function generator made it impossible to use a single J221 output to drive more than one trigger input.

After this was fixed, it turned out that the signal generator was now outputting a constant non-zero signal (depending on the output phase) in pre-trigger state.

It was attempted to eliminate this problem by using another gate signal for amplitude modulation. However, using amplitude modulation is mutually exclusive with externally triggered operation, so the function generator is running continously on its internal clock and the phases of the two signals are no longer synchronized. This means that the only way to ensure a fixed phase relation between the two signals was to connect them to a scope and manually tune the phase difference after every frequency change. This was considered impractical.

The next attempt was to trigger the function generator externally and set both phases to 0. The phase difference was then produced by sending the second trigger with a (frequency dependent) delay. This produced the desired signals while being zero in pre-trigger state.

At this point, a new problem arose. The partial integration of a pure sinosoidal actually introduced a constant offset. Therefore, even after finally getting the function generator to output the correct signals, the mode amplitudes still showed a constant drift.

It then seemd that it would be easier to keep using the equilibrium filter, but ensure that it would average over an exact period of the input frequency, so that this frequency would not been filtered out. This approach required adjusting the filter for every frequency, but still seemed to be the least complex option.

This test setup has been tested and seems to work. However, due to lack of time the actual frequency scan was postponed. Feedback experiments were then started based on the assumption that introduction of the constant term in the gain would improve the low-frequency response sufficiently.

Shots were taken with 70, 85, 100, 115, and 130 degree phasing. Not enough data is available to draw any conclusions.

A rather different idea to calibrate the amplifier response that did not suffer from any of the problems above only came up at the end of the day. Since the amplifier response is actually independent of the rest of the control system, it should be enough to test a single shell and amplifier on the bench, directly feeding it different sinusoidals without using any control computations. In this setup, it should also be possible to measure the phase of the produced magnetic field instead of the control coil current, so we would be able to include shell eddy current effects without having to rely on the plasma to mediate them. I am considering to do this calibration after the current feedback run cycle is done.

Friday August 17 2012 7:39 pm Niko 75319 - 75412 Feedback Phase Scan

A total of 66 shots with no feedback and feedback with 70, 85, 100, 115 and 130 degree phase were taken today. In this phase area, mode suppression is expected and was previously observed with 100 degree phase. 42 of the 66 shots have similar q profiles and can be used for statistical analysis which is yet to be done.

The current signal for FB02_C2 temporarily failed again today. The problem was resolved by wiggling the ribbon cable of the A14s.

Todays run was again done without a working OH heat lamp. However, a new bulb has been found and is ready for installation.

A few shots with relatively constant major radius near 92 cm were taken at the end of the day for Qian.

Monday August 20 2012 8:41 pm Niko 75414 - 75490 Feedback Gain Scan

The OH ignitron heat lamp was replaced this morning before the run, and the gate valve to Istvan's probe was closed and the probe removed some time before noon. Plasmas became reproducible enough to begin a feedback gain scan shortly after noon.

After about 31 shots were taken the plasma behaviour changed dramatically, preventing the gathering of more data for the gain scan. The last good shot is 75473. In these and all previous shots the plasma current rises from 0 to 4 kA in a few us after the OH start bank fires. Shot 75473 was taken 4 minutes later with the same parameters and without anyone entering the Tokamak room.Bank current traces for this shot are identical to the previous shots. However, the plasma current stays flat even after both OH start and OH electrolytic banks have fired. Only when the VF bank fires, the plasma current slowly rises to a maximum of 7 kA at 3ms. Major radius calculations give results above 100 cm. The e-gun filament is emitting light. All three e-gun batteries have been checked. To be on the safe side, the 9V battery in the circuit was replaced and the 225 V battery fully recharged. Neither changed the plasma behaviour.

The gas valves were checked, and the pressure was confirmed to drop during the puff.

When the VF trigger was moved from 1ms to 1.2ms, the start of the plasma current rise moved from 1ms to 1.2ms as well. When the VF trigger was moved to 0.5ms (before the OH), no plasma current developed at all.

When both OH and VF triggers were moved forward by 0.5ms, the plasma current rise time again stayed with the VF.

Moving the gas puff time from -95ms to -95.5 ms did not result in any changes.

It was found that if the duration of the gas puff is increased enough, the plasma current becomes negative with the firing of the OH bias bank, and then correctly rises with the OH start bank. In these cases the plasma evolution is still very different, and plasmas last no longer than 5ms (compared to 8ms in previous shots). Example shot: 75489. The threshold between the two cases (plasma current rise with VF, and plasma current rise with OH bias) is somewhere between a puff duration of 1.2 and 1.3 ms. Previous shots were taken with a puff duration of 1.1 ms.

Tuesday August 21 2012 6:11 pm Niko 75491 - 75569 Feedback Gain Scan

The unusual plasma behaviour that affected yesterdays shots was gone this morning.

A feedback gain scan was completed. The scan was performed at the phase that seemed most effective at mode suppression in previous experiments (85 degrees). It was found that the gain at which previous experiments were run was very close to the maximum. Attempts to further increase the gain generally resulted in disruptions shortly after activation of the feedback system.

The results of the gain scan confirm expectations. Without feedback, the mode amplitudes are strongest. With higher feedback gain, the mode amplitude decreases. With too much feedback gain, the plasma disrupts. At the highest non-disruptive feedback gain, the mode amplitude is at 78% of the no feedback amplitude.

In the data from both today's and previous runs, it was found that prior to the major radius crossing 92cm, feedback has effects only when running at amplifying phase. Feedback mode suppression only becomes visible once the plasma has crossed 92cm. It should be noted that it this time all toroidal arrays report uniform amplitude increases, so this is not just an effect of the plasma expanding in minor radius and coming closer to the upper and lower arrays.

Thursday August 23 2012 6:52 pm Levesque 75615-75718 Toroidal asymmetry

Made plasmas with toroidally asymmetric shell configuration to look for possible modulations in mode amplitude/frequency with toroidal angle. Shells in sections 1, 2, 6, and 7 were retracted for half of the run. A couple shots overlaid sufficiently with the desired reference shots (refs = 72468 and 73009, matching shots = 75629 and 75706), but most were different. A different reference shot type may be selected tomorrow due to difficultly in matching the original references. D_alpha light from the spectrometer was lower than usual throughout the day. Several "static fill" shots were taken during the day to check for issues with the puff system. Results were as follows. Shot 75698/9: puff=1000us, fill P=3.3e-5torr ; shot 75709/10: puff=1200us, fill P=3.0e-5torr ; shot 75711: puff=1500us, fill P=5.4e-5torr. The puff/fill discrepancy between the two shorter puff cases is concerning, and should be investigated with more static fills at some point. The drop in fill pressure happened despite the regulator pressure remaining at 21psi throughout the day. Niko and Pat also ran test shots to measure the transfer function of a full control coil and amplifier system on a spare shell (shots 75632-75689).

Friday August 24 2012 6:55 pm Levesque 75719-75771 Toroidal shell asymmetry

Continued running with toroidally asymmetric shell configurations. Plasmas were more reproducible than yesterday. The following configurations were used today: Sections 1,2,6,7 out 4cm; Sections 3,4,8,9 out 4cm; and Sections 1,2,3,4 out 4cm. Several shots with mode amplitude and rotation modulation have been observed, as measured by the TAr sensors. These modulations occur later in the shot, as q* rises to ~2.7 from below, though extensive analysis has not yet been done. These have not yet been correlated with the toroidal angles of retracted wall segments, and are not always observed. About 25 shots were close enough to the reference shots to allow direct comparison

Regarding the D_alpha emission problem observed yesterday: light levels jumped up after reinserting shells in sections 1,2,6,7, but not quite to levels from the reference shots, even with a longer puff time (1300us versus 1000us). Note that the spectrometer fiber bundle is between sections 6 and 7, and local shell changes may affect the reflected light seen or the interaction with neutrals.

Monday August 27 2012 7:13 pm Levesque/Rhodes 75772-75851 Toroidal Wall Asymmetry

Finished low-initial-q* experiments with toroidal wall asymmetries. The following configurations were used today: Sections 1,2,3,4 out 4cm; and Sections 3,4,5,6 out 4cm. Mode amplitude and rotation modulation were again observed in TAr sensors in some shots, but in-depth analysis remains to be done. This concludes the wall asymmetry experiments pending further analysis and modeling results. Noteworthy shots: 75834 and 75847.

Control coil amplifier response testing continued in the middle of the run day (shots 75794-75827) to verify the hall probe's frequency response.

Tuesday September 4 2012 5:01 pm Peng 76210-76268 const q round 2.5

* It is found that for such q profile the natural mode is smaller compared to other shots. ( for other shots, the mode amplitude decreases when q reaches 2.5)

* static phase, changing amplitude, perturbation is applied on such shots hoping that we could isolate the effect due to perturbation. However, it was found that for shots in such style, the mode amplitude does not quite feel the perturbation. ( the behavior is similar with/without the perturbation). Later, a phase flip is added to the perturbation, and it still doesn't have noticable effect on the mode amplitude. The phase flip does have some effect on the background trace of the signal on some sensors. The typical shot number is 76264 ( with flipped perturbation)

Thursday September 6 2012 7:16 pm Peng 76269-76340 Steady 2/1 mode

* Shot for steady 2,1 mode is reached.

* Roughly constant Ip, MR, and q for a duration around 3ms, then the trace varies. Star shot is 76320

* This shot style is finally reached around 5 pm so not too much sample points was taken. ( 76320~76340)

* It is not obviously seen that the mode gets locked by the perturbation, but shots with/ without perturbation seems to be different. Further analysis is needed to identify the effect of perturbation on the mode activity.

Sunday September 25 2012 8:05 pm Levesque 76345-76397 Clean up shots

First run day after the recent up-to-air period. The base pressure at the start of the day was 3.8e-8 torr, while normal operating base pressures have been around 1e-8 during the middle of run days. Plasmas were not the standard quality, as was to be expected. Initially plasmas had very erratic major radii, even with standard bank and puff settings. Some plasmas early in the day lasted up to 13ms. Long-duration plasmas had little or no x-ray emission according to the SXRT detector, very large R0, current ~6kA, and erractic q* between 2 and 4. The long shots had very large spikes in D_alpha light as seen by the spectrometer (~3-5 volt spikes, whereas normal disruptions give ~0.4 volts). The first few plasmas of the day seemed to look more whitish, instead of the usual purple. Further plasmas were not observed through a port by eye due to the lack of a control room sightline.

Performance for normal plasma shots improved throughout the day. Plasmas with the standard starting parameters and current ramp lasted up to ~3ms. Cleanup shots will continue tomorrow.d

The fast camera was active for most of the day, however its D_alpha filter was not on. Very interesting rotating high-m structures (m ~= 6-8) could be seen with the camera during disruptions and as plasmas were falling inward. Shells in sections 7, 8, and 10 were retracted for half of the day for the camera sightlines. Shell 7B is presently stuck in the IN position, as the motor spins without gripping the threaded shaft. This can presumably be fixed by tightening a set screw, but the correct size hex key could not be found today (the screw head is either metric, or is slightly smaller than a standard 5/64 english size, which we have seen before in a batch of poor-quality screws -- filing down a 5/64 hex key will work).

Monday September 26 2012 9:18 am Levesque/Peng 76398-76455 Clean up shots

At the start of the day, base pressure on the control room ion gauge was at 3.8e-8 torr again. This is not an improvement from yesterday. The RGA showed the dominant partial pressure to be nitrogen, with water and oxygen down by factors of 3.5 and 5.3 respectively.

Lowering the puff time to 200us (from 1250us) decreased the amount of D_alpha light at the startup slightly, but there was still more light than usual during the main plasma discharge. Eventually we returned to a more standard puff time of 1000us.

Explored a broad range of bank and puff settings, with little or no improvement from yesterday for most of today. At the end of the day, we were able to achieve plasmas that looked more typical after raising the bank voltages significantly. The highest-performance plasma achieved was 76451, with a breakdown current of 14kA and peak current near 19kA. The high-current shots also started having disruptions that looked more normal compared to previous disruptions. Reasonable plasmas still only lasted ~3ms, versus the standard 5 or 6ms..

Monday October 1 2012 7:20 pm Levesque 76671-76679 Testing SXR fan array

At the end of last week, we leak checked and tightened some bolts, then baked over the weekend including the GDC bellows which was not baked previously. The base pressure came down to 2.6e-8 today after turning off the bake. Partial pressures are still dominated by nitrogen, with oxygen down by a factor of 4.4 and water down by a factor of 5.5.

The main purpose of the run was to test the new SXR fan array and filter. So far I have only looked at the raw data, and it appears that 13 of the 16 chords are working well and are in rough agreement with each other. One of the working chords has about 10 times as much noise as the others. All 3 of the faulty chords saturate the CPCI (+10V) at the plasma breakdown. We have not done any troubleshooting yet to see what might be wrong with the bad chords.

Thursday October 11 2012 6:40 pm Byrne 76680-76704 cleanup

Shots were difficult to push outboard, eventually had plasmas centered at 94/95cm by reducing VF (110/3.8k) and increasing OH (300/6k/2k). Shots were taken with base pressure from 30-50nTorr. Base pressure was at 24 nTorr, up from ~20nTorr before opening. Most of the influx seemed to be water. Pressure is at 32nTorr and dropping, although it is not clear what value it will settle at.

Originally trying to increase OH alone led to plasma failing to break down. At 7kV OH St and 3kV OHB, there was no appreciable Ip. This may be due to the plasma not being able to superimpose its own field geometry on the stronger OH field, although I_oh still passes through 0 during "breakdown"

In all cases plasmas were shortlived, had low Ip and had small current ramps, only weakly affected by OH settings. Disruptions seemed not to be caused by resonant surfaces, as q varied greatly, up and down, over the course of a shot, before the plasma crashed (see shots 76702,76704)

All disruptions seemed gentle, as plasmas crashed over a longer time than target shots, and crashed from lower energy. Not sure why this should be so. One interesting shot is 76703, which despite high OH bank settings immediately crashed inwards, while still maintaining a fairly constant Ip profile, even with MR at around 88cm. not sure what to make of that.

Also, when setting up this morning, the cooling water was running, and the sign was flipped to open. When asked CTX was not using it. It was shut off sometime between setup and this afternoon, when Chris discovered it, most likely lunch as the room was not being monitored, and had not yet been darkened and locked up for running. No digitizers seemed to be affected, however, if you're using equipment (racks, cooling water, etc.) that serve multiple users, even if you were the only one using it when you turned it on, please ask around before you shut it off again.

We expect to work on the Thomson scattering Beamline tomorrow, with cleanup shots taking place around that work. Expect to setup in the morning, with hi-pot and most of the running happening after the work in the late morning/early afternoon. Hopefully the new Cryo will arrive tomorrow or Monday, allowing us to take shots at lower base pressure

Friday October 12 2012 5:37 pm Byrne 76708-76720 cleanup/stability/camera calibration

Theme of the run was cleanup shots. Banks were run with very low voltage on VF, and roughly the same voltages as the end of yesterday's run on the OH banks. When the OH electrolytic bank was raised above 300V, it failed to reach a full charge before the TF, and continued charging after the shot. It was dumped and the run was continued

The final four shots are all comparatively long lived, if very low current. VF settings were 2kV on start, and 70V electrolytic, with 5.7kV on the OH start and 270V on the electrolytic. Bias was left at 2kV as there was not time in the runday to vary it. However, it seems likely that the plasma *was* run in a positionally unstable regime for some or most of the shot. Major radius starts at around 95 and moves steadily outward to around 97cm, by 2.25ms, at which time the plasma crashes back inwards, only disrupting around 5ms.

Sarah was able to achieve framerates of 80k fps without the filter, but needs a slower framerate with the filter installed. This may require work to discern plasma dynamics.

Tuesday October 16 2012 5:07 pm Hughes 76721-76731 Magnetic Force Measurements

Aside from some crates-only shots to verify that the baseline signal was coming through properly, took TF-only shots to measure the force on a sample of magnetic material close to the vacuum vessel. After ramping up from 1.1kV, took 4 shots (76727-76730) at the usual 6.1kV. Analysis has yet to be done, but a preliminary glance in jScope showed a lot of shot-to-shot variation that I don't understand yet. More to come.

Friday November 9 2012 10:06 pm Levesque 76732-76782 Cleanup and shot development

Ran cleanup shots and shot development to try recovering our standard shot style and duration. Base pressure at the start of the run was 1.1e-8 torr. Standard-style shots started out short early in the day, then eventually recovered to durations matching those of shots from the end of September (with base pressures at 3.8e-8 and 2.6e-8 torr). Standard shots did not get longer after reaching these reference shots (ending at around 4.5ms). Tried low, medium, and high starting plasma currents (~4.5kA, ~9kA, and ~13kA) to explore discharge evolutions. Several low- and high-current shots lasted slightly longer than the present limit on standard shot duration, but returning to standard shots after these longer shots did not improve their duration. High-current shots appeared to have our normal disruption characteristics.

SXR emission did increase throughout the day, as seen by the high-field-side SXRT diode and the SXR fan array. Modes also became more coherent throughout the day, possibly in part due to larger amplitudes above the noise floor.

Had trouble getting below q* = 3 with the standard shots crossing from above -- shots that began with q* anywhere above 3 could not get to q* < 3 except transiently. Minor disruptions appeared to be caused by, or coincide with, 3/1 and 4/1 mode activity when 3 < q* < 4. 4/1 modes were especially more apparent than they had been in similar discharges before the recent diagnostic upgrades. 4/1 modes often burst to moderate amplitudes after minor disruptions, even when they weren't strong or obvious beforehand. Shots that initiated with q* ~ 2.5 at the end of the startup seemed to evolve as normally expected (but still short), and had large 2/1 modes when going below q* = 2.5. Large loop voltage oscillations, along with small plasma current oscillations, were seen during the 2/1 modes.

One of the new feedback sensors amplifier boards was installed in section 10. All channels in this section now appear to work and are consistent with each other. However, these sensors could not be directly compared with others since this was the only section with the shells retracted (for the fast camera). The run will continue on Monday, with shells in sections 7-10 retracted to improve the fast-camera view.

Monday November 12 2012 7:00 pm Levesque/Angelini/Byrne 76783-76841 Cleanup and shot development

Continued cleanup and shot development. Still unable to get long-lived or reproducible shots. D-alpha emission, SXR emission, and disruption characteristics are looking more like our standard though. The longest shot disrupted at ~ 5ms. Startup parameters were different than Friday's run due to the different shell configuration -- 4 sections were retracted today, while 1 section was retracted Friday.

Nitrogen was the dominant partial pressure before running today and Friday as usual, with oxygen down by a factor of 3.3 (Fri) or 4.6 (Mon). After the end of the run each day, much of the oxygen had combined into water as expected, leaving water as the second-highest partial pressure: Water down from N2 by a factor of 2.5, O2 down from N2 by a factor of 14. Over the weekend, the leak rate was high enough that the O2 partial pressure became secondary and large again.

The run will continue tomorrow, hopefully with the O2 partial pressure lower than this morning due to consecutive run days.

Tuesday November 13 2012 6:54 pm Levesque 76843-76878 Cleanup shots

Continued cleanup shots. Plasmas today improved significantly versus yesterday and Friday, and are much closer to standard operation. Minor disruptions frequently occur near 1.2ms, which also happened before the up-to-air under certain operating conditions. Was finally able to get plasmas crossing q*=3 from above, whereas in the past couple of runs, q* would not cross 3 due to minor disruptions or steady but fast decreases in R⁰ or I^p. The longest shots disrupted at around 6ms, and had reasonable evolution of all plasma parameters. 4/1 and 3/1 modes have become more clear as well. However, plasmas are still not as long-lived or as reproducible as they were before the up-to-air.

Nitrogen was the dominant partial pressure before running today as usual, with water down by a factor of 4.5 and oxygen down by a factor of 13.4. This was an improvement over the past couple of run days.

Monday November 19 2012 11:42 am Angelini 76879-76899 Inboard limited clean-up shots

The purpose of the run was to continue cleanup shots and attempt inboard limited shots. Each shot was run with a base pressure between 17 and 33 nTorr. On the whole, the shots were short-lived, but that may have been due to the settings for inboard limited shots. While the inboard shots were mostly unsuccessful, there are some interesting shots in which the q drops from 5 to close to 2 almost linearly with time. The poloidal modes are easily seen from the BD. I'd like to see if I can reproduce this BD from the fast camera data. The shots from 76894 to 76899 display this behavior. Shots 76895 and 76897 give especially clear BDs.

The fast camera seemed to be operating correctly. I've noticed "glints" of bright pixels occasionally during the course of a video that may be some form of pickup instead of something physical. This will need to be investigated.

Jeff says that one chord of the SXR fan array has been repaired and the other chords seem to be working. The SXR was turned on for all shots.

Toward the end of the day, the lack of air conditioning started to affect the data storage. Data in shot 76899 may be suspect.

Monday November 26 2012 11:57 am Qian/Sarah/Chris 76900-76911 Plasma shot

This is the run report for last Wednesday.

The objective is to get familiar with the procedure of running the feedback system and using the analyzing code, so that we don't have to re-invent the wheel when Niko left.

I've run the feedback procedure on several shots and system itself is running as expected. ( Plasma is not good though)

Then we tried to run the phase flip on some shots for the fast camera but it didn't work out very well.

For the last shot, OH elect didn't fire. We then dumped it and shut down the machine. ( The reason why it didn't fire is not clear.)

Wednesday December 5 2012 5:13 pm Levesque ****** Gasket replacement and GDC

The chamber was vented, resealed, and pumped down on Tuesday Nov 27th to fix leaks by replacing copper gaskets. The standard 10" copper gaskets in 6 sections were replaced with annealed copper gaskets. The remaining 4 sections that were not replaced include the 2 pump stands (with Teflon gaskets), section #2 (which was not found to have leaks, and could not be replaced quickly), and section #9 (which has not been opened since the shell upgrade in 2010). Note that annealed gaskets were not used after the up-to-air in September since we did not have them in stock. The unused leaky gate valve in section 5 was replaced by blank conflat -- this will be replaced with a better gate valve during the next up-to-air.

While the chamber was open, a brief inspection revealed that the fast-camera background was damaged from the September glow or standard plasma operation. Pitting to the black surface created random speckles of reflective dots. In spite of the speckling, the background is still an improvement over looking directly at the chamber, thus the background was left as is.

After 1 day of leak checking, we glowed with 90% D2 + 10% He for 2 hours, followed by a 100% He glow for 3 hours. Glows were initiated with fill pressures of 50mT. Adding a large inductor to the GDC circuit did not prevent sparks/arcs on the fast-camera black background. Reducing the pressure to < 1mT as read on the convection gauge kept the glow emission only on the plasma side of the shells, and prevented sparks on the fast-camera background. The GDC current only decreased by ~15% with the large decrease in fill pressure. The total time of sparking on the black background with the higher fill pressure was roughly 1 hour, versus 3 hours in the September glow. One of the differential roughing pumps was unintentionally unplugged before/during the glow, and wasn't discovered until the next day, which allowed a significant amount of oxygen back into the chamber.

The bake was on during the morning of Nov 28th through Dec 4th evening. One differential roughing pump seized-up during the weekend, and again allowed more air into the chamber for about a day or less. Both cryo pumps were regenerated (separately) during the bake. After the machine cooled down, the base pressure this morning was 7.5e-9 torr. Two of the differential pumping lines on spool pieces between sections 3 and 6 are suspected to be responsible for significant leaks contributing to the remaining base pressure.

Water resistors in the OH electrolytic bank dump were replaced by heating-element coils. This has decreased the dump RC time to ~8 seconds, versus the previous many-minutes decay time. The bank is now much safer following a dump or power loss.

Wednesday December 5 2012 6:31 pm Angelini 76912-76940 Fast Camera D-Alpha

The first run day after the up-to-air primarily produced short, but reproducible, 4 ms plasmas. The fast camera was used with section 10 open, but with the remainder of the shells left in closed position. This allowed for the shells to be conditioned from the run. While Abel Inversions will not be possible without the black background, the plasmas were cool enough that the D-Alpha filter could be used for a majority of the shots. A quick peek at the fast camera data shows some mode structures that may be able to be correlated with the magnetics data. Further analysis is needed to be sure.

The first shots of the day were unusually long-lived 15ms plasma-filament shots. The shots appear similar to a normal shot until about 3.5 ms where instead of falling into the typical disruption, the plasma breaks into smaller tubes. We were able to capture some of these shots with the fast camera and I can show them at the next meeting.

The new OH dump was tested and performs wonderfully.

Monday December 10 2012 6:08 pm Angelini 76941-76981 Cleanup

There was improvement in the cleanup shots today from last Wednesday. I started with the same run settings, and got plasmas that were virtually identical to the plasmas at the end of Wednesday's run. The plasmas continued to improve from there. The shots today were, on average, 1 ms longer than the ones last week. They are, however, still short at 4.5 to 5 ms. There seems to be an issue with channel #12 on the SXR array. Starting with shot 76955, it displayed strange data that didn't fit with the other channels. The issue is intermittent, though. The SXR data for shots 76974, 76975, and 76981 looks fine. The shells for this run were all inserted except for section 10. I'll open up the rest of the shells for fast camera analysis once we've moved forward from the cleanup phase.

Wednesday December 12 2012 6:22 pm Angelini 76983-77013 Cleanup

Cleanup shots continued today. There was still improvement in the shots. The length only increased on average about a 1/2 ms, but the modes became more coherent. Clear 3/1 modes can be seen on a number of the shots later in the day.

The SXR array channel was fine for most of the shots today. The problem with the channel is intermittent.

The plan is to continue running tomorrow.

Thursday December 13 2012 7:02 pm Angelini 77014-77054 Cleanup

The run today was successful. The shots are on average about 5.5 to 6ms long and are becoming reproducible. The edge q was also lower. Shots that reach q of about 2.4 are 77045, 77048, and 77053.

The SXR array worked for most of the shots today. Channel #15 had trouble earlier in the run day. This is a different channel from before.

Earlier today, Jim found a leak in the Cryo pump in section 7. This is expected to be repaired the next time the pump needs a regen.

Friday December 14 2012 7:55 pm Angelini 77055-77083 Cleanup/Fast Camera

The shots continued to improve. Starting with shot 77075, shells 7,8,9 and 10 were opened for the fast camera. At the end of the run day I was able to get shots whose behaviors were similar to shot 70000. The long shots are 77079, 77081 and 77082. 77076 and 77077 weren't long, but their q profiles got rather jumpy as the plasma reached q=3. The 3/1 mode is strong there.

Jim seems to have fixed the leak in the cryo pump. The base pressure was in the 1e-8 range for most of the shots today.

The SXR array was fine for the run today, but nothing has been fixed yet.

I'm in the process of creating stripy plots from the fast camera data. I'll show those at the next meeting.

Monday December 17 2012 7:00 pm Angelini 77084-77122 Fast Camera Phase Flip

The shots remained long throughout most of the day. I think it's safe to say the cleanup phase is complete.

The primary goal of today's run was to get phase flips working. While there are still a few things to work out, I was able to get the phase flip program running by the end of the day with Niko's help. I haven't done any fast camera analysis yet. I'm not sure I will see much because the q and MR were changing drastically through the phase flip for most of the shots. Tomorrow the plan is to run phase flips while the shot profiles are more constant.

The SXR array didn't show any lost channels today. That's not to say they won't have problems in the future.

Today's shot with the strongest 3/1 mode is 77104. Shot 77116 had the lowest q of 2.28. Unfortunately, I don't think that's reproducible.

Tuesday December 18 2012 5:29 pm Angelini 77123-77146 Phase Flip/Fast Camera

Today the majority of the shots were phase flips. The hope was that something would show up on the fast camera during the phase flips. While by eye, one can vaguely see a "stalling" of the rotation in the video, I have yet to find a way of displaying this numerically. I can show some of the videos tomorrow. The shot with the strongest locked-mode was 77132. 77137 and 77138 get honorable mentions. Shot 77131 had the strongest 3/1 mode.

The six feedback coil currents as digitized by the A14 were terribly off. Jeff assures me that the signal sent to the coils were correct, but they're being recorded incorrectly. I don't think a single shot today had them displaying correctly.

The SXR continued to work fine.

The base pressure was a little higher today, but the shots seemed to be fine. Jim mentioned that we may need to regen the cryo pumps.

Wednesday December 19 2012 4:56 pm Angelini 77147-77173 Phase Flip / Crate Failures

The goal of today's run was to run more phase flips and see how the fast camera responds. I got through about 9 functional shots today. After the meeting, the heat was too much for the server room, and the crates started failing. After being unable to cool them down properly, the run was ended early.

For the shots where the crates didn't fail, the six feedback coil channels operated properly for some of them. I don't know if the heating problem had an effect on them, too.

The phase flip run will continue tomorrow if the temperature will allow for it.

Thursday December 20 2012 2:48 pm Angelini 77174-77191 Crate failures / Phase Flip

The purpose of today's run was to continue phase flip shots with the fast camera. Unfortunately, the temperature problem in the control room and server room has continued to cause crate recording failures. The failures are reported from the basement, south and north racks, which lead me to believe it isn't the racks themselves with the problem, but the server. I ran some crates-only shots before setting up, and everything seemed fine. For the first few shots, the crates performed normally, but they started failing as the run progressed. Until the A/C unit is repaired, I do not think it would be useful to attempt any further runs.

2013

Thursday January 24 2013 6:31 pm Angelini 77192-77230 Cleanup

Today's run was primarily to clean up the machine after the break and ensure everything is working. By the end of the day the shots had cleaned up almost to where they were before the break. The plan is to attempt phase flips tomorrow afternoon if all continues to work well.

The SXR array lost channel 12 during shots 77209 and 77230. The problem is intermittent.

Some shots today showed strong 3/1 modes; 77222 and 77225 are both good. I'll attempt a BD on the fast camera data.

Surprisingly, the crates worked without any problems today, perhaps due to the cold weather. The leak in the A/C was fixed today, and the coolant is supposed to be replaced tomorrow.

Friday January 25 2013 5:18 pm Angelini 77231-77244 Standard plasmas

The goal of today was to do phase flips, but due to problems with caliban, I could not run the program. There may be a hardware problem with the computer. On Monday we'll attempt to debug the problem in depth.

The plasmas today were as good as the ones before the break. They would have made good candidates for phase flips.

I didn't observe any problems with the SXR array today.

Tuesday January 29 2013 5:51 pm Angelini 77245-77280 Debug

Yesterday the dual-channel PCI-e card from caliban was sent back to NI for repair/replacement. Until we get it back, the control coils cannot be used. The goal of today's run was to ensure everything was working for when the card returns. Unfortunately, I had problems all morning. One culprit was discovered to be the pulse amplifier for firing the TF crowbar. The signal was sent through another amplifier instead and that fixed the TF firing problem. Other camac problems were present which may be related to the bad pulse amplifier, but it is unclear whether changing the amplifier was the fix for those.

The ten plasma shots taken today were comparable to the last good shots from the previous week. Shot 77277 shows a rather fast 3/1 mode. Shot 77280 has a slower, but strong, 3/1 mode from 3 to 3.5 ms. Both may be clear to see on the fast camera.

The SXR had issues with channel 12 for shots 77279 and 77280.

The base pressure bottomed out today at 1e-9. This is the best vacuum we've seen over the past 5 months.

Tuesday February 5 2013 6:36 pm Angelini 77281-77325 Mode-seeking

The purpose of today's run was to get camera data of very clean 3/1 modes. In the future the fast camera's spacial resolution will be modified for comparison. Shots today with strong modes were 77314, 77318, and 77324. The 3/1 mode lasted from 2 to 3.5 ms cleanly in shot 77324. Shot 77312 has a short burst of a 4/1 mode.

The first shot of the day exhibited camac problems. A thorough investigation of the connection showed no problems and all subsequent shots went smoothly.

The SXR array continues to have intermittent channel issues.

The base pressure is holding steady in the low nanoTorr range. There seem to be no vacuum problems at this time.

Thursday February 7 2013 6:37 pm Hughes & Angelini 77327-77356 Shells Out 4cm

Several cleanup shots early in the day. Sarah started the run day, but the fast camera was being ill-behaved, so I took over to do cleanup shots. Experimented with shots that swing in and out (77342-77348) with particularly good results on the last two. From shot 77349 onward, all shells are back 4cm. From 77351 VF_ST timing is changed from 860us to 780us. Plasmas bounce around wildly in major radius until shot 77353. Jeff's favorite shots of the day are 77353-77355. 77354 sits at 91cm for a long while, and 77355 is very long. In both shots, q moves very smoothly through 3 from above. 77352 is interesting for a 2-1 tearing mode which causes a disruption, in which the plasma jumps in by almost 2cm, q jumps from 2.1 to 2.8, and a 3-1 kink flares up. Unlike its usual intermittent failures, SXR Fan channel 11 has been dead all day.

Friday February 8 2013 6:32 pm Levesque 77357-77378 Retracted walls

Continued run with walls uniformly retracted by 4cm. 3/1 modes appear to persist to lower q* values than for similar shots with walls fully inserted. Can see nice exponential growth of 3/1 modes for several shots with constant major radii from today and yesterday. Most shots had major radii in the up/down limited range for most of the discharge, with q* crossing smoothly through 3.

The fast camera was active for most of the day. All SXR fan array chords were working for all shots today. Midplane SXR signals from the active tomography chord have been lower than usual in comparison with the normal wall configuration, in spite of plasmas being closer to the diode. Loop voltages have been lower than in runs for the past month or so, but this is most likely due to the lower OH Electrolytic voltage setting. The retracted wall experiments will continue on Monday.

Noteworthy shots: 77362, 77364, 77372, 77374, 77377

Tuesday February 12 2013 3:50 pm Hughes 77379-77397 Ubiquitous Triggering Failures

Today's run was characterized by almost everything going wrong. Most of the run day was spent debugging the misbehaving basement J221s which failed to trigger the TF, the OHEL, and the OHST banks. All banks dumped safely, however. Additionally, there have been intermittent South and North rack failures.

Other than one anomalously successful vacuum shot in the middle of the day, we tried copying the J221 input-5 signal into the input-6 node in the tree, and this improved behavior somewhat: we were able to get all banks to fire on a vacuum shot with the TF lowered to 1100 and all other banks uncharged (77394), and then on a plasma shot at normal bank settings (77396). However, this was the only remotely normal shot of the day, as the VF crowbar failed to fire on shot 77397, causing the VF to ring.

We do at least have some interesting fast camera data from 77397, where we can see the plasma swing in and out, with the plasma current (and brightness) peaking and falling, due to the ringing of the VF.

After the VF crowbar failure, we decided to end the run day rather than risking damage to the banks, and removed the J221 input-6 hack.

77397 is pathological, but the nearest thing to a star shot for the day. 77396 is much more typical, but not very interesting.

Thursday February 14 2013 5:51 pm Hughes 77398-77428 Shells Back 4cm

An actual run day! CAMAC trouble with failure to trigger the OHST and TFST persisted in the morning until we repeated the hack-fix of copying J221_02:5 to J221_02:6 and J221_09:5 to J221_09:6. Good shots followed, but the South Rack consistently produced TR6841_16 and TR6841_17 errors. Additionally, one of the highest-number SXRF channels (14 or 15) was losing signal intermittently throughout certain shots for some unknown reason. The run day ended with another failure of the VFEL to fire, ringing the VF bank, and the decision to debug in the morning.

Plasma shots are from 77401-77428. Shots of particular interest are the following:

• 77408 - Extreme outboard plasma. Manages to push below q*=2 briefly.
• 77419 - Outboard limited plasma. Hovers at almost exactly 93cm for most of the shot, and then crashes inward sharply.
• 77420 - Outboard to up/down limited plasma. Major radius curve closely resembles that of the target shot, 70000, although the current is different, producing a different q* curve.

Most of the shots today were hampered in approaching 70000 or steady major radius due to the "early mode" where the plasma kicks inward at 1.2-1.3ms on every shot. One notable feature of the recent plasmas is a spike on the leading edge of the current trace which is much larger than the leading-edge current spike on shot 70000. This may be due to one or both of the OH and VF crowbar banks firing a little too late to prevent the current trace from turning over after the start banks peak, but this hasn't been investigated.

All recent plasmas also show SXR midplane traces markedly lower than that of shot 70000, but since their loop voltages are also lower, this may indicate a difference in impurities (that recent plasmas are cleaner) rather than a higher temperature. The triggering system will be examined before attempting to run with banks charged tomorrow. The following modifications to standard configuration are being left in place:

1. All shells are back 4cm from "fully inserted."
2. Of the J221_02:6 and J221_09:6 nodes in the tree, each has a signal cloning the corresponding output 5.
3. The VFST time is set 80us earlier than standard.

Friday February 15 2013 5:26 pm Hughes 77429-77451 Shells Back 4cm

Since yesterday's run day was ended by the VF crowbar's failure to fire, we began the day by testing the krytron triggering. Everything seems to work, although the previously mentioned 6th output trick is still in place. As noted before, the VFST timing is still 80us earlier than normal, and of course all shells are still back by 4cm.

The initial vacuum shot was 77430.

77432, 77440, 77449, 77450, and 77451 are good shots for comparison to 70000. 77440 is probably the closest in R_0 and q*. 77451 is also very close except during the period from about 3ms to 5ms, when it stays outboard longer and then falls in quickly at a later time.

Other interesting shots were 77437, 77439, and 77444.

In 77437 and 77439, q* is extremely stable for about 1ms, starting close to 1.5ms.

In 77444, a strong, steady mode exists from 3.5ms to 5.5ms.

As in most of the recent shots, the presence or absence and the strength of the early major-radial-inward jump at ~ 1.3ms is the strongest determining factor in the traces of R_0 and q*, far outweighing effects of bank settings. It may be worth looking at a population of recent (i.e. shells-back) shots and a population of shells-in shots, and comparing the rate of occurrence and the amplitude of this radial jump.

As of shutdown, the machine is still pretty well-behaved, so the longer startup procedure may not be necessary on Monday.

Monday February 18 2013 6:45 pm Angelini 77452-77488 Fast Camera Shell 8 Asymmetric

The goal of today's run was to determine what effect the shells have on the fast camera data. Once things were working, only the bottom shell in section 8 was fully inserted with the the top shell and all the shells in the other sections extracted 4 cm. It seems that just that one shell has an effect on both the plasmas themselves (leading to a modification of the VFS timing to 760) and the fast camera videos. Once the shell was inserted, all videos showed the long bands, even when the plasmas were more inboard-limited.

To counteract the continuing Camac problems, the two TR6841 modules were deactivated and disconnected from the south rack crate. No further camac errors were occurred during the day.

The PCI-e replacement card was installed in the caliban computer in the south rack. All the RTM-T modules were correctly identified and initialized. Some tests of the control coils were made at the beginning of the day. A modification to the do_awg program was made to address the missed samples problem. Initially this modification was causing spiky currents on the control coils at about 5.5 ms. That error was corrected and the do_awg program now gives reliable phase flips.

During the insertion and extraction of the shells in section 8, an issue with the top shell motor was discovered. The set screws weren't well-placed and the motor was spinning without moving the shell. The motor was removed, repaired and replaced. The top shell in section 8 moves properly now.

The SXR fan array dropped channel 15 earlier in the day, but it seems to be working again.

The run tomorrow will continue the shell effect studies. Good shots for today are 77483, 77487, and 77488. 77483 and 77488 are outboard limited, while 77487 is more top-down limited. All the shots show the long stripes in the fast camera data.

Tuesday February 19 2013 6:53 pm Angelini 77489-77524 Shell 8 effects on fast camera

At the start of the runday, the base pressure was up in the 10^-6 range. The cryo pump in section 6 had shut down and released its trapped gas. After it was gated off, the base pressure came down quickly and it was possible to run. The base pressure throughout the day were higher than typical, but the run was able to proceed.

The goal of today's run was to see what effect the bottom shell in section 8 may have on the fast camera data. At the start of the day, the shell was fully inserted. It was pulled out in 1 cm increments throughout the day. When the shell is inserted, both outboard and inboard-limited plasmas show long banding in the fast camera data. As the shell is set further and further out, outboard limited plasmas are much more difficult to produce. The inboard-limited plasmas show less banding and more of a pinwheel. Inboard star shots are: 77496 (bottom shells in section 8 fully inserted), 77499 (retracted 1 cm), 77505 (2 cm), and 77517 (3 cm). As outboard-limited shots became more difficult to produce as the shell was retracted, further study is needed to form a coherent set of comparable shots.

It is not yet clear whether the proximity of shell 8 to the fast camera view has an effect on the banding in the data. At the end of the day, with shell 8 withdrawn 4 cm, the bottom shell in section 4 was fully inserted. A few shots were taken, but more runtime is needed before any conclusions can be drawn.

There were no camac problems to report. Removing the TR units in the south rack seems to have fixed the problem.

The SXR array seemed to work fine today.

Thursday February 21 2013 6:47 pm Angelini 77525-77550 Fast Camera Shells Retracted

Today's goal was to see the effect the shells may have on the fast camera images when the plasmas are outboard-limited. For the first half of the day, the bottom shell in section 4 was completely inserted while the rest of the shells were retracted. The second half of the run was with all the shells retracted 4cm. I was able to get a few outboard limited shots in each shell configuration, and some shots which transformed from being outboard to inboard limited.

Even when the shells are fully retracted, it is possible to get long-banded fast camera bd images. These images are harder to find because outboard-limited shots often lack the coherent modes typically seen in plasmas which are top/down-limited. As the plasmas move inward, these banded images start to appear more like pinwheels.

The SXR array sporadically lost channel 11 during the day. By the end of the day it'd come back again.

The cryo pump in section 7 had been regen'd. The base pressure held steady in the low 10^-8 range throughout the day as normal.

Star shots for the day:

77535 and 77536 when shell 4b was inserted. They were more outboard limited although the video for 77536 didn't come out properly.

77543 and 77546 when all the shells were retracted. Shot 77543 follows shot 71000 quite closely.

Friday February 22 2013 5:30 pm Angelini 77551-77590 Fast Camera Shells Out Phase Flip

The goal of today's run was to try phase flips with all the shells open. Some of the shots during the first half of the day were successful. In the afternoon the plasmas became less dependable. I plan to continue the run on Monday.

The phase flips were most successful when the plasmas were run outboard-limited. This is possibly due to the proximity of the control coils, but also because outboard-limited plasmas tend not to have q_* values near a rational surface. Good shots for 0-degree and 90-degree phase flips are: 77557, 77560, 77563 and 77566. Further runs are needed for other phase angles.

A few of the plasmas went very far outboard and as a result had about 0.5 ms at a very low q_*. Two shots in particular are: 77583 and 77584.

The SXR array lost channel 11 intermittently throughout the day.

The vacuum was great today. The pressure at the start of the day was 3e-9 and remained in the low 10^-8 range throughout the run.

Friday February 22 2013 5:45 pm Angelini see previous addendum

I'd forgotten to mention in the last report that the make_tree.py script has been updated to reflect the removal of the TR6841 devices from the south rack. It also has output 5 duplicated in output 6 for the two J221 basement devices. It seems that workaround is still necessary because the first plasma shot of the day failed to trigger properly without it. The script still uses the previous gas puff and VFS timing, so those need to be updated manually in the tree whenever make_tree is run.

Monday February 25 2013 5:15 pm Angelini 77591-77637 Fast camera shells-open phase-flip

Today's run continued last Friday's open-shell phase flips. The shots continued to vary just after breakdown making reproducible shots rather difficult. I doubt I'll be able to take enough shots to make any statistical statements about phase flips with the shells open, but I hope to have enough to give a picture of what the fast camera sees under those circumstances. Tomorrow I'll continue the run, but after that the tests will be closed-shell. I'll compile the shots most representative of their phase angle and show them on Wednesday.

The A14_14 in the north rack threw some initialization errors during "prepare". Re-running the "prepare" phase seems to clear the error and there are no errors during the recording or analysis phases after the shot. I don't believe there has been any loss of data.

The gas puff valve was set a bit too open so the pressure read 22 psi. It was lowered, but now displayed about 20.5 psi so tomorrow it'll probably be adjusted again. I increased the puff time at the very end of the day, but it didn't seem to have a noticeable effect on the shots (they were already unpredictable by that point).

The SXR continues to have intermittent problems with channel 11.

The vacuum is still performing well. The starting pressure was 3.9e-9 this morning and it stayed in the low 10^-8 range all day.

Tuesday February 26 2013 5:07 pm Angelini 77591-77682 Fast camera open-shell phase-flip

Today's shots were rather unreproducible. As far as I could tell, there were no major hardware problems. With the same shot settings, the startup for the shots between 1 and 1.5 ms was very similar, and then after that the shot profiles diverged wildly. The early mode seemed to have a very strong effect today.

I was able to get enough shots to have a basic idea of what a phase flip looks like with the fast camera at different angles. I'll show the images in the meeting tomorrow.

The SXR continued to have problems with channel 11.

While we've had recording problems with the A14 in the north rack for the first 6 control coils, the problem seems to have spread to 6 others. Now all the coils in sections 1,2 and 3 show the phase flip late and spread out. It may be worth checking to see what the problem might be.

After Qian's latency run, I intend to run phase flips with the shells fully inserted.

Wednesday February 27 2013 5:06 pm Qian 77638-77639 crates_only

TR modules are put back in and were not producing errors for the crates only shot.

There are some driver issues that the feedback code could not be compiled and run. Will plan to resume the test after this driver issue has been sorted out...

Thursday February 28 2013 5:34 pm Angelini 77686-77713 Closed shells

The goal of today's run was to attempt phase flips with the shells closed. All of the shells have been fully inserted, except for the shells in section 10 (for the fast camera. Unfortunately, the shots weren't consistent enough for phase flips. Even if that weren't the case, there is also currently some problem with detecting the RTM-Ts in the south rack computer, so phase flips wouldn't have been possible anyway. Instead of running tomorrow, I'll be taking the time to debug the problem.

There were two shots whose edge q started above 3, neared 2.2, and then crossed back up over 3 again before the end of the shot: 77708 and 77710.

For all of the shots today, the SXR array was working fine.

The vacuum bottomed out at 1e-9 this morning and many of the shots had a base pressure in the 10^-9 range.

Thursday March 7 2013 11:50 am Angelini 77714-77753 Closed shell phase flips

This run report is for last Monday, the 4th. I was unable to submit the form at the time.

The goal of the run was to try phase flips with the shells closed. In the morning the plasmas responded well to the RMPs, but in the afternoon I couldn't seem to get the q where the plasmas would respond well. I'll try some different bank settings tomorrow, and then I'll try increasing the control coil amplitude. For the shots with passable phase flips, I was able to see some locking behavior as I did with the shells open. The "good" shots for the day seem to be 77723, 77726, 77733, and 77741.

While I am seeing the locking behavior, the appearance of the flips are more of the long banded style rather than the pinwheel style. This is consistent with what I saw for the shots with natural modes.

The SXR array temporarily lost channel 11 again today. It came back again at the end of the day.

The vacuum's at its 1 nTorr low. The base pressure was in the 8 to 10 nTorr range all day.

Thursday March 7 2013 5:56 pm Qian 77754-77774 crates_only

Feedback latency is tested with osciloscope and the latency is 20us.

Another thing I found is that there might be some mapping error in the feedback signal. When the program is set to take a one-to-one from whatever input to output channel, according to the map, Input FB02_S1P should be sent to control coil FB02_C1, but the signal was instead sent to FB09_C2. If I swap the two SCSI cables on the breakout board for control coils, then FB02_S1P channel would coorespond to FB02_C1. However, individual check with the awg shots suggests that the mapping for control coil should be correct and those two cables should not be switched. Then, there might be some mapping error on the input side. Haven't had time to check that.

The SCSI cable connection is restored to what it was on Tuesday.( Which I think is the correct connection it should be.) But it might worth double checking that the phase-flip is as expected for Friday's run.

Friday March 8 2013 5:32 pm Angelini/Peng 77775-77816 Fast Camera Phase Flip Closed Shells

Today's run focused on phase flips. While the run day was plagued with a succession of small issues, I was able to get shots at the remaining four toroidal angles to complete the phase-flip collection plot. The camera BDs for some of those shots were a bit shaky, but they do show the phase flip. Toward the end of the day, I saw some locking in the first half of the phase flip, but the plasma didn't seem to want to lock during the second half.

The fix to the cooling water system was successful. I didn't have any basement rack errors at all during the day. The pressure on the high pressure valve was 190 psi. We should keep an eye on that.

The TR6841 modules were connected and activated in the tree for all shots today. There were no problems with them. Because they are needed for latency checks on the GPU system, they should stay connected unless problems arise again.

I got more intermittent problems with the North Rack A14_14 module. Sometimes it would fail to initialize. Re-initializing would fix it. On a few shots it failed to store. It wasn't detrimental to the run, but it could continue to cause problems.

The beginning of the day marked issues with the control coil amps. 21 of the 40 fuses needed to be replaced. It seems the fuses we have are different from those recommended on the wiki and are more prone to breaking. For the future, the correct resistors will be ordered.

There was an issue with the control coil signals being switched in section 2. Qian was able to resolve this before the phase flip shots.

I had an issue with the OHE bank when I tried to set it above 320 V. While the sticker recommends keeping it below 350, it seems the voltage cut-out is set to somewhere right above 320. Once the OHE was turned down, everything ran correctly again.

The SXR channel 11 dropped in and out throughout the day.

The vacuum started in the low nTorr range and remained in the 10-15 nTorr range for most of the shots today.

We reached shot 77777 today. It was a simple, short cleanup shot. The longest shot of the day was 77811, which was similar to shot 71000.

Monday March 11 2013 6:42 pm Angelini 77845-77881 D-Alpha filter Fast Camera

The goal of today's run was to see what effect the D-Alpha filter had on the videos. I was able to get a few pairs of similar shots with and without the filter. The primary issue is that there is a lot of light lost with the filter in place. The mean intensity of the light with the filter is about 15% of the intensity without the filter. Even with the substantial light level difference, the dominant modes were similar with comparable shots. However, the "checkerboarding" did show up when the light intensity dropped low enough. To test if this was due strictly to the intensity level, I took some shots at the end of the day without the filter and with the fstop closed down. When the fstop is set to about 6, the checkerboarding becomes apparent. I suspect what I'm seeing is noise from the CMOS when the signal no longer dominates. I'll show some images on Wednesday to demonstrate this.

The first vacuum shot of the day failed to fire the VFE. The connection to the grounding cable was tightened and the problem did not reoccur.

The SXR continues to sporadically lose channel 11.

The vacuum is still excellent.

Monday March 12 2013 5:41 pm Angelini 77882-77910 D-Alpha / Disruptions

The first half of the day was spent taking D-Alpha filtered shots with shells 7, 8 and 10 open. Shell 9 was initially left inserted and later retracted. It seems that when shell 9 is in place, there is a greater amount of light emitted on the outboard side of the videos. This could indicate some plasma-wall interaction. After retracting shell 9, the overall light level with the filter dropped into the noise range. The remainder of the shots were taken without the filter.

The second half of the day was spent adjusting the frame rate and exposure level of the camera to see if the post-disruption modes could be captured. Because those modes only last about 100 to 200 microseconds, the frame rate had to be set to 121000 fps. This rate isn't achievable with the 128x128 resolution. The 128x64 resolution used ends up cutting off the top and bottom of the image, but because the plasma is so small after the disruption, most of those cut pixels wouldn't carry any data. I can get about 20 frames with those settings. Unfortunately, these settings prevent any light from being captured during the remainder of the shot.

For once, there are no major problems to report. The SXR is still losing channel 11 and the vacuum is optimal.

Monday March 14 2013 7:35 pm Angelini 77911-77953 D-Alpha and Color Filters Fast Camera

Today I continued trying shots with and without the D-Alpha filter for comparisons. I also moved shell 9 back in again for the color filter comparisons.

Comparing close-to-identical shots, I'm still seeing similar behavior with and without the D-Alpha filter. I'm confident the phase flip q-scan can be done without the filter. I also have shots I can use to compare the light in the dark region. Abel inversions on comparable shots with and without the filter give similar profiles.

At the end of the day, I ran shots with the color filters to attempt to see the balance of green, blue and red light. The filters are subtractive, so for each of the primary colors I needed to use two filters stacked. The red light gave the strongest response, followed by blue and then green. A reconstructed 3-color image in Matlab displays rather reddish plasmas. I did the color filter tests with shell 9 pushed in so I could see if there might be a difference in how much light comes from the shells. There wasn't. The BDs showed very similar modes for each of the colors. It may be worth repeating this experiment with the shells in different configurations.

The base pressure was up a bit today. The A/C in the test cell wasn't turned on this morning, so that may have affected it. The plasmas were still mostly consistent, although the running base pressure was between 12 and 19 nTorr.

I don't think the SXR channel 11 was on at all today.

The shells are all pushed in except for section 10 in preparation for phase flips tomorrow.

Monday March 15 2013 6:01 pm Angelini 77954-77972 Breakdown Failures

Today's run suffered from breakdown failures. While the shots early in the run were fine, after a while they completely failed to break down even with identical bank settings. Some troubleshooting steps were taken:

• A shot was attempted with double the standard puff time (2500 instead of 1250).
• A static fill was done. The pressure filled to 3.4e-5 Torr.
• The e-gun battery was checked and recharged. It reads 12.6 V.
• The connections to the e-gun were checked for continuity.
• The puff valve battery was checked and its voltage was 237 V.
• The signals on the coils for the failure shots were compared with the vacuum shot at the start of the day. No difference was observed.

The only possible abnormality in the diagnostics is a signal spike in the soft x-ray midplane detector at the time the breakdown should occur. This only appears on some of the failure shots.

Further troubleshooting will continue on Monday.

Friday March 22 2013 6:04 pm Angelini 77885-78036 Fast Camera Phase Flip q-scan

The issue with last week's run turned out to be a short in the e-gun. The e-gun was repaired and the filament was replaced. Yesterday and early today the e-gun was baked and pumped out. The vacuum didn't seem to be too affected by this. The current on the e-gun is set to 10 amps.

The first half of the day was spent on cleanup shots. The plasmas weren't too different from how they typically behave after a week without running. The pressure started at 4^e-9 and reached around 1.7^e-8 Torr for many of the shots.

In the early afternoon the shots were consistent enough for phase flips. The shot style involved a q somewhat above 3 at the beginning of the flip and typically ended a bit below 3 after the second half of flip. The closer th edge q was to 3, the more quiescent the dominant temporal mode of the fast camera data seemed to be. I'm going to try to classify "quiescence" using standard deviations and see if there's a clear relationship with the edge q.

I'll continue with the phase flips on Monday to try to increase the dataset.

Friday March 25 2013 2:47 pm Angelini 78039-78073 Fast Camera Phase Flips q-scan

This morning we had a small water leak by the north rack on the low pressure side. Nick was able to tighten a clamp and repair it quickly. The water luckily flowed away from anything sensitive and dried soon enough for the run. The high pressure valve did read 200 psi today.

Today's run continued the q-scan from Friday. I have a nice solid set of shots for analysis. A quick look shows there may be a q vs quiescence trend during the second half of the phase flip. I'll show the scatter plots on Wednesday. I intend to continue with this tomorrow, but with the timing of the phase flip changed so it's later in the shot. The lack of a clear trend during the first half of the phase flip might be caused by the lingering effects of the early mode.

The base pressure was up a little today. It started at 3^e-9 Torr and most of the shots were in the 15-17 nTorr range.

Friday March 25 2013 7:26 pm Levesque 78074-78087 Perturbations during startup

Ran shots with control coils energized in a few different configurations during startup to see if breakdown would be hindered. Peak control coil current in the strongest cases was ~40amps.

Applying a -3/1 field (78076-78080) with an amplitude of 20amps did not dramatically affect the breakdown. Using a 40amp amplitude reproducibly changed the startup, with breakdown currents ~10% lower and plasmas rapidly crashing inward. This could possibly be compensated for with different bank settings, but I didn't explore that.

Applying an m/n = 0/5 field (78083-78086) with amplitudes of 40amps did not hinder the breakdown at all -- plasmas mostly behaved as if the perturbation wasn't even there. This is good news for the proposed ferritic wall upgrade, which may produce strong n=5 error fields depending on the wall thickness.

6 control coil fuses blew during shot 78086. These were replaced after the following shot.

78087 had a 40amp -1/1 field on during startup, but 6 of the 40 coils had blown fuses. Breakdown was not noticeably affected.

Tuesday March 26 2013 5:38 pm Angelini 78088-78127 Fast Camera q-scan phase flip

Today the q-scan of the fast camera phase flips continued. This time I shifted the timing for the phase flip to run from 3 to 5 ms with the flip at 4ms. It seems to have solved the issue of the first half of the phase flip being too close to the breakdown-induced early mode. However, the later half of the phase flip often took place with the major radius (or the q) changing more rapidly. Subtracting out the shots with problems, I should have enough of a data set to find any trends.

There are no major problems to report today. Even the SXR channel 11 is working.

The base pressure was at 4e-9 Torr and the shots were again in the 15-18 nTorr range. This is a little higher than before, but doesn't seem to be causing problems yet.

Thursday April 4 2013 7:04 pm Angelini 78183-78241 Cleanup and Fast Camera Phase Flip

As expected after a week without running, the plasmas today were mostly terrible. They only started cleaning up after 4pm and as a result, I only got a handful of good phase flip shots. After my talk tomorrow, phase flips with a control coils amplitude of 20 Amps will continue. There weren't any difficulties at that amplitude and even the A14s displayed the flip properly. In the past the first 6 control coils would display with a shifted and extended timebase, but that wasn't present on any of the phase flip shots today.

During the high pot today, there was arcing between one of the TF magnets and the port on the top of section 4. The insulating sheet was shifted to address the arcing.

The first few shots of the day were OH-only to check for problems. Everything behaved correctly.

The base pressure at the start was quite low at 1.9 nTorr, but most of the cleanup shots had a higher-than-normal pressure in the mid-20 nanoTorr range. After the shots started cleaning up, the base pressure dropped closer to the typical running range. I'm hopeful for a good run tomorrow.

All the SXR channels were present for each of the shots taken today.

Friday April 5 2013 7:32 pm Angelini 78242-78289 Fast Camera Phase Flip 20 Amps

Today's run continued the phase flips with coil amplitude of 20 Amps. After some issues, I finally got consistent shots at the end of the day. Monday I'll be moving to larger amplitude phase flips and hopefully concluding this run campaign.

The base pressure came down to normal in the early afternoon, but a breaker tripped and the cryo pump in section 1 was turned off. There has been discussion of moving the cryo to its own breaker to prevent other electronics in the test cell from throwing the switch. The base pressure came down again after some cleanup shots.

The A14_14 in the North Rack threw errors all day during the Init. I had to run the Init multiple times for each shot to get it to activate. Fortunately it didn't show errors during Store, but it is still worth checking out.

The SXR array continues to show all channels. Did someone fix it?

Monday April 8 2013 6:34 pm Angelini 78290-78366 Fast Camera Phase Flip / Disruptions

Today's run was a good one, which made up for last week's poor runs. The goal was to round out the phase flip data set. Shots with a CC maximum amplitude of 20 Amps were taken with lower q values and shots with a maximum amplitude of 35 Amps were captured as well. The high amplitude shots almost unanimously disrupted before 4ms. While this didn't add anything to the phase flip dataset, they consistently disrupted, so I was able to capture disruptions with shells 7 and 8 removed. For the last few I used the D-Alpha filter as well.

I'll have analysis for Wednesday showing the effects of a larger amplitude phase difference on the fast camera data. I'll also show some of the disruption shots from the end of the day.

The vacuum was a drop higher today, probably because the test cell was warmer. Shots ran in the low 20 nTorr range.

The SXR array again seemed to be fully functional today.

Because of the high amplitude current pushed into the control coils, I needed to replace a few fuses in the amplifiers. Only one coil was down at a time, and the failures were few and far between.

Wednesday April 10 2013 5:41 pm Stoafer/Levesque 78367-78369 Thomson Setup

We used crates-only shots to setup the Thomson scattering diagnostic. This included laser and polychromator scope timing. During the run day, laser alignment was also performed.

The run has been very successful so far. We were able to detect stray light from within the chamber from a laser pulse in the polychromator -- this is a promising result so far.

Some tests of the stray light were also performed with the following results. In normal operation, with the collection port opened and the viewing dump open, we saw about 80mV of signal. With the collection port closed, we saw about 30mV of signal, showing some light entering the collection lens from outside the chamber (presumably originating from the beam dump). This was after we enclosed the collection lens in foil to reduce this stray light. When the collection port was open and the viewing dump was closed (no viewing dump), we saw about 550mV of signal. This showed that the viewing dump reduced stray light by about a factor of 7 with our current arrangement. There is a lot of room to improve stray light reduction, including sources outside the chamber.

We will continue the run tomorrow and plan on running plasma shots in the afternoon.

Thursday April 11 2013 7:31 pm Stoafer/Levesque 78373-78400 Thomson Setup

Plasma shots were taken with the Thomson scattering diagnostic for setup and proof of principle purposes. Our first plasma shot recorded successful Thomson scattering data, although signal was rather low.

Throughout the day we attempted to adjust the collection lens alignment, plasma parameters, and laser timing to gain signal with some success. We were able to obtain relatively hot and dense plasmas by the end of the run day, but are still limited in TS signal.

The run will continue tomorrow to optimize collection lens alignment, polychromator alignment, and laser alignment.

We have seen evidence of poor laser beam profiles, which limits the amount of energy we can put into the beam. This will be looked into once the run campaign is finished. For now, we will run with lower energy pulses.

Friday April 12 2013 6:40 pm Stoafer/Levesque 78403-78418 TS setup

We continued in the TS system optimization focusing on the collection system alignment. A good plasma for Thomson scattering has been produced with good reproducibility. Adjustments in collection lens location and fiber bundle position behind the lens were made. We found a significant change in scattered light signal for the fiber bundle position relative to the collection lens, whereas movements of the collection lens relative to the chamber showed less significant changes in signal. Therefore, the next run will begin with optimizing the position of the fiber bundle for collection focus. A shot with good Thomson signal is 78417. A great plasma with mode activity is 78414.

Monday April 15 2013 5:46 pm Stoafer/Levesque 78419-78464 Thomson Setup

The TS system was setup further for optimizing the system. The focus position of the collection fiber bundle was optimized for signal collection. The collection lens was repositioned twice without much of an effect on the signal. Further optimization will take place offline with work on laser power, laser spot, laser alignment, polychromator alignment, and more. A good, reproducible shot was developed and a scan of laser pulse timing was performed. This was to measure the temperature and relative density at various points throughout the plasma. The TS data will be analyzed. Bumping the soft x-ray fan array during collection lens adjustments affected whether certain channels were working or not.

Thursday April 18 2013 8:43 pm Qian 78486~78551 plasma shot

feedback system is tested today with several parameters briefly tried. While some fb shots seems to give positive results on the fb control. It could also due to the shot to shot variation.

The overall shot style does not give a very strong mode when q varies from 2.8 to 2.4. certain gain and phase are tried which seems to give a mild growth or suppress depending on the phase, but the shot is not very reproducible and i cannot confirm such behavior.

one can check 78539 and 78552 and compare with the neighboring shots.

further tests will be make tomorrow.

Friday April 19 2013 8:15 pm qian 78553~78644 Plasma shot

Since the q towards 2. style shot is itself very quiet during much of the time. It is not very suitable for conclusive feedback result. Today's shot is mostly focusing on the q around 3~2.8 shots. The plasma is not very reproducible so the results are not conclusive yet.

Gain for yesterday is found to be actually too high. Lower gain feedback gives less dramatic change in the perturbation (like phase flip would do) and gives better performance. Many fb shots are found to be suppressing the mode overtime. But shot to shot comparison is yet to be made to give more reliable results. A phase scan is performed. (sample for each point is low though) and gives frequency change and amplitude change. Detailed analysis is yet to be done.

Overall, the feedback definitely does some thing to the plasma. But to suppress the mode,if not impossible, the phase window is at least narrower than expect, and needs moderate gain.

Monday April 22 2013 2:23 pm Hughes 78645-78648 Attempt at repairing some sensors...

Jeff and I replaced three array boards and two feedback boards, then took a brief series of vacuum and plasma shots to test those channels. The good news is: 1) The shell 8 topmost poloidal sensor is now working! 2) The other sensor faults that Jeff and I went after apparently aren't due to board problems, so we have lots of good boards. The bad news is: 1) The new array boards are awful. The gain is low by a factor of 5-10 and they have tremendous amounts of noise compared to the other boards. All three array boards will be taken out and replaced with the old counterparts. 2) The shell 6 lower-midplane poloidal sensor has an in-vessel short to the chamber. 3) TA02_S1P, PA1_S01P, and PA2_S14P all look bad compared to other sensors on sharing a board, meaning the fault is probably in their cables or in-vessel. To do: 1) We should add FB06_S2p to a list of things to look into at the next up-to-air. 2) We should investigate the cabling of TA02_S1P, PA1_S01P, and PA2_S14P to see if their problems are in- or out-of-vessel, and decide what to do about them. 3) We need to figure out what we're doing about the lousy new array boards. 4) I'll replace the new boards with their old counterparts, and try to just do plasma shots as per usual tomorrow (Tuesday).

Tuesday April 23 2013 7:09 pm Hughes 78649-78695 Even-Numbered Shells Out

Shots of interest: 78649 - Initial vacuum shot, modeled on 77400 78686 - Closest shot to 70,000 78694 - Characteristic shot type of the day Shots 78651-78659 were spent on deliberate cleanup shots, trying to crash the plasma out toward the shells. Shots 78660-78671 were spent attempting to reproduce shot 70,000 with all shells still in, to find the right neighborhood in bank settings. This was only marginally successful (I suspect the plasmas were still cleaning up). Around 2pm I decided to switch over to the intended configuration with the even-numbered shells retracted by 4cm from their fully-inserted positions (as estimated using the plastic stops that establish maximum insertion of the shells). Shots 78672-78695 were spent attempting to reproduce shot 70,000-style plasmas (based on R0 and edge q traces). The timing of the VF bank firing was modified a few times during this shot series. Shots 78672 and 78673 are at the normal timing (860us); shots 78674-78675 and shots 78679-78695 fire the VF 40us early (820us); shots 78676-78678 fire the VF 60us early (800us). A major factor in plasma performance has been the 'early mode' tendency to jump or deflect in major radius at around 1.3-1.7ms, producing general irreproducibility. However, there commonly appeared a shot style similar to shot 78694 characterized by a very flat R0 around 93cm, a shot length around 8ms, and an edge q profile similar in shape to but displaced away from 70,000. Although it may not be easily comparable to 70,000 style shots, we may be able to use this as a target style for half-shell runs. I'll have to do some analysis Wednesday and figure out where to go next. As of the end of he run day, the even-numbered shells are still retracted. The VF timing has been returned to its usual condition. Some interface thoughts that came up during the run: We should really look at adding an "advanced settings" button to the Dispatch Control to avoid having to muck about in traverser and risk damaging the tree by accident--I'd suggest it should include a "restore defaults" button, as well. We should also try to keep all the changes applied to the current shot, rather than half of them affecting the current shot and half affecting the next shot.

Thursday April 25 2013 5:34 pm Hughes 78696-78743 Even-Numbered Shells Out

The model shot family for today's run was 78674, 78682, and 78694. Shots falling between these were selected; many were close but discarded, and these were marked in terms of how they related to the target family.

Noteworthy shots:

78696 - Daily initial vacuum shot.

78699, 78702, 78707, 78712, 78713, 78714, 78722, 78735, 78738 - Good (or at least okay) shots within target family.

78705 - Characteristic type of shot with R0 a bit too low and q a bit too high.

78706 - Characteristic type of shot with R0 a bit too high and q a bit too low.

As the latter half of Tuesday's run, all even shells are retracted 4cm from their fully-inserted position. Also, the VF_ST time has been set 40us earlier than standard to 820us; as of the end of the run, this has been returned to the standard setting.

Also, the array amplifier board Nick repaired has been installed in amp box 2, Bot3 position to test its performance and to attempt a repair of PA2_S14P. The sensor's signal seems like it might be better, but it's still strange. The board, however, appears to function essentially like any other board, so it seems to have been repaired successfully.

I now have 16 shots fitting a fairly tight envelope in R0 and q, plus at least half a dozen more good ones if the standards for grouping are relaxed. A lot of plasmas weren't awful, but weren't terribly good either. If someone desperately needs Friday, I can probably sacrifice it to more urgent goals (besides which I need to do analysis at some point). If someone wants Friday, please let me know tonight, so we can figure out who's doing what in the morning.

Friday April 26 2013 6:18 pm Hughes 78744-78789 Even-Numbered Shells Out vs. All Shells In

Today's run was intended to compare even-shells-out operation with all-shells-in operation. The main upshot is that each configuration is pretty resistant toward the other's shot style.

Noteworthy shots:

• 78744 - Initial vacuum shot

• 78752, 78761 - Similar to 78674 target

• 78764, 78772, 78777, 78785, 78787, 78789 - closest shells-in shots to 78674 target.

Jeff and I spent awhile playing with bank settings trying to get the plasma to stabilize at around 93cm the way it was doing with half the shells back, but it didn't seem happy doing this long enough for the current to rise and drive q down to a similar region. I also experimented with letting the plasma heat up and then swing out to drop q, at the expense of R0 being somewhat higher.

I suspect the issue may be that the shells which had been retracted haven't really seen cleanup shots since Tuesday afternoon, and that this combined with increased recycling is making for much dirtier (possibly denser?) plasmas. Even with the plasmas sitting a good bit farther outboard than in shot 78674, the SXR midplane sensor sees a considerably higher signal, which seems to corroborate this hypothesis, at least at first glance.

Although nothing came out looking quite like the 78674 shot style, we may be able to at least vaguely compare the behavior of the above-listed shots. It may be more fruitful to compare them against the 78706 shot style, which had somewhat lower q and an earlier q minimum, and tended to sit on a higher R0.

Monday April 29 2013 6:09 pm Angelini 78790-78855 Fast Camera Amplitude Scan

Today's plasmas were rather unreproducible. That meant I didn't get many shots for the amplitude scan, which requires a tighter q range. The plasmas weren't that bad so I'm hopeful for tomorrow's run. I may need more days than I'd originally planned.

The base pressure was typical for a Monday at around 3.5e-9 Torr.

At the beginning of the day the RTM-Ts were unable to initialize. Both the init_rtm-t.sh script and the do_awg.cpp program had slots 101 and 102 swapped. Flipping them back allowed caliban to properly communicate with the RTM-Ts. I was unable to find any hardware problems, but changing the software slot references may be necessary in the future. There were no problems with the phase flips after the RTM-Ts initialized properly.

Tuesday April 30 2013 6:20 pm Angelini 78856-78905 Fast Camera Amplitude Scan

The amplitude scan continued today. The shots were somewhat more reproducible and a general trend is beginning to show. I'll show the plots tomorrow.

A few issues came up today. First, the battery backup for athena seems to have a low battery. The workstation just lost power at one point during the run. It's been plugged into the surge protector side.

The other issue was with the CPCIs in the south, west and north racks. They stopped initializing in the afternoon. We tried power-cycling the ones in the south rack, running crates_off and crates_on and restarting the actserv_8002. After that they initialized properly again, but I'm not sure what fixed the problem so it may reoccur.

The north rack A14_14 continued to have problems. It had trouble initializing and during a few shots failed to store its data (even when it initialized without error).

The vacuum is still good. The base pressure at the start of the day was 2.2e-9 Torr and it remained in the low teens throughout the run day.

Thursday May 2 2013 6:30 pm Angelini 78906-78955 Fast Camera Amplitude Scan

The run today was somewhat more successful than the previous days. While I was able to fill out the charts a bit more, it didn't seem that I could get any clean shots with edge q lower than 2.9 when an RMP with a maximum amplitude greater than 25 Amps was applied. They all disrupted right at the phase flip. Despite that, the plasmas were more reproducible today, perhaps because the OHS was set 10 kV lower.

The plots so far still show a saturation for shots with an edge q close to 3. I'll try a few different ways of plotting the data to see if I can find a good q value to mark the division between shots with saturated modes and those without.

The vacuum continues to stay low at 3e-9 Torr and there were no other problems to report today.

Friday May 3 2013 7:55 pm Levesque 78956-78996 Fast camera shot variety

Ran with several very different shot types in order to expand the database of fast camera videos. Shells in Sections 7, 8, and 10 were fully retracted for Shots 78956-78961, then Sections 7 - 10 were fully retracted for the rest of the day. This exposed the black background to allow Abel inversions of the fast camera data. Shot groupings are as follows:

78961-78971: High start-current: ~12-17kA startup current, versus our typical ~8-10kA in most recent runs. Noteworthy shots are 78961, 78965, and star shots 78969, 78970. Several shots had very large 2/1 modes. The fast camera showed typical emission behavior in preliminary analysis.

78977-78983: Double-puffs. These shots had typical or short initial puff durations, then had an additional gas puff during the plasma (at 2ms or later). With the second puff duration programmed to be as short as 10us, the fast camera saw bright emission (about the same intensity as the breakdown) at the high-field-side for ~2ms, or until the disruption. Note that we are not sure how quickly the puff valve closes (i.e. it might actually be open longer than 10us). The 2nd puff emission typically appeared in the fast camera view about 0.3ms after the valve was set to be opened. Extra emission in the normal spectrometer view took longer to appear and was not as strong; the emission slowly rose starting ~0.8ms after the puff. Note that the puff valve is at the high-field-side of Section 9, the fast camera views tangent to this region, and the D_alpha spectrometer views a vertical chord at Section 6.5. The camera/spectrometer discrepancies suggest that the extra neutral interactions are localized near the puff valve for a substantial amount of time, and do not quickly spread around the machine in comparison with the equilibrium evolution. The loop voltage began rising 0.5ms after the puff, accompanied by a decreasing plasma current, suggesting that the plasma was cooling down. The midplane SXRT chord also had a delayed rise from the puff. Plasmas disrupted within ~2ms of the second puff.

78984-78993: Drastically different VF/OH bank timings. These shots are very bizarre, and produced some exciting fast camera videos. Small plasmas are seen to move vertically -- most start near the top of the machine and expand/drift downward to being vertically centered. 78984 broke down near the bottom of the machine, then the plasma rapidly moved up to the top shell over ~0.1ms before later expanding downward. Shot 78987 has an exception video including filamentary structures during a double puff. 78988 had a triple-puff, but the third puff didn't have a clear effect the could be unambiguously separated from the standard behavior of those shots (many bright flashes after a slow decay of emission). Star shots are 78991 and 78993, which had clean 4/1 and 3/1 modes respectively after the initial mayhem.

78994-78996: Targeting Niko's feedback shots (similar to 74780). Didn't have much success here, but ran out of time after only 3 shots. It looked like it may not be possible to reproduce the target shot with the 8 shells retracted.

Plasmas that were transiently far inboard or far outboard appeared throughout the run.

Tuesday May 7 2013 5:42 pm Rhodes 79037-79087 Phase Flips

Continued yesterday's excitation of internal resonant surfaces.

First few RMPs after 5ms did not work because of limitation on control coils. They are now set to sample 11ms into shot.

Today obtained long lasting plasmas with lower q*, closer to 2. RMP was applied towards the end of the shot, when the major radius was between 90 and 92 cm. Amplitudes were kept low with the hope of seeing a torque but not complete locking of internal mode. This remains to be seen.

Thursday May 9 2013 6:59 pm Angelini 79088-79130 Fast Camera Gas Puff Imaging

The purpose of today's run was to capture modes and disruptions with the shells fully retracted. At the beginning of the day, the standard exposure (9 microseconds) was used with shots naturally disrupting from high mode activity. We attempted to get the shots to disrupt near the midplane with some success. After that, the exposure was dropped to its lowest (1 microsecond) and a second puff was fired in the middle of the shot. Shot 79109 was the only shot which didn't disrupt immediately following the puff. Because of the limits of the puff circuit, there doesn't seem to be a way currently to lower the strength of the puff and prevent the immediate onset of a disruption.

Shots with edge q less than 2.5 were attempted in conjunction with the gas puff. It seems there is still 3/1 mode activity even with very low q, and the disruptions were primarily due to that mode. Some 3/1-2/1 hybrid modes were seen in the magnetics. I'll attempt to pull some of this out of the fast camera data for next week's meeting.

Strong condensation was seen on the cooling water line to the basement rack during shutdown. It doesn't seem that any electronics are in danger of shorting due to the condensation, but please keep an eye on this as the weather gets warmer.

The loop voltage diagnostic is currently disconnected in section 10 for the tests tomorrow.

Friday May 10 2013 5:52 pm Byrne 79131, 79132 Vacuum/crates only

The Loop Voltage output box failed today on the first shot. The first shot was un-divided, leading to an overvoltage condition in the CPCI at the firing of the OH start bank.

Rather than BNC feedthroughs, the LV coil was fed directly into the divider box. Torques to the box shorted the internal voltage divider, either through moving the internal wires relative to each other, or breaking a solder joint leading to loose wire contacts.

A crates only shot was taken with a low level sig-gen sine wave input after this to ensure that the CPCI was not damaged. This was an identical signal to one input yesterday, and the digitized signals were equivalent before and after, with a high degree of fidelity. It seems that no damage was done to the digitizer. Furthermore, the voltage divider has been re-constructed using better solder joints and BNC feedthroughs to eliminate mechanical stresses on the electronics. Bench testing (9v Battery) shows equivalent voltage division to before, but has not been tested on the LV coil as yet.

Repairs to the voltage divider meant that pre-positioning shots could not be taken, as facilities was working in the cap room AC the rest of the day. It was decided to begin moving the coil without shots, and compare the post-positioning shots to vacuum shots taken during earlier runs.

The loop voltage re-positioning is 40% done (The coil is hanging from zipties on 4 of ten braces), and work is accelerating as methods for reliably threading the LV coil and attaching the ziptie holders to the radial braces are developed. There is no reason to suspect that one more day would be insufficient for completion.

I have two finals on Monday, so will be unavailable to work for much of the afternoon. I don't think it would be a good idea to have work-studies on the machine unsupervised, and the only qualified people in the lab with experience up on the "roof" are Jeff and Jim. I don't want to let this job run to Tuesday, or require other people to step in and do my work for me, so I'm hereby requesting clearance to work Sunday to get a jump on the threading.

There is little to no risk of harm to myself or the machine, as all tools I will be using are light, plastic, and unpowered. It will be necessary to do a safety check when the work is done regardless of how, so all Sunday work will be inspected before the machine is brought back on-line.

While repositioning was always going to be difficult, the expected "unexpected problems and delays" have arisen. Though they have been dealt with, I'm recording them so that a post-positioning safety inspection takes full account of them.

Certain PVC covers are positioned such that they completely occlude the TF-chamber gap. They had to be removed. So far 5 (of 40) covers have been removed. They have been labeled with masking tape as to which TF (case and side) they belong.

The loop voltage coil, as it turns out *is* fixed in place. It is run though one tab in one chamber piece. It is a rather tight fit, so while care is being taken to pass the wire through gently, some nicking in the second, outer plastic insulation has been noticed. It is the same type of insulation used for the flux coils (thick, translucent plastic, not heatshrink).

I'd like to show this to someone knowledgeable to determine whether there is still enough of a safety margin with TF arcing. If it is not, the only real option I see is to completely remove the LV coil, strip the outer insulation, re-insulate, and install. Or simply insulate a new length of wire and replace the LV coil, as it may be tough to straighten it enough to swap jacketing. In any event, the coil will have to be completely pulled through the tab, so after inspection, that work will continue.

Two radial sensors, on chambers 3 and 8, which were hung around the inboard midplane conflat flanges, were removed to make way for the re-positioning (and later, the Shaping coil holders.) One was disassembled and removed intact, but the thin wire connecting the sensor was broken due to rough handling. The sensor wire on the other one was already broken before work began. The conflat nipple at the second sensor was connected to the spare puff valve. Short of gating the CF off, and removing the attachment, the only way to remove the sensor was destructively. Since it was already broken, it was destroyed.

A small hand saw was used, and the sensor was split in two and removed. There is significant plastic sawdust at the point of removal. Though it is not an arc hazard, the plan is to use canned air to blow it to the ground and vacuum up what can be reached. This has not been done in case there is a reason not to do so, or a better way to remove the plastic dust. The sensors themselves have not been used on the machine for many years, were not connected to any digitizers, and the general state of neglect and disrepair they were found in means that they should not be missed.

Pat

Friday May 10 2013 6:36 pm Byrne 79131, 79132 Vacuum,Crates

On Shutdown, it was noticed that the feedthroughs for the ion gauge have come disconnected. Two of the 7 feedthroughs (at 11 and 1 o'clock on the face) pulled loose. Rather than risk damaging the ion gauge, by reattaching them, they have been left as is.

Tuesday May 14 2013 6:06 pm Byrne 79133-39 LV calibration

Took a few low level OH shots to check for loose tools. None were found.

Took a VF only and OH only shots with identical bank settings to previously taken shots. In all cases, the gross features are the same, and it seems that noise has been reduced. This may be due to better electrical coupling at the voltage divider.

A full vacuum or plasma shot has not been taken, but all indications are that performance will be as good or better than it was before.

Byrne

Thursday May 16 2013 5:14 pm Qian/Dov 79139~79179

Some clean up shots are performed today. Plasma shot traces seems normal towards the end of the day. A OH coil connector over the top of the machine was found to be torn broken after OH elect was set to 400mV. Jim fixed it and it works fine so far. So, it seems not safe to set OH elect that high. Plasma is not reproducible for the current bank settings. Further adjustment need be made so that the result of fbtest would be easier to identify.

Friday May 17 2013 4:53 pm Qian 79181-79234

Today's plasma is reproducible. Plasma response is quite distinguishable with different phase set. It seems that when the phase is set for suppression, the mode will drop, but the equilibrium starts to vary heavily. (Another way of saying it is: when a fast mode get suppressed, a slow mode get excited.) This seems to be consistent with the gain-response diagram.

I will test some other possible improvement for the feedback algorithm on Monday.

Monday May 20 2013 4:57 pm Qian 79235~79292 feedback shots

Further improvement with the feedback algorithms are made and tested today.

Demonstration shots: 79269,79270,79271, 79291,79292 where 79271,79291 are reference shots without feedback and others are feedback shots with the most promising parameters tested so far.

Friday June 7 2013 4:26 pm Peng 79548~79593 plasma shot

Shots with slowly rotating perturbation are tested today. With freq around positive 0.3~0.5 KHz, the plasma behaves similarly to those suppressed fb shots.

With only section 6~10 of the control coil turned on, this behavior is not seen. This effect is also not seen for negative freqs.

Monday June 24 2013 5:14 pm Levesque/Stoafer 79601-79632 Cleanup, SXR fan testing, Thomson scattering

Cleanup shots following up-to-air for Thomson scattering Rayleigh calibration, SXR fan array filter replacement (now 100nm Al instead of 400nm Be), and Section 5 10" conflat replacement with a gate valve. The bake was on for 1 week. Base pressure at the start of the day was 6.9e-9 torr, which is better than the last up-to-air, and was about as good as expected. The partial pressure is dominated by N2, with water and O2 about equal and down by a factor of roughly 3. The TF cooling/insulating oil level was very low at the start of the day, meaning that the leaks are pretty severe, though the oil hadn't been refilled in the last 2.5 weeks.

The SXR fan array signals are heavily saturated, as was expected from the filter having more transmission at lower UV energies. The gain will need to be reduced by changing resistors on the amplifier boards. Three of the channels are not working -- this will be investigated. The disconnected m=3 Rogowski signals looks the same as one of bad SXR chords, even though this Rogowski signal is usually zeroed out, suggesting there is some connection between the two strange signals.

Thomson scattering was active for the last 2 hours of the day. All TS data were collected at or before 4ms. Observed temperatures were in the range of 10eV to 100eV, depending on the shot parameters. High versus low temperature shots were in agreement with x-ray emission. Densities were in the range 1.8 to 3.4e19m^-3. Cleanup will continue tomorrow along with more TS data collection.

Everything with spitzer and the run seems to be working fine, except one minor issue with the auto-updating of plotting shot parameters. This is being investigated.

Tuesday June 25 2013 5:57 pm Peng_Stoafer 79635~79425

A good many of shots are taken today. Shot duration has been close to that before up-to-air. The Thomson scattering diagnostic was used in the run today. The laser worked well throughout the run. We were able to make reproducible shots which we could do time scans on for temperature and density; more to come. Temperatures ranged from 60-150eV on normal shots. The density ranged from 1-3.5 e19 m^-3. A more thorough scan of shots will be performed tomorrow.

Wednesday June 26 2013 6:41 pm Stoafer/Levesque 79726-79795 Temp/Dens scan

The run today was aimed to use Thomson scattering for a temperature and density scan of various shot styles. Overall, the day was successful with reproducible shots and reliable Thomson system.

The first shot style was similar to yesterday's with a current ramp from 11 to 14kA and constant major radius around 92cm. This was a very reproducible shot on which a full scan was performed from 2-7ms. Densities ranged from 1.5e19to 3.6e19 and temperatures ranged from 70 to 190eV.

An attempt at low density shots was taken by lowering the gas puff time from 800us to 150us. The shots were not reproducible and densities were not much lower than the previous shot style, so this scan was not carried out. Densities ranged from 1.5e19 to 3.2e19, but the values were similar in the previous shot style for the time windows sampled.

A shot with high current was developed, based on previous shot development, to look at a new regime with higher Greenwald density limits. The shots have a current ramp from 14 to 18 kA. These shots are not very reproducible, but we are working on performing a scan regardless. We find these plasmas to be very hot with a relatively reliable measurement around 170eV (see discussion about high temperature readings below). So far densities range from 2.6 to 3.6, but we have just begun the scan; this scan will be continued tomorrow.

Currently, we are only able to use two TS channels for every shot, which creates a problem in relying on the data. We have found that this is especially true for high temperature readings (above 150eV). However, at high temperatures we are able to use a third channel as well. We will start to include the use of this channel when possible and look into improving the analysis of the signals (there is evidence of systematic noise).

The plasmas are looking very good and lasting a long time. Star shots include 79744, 79768, 79771, 79783, 79787, 79794.

Thursday June 27 2013 5:48 pm Stoafer 79796-79866 TS scan of high current

The goal of today's run was to perform a TS scan over time of high current shots. This shot style showed much higher temperature and density, as expected, which was also seen in the soft x-ray signals.

The shot style was not very reproducible, so not many samples were taken at each time point.

I will continue to work on analysis of the TS signals based on the discrepancy of using the high temperature channel or not.

Friday June 28 2013 9:00 pm Angelini 79867-79880 Fast Camera

Today's run was supposed to focus on collecting videos of long shots with more quiescent modes for comparison with prior shots. While I was able to get a few clean shots, a leak in the cooling water pipes (high pressure side) above the North rack sprang a leak and we ended early. A collar around the pipe was tightened, but a more thorough check and repair will happen on Monday. A potential contributor to the leak may have been a further increase in pressure. The gauge now maxes out.

Jeff has started replacing the resistors in the circuit for the SXR array. The new resistors will fix the over-saturation problem.

There is a small leak in the cooling water for the ignitron near the TF bank. It seems to be small enough not to be a critical problem.

The vacuum is holding steady around 7 nTorr.

Monday July 1 2013 7:50 pm Angelini 79881-79918 Fast Camera

The goal of today's run was to try to get tearing mode shots. Toward the end of the day I was able to get some shots with a constant major radius and a q just above 3. I'll do some analysis and see if the fast camera sees anything different from previous shots with the q just below 3.

Both the water leak near the north rack and the leak in the ignitron in the basement have been fixed.

All but two of the SXR array channels have had their resistors replaced and are now at a lower gain. The SXR plot now tracks much more closely with the major radius.

The base pressure seems fine. It started in the mid nTorr range and reached 25 nTorr during the run.

Tuesday July 2 2013 5:54 pm Peng/Levesque 79919-79967 RMPs and SXR data.

More shots with rotating perturbation is done today. To affect the Bp pattern, one needs the (3,1) mode perturbation. (0,1) or (3,0) perturbation doesn't have much effect. This rules out some possibilities of the explanation

More test with rotating frequencies is done, but the mechanism is still not known. Will try to figure out some model before doing further tests

At the end of the day, ran shots with widely varying major radii in order to observe movement of the SXR fan array emission centroid. In at least one case having major radius above 94cm and q* near 2.1, a 2/1 mode in the magnetics was accompanied by an apparent inversion radius in the SXR data.

Friday July 5 2013 2:51 pm Rhodes 79968-79992 RMP

The goal was to calculate m number of the mode based on the phase shift between control coil signals (used as sensors). This proved difficult to do within reasonable error.

Then could not get 'do_awg' to initialize the ACQ for resonant perturbations. Problem not resolved.

Turns out there's an early campus shut down today. Will try again Monday.

Monday July 8 2013 6:39 pm Rhodes 79993-80053 RMP and low q

Took a few RMP shots with q*>3, trying to excite 3/1 tearing mode activity. Then spent most of the day trying to create reproducible low q* shots near q*=2. Had a few good shots but none were reproducible, which made it difficult to apply RMP.

Tuesday July 9 2013 3:58 pm Qian 80054-80100 plasma shot

More low q shots were tried today.

Slightly delaying the VF start timing will result in a bigger major radius and lower starting q. The shots are not very reproducible, which makes it hard for comparison with/without perturbations. Increasing the puff timing would reduce the possibility of minor disruption, but will also make the MR falls in quickly.

80083, 80086 is a pair of shots that have similar equilibrium traces. The 0.5kHz rotating perturbation does not seem to much effect in this case.

Thursday July 11 2013 6:23 pm Stoafer 80101-80132 TS scan of double puff

The goal of the run was to obtain a T_e and n_e evolution of a discharge with a secondary gas puff. The puff occurred at 4ms into the shot and TS measurements were taken before and after the puff. Increasing densities were seen after the puff with a crash 1-2ms after the puff. Plasmas with and without the second puff were very reproducible all day; without the second puff, plasmas were lasting onwards of 10ms.

The TS system was relatively easy to get up and running after it had not been used in 2 weeks. The alignment did not seem to change significantly (no alignment adjustments were made in order to retain the alignment of the Rayleigh calibration). The beam dump had shifted out of place; a problem which will be addressed.

Friday July 12 2013 4:37 pm Stoafer 80133-80187 TS scan of outboard plasmas

The theme of the run was to obtain the T_e and n_e evolution of discharges that start outboard limited (around 95cm) and progress inwards. The plasma current was ramped to increase the temperature throughout the discharge.

Plasmas were relatively reliable after a target shot was obtained, so a TS scan was performed.

Target shot was 80164.

In the beginning of the day, the VFE bank would not charge. We found the issue to be that the shorting bar was in contact with the VFS shorting bar, which share a bucket. Future issues will be avoided by adding another bucket.

Thursday July 25 2013 8:02 pm Hughes 80198-80212 Ferritic segment force testing

Success! Starting with the TF at 200V (~100G) we gradually ramped up to higher fields, using the lab camera to record video of an edge-on view of the control coil end of shell 9T.

Below 4kV (~2kG) the deflections of the shell were too small to measure clearly, since the edge-finding algorithms all have some finite precision, especially when tracking a moving object with imperfect focus and finite contrast. This gives us a roughly estimated error of +/-0.3mm on most shots (about 2 pixels). This error estimate is, if anything, a conservative underestimate, to avoid shrugging off deflections as merely reading error.

At 4kG, we observed a deflection of roughly 0.6x0.15mm (machine-radially and vertically, in that order), which rose to about 1.67x0.6mm at 6.5kV with all other banks also firing at maximum normal settings. This is well below the 2.5mm limit we had set. Although analysis is slightly complicated by changes to camera position between shots, there is no evidence of any change in resting shell position from the beginning of the run day to the end (4 shots at or above full normal field). Any changes are well within the conservatively estimated reading error of the edge-finding algorithm.

Installation was occasionally frustrating and not always pleasant, but went fairly smoothly, and some aspects can be easilly improved with better planning (e.g., preparation of a smaller hex key with a ball tip).

The greatest standing concerns are the buckling of the prototype's shimstock panels when on the bench (which may or may not be an issue when installed), and avoiding the chamber flange when retracted. As to the first, we may need to add toroidal ribs to the design; this would also simplify assembly in some ways, which might make it easier to shop out construction. As to the second, we may have to choose whether to lose a centimeter or so of either retraction or insertion, unless another acceptable design solution approach can be developed.

Images will be uploaded to the wiki before the next meeting.

Tuesday August 6 2013 6:36 pm Peng/Levesque/Byrne 80229~80301

Today's main purpose is to do clean up shots. This is the first run day after a 3-day up-to-air followed by a 1-week bake.

Several chords in the SXR fan array were fixed during the bake last week. Two more gain resistors were changed to reduce the gain to an appropriate level; now only one chord is left with the high gain (it's the first channel, which is at the innermost tangent radius). The power connection for the first chord's amplifier board was repaired, which seems to have fixed the intermittent low-saturation problem on that channel. Two chords were fixed by reconnecting the crimp-connections at the vacuum feedthrough, within the copper bellows.

There are now only two apparent issues with the fan array: 1) Chord #1 has the higher gain by a factor of 8, which can be corrected in software when the signal doesn't saturate, and 2) Chord #15 appears to have reduced sensitivity, which is likely related to a poor connection upstream of the amplifier board -- that ground side of that channel has a higher-than-normal resistance to ground.

The voltage induced on the shaping coil was measured during high-power OH-only shots.

The base pressure this morning is around 1e-8 torr. Mass mainly around 4, might by He or D2. A good many clean up shots were taken today and the duration of plasma increased gradually during the day. The clean up shots will continue tomorrow.

Athena and other workstations got frozen several times today. Luckily, it hasn't caused any auto-discharge on banks. This problem is under investigation.

Thursday August 8 2013 5:42 pm Qian 80303-80382

The main purpose of today is clean up shots.

Cryo pump finished regen later in the afternoon and the base pressure is brought down. At the end of the day, the pressure is 2.4e-8 torr.

Plasma got good enough in the late afternoon and some feedback test is done. The plasma is not very reproducible so the drift subtraction is not working verywell. Highpass filter version, which suffer less from drift effect is firstly tested. No suppression effect has been observed so far. Nor is any slow mode being excited. Further tests are expected to be done tomorrow.

Friday August 9 2013 6:20 pm Qian 80383~80451

clean up shots and feedback tests are performed today.

base pressure is 6e-9 at the beginning of the day. puff time was decreased to 100 and the plasma breaks down normally. Shots similar to those before up-to-air are obtained. This suggests there might be some thing loose about the puff valve.

further feedback test with high pass filter is done today. Without high pass filter, the feedback would excite big slow mode at around 95 phase shift, which was observed before. With high pass filter at around 45 phase shift, the mode would be slowed, but one do not see a big dominating slow mode as before. The mode is not obviously suppressed. a rough phase scan does not seem produce much effect.

Monday August 12 2013 5:34 pm Peng 80452~80514

feedback tests are performed today.

The main algorithm tested are highpass filter version. It was found that at a narrow region of phase shift parameter, the plasma is affected a lot. A slow mode is still excited despite the highpass filter. But the overall mode is slightly suppressed as in contrast to those with only Bp feedback previously.

BpBr version is tested later. But the plasma is not very reproducible later in the day and the drift subtraction is not working very well.

The AC in the machine room seems to be not working, and the temperature is getting warm .

Tuesday August 13 2013 6:13 pm Peng 80515~80579

feedback tests are performed today.

It is found that the slow mode is self-excited by the control coil when the mode is slowed down. So the omega dependent gain parameter is adjusted to make sure that the gain decreases at low omega. Also, since the Br mode is quicker in detecting the slow mode, mode coupling setting was used. The phase shift introduced by high pass filter is taken into consideration and compensated. When the feedback is on, the mode frequency would still vary, but the control coil would no longer excite severe event that leads to disruption. Mode suppression can be observed in some shots, but not much.

Thursday August 15 2013 4:31 pm Levesque 80580-80602 Problems running

Tried to run far-outboard plasmas with R₀~96cm, but encountered too many problems with the run. Only shots 80581-80587 were plasma shots; the rest were from trouble-shooting. I was not able to achieve the desired plasmas before problems ended the run day. The final gain resistor in the SXR fan array amplifier box has been replaced, so all channels now have the same gain. Note that chord #1 is exceptionally noisy even with the lower gain though, so we may want to change this channel back or hunt down the noise.

At the beginning of the day, the VF Start bank was charging much slower than usual. It was slow enough that the bank would not fully charge before the TF completed. We were unable to find significant problems around the bank itself, and ended up raising the current control on the VFST power supply. This current control knob may have been accidentally lowered by a passerby in the control room. The bank appears to charge at a reasonable rate now.

The banks failed to fire after they fully charged for shot 80588. The data storage cycle initiated, but the triggering pulse was not delivered. Upon storing, all CPCIs and the two TR modules in the south rack failed to store, and all successful stores happened suspiciously faster than usual. Subsequent crates-only tests all had the same problem: the PULSE_ON delivery was not able to fire the puff valve or krytrons, and the same devices always failed, despite all INIT events reading success. Restarting various parts of the system did not help.

Thermal problems in the south rack are the suspected culprit. The air conditioner in the machine room hasn't been working for the past week, and the south rack was too hot inside. The solenoid valve delivering cooling water to the south rack was found closed due to lack of power. We found the unplugged connection, restored power to the valve, and the chiller started blowing colder air. If the next run day proceeds without issues, then this was probably the problem. If not, more trouble-shooting will be required.

Monday August 19 2013 6:44 pm Levesque/Peng ****** crates_only

Some of the crates_only shots are tested later this afternoon to prepare for tomorrow's run.

CPCIs were still producing errors during the storage phase. We did further check in the basement and found standing water in the basement rack, and a powers trip in it was not working. After replacing the power strip, the storage works fine. The source of the standing water is not clear.

Some of the bank cryotrons are not working and need to be looked into tomorrow.

Tuesday August 20 2013 6:01 pm Peng/Levesque 80609~80657 plasma shot

The standing water in the basement rack is due to the slight leak in the water pipe going through the fan. It is fixed and water is mopped out.

Feedback system is tested today with passing the total mode amplitude through a lowpass filter. The result is quite promising.

comparison groups: 89631,80639 and 80650,80651,80653,80657

each group have similar shot styles. and 631,653 are reference shots. One can choose their own mode amplitude analysis tool and the suppression should be seen during the feedback is on.

Thursday August 22 2013 6:30 pm Peng 80658~80736

feedback tests are continued today.

While the frequency variation due to feedback algorithm is seen. The suppression cannot be seen conclusively. Some phaseshift calculation check will be done later to see if the code's output is expected.

Monday August 26 2013 6:20 pm Peng 80744~80798

Feedback test is continued today. an invert transfer function for the amplifiers are constructed later today and makes the actual control coil current more like the requested one. a brief phase scan is done but the mode still cannot be suppressed. Phase acceleration/ deceleration and mode amplification is seen at corresponding phase shift. But the suppression phase shift still have a frequency oscillation.

The frequency oscillation will be further looked into and see if adjustment can be made.

Tuesday August 27 2013 6:19 pm Byrne 80799-80825 Rogowski Calibration

The Rogowski seems highly susceptible to noise.

The function generator's signals passed with no trouble and integrated well last night, so the cabling is continuous and the integrator and software work well.

The first shots of the day proved unrecognizeable as the VF trace. A huge noise pulse at the firing of the VF was the culprit.

The noise existed with a 50 Ohm terminus at the end of the cabling, rather than the rogowski

The noise was reduced, but still significant when the Triax cable connectors were tightened to provide better shielding.

The noise persisted no matter the level of the banks (down to 20V VFE, 200V VFSt)

The noise occurred at different times depending what banks were on.

The noise persisted when the original rogowski was used to measure the OH rather than VF current, and when a random other rogowski was hooked up to the VF cabling and the current was measured through that

The last three suggest there is still a problem in the interference shielding. I re-inspected, and rebuilt in a few cases, the triax connectors. This had no effect.

I think this means I am officially out of my depth. The noise is definitely coming from the krytrons, and I seem to have hit the limit of what I can do to the connectors to make things better. There may be breaks in the shielding away from the connectors, both cables were inoperable when I first found them, but it will be difficult to inspect the full length of both of them.

The fact that the first two random triax cables we selected were both initially unuseable has me hesitant to try replacing them with others from the lab, as even pulling them enough to check for continuity and proper insulation will be a full-day job.

We can look at what at what I was able to do tomorrow, and with the understanding that signal levels will be ~3 times what I saw on the VF and that the noise does not scale with signal, we might decide it's ok.

If it's not, I'm going to need some input. Personally I would vote for making brand new Triax cables and re-running them, it's the only way to be sure there's nothing wrong with the cabling. The shunt into a Jensen is another option, but we'll have to see if it can handle the load.

Wednesday August 28 2013 3:50 pm Peng 80827~80849

some feedback tests are done today. With the improvement on control coil currents, it's now much less likely to excite a dominant slow mode. The modes could still be slowed down at some phase shift, but it is higher than 2kHz.

Thursday August 29 2013 6:53 pm Peng 80852~80913

feedback tests are done today. While amplification, slow and acceleration can be observed at corresponding phase. The suppression is still not ideal. Slow mode would always arise. The slow mode does not keep increasing and becoming dominant. But we cannot get rid of it. Typical shots are 80911 and 80913.

Thursday September 5 2013 5:34 pm Peng/Levesque/Byrne ******

Triggering system for the banks are not working properly today. This might due to the high temperature in the server room and the Kamac is probably producing errors. A facility guy is sent in today to look at the AC problem

The triggering system are left to cool overnight to see if problem goes away tomorrow.

Friday September 6 2013 6:06 pm Peng/Levesque 80975~81039

Banks are triggered normally today

feedback system with Kalman filter are tested today. The slow component cannot be filtered out very well. The Bp/Br ratio for the slow mode changes when Kalman filter option is on and this makes the fast mode detection not working as expected. Reasons for this will be looked into.

Tuesday September 10 2013 6:05 pm Byrne 81050-81065 Shaping Coil Discharge

Low shot numbers since many shots were crates only with multiple pulses.

81055-81065 are shaping coil shots. 81063 and 81064 the bank was charged but deliberatley not discharged to look for susceptibility to tripping by bank room EMF interference. None was found.

Bank was stepped up to a 200V Start, 80V crowbar discharge. After an initial shot with myself in the bank room to get the timing right, all other shots were performed with the cap room empty and locked. Crowbar bank voltage will be roughly +- 5V from 80 in any given shot, but current traces with the rogowski show good agreement from shot to shot.

PA pickup shows ~130G on edge, and very localized, as would be expected.

Pulse shape looks similar to what was predicted by SPICE, but the resistance measured is too high to give the pulse shape we are getting. This may be due to the difficulty of measuring order 10^-2 Ohm resistances. Inductance was measured at 40uH, rather than the 49 predicted, but that is still quite close.

More analysis work is required on the Rogowski output before we can compare it to simulations however.

But as of now we have a working, chargeable, dischargeable shaping coil system. Remaining work is mostly learning how to use it and interpret the data from its use.

Byrne

Tuesday September 17 2013 4:50 pm Byrne 81066-81077 VF qualification

We ran the VF coil up from 50V El/1kV St to the maxima of 350V El/10kV St. The purpose was to ensure that the coil had been put back together properly after it was disconnected from its leads to allow access for shaping coil work.

In no case were there any problems with the coil. All shots fired, and the traces on the VF current monitor look fine.

We did lose some time trying to track down the cause of a trigger failure. As it turns out, the crate in the server room needed a reboot. This crate is never turned off under normal conditions, and we saw last week how extended operation can degrade the Camac systems. With the AC out, the problem is compounded.

After a reboot, we get the same errors on calling 'crates_on' that we used to get before an upgrade, which requires the re-running of the program. During this time we never had trouble with the server room crate, suggesting that it was a) reset at the end of the day when it was hooked into Matterhorn, and b) that the resetting is what prevented the errors.

We can talk about it at the meeting, but it might be worth it to turn that crate on and off with the others at the beginning and end of the run day. It can't really cause harm so the worst case scenario is the present status quo.

Wednesday September 18 2013 6:30 pm Byrne 81078-81110 Shaping Coil!

The run day ended when a start cap blew. We were running the coil up to a 600V/200V start/crowbar shot. This was done with charge solely on the coil banks. As the start bank reached 600V, a high leakage current was noticed. Just as the power supplies were cut, a loud snap was heard. Checking the bank, one cap was found with charring near its blow-off valve. The MOV was blown, as was the fuse. This is the failure mode we had expected and designed for.

Since the caps are individually rated for 450V, there would have needed to be greater than a 3/1 charge imbalance for an overvoltage on any cap. Initial, low current burn in tests did not indicate this would be a problem, nor did tests up to 200V with the higher amperage power supply we currently use. No autopsy has been performed, but at the moment, I would put it down to a 30 year old cap with high leakage cascading to a short failure. If the problem is limited to that one cap, we have multiple replacements ready to go.

On the lighter side, several shots were taken with the banks charged up to 400V/160V. These shots show a significant departure from our usual modes of disruption. Rather than an inward crash of several milliseconds, the plasmas suddenly slam inward over the course of tens of microseconds.

The other characteristics of the shots do not seem markedly different from what we normally see. Shots tended to disrupt around 5.5ms, but we were still in the clean-up phase of the run due to Jorway problems early in the day. Shots tended to last longer as the voltages were developed and as more shots were taken.

No analysis has yet been made, but start shots are 81098 (1st shaped shot settings 200V/80V), 81106 and 81108 (400V/160V)

Pat

Thursday September 19 2013 6:24 pm Qian/Dov 81111~81167

Feedback shot tests are performed today.

for state-space model, some problem was found in the matrices which overflow the output on the control coil currents. Several fuses were blown and replaced, the coils seems fine. The units of the model will be further checked to make sure the output is reasonable.

for single-helicity model. algorithms using four array individually were tested today. Modes can be accelerated and decelerated, yet the suppression is not reliably achieved. While the fast mode decreases, a steep gain curve edge depending on freq is made such to prevent the feedback acting on itself once the fast mode is suppressed. This results in some pulse-like suppression.

Friday September 20 2013 5:50 pm Byrne 81168-81177 Shaping Coil

After determining the problem was the MOV's (which are used to prevent overvoltage as fuses are to prevent overcurrent) low voltage rating, the banks were fired in isolation, starting at 120V/120V, and stepped up gradually to 250V/250V.

From there we increased the start bank to 500V, for a 500V/225 V. All crowbar bank shots beyond the one 250V shot were taken at 225V.

The bank fired fine, with no problems. The crowbar bank is now completely qualified for operation up to it's max rating of 250V, and the start bank up to 500 of 900V. The GDC power supply's voltage dial has been set to a level to top out just below 250V.

Higher rated MOV's are en route, and can be replaced as soon as they arrive and access to the bank is available. I have a rough algorithm for determining the proper post processing and gain, which would be good to put into the tree, so that those interested have more than the raw data (North rack cpci input 96) to look at. If anyone knows how to do that and is willing to show me, I'd be much obliged.

Pat

Tuesday September 24 2013 6:52 pm Qian 81178-81219

some static perturbation test was done today to check the Bp,Br mode response to the control coil. With static perturbation and phase flip, the phase difference between Bp and CC mode is roughly -55deg and Br/CC is roughly -150 deg. This was used to obtain the expected phase shift parameter for feedback system, which is -35 deg. This is within the phase window where we saw fast mode got suppressed with slow mode arising.

Tuesday September 24 2013 7:02 pm Byrne 81220-81227 Shaping Coil

Shots were taken at the end of Qian's run to test the new MOV's

Banks were progressively charged up to 250V/900V, and discharged. In all cases, the banks held their charge well and discharged properly.

Finally a shot (81227) was taken with plasma and a fully charged bank. No shot development was done, and a rather dramatic disruption occurred. Within .5ms the plasma had basically crashed inwards.

Python code has been developed to give us the actual current in the coil. Looking at that shot, the current seems to be briefly in excess of 9kA.

We were expecting a max current of roughly 8kA. This could be due to the lower than expected inductance, or it could be that the resistance of the coil is also lower than expected. We have not been able to accurately measure it yet.

The phidget has been giving us some trouble, currently it does not measure voltage on the banks accurately, and thus does not stop charging when the specified voltage is reached. Nick and I are looking into the code and expect to find a solution soon. We also need to modify the Take_shot script to allow entering of the bank voltages into the tree for recording. If anyone has any familiarity with this, your help would be appreciated.

At the moment, the shaping coil can be used. It can be charged (manually) to whatever voltage you'd like and be discharged at a set time (currently set by the fast camera timing in the tree) in the shot. Current is recorded (on CPCI channel #96 in the north rack, currently no dedicated tree node), and viewed (using my code). We should be eliminating each of those inconveniences one by one in the upcoming days.

Pat

Thursday September 26 2013 6:19 pm Hughes 81228-81279 Odd Shells Out

Not the best run day ever, but not the worst. This run featured the odd-numbered shells retracted 40mm +/- 0.15mm, and the target shot has major radius steady just inside 92cm, with a current ramp from 11kA to 15kA. Attempts at the second shot style (12kA to 16kA) didn't accomplish much, mostly due to intense early-mode activity making plasmas extremely capricious toward the end of the day. Since the real objective isn't to raise current, but to reduce q, I may try adjusting the TF tomorrow, instead. We'll see.

The VF was set early by up to 170us to prevent the plasma flying out to the 94cm region where it likes to hang out. Plasmas had okay breakdowns, but unpredictable and messy early behavior.

There was also a failure to dispatch jobs on the first shot (81228), but this resolved itself after restarting the takeshot program. I suspect tokamak elves were involved.

Shots of interest:
81229 - Daily Vacuum Shot
81255 - Probably the best shot of the day
81251-81262,81265,81273 - Good enough to be used

At the current rate things are going, I'll probably need Friday just to finish the set of three q traces for the half-out configuration. Maybe things will go super fast and every plasma will be awesome, tomorrow, but I wouldn't put a lot of eggs in that particular basket.

Friday September 27 2013 6:21 pm Hughes 81280-81326 Higher-current, steady maj. radius

The objective was to continue the runs from Thursday (9-26) pursuing higher-current, lower-q shots with major radius steady at 92cm. This didn't turn out terribly well. Intense early mode activity, possibly instigated by plasmas breaking down with q just above 3, made plasmas extremely irreproducible and generally uncooperative. The few shots close to the target shot of this style briefly hovered nearer to 91cm; some were not very steady at all.

Later in the day, another series of shots was taken with the start current raised so q at breakdown would be slightly below 3, and the current ramp would evolve the edge away from the resonance. This had at some success, producing a couple of shots with much better-behaved plasma current and R0 traces.

Notable shots:
81280 - Start-of-day vacuum shot.
81282, 81294, 81298, 81302, 81304, 81306 - Tolerable current traces, but R0 doesn't match target shot very well.
81325,81326 - Best shots of higher-current style.

Since Pat needs to do setup work with the phidget on Monday, I'll do some more running then, and see if I can get some more decent high-current/low-q plasmas, as well as some high-q shots. If there's time, I'll also experiment with changing the TF.

Monday September 30 2013 6:13 pm Hughes 81327-81399 Odd Shells Out + Different TF Settings

Today was rather more productive than Friday! A couple of decent high-current shots were taken, as well as several pretty good lower-current/higher-q shots, with q_min around the 2.8-2.9 neighborhood. There was also some time to play around with TF settings and see how the plasma handles it.

Last things first, with only a few shots, I was able to get a plasma with the TF at only 4.6kV, 75% of our usual 6.1kV setting. Making a halfway decent plasma required cranking up the OH quite a bit, probably to compensate for reduced confinement. There seems to be a general trend that you want to raise the OH *roughly* by the same proportion as you drop the TF, at least as a rule of thumb for starting with something that looks like a plasma.

Shots of note:
81327 - Start-of-day vacuum shot. There's some peculiar noise on the loop voltage monitor, but it seems just to be an unidentified curiosity.
81333, 81338 - Marginally acceptable high-current/low-q shots.
81365, 81376, 81386 - Best low-current/high-q shots.
81368, 81373, 81385, 81388, 81389, 81391 - Reasonably good low-current/high-q shots.
81396, 81398 - Best shots with significantly reduced TF.
81399 - TF just above 60% of normal; didn't even break down! This *may* be the limit.

Anyway, I've got a couple dozen decent shots to chew on, so analysis will come later. The odd shells will be inserted as per normal configuration, the VF start time will be reset to 860us, and the puff time will be returned to 200us as it's been during recent operations.

Tuesday October 1 2013 6:19 pm Qian 81401~81461

Feedback tests are done today with low q shots. Modes would interact with perturbation, but are not suppressed. The response can still be seen with half the gain we used to set, which might suggest we were operating with too much gain.

Thursday October 3 2013 5:55 pm Byrne 81462-81464 Phidget qualification

Some harmless flooding was seen today. During the day, the ignitron water was turned on, as the phidget is wired into our system to the point that even bench testing requires all interlocks closed. At the end of the day, as I was tuning on the basement crate power I noticed a large puddle forming under the OH power supply.

It turned out that the trough was slightly blocked on the other side of the wall separating the drain from the cap room by a mop, and that the ignitron outflow hoses were directing their water outside the trough to begin with. Nick and I fixed the problem, although we have no idea how the hoses could have ended up like that between the last runday and today. No damage.

As of today, the phidget and its programming are working properly to prompt for operating voltages, dump, switch power, cut off power when desired voltages have been reached.

At the end of the day attempts were made to fire the coil using the phidget. The south rack Jorway outputs would not trigger, however so we were unable to fire.

However, since the point of this week has been to get the phidget up and running, we can declare success. The banks were charged to the voltages we had specified at which point their supplies were cut. The banks dump when the phidget sees a dump signal. The interface is still a bit wonky, but that has more to do with my coding those bits, instead of Tyler. If we can get the Jorway running tomorrow I expect the bank will be fully integrated into our system. Next step will be pickup subtraction in our IP, MR, and Cos(3-theta) rogowski coils. Pickup is very low, so if a Daisuke-style response function can't be hacked together quickly, we'll just do a proportional subtraction. Should not take more than tomorrow.

Friday October 4 2013 6:06 pm Byrne 81465-77 Shaping coil

The Jorway problem from yesterday turned out to be intermittent. Was able to run all day no problem.

Phidget interface code still ahs some bugs to be worked out, but as of now, accepts voltages, charges, cuts off, adn gives realtime graphical data as to bank voltages

problems with the rogowski (it was routed incorrectly) slowed us down, but some good flattop shots have been taken. It is my understanding that step-function like shots are the easiest to model using response functions.

Pat

Tuesday October 8 2013 9:58 pm Levesque 81478-81521 Far outboard plasmas with low q*, RMPs

Ran far outboard, low q_* plasmas to study internal RMP effects on 2/1 modes with dark SXR fan array chords on the high field side. Having dark SXR chords on one side of the plasma implies that there are no emission contributions at minor radii outside the tangent minor radius of the last bright chord, which allows easier interpretation of SXR mode data. Target plasmas had R₀ near 96cm and q_* near 2 for 1ms or longer.

Most shots were unusable for the present study due to minor disruptions at low q_*, but there is a good dataset of usable shots. Successful discharges showed magnetic and SXR evidence of the 2/1 mode locking to the applied perturbation. Without applied fields, large oscillations in the entire emissivity profile are seen at the frequency of the magnetic-observed 2/1 mode. Additionally, a ~40kHz mode is often seen simultaneously with the ~8kHz 2/1 mode.

Noteworthy shots:

81498 and 81519 -- Good reference shots with no applied RMP.

81509, 81511, and 81506 -- Good RMP shots with CC currents peaking around 18amps. Phase angles were 0^o, 90^o, and 180^o respectively.

81489 -- Very clear high-frequency (~40kHz) modulations in the SXR data during 2.8-3.4ms.

Wednesday October 9 2013 7:42 pm Byrne/Levesque 12860-12869 crates only - tree updates

The Take Shot Make Tree and Shot Browser scripts have all been updated to reflect the new addition of the Shaping Coil.

From here on, if you'd like to see the output of the shaping coil, it can be found at .Sensors.SH_CURRENT. The timing of the bank is stored at .timing.SH_ST. and the voltages for any given shot will be stored in .metadata.SH_ST/SH_CB

Since the shaping coil has the capability of being fired at an arbitrary time during a shot, it's most likely this will be taken advantage of. Therefore in addition to the bank voltages, you can also browse the Start Bank's trigger time (the CB always happens ~.8ms after) in the shot browser.

The Take_shot frontend now has a checkbox for the Shaping coil, which allows you to run without triggering the switch, (please do, since the eventual plan is to move the phidget, with it's myriad garbage warnings and errors, to an automatic linkage with the tree.) a text field similar to that for the gas puff and VF timing for the shaping timing, and text fields for the voltages. You will still need to call the phidget program and manually insert the voltages to get the banks to actually charge, at the moment, however.

In looking around the top of the machine, we found that one of the leads was zip-tied badly, and not canceling the bundle - crossovers. This was fixed. The main source of error, the SXR passage, may be harder to fix though. Fortunately it is a strengthening of the field, so a nominally diverted plasma will remain so, but with an N=1 error overlaid. It seems likely this is due to the bundles being forced apart rather than pushed closer to the chamber, but it hasn't been thoroughly investigated yet.

Pat

Thursday October 10 2013 6:15 pm Byrne 81532-81552 Shaping Run

Very interesting data for our first real shaping run.

Lost the first half of the day to a series of brain farts. Jscope error was due to a typo (not the period/colon thing, apparently either will work) and subsequent noise discovered in the basement was just intrinsic noise caused by a bad choice of frequencies and gain to integrate/amplify out of the func-gen.

The first shaped plasma shots are 81544 and 81545. We were getting quick inward crashes as soon as the coil was turned on. Dropping the VFE and raising the VF ST and OHE (Shaping begins at 2ms) we were able to create a mid-plasma outward push to counteract the shaping coil's pull

The plasma was *extremely* repeatable, for HBT-EP, and shots 41548-52 are the star shots.

49 and 50 are unshaped shots. They fly outwards and have a sharp inward crash as soon as q~2.1. The shots were bookended by 48, 51, and 52, which fly out more slowly once the shaping is on, and on reaching q~2.1 hover for nearly 1ms before a much more gradual crash inwards than the unshaped shots. There's still no pickup subtraction in the MR rogowski, but a strong 2-1 mode is seen on the HD arrays for the entire time q is indicated to be near 2.

Zero crossings on the both PA arrays' poloidal sensors suggest the plasma is diverted in these shots, though Ish is ~4.5kA. This makes sense, as my simulations predicted diversion would be easier with an outboard plasma, or a low current (Ip was ~10-11kA in all cases) or by implication a real plasma, as outward shafranov shifts are not modeled in my simulation. Since we had all three, it's a definite possibility.

Definitely will be investigating this further tomorrow

Pat

Friday October 11 2013 6:30 pm Byrne 81553-81580 Shaping

Went looking for a higher IP shot so that I wouldn't be so dependent on extreme outboard plasmas, and was unable to find one.

Star shots are 81563 and 81565, low q, briefly dropping below 2 in 65 before experiencing a crash. The crash is similar to an unshaped shot, but with a q about .15-.25 lower. I'll have to look at the modes and any sources of pickup on the relevant diagnostics to see if we actually are getting a lower q.

81578 and 81579 are interesting in that the plasma is stopped right at or below q = 3 for a prolonged period with shaping. It would be interesting to compare these shots to another run's unshaped ones to compare multimode specta. (if I were to not shape that particular shot it would fly outboard and have a q=2 crash, so I don't have an apples/apples comparison from circular to shaped yet)

Pat

Monday October 14 2013 7:09 pm Byrne 81582-81609 Shaping

Attempted to get a steady shot around q = 3 to observe mode behavior/differences from unshaped plasmas.

Started seeing large jumps in current in shaping coil at ~15ms, beginning at 81604. Checking the banks, found that the mylar wrapping at the leads had come loose. At the time of work, these leads were tightly enough captured between TF and chamber that mylar sheets had to be slid in, rather than wrapped and taped around. Modifying the leads Friday seems to have loosened them and the mylar had fallen loose. We are operating on the assumption that this is the location of a short to chamber ground, though we will have to test it to be sure. Mylar was wrapped around the central bundle and taped to the cable to keep it in position.

Spacers were inserted in the 2-bundles to hold the wires stiffly in the coil holders. These were found to be inserted in the wrong location poloidally in some holders, and were removed, and partially replaced (the job will have to be completed before the next shaping runday.) This will increase the spacing between the central bundle and the mis-spaced 2-bundle. No prediction on what this will do to the assymetries of the coil.

Tuesday October 15 2013 9:06 pm Qian 81610~81686

feedback test with inverted cc-Br transfer function has been tested. frequency dependent gain compensation is no longer needed with this feature. Phase scan is performed and it seems that the plasma response also depends on the major radius a lot. with the same settings, there can be big response or even no response depending on the major radius varying from 90 to 92 cm. So the relation between response and phase shift can not be easily seen. Further analysis will be done.

Wednesday October 16 2013 6:39 pm Byrne 81689-92 Short-location

Three shots were taken, Shaping only, TF only and TF + Shaping. On the third, the current bump was seen. The scope had triggered (on what I'm not sure, the level was set at 1V, and there should be no noise across the resistor, but I was using the old, beat up o-scope.)

A fourth shot was taken with people in the machine room, looking and listening for sparks. Dov located the spot as the leads. Large gouges were found in the insulation.

The twisted pair was secured only at the outer, lower VF coil and at the entry to the coil proper. It seems that it had been jerked around (and possibly unwound, as there was a large un-twisted gap on the line to the coil itself. pics will be sent out in a separate email.

Lugs are on order. When they arrive we will cut each wire in the twisted pair at a different location, along the length where they pass through the floor. This will isolate each break from the other, and from accidental access by anyone near the coil. The coil-end will be drawn up and through and inspected for any other cuts. It will then be wrapped in mylar, taped, and heatshrunk. Running the heatshrink down the length of a twisted pair will be difficult, but the large amount of oil leaking at its location should make it possible.

While we wait for the lugs to arrive, I will machine a plank of wood left over from the bank construction. It will screw into the G-10 footings the TF feet attach to. It will be located beneath the chamber, and run from the inner to outer footings. The twisted pair will be lashed to this plank with zip-ties, to keep it clear of sharp objects, and hold it securely long term.

Pat

Monday October 21 2013 7:55 pm Byrne/Levesque/Peng 81695-81750 Shaping/cleanup

Shots 81696-81701 went well. They were taken to make sure the coil was properly attached, polarised, and measured. As the bank voltages were stepped up, it seemed that the crowbar bank was having difficulty charging to more than 200V, even with large (2A) currents. Also the phidget seemed to be reporting large noise on the banks (127V on a 100V shot, 650V on a 700V shot, both on the Start bank) Again, the actual voltage as directly measured with a DMM shows no such noise spikes.

A cause for the inability to charge the crowbar bank has not yet been discovered, but the plan for tomorrow is to: A) test to make sure the phidget is reading the voltage correctly B) Discover the source of the leakage current. - a shorted cap should be immediately removed by its charging circuit.

Jeff subbed in in the afternoon to take some shots, but found that the differential roughing pump near section 9 was switched off. This was probably kicked off while work was being done locally on the shaping coil. The pump seized up upon turning it on. Jim replaced the pump with a working spare.

Also, the small pump stand cryo was turned OFF, and it's gate valve was OPEN. "Cold head" switch was on, but "Compressor" switch was off. Checking the temperature shows the head at room temperature, and there is no condensation outside the pump chamber, meaning that it was off at least overnight (probably over the weekend, since today is Monday) and was outgassing into the chamber. It looks like it would be very difficult (but still possible) to accidentally turn the cryo off -- could the switch flip to the off position if there is some problem with the pump or power?.

Jim says the temperature on the other cryo pump is okay.

Present status: Large pump stand cryo is pumping, small pump stand cryo has been chilled and is pumping. Baseline pressure is 30nTorr, and RGA shows roughly equal parts D2, H2O, and N2.

Rest of runday was taken up with cleanup shots. Base pressure dropped significantly from 80nTorr to 30nTorr, and plasmas were made with lifetimes of ~5ms, and currents of ~12kA. These were outboard-scraping cleanup shots, so a centered plasma should be able to do better. A night of pumping should help as well.

These plasmas are similar in behavior to my shaped shots, so we can probably use them to take some data with the new coil, if the problem with charging can be found and worked out.

byrne

Tuesday October 22 2013 6:30 pm Byrne 81751-89 Shaping/cleanup

Day began late due to troubleshooting the charging issue in the crowbar bank. This turned out to be a holdover from the original code that was not corrected, and causing the bank voltage to be mis-measured. It was fixed, and the banks work well enough, within the bounds of the phidget noise. This work enabled us to take another look at the bank voltages during charging, and they are definitely not as spiky as the phidget makes them out to be. No ideas on how to fix it yet.

After that came shot development. A high OH, low VF shot, with a late (1 ms) VF start was developed. This gave a steady (~6ms lifetime), far outboard plasma. Shaping with 2/3 max current (600V st, 145V cb) gave a series of plasmas with q dipping to 3 +/- 0.1 for about .5ms. These shots are 81784-89

These shots will be studied to look at their multimode spectra. If anyone else has unshaped shots with similar behavior in their portfolios, i.e. outboard, with q =3 , I'd greatly appreciate being able to compare the two.

Other strange behavior was the plasma suddenly jumping *out* when the shaping coil was turned on. It may be uncorrelated, but it does happen frequently. the last shot of the day -89, gives a good example of that. Will have to check to see how the fields and fluxes develop during the shots to get an idea of why. Plasmas still crash inward quickly at the end of these shots, and strong shaping definitely causes a centrally located plasma to fall inboard to disrupt.

Base pressure was still high today, 10nTorr equilibrium, 30-40nTorr when taking shots. Jim said that both cryos may need to be regenned.

Pat

Wednesday October 23 2013 9:35 pm Levesque/Byrne/Hughes 81796-81844 Cleanup and inboard SXR shots

Cleanup shots were taken to recover from the poor vacuum quality earlier this week. Although the base pressure is still a little high, we were able to return to standard plasmas by the end of the day.

Fired the shaping coil only near its maximum rated voltage with 4 people observing around the machine to look for coil motion and listen for sounds of coil impacts. Substantial noise and motion were observed. Decided to hold off on further shaping shots until a large chunk of time can be dedicated to improving components around the coil after/during the APS DPP conference. Analysis in the near term will be done using the existing data. More detail was given in the email that Pat sent around earlier today.

Made far-inboard plasmas to collect SXR fan array data with little or no emission in the outermost chords. This is a companion experiment to shots taken on October 8. Was able to get modes in plasmas that were transiently as far inboard as R₀=86-86.5cm. Saw clear 4/1 modes in the magnetics with q_* slightly below 3, and 3/1 modes with q_*=2-2.1, suggesting that there are problems with the major radius and/or q_* computations this far inboard. Problems with the major and minor radii measurements are likely for plasmas this far inboard since they are far from the well-calibrated radii. SXR data has not yet been analyzed in depth.

The TF charging electrode plate has been fully eroded away, and needs to be replaced as soon as possible. The TF bank can still be charged now, but the threaded rod which normally holds the electrode is eroding away. Note that the electrode was seen to be very thin but still present a couple of weeks ago, so it hasn't been missing for very long.

Thursday October 24 2013 6:42 pm Hughes 81845-81900 All Shells Out

Today's run was a followup to my previous run campaign (shots 81228-81399 from 9/26-9/30) with the odd-numbered shells retracted. After a couple of attempts to run with a full high-density poloidal array inserted (first shells 8T&B, then 3T&B), it was determined that inserting only a single pair of shells made achieving breakdown difficult or impossible. Jeff suggested manipulating other settings (OHB>2kV, driving control coils, etc) may help achieve breakdown, but I opted to run with all shells out and figure out a way to do mode analysis later.

I took shots in the medium-current and high-current regimes from the previous run campaign, characterized by shots 81255 and 81325, respectively. Noteworthy shots of the day are:

• 81845: Start-of-the-day vacuum shot
• 81853, 81856, 81857, 81874: very good 81255-style shots
• 81900: only very good 81325-style shot
• 81891: comparable, but very short-lived

Other reasonably good 81255-style shots:
81854, 81855, 81858, 81866, 81867, 81869, 81870, 81871, 81875, 81876, 81877, 81878

Other reasonably good 81325-style shots:
81881, 81882, 81883, 81885, 81887, 81889

The most obvious trends in this run compared with the previous (odd shells out) campaign were:

1. needing to set a significantly earlier VFST time (up to 100us earlier) to keep a similar initial major radius, and
2. plasmas almost uniformly lived much shorter lives, generally disrupting at 4-5ms, compared with 6-8ms as seen in most shots of the previous series.

This reduced lifespan may be due to increased mode activity; the m3 Rogowski signal seemed more lively than usual, and there was no evidence of any other cause. This obviously begs analysis. As is, it's often difficult to identify a poloidal mode number even by eye, and codes are almost always worse at it. I may be forced to look at "n=1" modes rather than sorting them by m-number, unless I can do something very clever with the inboard (ergo, stationary) halves of the two poloidal arrays.

Friday October 25 2013 4:40 pm Hughes 81903-81941 All Shells Out

Today's run started a bit late, and ended a bit early due to no one being available to co-pilot, but got a pretty high density of good-quality shots. All shells were kept retracted throughout the run. The first hour or so of the run was dedicated to filling out the high-current (81325-style) shot library, and the remaining shots targeted the low-current shot style typified by 81376.

Notable shots include:

• 81903: Start-of-day vacuum shot
• 81908: Failed to breakdown for unknown reasons
• 81909,81911: Pretty good, if not perfect, 81325-style shots.
• 81938: Perhaps the best 81376-style shot of the day.

Other high-quality 81376-style shots from this run day include: 81923, 81924, 81926, 81928, 81932, 81933, 81934, 81938, as well 4 medium-quality shots that could be used in case more data is needed.

Plasmas tended to live longer than the high-current (81325-style) shots, with generally lower m=3 Rogowski signal, but didn't live as long as the characteristic shots of the previous only-odd-shells-retracted run campaign, typically disrupting between 5-6.5ms.

I am planning to try using an angle-correction algorithm to compensate for the angular displacement of the retracted high-density array sensors; hopefully this will make the poloidal structures easier to analyze. Analysis will follow in the coming week.

All shells will be fully inserted by the end of the [run] day.

Monday October 28 2013 6:09 pm Qian 81942~82024

feedback tests were done today. With MR around 92cm and q below 3. More mode suppression can be seen. But since the shots are not very reproducible and q often goes to 2.5. It cannot be completely assured that the mode suppression is not simply the effect of q trace. More analysis will be needed.

Wednesday December 11 2013 6:35 pm Byrne 82054-82095 cleanup/shaping

Run was delayed due to problems with the crates.

Firstly, the west rack was not initializing when crates_on was run. A fiber optic cable was found between the door and the crate. When jiggled, it seemed to cause the no sync light to light up. The power on the crate was also cycled. When crates on was next run, everything initialized, and worked fine all day. The fiber optic may need replacing, but we'll keep our eye on it. as we continue to run

Next, the CPCIs were not digitizing. After several attempts to diagnose the problem, it was found in Traverser that all the inputs were disabled, which was why no errors were returned by the shot dispatcher. make_tree was run and everything was well. It's unclear how this happened. If anyone has any ideas why the cpci's would have been disabled in this way, please let the lab know.

Most of the rest of the day was spent on cleanup. Most of the shots were the typical shots, low current, short life, high light output, but some, for example 82074 and 82076 had extremely long lives, and an *outboard* crash which is usually not seen, following large MR fluctuations. I_p decayed in stages, stopping for ~.5ms between each decay. The bank settings were the same as for the shortlived shots.

Eventually the shaping coil was turned on, and the plasma seemed to crash inwards after about .5ms. The mode structure was analyzed, and it seems to be due to a strong mode locking. 82089 and 82090 are good examples.

Reducing the startup current, so that the slower crowbar bank was forced to push the shaping current to its max value also seemed to allow the plasma to develop more naturally. The effects of wall eddies as the shaping field soaks through, or simply the rapid imposition of the shaping field (timescale of the wall, rather than the crowbar bank) could be the cause of the early disruptions.

Shots had been taken recently with 250us long gas puffs. This was continued on this run, but towards the end, longer puffs were taken. 350us was very similar to 250, but 1250us, the tree default led to higher post-breakdown I_p, higher initial Dalpha and larger SXR emission during the bulk lifetime.

Pat

Thursday December 12 2013 6:21 pm Byrne 82096-82128 Cleanup/Shot development

Forgot to turn on the oil pump during setup this morning. Was called to my attention by Jim close to 5pm. Plasmas ran well all day, and on inspection the only thing we could find was that the cases were slightly warm to the touch, due to the oil not being cycled through. Sorry everyone, I'll try to be more mindful in the future.

The e-gun battery had not been charged in a long time (probably pre-APS) and died after the first plasma shot of the day. It was recharged for 20 minutes to get back up and running, and another hour during the holiday party. It seems to have died again on the last shot of the day, as all banks fired, the gas puff and e-gun were on, but no breakdown occurred. It is recharging now and should be at full charge by tomorrow morning. It seems a waste of time to charge the e-gun after each run, but it is probably worth it to save the time that is wasted when it eventually runs out.

The first good shot of the day was dialed into the same settings as the first shot of yesterday. In all indicators (plasma life, IP current, Dalpha and SXR) the plasma was similar but better than the previous day. However, trying to match other reference shots of Sarah's and Qian's gave shots that disrupted earlier, at lower Ip. We are progressing, but still have some way to go to clean up the machine.

I was not able to develop a good shot today, though I am somewhat close to the shot used in my APS poster (81597). The e-gun dying interrupted development of the shot. Like the pre-APS run, it demonstrates that the plasma can survive for periods when the shaping current is ~6kA.

Shots taken with a slower risetime yesterday and today gave reason to worry about that, as they disrupted later, but still at around the same current as "step function" shaping shots

The idea is to get a plasma that sits stably at q~3 for a millisecond or so. The reference shot was repeated several times, so there is no reason to expect that we can not return there with time.

Tomorrow morning will be devoted to finishing installation of the new water cooling system. It should not take too long, perhaps an hour to 90 minutes. I'll continue with the shaping run then. If shot development goes well, and I get a database of a few promising shots, I'll try reversing the polarity for a few "bean" shots.

Pat

Friday December 13 2013 4:55 pm byrne 82129-82158 Shaping

Run went significantly better today than previous days. Not much time to run with morning work, colloquia and APAM friday, but several interesting shots were taken.

Most important result is shot 82158. This shot was taken with the start bank at 810V, and the crowbar bank at 190V. This is the practical maximum (assuming we don't go above 90% rated voltage) for 'flattop' shots. The crowbar bank can be raised, but the shot will peak later, and will be changing appreciably throughout the shot.

The plasma did not disrupt immediately (this is so far the only shot we have taken that did not, at these shaping current levels). The plasma ran very far outboard. This has two interesting effects. One, the plasma is very likely diverted, as the shaping current is ~8kA and MR is > 94cm. Two, the large MR led to a low q. The plasma sits at q~2 for almost 2ms with little trouble. Shots taken with all settings but the shaping the same have quick disruptions right at q~2 and crash. Several shots of this type have been taken, with varying shaping currents.

82152 was also interesting. An 'early mode' disruption reduced its MR before the coil fired, meaning that its q was very close to 3. This was the original target, before the difficulty of getting there, the interesting q=2 behavior, and the time left in the day conspired to look at shots as described in the previous paragraph. Shaping current was 110% of that used in the q~3 shots analyzed for APS. Will try to get a q = 3 database Monday.

As the comparison of shaped and unshaped q~2 plasmas showed, if we want stable equilibria in both cases that are similar to each other, we will have to develop them seperately. So far I don't have an unshaped plasma that sits near where the shaped plasma sits, with similar q. Getting that might be tricky, while a shaped q~3 seems to be hard to get to, which would be easier in an unshaped case. Work will be done on all these fronts tomorrow.

Have a good weekend.

Pat

Tuesday December 17 2013 7:09 pm Byrne 82159-82251 Shaping

This is a run report for Tuesday the 17th, and Monday the 16th, on which I forgot to send out a run report.

Yesterday's work on developing a shaped shot that had a steady q = 3 did not exactly pan out. I was hoping to have analogues to the unshaped shots with long lives, and steady q's that did not vary from 3.1-2.9 over multiple milliseconds. I was able to get many unshaped shots with this behavior.

The best I was able to develop have been shots that come close to q =3, but have more variable behavior than I'd like. shot 82200 and -05 are a good example. I should be able to look at these shots and get some information out of them. I would like to continue to develop the shots at a later time, but I think it would be wasted effort until I have a chance to analyze my data from this week's run.

Today was dedicated to doing a few 'bean' shots. This morning on inspection the leads of the coil were found to have snapped two of the zipties holding them together. They were tightened and trimmed using the only gun that could fit in the available space, which may have been too heavy duty and over tightened them. Additionally, when installing replacements, one of the zipties cracked in half as I was tightening it. The zipties we used may be old and becoming brittle. This will be observed carefully in the days ahead.

The bean shots were interesting, they tended to fly outwards when the shaping coil was turned on, and did not show the 'hard crash' in MR at disruption that the shots shaped with positive polarity do. Only a few shots were taken, the initial shots were shaping plasmas that tended to go down near q =2. The bean plasmas showed a tendency for a minor disruption at higher q levels, while normal shaped shots seemed able to accomodate lower q than unshaped. This is obviously speculation as no real analysis has been done.

Next a plasma with a normal q of ~ 3 was shaped, and again, the plasma was pushed outwards, giving it lower q than an unshaped shot. However, two shots, 82243 (unshaped) and 82247 (shaped, with higher VF settings) are somewhat similar in their deveopment at 3-5ms, so this might be an interesting area to look at.

This will probably be my last run for a week or so as I sort out all data analysis and the other related work. If anyone else needs machine time, the coming days will be available.

Pat

Thursday December 19 2013 5:45 pm Byrne 82253-82291 Cleanup

Problems with the CAMAC highway kept us from starting up until 1pm. The u-port in the basement rack was apparently not receiving a signal from the north rack. a spare f.o. cable was found in the conference room closet. Replacing the cable from the north rack to the basement rack gave was found to fix the problem. However, re-replacing the cable and cycling the power made the highway work as well. Unplugging and replacing this original cable and cycling power had no effect. I have no idea why transiently inserting a new cable should fix things.

One of our first ideas might have been that the crates were overheating because the coolant water is still under construction. The large fan was plugged into the same circuit as one of the cryopumps. Some time around 1:30pm the circuit tripped. Work was done on the shaping coil charging circuit first, which meant that plasma shots were not taken until 2:30. It was at this time that we noticed pressure was at 10-5 Torr.

The cryopump on chamber 1 was on the breaker and lost power. Unfortuately the gate was not. So the cryo warmed to 100K before we caught it and outgassed into the chamber. At the same time, the gate on the other cryo WAS on the circuit. So one cry was off, warming into the chamber, while the other was gated off. I think we should definitely have our failsafes on the same circuit as what they are supposed to ensure fails safe.

Fortunately the problem was not allowed to progress for too long. The pressure has been coming down steadily, especially after the one cryopump cooled sufficiently to open it back up. As of this writing, the pressure is in the low 10's or nTorr ~(20-30) by the time a new shot is taken. Shots have been steadily improving over the course of the day, and will probably be useable tomorrow, especially as the day progresses.

Star shot is 82291. Delayed the VF by 200us. Plasma starts outboard at around 95cm and q begins under 3. Tons of development is needed to make this a good kink-mode shot (and then to make another shot, which is also shaped.) but still.

Pat

Friday December 20 2013 6:11 pm Byrne 82292-82340 Shot development

The of the day was trying to get shots with a q that sat below 3, while Ip still had a strong ramp. Late start to day kept us from taking many shots, but we got close to an unshaped equlibrium that looks good.

Star shots are 82338 and 82307. Slightly different settings, but they illustrate the basic idea. We would like to move the steady state q from 2.25 to 2.5+.

Plasma was highly resistant to this development. 82307 was reached around 2pm, but not much progress was made beyond it. A lower Ip to start, with the same ramp should be a pretty good start, but the current was not greatly changed by increasing OHSt (varied from 6-9kV over the course of the run)

A big change was in the puff time. Decreasing the puff time to below 1ms gave low Initial Ip plasmas, but strong ramps, while puffs longer than that (standard is 1250us) gave lower initial Ip, but flatter ramps. Effect seemed bimodal, as decreasing puff steadily did not give a continuum of Ip behavior, but that may be due to insufficiently statistical numbers of shots taken. This is where the investigation will resume on Monday.

Have a good Christmas and safe travels if you're taking off this weekend.

Pat

Monday December 23 2013 4:33 pm Byrne 82341-82404 Shot development

During hi-pot, a clicking sound was heard. It was established that this was the breakers tripping again. They were found and turned back on, and there was no change in the base pressure.

Base pressure is significantly higher than Friday, nearly 2X10^8, vs 5X10^9 before. This is likely due to the increased temperature today vs. Friday. AC is still out.

Shot development was hampered by the battery for the gas puff dying around 2pm. It took about a half hour to diagnose, and once the new power supply was inserted, the plasmas were differently behaved than in the past. This would explain things seen in previous days, like small changes to gas puff timing leading to largte changes in plasma behavior.

Pat

Friday December 27 2013 5:43 pm Byrne 82405-82466 Shot development

Today's run delays were:

Despite the dispatcher not showing any errors, the shot triggers for OH,VF, and SH banks did not fire. It's my understanding that the TF trigger comes from a different location. This bank fired. Cycling the crates on and off had no effect. Eventually athena was cycled, and the triggers worked.

After that a shot was taken, the same moment of which a breaker (not the one in the machine room that has been giving us problems) tripped, cutting power to the control station. The cap room was checked, everything was fine. The tripped breaker was found in Nicks room. No idea on the cause. This shot, again, only fired the TF, but after power was restored, the banks all fired well.

The machine seems to have been dirtied by its days inactive. the last shot of Monday was taken several times, but I was unable to return the plasma behavior to what we saw then.

Attempting to run some aggessively outboard cleanup shots, I discovered that the change in Ip with the gas puff duration seems to have been a function of the dying battery we were using. Previously, a gas puff of 250us would start with significantly lower Ip Dalpha, and the Dalpha spike would be noticeably delayed. After replacing the battery with the power supply, there was no difference between a 1250us shot from before, a 1250 us shot after or a 250us shot after. This suggests that the battery had trouble fully opening the gate during a short puff.

This removed a "knob" for control of the plasma, but reduced the space in which to search for our shot. A few shots were taken with delayed VF start(to increase initial MR), increased OHE (IP had a 'saddlebacked' droop between startup and crash, almost certainly an effect of the dirty chamber.) boosted VFE (to hold the plasma, which starts further out, at a constant MR for as long as possible)

Star shots are 82457, 451,446. These shots all have q just below 3 for roughly 1.5-2ms. 457 in particular is very flat below q = 3. There is some early mode activity in this shot though (perhaps 4 in 5 shots today displayed early mode activity, scrapping most) so it was not repeatable.

The 'early mode' in my case often seems to come about by approaching the q = 3 line too slowly. The faster you punch through, the less likely the mode becomes. However, for me this means q is much too low after the plasma establishes itself. I'd like to find a similar equilibrium without the early mode, but I'm unsure if anyone's found shots that are less likely to have them (so far I'e only heard of shot styles in which they are more likely)

At least I have a few q = 3 shots, at early times, so the current gradient should be high, unshaped. Shaped will be its own problem, and post New Years, the machine will probably need another day of cleanup. At least the base pressure is down below 10^8, thanks to the low ambient temperature. The weather should be extra cold next week, so that should help.

2014

Thursday January 2 2014 7:10 pm Byrne 82467-82516 Shot Development

As usual, problems with the crates on startup. No visible source of error. This was anticipated, though, and the crates_on script was the first thing I did. Jeff rebooted the system while the rest of the startup checklist was run down. No time was lost. It stinks that the problem still exists, but it may be worth while to check that crates on will not throw an error first thing in the morning to avoid missing the time it takes to reboot either Spitzer or Athena.

Machine was dirtied up again, as expected. Eventually was able to return to the start shots of last Friday. Apologies for my mistake, but I entered the numbers in the book incorrectly, and that made it into last run report. I mentioned a star shot as 82446, but it was 82445. Not only was I able to replicate the shot to within the limits of a dirty chamber (short life, so shorter time below q =3), but I was able to do so in a repeatable manner. Previously, up to 5 or 6 shots were required to get one that was not scotched by an early mode.

These shots are born just below q =3 and stay roughly constant for up to 1.5ms, but IP ramp is almost non-existent. We see the 3-1 mode pre-existing, but then declining in amplitude, which may be due to the current profile relaxing after startup in the presence of constant IP.

Since we were sure that we can get more data on those shots if need be, the decision was made to try to replicate Niko's thesis shots. Reference was shot 74780. These shots are far outboard (which is helpful for future shaping) low Ip(t=0), and long lived, with a constant dIp/dt. q is constant at ~2.25, but a growing 3-1 mode is seen. It happens late, though when Ip has pretty much peaked. The idea is to run the plasma slightly more inboard to to increase q across the shot. These shots were excellent cleanup shots, as can be seen if you stack the IP traces of shots 82502-82506. Leaps and bounds in terms of lifetime, and when the shot style of -445 was run, it too lived longer and less disruptively.

To try to hold MR out Jeff and I experimented with using the control coils as a pseudo-VF coil. The coils all had the same step function voltage spike, with the polarities switched between to and bottom. Shots 82513 and 82514 have "anti-VF" pulses at different times and amplitudes, but in both cases, we see a very strong effect. That the plasma is far outboard, and so is strongly coupled to the control coils doesn't hurt. We definitely have the ability to arrest the plasma's inward crash, or even draw it outwards, though how finely we can control MR, and thus q, and for how long, remain open questions

In other hardware news, the plasma is definitely insensitive to gas puff duration now that the new power supply has replaced the old battery. Puffs of duration from 1250 to 250 us were taken, and very little difference in Ip or Dalpha at breakdown were seen. Jeff actually has a shot development primer on the wiki that demonstrates the effect that varying puff time used to have.

When the puff time was reduced below 250us, the plasma did not breakdown initially. However IP did rise after about 1ms. I don't know why the OH pulse wouldn't cause breakdown but OH+VF would, but that seems to be the case as you can see in shots 82496,82498, and 82516. This last shot was taken with a gas puff of 250us. I left the power supply on after Friday and its battery died. We gave it a quick charge during set-up, but it probably died. It seems that without enough juice, there is some effect on it that leads to these types of shots. So if you want to fire a non-fast startup shot, the option seems to be on the table. At any rate the Gas Puff power supply will probably need recharging at least monthly, assuming we remember to turn it off at the end of the day.

Pat

Monday January 6 2014 6:16 pm Byrne 82517-564 shot development

Star Shots are 82553, 82545.

Replicating the same style of shots used by Niko for his thesis. As before, the cleanup nature of the shots meant that the equilibrium was highly dynamic from shot to shot. I was able to get several shots, of which the star shots are good examples, where q was below 3 by 2ms, and rose steadily from 2.5 or so back up to 3 over the course of 3 ms.

Attempted to switch to shaping but there were problems with the phidget controller. The shaping coil fired well during the vacuum shots at the beginning of the day, but when the coil was energized for the first shaped shot, 82547, it only fired with about 1/3rd the expected current. The bank gauges indicated full charge.

Attempting to repeat the shot, the interface software correctly identified when the start and dump buttons were pressed, but neither interlocks on the dump, nor the two Shaping power supplies were thrown, suggesting that the problem lies in the phidget controller rather than any one of the three other instruments. I was unable to diagnose or repair the problem by myself with the time allowed (this was at about 4pm) so I continued to take shots to build a database.

The plasmas began falling in slower and slower, keeping q too low. I tried to fix this by boosting VFE, but that was a was as the VF's Ohmic drive cancelled any gains in the MR crash with higher IP. I should hve reduced OH instead.

The crate highway turned on with no trouble for the first time in several runs.

Pat

Tuesday January 7 2014 7:00 pm Byrne 82565-82597 Shaping

Morning was spent debugging the phidget. No problems were found in hardware, and we were able to use software snippets to perform each phidget operation independently. It seems that the phidget is able to get into a state which the normal operation of the software is unable to get it out of. This can be fixed by a hard reset - removing and replugging the usb cable. On the bright side several errors and possible problems in the phidget code that we have not encountered were caught and fixed, on the dim side, we weren't able to fix the problem that we currently have, but at least we have a way around it

Star shots are 82586, 82589 for large m=3 activity, as compared to an unshaped shot, 82588. Have not analyzed, only looked at stripey plots and m=3 rogowski.

Other shots of interest are the run from 82590-97. This was a shaping scan. All shots were flattop-style, each one involved stepping current up by 1kA per shot, from 1kA to start. Saw a 1ms drop in plasma lifetime from 3kA-4kA of shaping, and from 5kA to 6kA of shaping, vs .25ms per kA in others. 82593 and 82594 (4 and 5kA) have a nice q ramp, similar to the model unshaped shot, 82545, which will be helpful in comparing the effect on the MHD.

See you all at the meeting tomorrow,

Pat

Friday January 10 2014 5:58 pm Byrne 82598-82642 Shaping

Base pressure was at 1.3x10^-8 Torr at the start of the runday. It has been rising steadily, despite the recent cold. I think it's likely that one or both cryo's ought to be regenned.

The hook-and-loop type velcro on the new PVC has begun to come loose. The installation of the shaping coil holders meant that we needed to replace the old style of PVC installation with a two-piece system to allow for removal. This was done on the old, grey PVC when the coil was installed. The loop section of the velcro, which has been on since the upgrade in August, is still sticking well, likewise, the hooks, which were on from august until we replaced the PVC in November were fine. It's likely the hook velcro on the new PVC was a different type, that was not high-adhesive. I have ordered high adhesive Velcro from McMaster and will replace it all when it arrives. For the meantime, the panels were stuck back on, as the adhesive still had some tack, but not enough to serve long term

The high base pressure and the cleanup style shot I am shaping meant that the equilibrium was highly variable. Only the last 5 shots or so did not develop from shot to shot. This made comparing the effects of a shaped and ushaped shot very hard. The intention was to repeat each of the shots in my shaping progression from Tuesday's run at least twice, so that I could speak with some confidence as to the effects of shaping. I have 100V and 200V done, and intend to shape up to 800V, so much remains to be done.

Star shots are 82625 and -34, simply because they are unshaped shots that are very similar to my reference shot, increasing the database for unshaped plasmas.

Monday January 13 2014 5:02 pm Byrne 82643-82669 Shaping

Base pressure started at 12nTorr, but increased steadily throughout the day, until after a long break it had settled at 30ntorr. During running at the end of the day, the base pressure was seen to be at settled at around 50-60nTorr between shots. Cleanup shots had little effect with the base pressure this high.

The cryos were at 30K and 18K when measured. 30 Kelvin is pretty high, and the pump should probably be regenerated once the liquid nitrogen on order arrives.

So not much to report on the plasma front. Several machine issues to report, though. One of the OH charging resistors was found to be too low by roughly an inch. It must have been too low for a long time, and it's not obvious how the banks were able to charge during all this time. The reason the banks were not able to charge, however, was found to be a tripped breaker in the bake oven room. This was fixed, but not before we opened the supply box for the TF, which required work to put back together.

Also, in the last few run-days, a strange noise spike has been seen in the loop voltage sensor at around 4ms. Sometimes it occurs, sometimes it doesn't. When it does, the time varies from 3.5-6ms. The shaping and VF coils have not been shown to have any effects. When the OH or TF is off, it does not occur, but when both are, it does. It's not clear why this should be so.

Also, after the TF/OH banks were worked on, we re-high-potted. A low, frequent popping sound was heard near the feet on one TF case at chamber 10. We were unable to isolate exactly where it is arcing and what to, but the power supply had no trouble holding voltage, and we've run with a similar issue before with no problems. Nevertheless, we should monitor to make sure it doesn't get worse, for the obvious reason but also because it would make it that much easier to isolate and fix.

Also, I finally sat down with the pickup subtraction code, and it seems like 1) Qian's pickup subtraction is more accurate than the current method, at least when trying to zero out a pickup signal. It stands to reason that a full plasma shot, which is harder to quantify the performance is just as good. 2) The differences in the MR can be rather large, up to 1cm. In shots like mine, where MR is far outboard, this is a large effect on mr, and thus on q. I'll be putting some plots together and sending them out for perusal tonight.

Pat

Wednesday January 15 2014 6:36 pm Byrne 82681-82728 Shaping

Spent most of the day developing the shot again. Fortunately I was able to get back to the reference equilibrium for the last few shots.

The second Cryo was turned on around 5pm, as of this writing (6:30) the base pressure is around 24nTorr. Since the pressure after firing is usually much higher than the base pressure in the morning, it's safe to say our base pressure will be significantly lower tomorrow.

82722-27 are shaped. The coil was run with 800V on the start, and 200V on the crowbar, giving 8kA of step-function like current. IP never rises above 10kA. There is no doubt the plasma is diverted in this case.

Again, large amplitude is seen on the m=3 rogowski, though I haven't done BD analysis. 825, 6 and 7 have the SH start delayed by 1.5ms to see what effect that would have. The coil essentially came on as the plasma left it's low q region naturally. m=3 rogowski signal is more muted in this case.

The last shot of the runday had a louder snap when the banks fired. The only thing I was able to find was that a few of the zipties had broken at the bundle, which is reasonable given the high currents involved. Nothing was seen on the ISH trace to indicate any shorting.

If the pumps continue to cooperate, it should be much easier to return to the unshaped equilibrium, so there should be a much larger cohort of shaped shots.

Pat

Thursday January 16 2014 6:38 pm Byrne 82728-82774 Shaping

HUGELY improved performance now that the base pressure is down in the single digit nanoTorr range. Returning to the reference equilibria took 4 shots, as opposed to 3-4 hours.

First shaping only shot saw the rare 'Start Only' discharge. According to the phidget, the crowbar bank charged, and on the next shot, the bank was found to be completely discharged. Not enough time had elapsed for the crowbar bank to have discharged through the dump, but according to the rogowski it had not discharged through the coil. There is no obvious damage to the bank, and nothing in the current trace suggests a short. This was also the only instance of this shot seen all day. The shot was immediately repeated and the bank discharged properly.

A large database of high-power shots was taken. The first six shots of the day were 4 700V-175V shots and two 800V-200V shots. After this the leads were inspected. They had been double ziptied, and at one of the locations had broken both ties. The OH leads on top of the machine had also come loose. The OH was bound more tightly, and at the end of today, has not broken loose. The Shaping coil was re-bound and shots were taken at 600V-150V instead. The zipties on the SH leads are also fine at the end of the day. We will probably need a new solution if we hope to put >6kA through the coil, though Nick and I have some ideas.

At any rate, we now have a multi-shot cohort for 800/200, 700/175 and 600/150. Analysis to come.

At the end of the day, I moved the shaping start back to the same time as the VF start - 1ms. The expectation was the plasma would disrupt before forming, but we actually saw an improvement in performance. The plasma lived longer and fell in more gently than when the coil was turned on to a full formed plasma at 2ms, though it still ended in a hard MR crash. These shots all had trouble getting through q of 3 though. It may be worth investigating a bit more.

On inspection, there seems to be a 50us delay in firing the SH coil vs the VF coil. The last shot of the day, the VF start was moved back 50us, but when the values are backed out of the tree, they seem to rise at almost the same time. Anyone know why this should be so? Are the krytrons that much faster than an SCR?

Pat

Tuesday January 21 2014 6:56 pm Byrne 82775-82832 Shot development

This report covers Friday's run as well as today's.

The run Friday began with the no-breakdown 'triangular IP' plasmas. The gas puff was orignally set to 500us. This had worked previously on Thursday, but on Friday (The first day to begin with the AC system fixed) it did not. Increasing the Gas Puff turned out to be necessary. This *may* be because the background pressure was lower, requiring more gas? How much of the background would be deuterium, which I would imagine the OH is optimized to ionize? I can't imagine a background pressure that was mostly nitrogen or oxygen would have the same paschen curves as hydrogen.

After this, most of Friday was spent trying to get plasmas below q = 3. This was seen on Thursday, but developed as the machine pumped down, and was not recognized as a feature of the new equilibrium. Plasmas are subject to small, frequent bursty activity as they approach q = 3 from above, keeping MR low and preventing the plasma from breaking below q = 3. 82794 is a good example, or 82821, which is from today.

Anecdotally, the reason for this was the aggressive current ramp. Backing off the OH did not have much effect, though raising OHE to 300V did allow us to get the q =3 surface outside the plasma. It was at this point that q =2 crashes became an issue, as the plasma would go too low once through q = 3. Eventually we were able to get unshaped shots that passed through q = 3 and lived, quiescently, between q = 2 and q = 3.

Friday's run started after lunch. I had a class in the afternoon, so the theme of the run was about re-conditioning the chamber after the 3-day weekend and getting back below the q =3 barrier. 82831 and 82831, the last two shots of the day, look pretty good on that front. We can resume shaping tomorrow.

Friday morning was spent cleaning up the cap room. The oil leaks may be getting worse, or the neglect may have reached a tipping point, but there were 1/8" deep standing pools of oil beneath the krytrons for the OHE bank, and near the barrels of oil. The metal walkways, which are bolted to the floor, have piles of saturated absorbent beneath them.

The oil filters, and the line, to them seem to be a big culprit. I suspended a catchment bucket beneath the filters with wicks to direct the oil, and moved a wick from the top of the filters' feedline to the low point, as plenty of oil seemed to be leaking past it anyway. This seems to be helping, at least the catchment is about 1/8 full after today's run. I think there's probably still a leak above the OHE ignitrons (which may be the PVC routing pipes at the southwest end of the Tokamak) which we'll have to watch. The loose oil and absorbent was covered with dry stuff to prevent leakage in the garbage, swept up and replaced with new absorbent. Conservatively, about 50 lbs of absorbent and oil came out of the basement. I'll try to be around to help Glenn, our custodian, cart it out tomorrow morning, but if I'm not around, please lend him a hand. It's in the trash near the sink.

Maybe we should talk during the meeting about who's responsibility it is to tend to the oil/absorbent in the basement. If it were part of startup, that might help people find the time. It's dirty work, but if it's done more often it shouldn't be that big of a deal.

Pat

Wednesday January 22 2014 6:25 pm Byrne 82834-82852 Shells Out Shot Development

Today we retracted the shells to full-out positions, and began running to develop a shot.

Interesting shots are 82835, as compared to 82832. This would be the first and last shot of today and yesterday, respectively. It illustrates pretty starkly how much radial stabilization we get from the walls.

Usual trouble with developing a shot. Plasmas were either highly susceptible to activity at q = 3 or fell through to q = 2, where they would suffer one big disruption, which would then lead to the ultimate death of the plasma within 1 or 2 ms. Shot 82839 is an interesting one. The plasma fell below q =2, and then held there, steadily, for about 1.5ms, before a minor disruption kicked it above q =2

At the end of the day, we were able to get shots that fell to q ~2.25 and were stable, so we began shaping. The best shot of the bunch is 82851. This shot has *no* activity as it falls below q = 3, and when the shaping coil fires, at 2.75ms, it rises pretty linearly to q of 3 and sits there, just before disrupting.

Pat

Friday January 24 2014 1:17 pm Byrne 82853-82877 Shaping

Computer trouble on friday aborted the run. This report is for Thursday, which was not sent out.

The computer trouble seems to be spitzer related. The TF phidget does not initialize when the take_shot_hbt script is run. running th script with the phidget disabled shows that every CPCI in the crate chain is not initializing. Everything else works. Not sure how to debug this, and the small amount of available time on Friday between morning and afternoon talks means that I'm going to see if this straightens itself out the way it seems to have broken itself.

As for Thursday, shots began perfectly with no development! The repaired AC and regenned cryo are working wonders in terms of preventing wasted time. Shots were taken shaped. Currently, the unshaped equilibrium does goes below q=2 and does not rise to q=3, while shaped shots do. We'll try to fix that later.

The shaping start bank had its voltage dialed down to cause a longer rise to peak current. After checking with the flux simulation code, it was found that at a certain point the plasma stopped being diverted. The voltage was brought to just to the point of marginal diversion, and the Shaping coil was fired progressively earlier (when the plasma was further out and IP was lower). These shots should be interesting to look at as they take place right at, and on either side, of the diverted/limited transition, and all disrupt near q =3

pat

Wednesday January 29 2014 5:27 pm Byrne 82881-82920 Shaping

No plasma shots were taken today. After some troubleshooting, the problem was found to be the e-gun filament. The bias voltage on the filament would drop as the heating current was turned up. This would be slow at first, with a large discontinuous drop around 6 Amps. The normal operating current is 10 Amps so this is roughly a 3-fold decrease in filament temperature. A few runs were taken with different filament temps and biases, in the attempt to get useful plasmas. Voltages of 167V, 100V, 60V and 14V were all taken, and in no case did the plasma ionise. This is a current range of 5-7A. Nominal voltage is 171V.

We have not been able to identify the cause of the problem, but have confidence it is limited to the e-gun itself. A power resistor was hooked to the e-gun supply, and bias was not found to vary with current. The filament itself, which had the form of a tight solenoid when installed is extremely kinked now. Pictures will be sent to the group through normal email.

The gas puff problem, however, was found to be a dead 9V battery in the power supply. I had not had anything to do with this battery previously, and did not think to check it yesterday. The shots that did not break down WERE seen to have a pressure spike, so that issue at least seems to be settled.

This seems to be the end of the shaping campaign for now. We'll check the operation tomorrow once we've retracted the e-gun behind the gate valve, but if that doesn't work the only option is to remove the e-gun and replace the filament. This will probably end up dirtying the vacuum to some degree when it is replaced. We will still take a few shots to ensure everything is ok for Paul's campaign.

The phidget problems were solved by attaching the shaping phidget through USB to spitzer. The drivers used to operate the phidget are located on Spitzer and Nick suggested network issues might be to blame, especially considering the errors are generally related to checking the phidget state, as well as how often these calls are made. Two vacuum shots were taken with the shaping coil, and though this is a VERY unrepresentative sample, unplugging the phidget and re-plugging to Athena was tried right before this, and did not fix the issue, so it was not the hard reboot that returned the phidget to operation.

Shots taken on Tuesday (Sorry for not having sent the report out last night) were 82881 and 82905. The first few shots were an attempt to get q below 3. 82891, we had a pretty good shot. By 82896 we noticed the Dalpha, initial Ip and SXR were dropping, and began to boost the gas puff time, but as it became clear that we would not be able to keep it operational we ended the runday.

Plasmas should still be possible Friday, though what condition the vacuum will be in after the repair to the filament, I couldn't say.

Pat

Friday February 14 2014 4:45 pm Hughes 82949-82953 Vacuum shots testing Fe wall effect on sensors

First shot of the run was TF only to check for unexpected noises. No problems were observed. All other shots were in reference to shot 82681, a vacuum shot from one of Pat's recent(ish) shaping runs. 82950 is TF and VF only. 82951 is TF and OH only. 82952 is a full standard vacuum shot. 82953 is identical to 82952, but with the TF reduced by 80% to increase the residual permeability of the ferritic wall by 25% from ~8 to ~10. Since the shaping coil and glow discharge use the same power supply, which is still hooked up for GDC use, and the run being delayed by the shell motor controller breaking yet again, we decided to put off shaping coil vacuum shots until another time. -- Paul

Friday February 21 2014 5:35 pm Hughes 82954-83015 Cleanup shots after Fe installation

Since this run has been cleanup shots (and will continue to be cleanup shots, unless plans change), there isn't a huge amount to report as far as star shots. Machine performance is pretty good for being right after a long up-to-air with lots of material installed, followed by just a week or so of baking.

My shots are 82954-82968 (some vacuum shots, some attempted plasma shots without breakdown), 82974-82983, and 82992-83015. Pat took over cleanup shots 82969-82973 on Thursday, and Qian and Ryan took crates-only shots 82984-82991 today.

Base pressure Thursday morning was about 7.2nTorr and up to 9.4nTorr this morning (Friday). Not sure what accounts for this change.

Noteworthy vacuum shots:

• 82957: initial full vacuum shot of the day (Thursday)
• 82958: ~80% full TF
• 82959: ~55% full TF
• 82960: ~55% full TF, ~150% previous VF settings
• 82974: initial full vacuum shot of the day (Friday)

Noteworthy plasma shots:

• 82971: Plasma shot, strange step up in SXR midplane, highly visible snake/n=1 behavior on fan array!
• 83002: Very healthy, relatively long-lived plasma, IP over 20kA!

While shutting down I'll turn on the RGA, record vacuum status before I leave, and check it again on Monday morning to see how the vacuum has been evolving since we shut off the bake.

Wednesday February 26 2014 6:12 pm Stoafer 83016-83028 Multi-Point TS

After installing the new collection optics and aligning the laser, some plasma shots were taken to test the new multi-spatial point Thomson scattering system. On shot 83028 we saw TS light in three spatial channels (light not in the laser wavelength channel)! Optimization of collection optic alignment still needs to take place. We had a few issues with start up including the puff valve battery not being charged and the strangely behaving e-gun. It seems that setting the current of the e-gun to 8A gets breakdown, even though the measured voltage is low. Breakdown did not occur when the current was set to 6A and the voltage was around 200. There is too much stray light in the system currently, which needs to be reduced for proper running of the TS system. Also, I had trouble opening the shutter to the viewing dump. This was resolved by moving the bottom shell in section 6 in. Although it is designed to be able to open and close with the shells retracted, it seems to get stuck on the shell. For now, please do not fully retract the bottom shell in section 6.

Monday March 3 2014 6:25 pm Hughes 83059-83101 Cleanup and Shot Development

Today's run went pretty well despite some snags here and there. In particular, the puff PSU battery hadn't been charged, and dropped to 7.8V during the run, wasting a good chunk of time on lost shots and recharging.

With a bit of tinkering and fairly similar settings (with the exception of OHEL raised by 20%), it was possible to reproduce shots of the style for which 81932 was developed with a steady R0 between 91 and 92, a fairly steady current ramp, and q starting near 4 and dropping below 3 at around 3ms.

Some especially good shots with fairly arbitrary grades:

• Grade A: 83093, 83099
• Grade B: 83083, 83085, 83097, 83098
• Grade C: 83084, 83089, 83091, 83101

Here are some other notable shots:

• 83059: Initial vacuum shot of the day
• 83076: q sits right at 3 for an extended period
• 83077: peculiar mode activity that looks like a strong dominant m=2 mode at q>3
• 83086: maybe killed by ideal kink ~2.2ms, initial e-folding time on m=3 rogowski seems much too fast to be a wall mode...

Several shots died with abrupt inward crashes at around 2.8-3.2ms, suggesting higher than normal disruptivity. Loop voltages have been about 20% higher than otherwise comparable shots from October. SXR midplane signals have also been noticeably higher, suggesting the plasmas are still not terribly clean. Elevated base pressures seem to tell the same story. I think the next step is to insert the non-ferritic walls for continued cleanup and shot development.

Tuesday March 4 2014 6:17 pm Hughes 83102-83159 Cleanup and Shot Development

Another day of cleanup shots and testing how willing plasmas are to behave as they did prior to the up-to-air. All even-numbered (i.e. non-ferritic) shells were inserted, and all ferritic shells retracted for this run. Plasmas generally were less similar with even shells in than with all shells back, perhaps due to greater impurity injection due to proximity to the wall surface. After seeing this, shot style changed from pursuing 81255 to producing plasmas that flew far outboard late in time. A final 81255-style shot was taken at the end of the day for comparison. None of the shots of the day are really great for my purposes, but there is some lively activity if anyone wants to dig through for analysis.

Noteworthy shots:

• 83103: Daily vacuum shot.
• 83137: A relatively good reference shot in the style of shot 81255.
• 83139-83158: Purely cleanup-oriented series with low VFST and high OHEL.
• 83159: Final shot of the day. Has higher TF than 83137, hence the higher q*.

Jeff also did some analysis during the day and sent notes on observations: 83140: 3/1 mode grows almost instantly during 4.0-4.1ms, from zero amplitude to pretty strong (auto-scaled PA2 plot is near +- 13G limits). This could be a good example of a mode growing on an ideal time scale (e.g. the ideal kink mode going unstable). q* is near 2.4 at this time.

83147 (and maybe others around this shot): looks like higher-order mode activity is dominant, at least from 2-3.5ms, while using boxcar smoothing. I think this is due to the 3/1 mode locking. Plotting 1.5ms onward with standard contour plots, it looks like there's a 3/1 mode that slows down and probably locks, then disappears from my analysis since it's treated as equilibrium.

Both 83137 and 83159 show a rolloff of plasma current starting around 3ms, as well as higher loop voltage and SXRMP signals than the target 81255 in spite of very similar IP and R0. This may just mean a lot of cleanup shots are needed.

Thursday March 6 2014 6:59 pm Hughes 83160-83262 Cleanup and Shot Development

Today was mostly dedicated to aggressive cleanup of the stainless (even-numbered) shells, with a bit of shot development at the end of the day to test progress of the cleanup shot campaign.

Cleanup shot style (83162-83237) was intended to give the plasma some time to heat up and then throw it outboard; shot development (83238-83262) targeted shot 81255 and its relatives.

For a series of shots longer than their immediate neighbors during the day, the time in ms at the current-peak of the disruption were:

shot: 83168 83174 83181 83193 83220 83231

t(ms): 4.99 5.50 5.87 6.43 6.65 6.86

Noteworthy shots include:

• 83161: Daily full vacuum shot
• 83168: First *good* plasma shot of the day
• 83250,83253-83262: Decent to good agreement with 81255. Of these, the best are 83253, 83254, 83256, 83258, 83260, and 83262. 83253 is the star of this shot family.

Although all of these shots followed the current, R0, and q* of shot 81255 quite well, none of them lasted as long after the loss of a steady major radius/decline of the plasma current. Although still somewhat higher, loop voltage and SXRMP traces were much closer to the target for a magnetically-similar plasma today. Loop voltages, for example, tended to be 1-1.5 volts higher, rather than reaching up to 4 volts higher. There are still large (2-4V) jumps in loop voltage, which correspond to changes in plasma current as well as SXR fan brightness.

Tomorrow morning, the new stoppers for the ferritic shells will hopefully be ready. The plan, then, is to insert the ferritic shells and undertake another aggressive cleanup program.

Friday March 7 2014 7:26 pm Hughes 83263-83334 Cleanup shots with Fe walls inserted

We finally started cleaning up the new Fe walls! That's pretty much the big news for the day, as it was otherwise just a string of cleanup shots without a great deal of time in-between for analysis, the focus being to produce lots of hot, outboard, long-lived plasmas. Most had their disruption current peaks between 5 and 6ms; not bad for cleanup shots, although not fantastic.

Base pressure at the start of the day was definitely down from previous run days, so the barrage of cleanup shots approach employed yesterday seems to have paid off. Today saw 68 plasmas, down from yesterday's 100, due to various interruptions. Shots 83265-83281 were shots in the same configuration as yesterday (even/nonferritic shells in, odd/ferritic shells back) to establish a baseline, make sure the nonferritic shells were clean, and buy some time to find the missing spectrometer lens mount.

Noteworthy shots:

• 83264: Start-of-day vacuum shot.
• 83267: Longest shot of the day, disrupting at about 7.6ms.
• 83300: Possibly longest-lived shot after inserting ferritic walls, disrupting at about 6.7ms.
• 83316: No breakdown, but not sure why...

We had hoped to use the spectrometer to check for the appearance of iron/cobalt or other impurity lines when the ferritic walls were inserted, but a mounting clamp for the spectrometer lens has gone missing, so we were unable to do so. Jeff has been looking at other ways to set up spectrometer optics that might be usable in upcoming run days.

PS: On a maintenance note, Qian observed that the south rack doesn't seem as well cooled as it ought to be. I checked the thermometer on the high pressure water line, and it was at 50F, which ought to be pretty cool. I'm wondering if something might have been altered or valved off (wholly or partially) during the work on the new cooling water system.

Monday March 10 2014 6:45 pm Hughes 83335-83380 Cleanup and Shot Development (Attempted)

After a few cleanup shots, shot development was started based on shot 82545 as a general model. This did not go very well. Shots either went completely awry or suffered minor disruptions around 2.5ms, followed by another disruption around 4-6ms destroying the plasma. Lower-puff shots were also tried, as low as 125us, resulting in very erratic starting plasma currents, generally jumping between ~6kA (target) and ~3kA.

Toward the end of the run, this behavior began appearing on higher-current shots. An attempt to counteract it by raising the puff time to 1ms didn't help, and resulted in a plasma with almost no breakdown. The e-gun bias battery and puff PSU battery have both been checked, and were still well charged. The 9V in the puff circuit has not been checked, and any other problem in the puff or e-gun systems could be at fault, as well.

The disruptions tended to occur at a q* of around 2.4-2.5, abruptly shifting from a modest 2-1 mode to a strong 3-1 as q* jumped up to 3 or higher with an inward jump in R0 of about 1.5cm, after which the strong 3-1 mode would generally fade away over about 500us. This *might* be a sudden 3-1 external kink, but the 1-0 structure of the sudden inward jump dominates all other signals, and I can't offer a totally convincing reason why the 3-1 kink would wait so long to appear.

Pat took a few shaping coil only vacuum shots during the day (83346-83348), and Jeff worked on getting spectrometer data.

Shots of note:

• 83336: Start-of-day full vacuum shot.
• 83352: Exceptionally clear 4-1 mode transitioning into a 3-1 mode as the plasma crossed the q*=3 boundary. Mostly interesting as a nice example of a shot supporting our existing q* measurement.
• 83365-83374: Experimenting with shorter puff times.
• 83376: Plasma with the latest minor disruption at ~3.92ms after sitting at q* ~2.5 for nearly 1ms.

Two leads apparently connected to the spectrometer came loose during attempts to diagnose the failure to get breakdown. I'm not sure how or where those plug in, so I'm leaving it as-is. Obviously we'll need to fix some things before we can run tomorrow.

Thursday March 13 2014 6:55 pm Hughes 83387-83426 Shot Development for FWM

Not the greatest run day. Several shots and perhaps two hours or more lost to chasing intermittent and poorly-understood breakdown problems. Targeted plasmas with a strong current ramp (>2MA/s) that start outboard limited (R0~93-94) and swing inward so that q* approaches 3 from below.

Shots that had good breakdown showed the same behavior seen on Monday where a moderately strong (perhaps a few Gauss) 2-1 mode would develop around q*=2.5, followed (sometimes after sitting quietly) by a minor disruption which kicks the plasma inward and kicks q* up just above 3, leaving a 3-1 mode which fades away over 0.5-1.0ms.

This suggests all the modes are internal, at least while they're relatively steady. The disruption could be caused by a severe internal mode; it could also be the sudden onset of a 3-1 external mode which survives and fades away after the q* jumps up and the mode becomes internal.

After shot 83408, the non-ferritic (even-numbered) shells were retracted. Other than the breakdown trouble that showed up and lasted until the e-gun heating current had been tweaked, there was no obvious effect, based on the 10 or so successful plasma shots that followed.

Noteworthy shots:

• 83389: Daily full vacuum shot.
• 83409-83415: No breakdown or very weak plasma.
• 83409: Characteristic weak plasma with almost no start I_P, but a mostly-normal I_P ramp. Very mysterious...

The cause of the breakdown problems hasn't really been resolved, although fiddling with e-gun heating current seems to (sometimes, maybe) help. Once we're past the IAEA deadline, we should probably make a scan in e-gun settings and see what effect it has, if any; it sounds like there have been some cursory efforts to do this while trying to fix it, but not as a dedicated experiment to understand the e-gun's peculiar behavior.

The spectrometer was set up to record for shots 83416-83425, but it's hard to tell whether any of the data is useful until I can sit down and put it into an analysis code. I'll try to get that done tonight and send out a report on what I see.

Friday March 14 2014 5:37 pm Hughes 83427-83451 Shot Development for FWM

Mostly bad news on this run. After hearing some fairly loud snapping sounds which were initially identified as cable ties breaking on the OH leads, we discovered an OH lead had broken its lug and was arcing across the gap (and also breaking cable ties). Fixing this took a big bite out of the run day.

After the repairs, several plasmas were made with the same configuration as before, having the non-ferritic walls fully retracted and the ferritic walls inserted to their new stoppers. Then on a hunch, I retracted the ferritic walls by 5mm. Comparing these two configurations with a few shots (8+ each), the limiting lower q boundary had been about 2.5 with the ferritic shells fully inserted, but this dropped to around 2.25 by retracting the ferritic shells slightly, a boundary very close or identical to the lower q limit often seen by Pat and others in this shot style.

This suggests to me that with the insertion limits as they stand, at least a part of at least one ferritic wall is acting as a limiter, so the real q* of the plasma is 5-10% lower than our calculated q* suggests. Without a deliberate study of each wall's effect, which could easily take a solid week to do right, I'm not sure what to recommend other than keeping the ferritic shells 5mm back from the insertion stoppers until we have more information.

I'll compare against some older shots, but I suspect we don't need to retract farther, and may be able to re-insert some shells some amount. I have taken spectrometer recordings of both configurations (although as usual the recording may or may not show anything), so I'll look and see if there are any suspicious lines in one that I don't see in the other.

Noteworthy shots:

• 83427: Daily full vacuum shot
• 83429,83430: Healthy plasma shots in spite of the broken OH lead connection
• 83437-83441: Spectrometer recordings with Fe walls fully inserted. q* hits limit at ~2.5-2.6
• 83442-83446: Spectrometer recordings with Fe walls retracted about 5mm. q* hits limit at ~2.15-2.25

As with Monday's run, today's run ended with plasmas suddenly refusing to break down. I suspect it's an e-gun issue, because I remember we had set the e-gun heating current to 7.5A before, and when I looked, it was down to 5A, and the filament seemed dimmer than usual. Maybe someone fiddled with it and didn't tell anyone; I don't know. Someday we should probably do a proper characterization of the e-gun behavior and propensity for failed breakdown in the I_heat-V_bias parameter space.

And that's the run. Happy Pi day, everyone.

Monday March 17 2014 6:16 pm Hughes 83452-83509 Shot Development for FWM

Today was mostly spent attempting to get a different kind of shot that starts out around R0~94 to drive q^* down quickly to 2.5 or lower, rather than the gradual shots before. These efforts did not meet with a great deal of success. Only a few individual shots produced anything like the q^* trace desired, with especially significant shot-to-shot variability, very vulnerability to early disturbances around t=1.3ms, as well as difficult to predict response of the plasma to changes in bank settings.

This may be due to operating in a regime where insufficient VF pressure can cause the plasma to lose energy quickly to the limiter and then crash in. The strong current ramp may also have influenced the vulnerability and unpredictable response at early times.

Noteworthy shots:

• 83452: Initial vacuum shot
• 83491, 83499, 83506: By far the best three shots of the day with a good current ramp (around 1.8MA/s) and a rising q^*, although their R0 traces are not particularly similar. However, at a glance, all I see is 2-1 activity on the stripy plots. I'll see if I can tease out anything else of interest, though. The first two were achieved by tweaking the banks continually in response to plasma behavior; the third came about by targeting one of Niko's old shots (74780).

As has been a recurring theme lately, the run day was ended by plasmas suddenly no longer breaking down. Since it was almost 6pm already, it wasn't a catastrophe, but it is a somewhat distressing trend. So far the only fix I know of is going home and coming back in the morning.

Wednesday March 19 2014 6:13 pm Hughes 83510-83527 Shot Development for FWM, and prelim e-gun & puff testing

Today was a mixed bag of trying to get a really good shot for FWM study, plus some dabbling with the e-gun and puff. While we made a few interesting and rather wild plasmas, perhaps the most important result of the day has been finding that the puff system is highly erratic across the run day.

Notable shots:

• 83511: Daily fully vacuum shot
• 83519: First plasma of the day, with 4000us (4ms!) puff
• 83522: Typical plasma from today, with 3000us puff.
• 83527: Last plasma of the day, with 2000us puff.

Comparing shots 83519, 83522, and 83527 and looking at the bias current (the negative plasma current preceding the OH start), what stands out is that both the 4ms and 2ms puffs showed about 3 times more bias current than the intermediate 3ms puff, and their bias currents also peak about 100us earlier than the intermediate puff.

This suggests something highly erratic in the puff system over timescales on the order of a run day. We don't know yet whether this is an electronic or mechanical problem. I would suggest we dedicate a run day to hooking up all the signal sources to record their performance, and doing a series of fill tests at different points in the day. We can then compare shots with similar puff settings and different fill performance, and see where in the system the behavior begins to diverge.

As things stand, we don't even know whether the puff is reasonably stable shot-to-shot except that in a 3 shot series (83516-83518) the fill increased with longer puff times, although not in a particularly linear way:

Time vs Fill

Shot no.	Puff Time (usec)	Fill Press. (uTorr)
83516	1250	22
83517	2500	37
83518	4000	56
83526	3000	68

I suspect we will have to lock down the issues with the puff before we can pursue reproducible plasmas, but we'll see what can be done tomorrow.

Thursday March 20 2014 6:32 pm Hughes 83528-83568 Shot Development for FWM, puff testing

Today began with some diagnosis of the peculiar puff behavior observed yesterday. The results weren't absolutely conclusive as to the problem, however, and the puff seems to have behaved itself throughout the rest of the run day.

The only sign of a possible puff issue while making plasmas was variability in the bias-stage negative plasma current amplitude, varying between about 300A and 160A with a general decreasing trend from the beginning of the run to the end. Jeff has said this is within the expected variability, but since we haven't tried to characterize the puff before (or so I understand), this might represent an inherent variability which becomes intermittently worse.

A few plasmas with q* that drops quickly to around 2.5 and then sits relatively steady and/or rise toward 3 were made, but at first glance, none of them seem to exhibit strong 3-1 activity near R0=92. Although there is often reasonably strong mode activity, it's always 2-1, as far as I've seen, even though the current ramp has been pretty consistently in the 1.5-2+ MA/s range. I'm not sure what to make of that.

Notable shots:

• 83530-83540: Static fill testing
• 83542: Daily vacuum shot
• 83551: A decent "dip" style shot
• 83556,83559,83560,83565,83566: Decent q*~2.5 shots in the style described above

I'll also send around a set of plots of static fill data from yesterday and today, showing that the puff has some unreliable behavior, needing at least three fit lines to fit its behavior reasonably well.

Monday March 24 2014 6:05 pm Hughes 83569-83593 Shot Development for FWM

Another day of trying to wrangle the shots that jump sharply to q*~2.5 and rise up toward 3 from below. Still no very promising results. A lot of 2-1 modes blending into a variety of hybrid 2-1/3-1 structures for reasons that aren't entirely clear. The high dR0/dt when R0 is near 92 may contributed to this messy mode structure. Big 3-1 modes mostly showed up only after the plasmas fell in to R0 around 88 or lower, where wall coupling is comparatively quite poor.

No really great shots were made, although there were a few where the q* stayed between 2.5 and 2.7 for most of the shot. Even modest (~15V) changes to the OHE setting in either direction would produce plasmas that fell sharply inward, suggesting the run campaign has led to a very strange, ill-behaved regime, where gradual development will be difficult. I'm planning to try a different shot style, tomorrow, and see if that's easier to work with.

Notable shots:

• 83571: Daily full vacuum shot
• 83576,83577,83578: Reasonably good steady q* plasmas

Some of the other plasmas were passingly interesting for various reasons like peculiar SXR fan plots, but not really what I've been looking for.

Thursday March 27 2014 7:35 pm Hughes 83594-83632 Preliminary Fe/non-Fe Comparison

We finally had a run day comparing ferritic vs. non-ferritic shells with the same shot style! This was done using steady major radius shots of the style I developed last summer. This shot style was combined with phase flips to test stability to an imposed -3/1 (with some sizeable n=5 sideband due to only energizing the inserted shells); the phase flip started at 3.0ms, flipped at 3.5ms, and ended at 4.0ms. Due to lack of a copilot much of the day, there are many fewer shots than I would have wanted, and I haven't been as picky about them as I'd like, but there are some interesting correlations.

I first took a series of shots without changing the preset amplitude of the waveform, which came out to about 1V (roughly 10A). A number of these shots disrupted around 3.75ms. The average disruption time (as measured by the current spike) for this family of shots is 4.37ms (7 shots). Interestingly, the plasmas that fell inward to 90cm or farther tended to live longer.

Observing this trend, I tried a few shots at 5A (0.5x amplitude). These had a mean disruption time (MDT) of about 5.08ms (4 shots), ignoring one that fell out past 93cm and died earlier than the other 4 of this type.

I then reversed the shell configuration so that the ferritic shells are now retracted to their full safe retraction (5mm in from the old stoppers) and the non-ferritic shells are 13mm back from full insertion to compensate for the ferritic shells having been between 10 and 15mm back from the old insertion stoppers; this means the stainless shells were (on average) at roughly the same radial positions in both configurations.

All the shots after reversing the configuration were relatively long-lived. Shots with amplitude x1 had a MDT of 5.62ms (6 shots); another set with amplitude x2 (not 1/2!) had a MDT of 5.47 (3 shots).

Obviously, 3 to 7 shots in a set do not great statistics make, but there does seem to be a trend. To put it another way and sneak in an XKCD quote, "Correlation doesn't imply causation, but it does waggle its eyebrows suggestively and gesture furtively while mouthing 'look over there'." While this isn't a huge result, it definitely suggests ideas for the next FWM campaign.

Notably, Jeff and I didn't see evidence of big mode excitation or locking on the Fe-in/SS-out shots, although I was only looking at the stripy plots, and Jeff can say better than I what analysis he used (or if I misunderstood what he said). I honestly don't have the experience with RMPs yet to make a meaningful statement about this result, other than that it strongly suggests greater disruptivity (and hence reduced stability) in the presence of the ferritic wall.

Notable shots:

• 83596: Daily vacuum shot
• 83605: Nice plasma in the style of shot 81255 (although might not be directly comparable)
• 83607-83622: Plasmas with RMPs; Fe-in/SS-out. 83607-83615 and 83621-83622 have 10A on the control coils; 83616-83620 have 5A
• 83624-83632: Plasmas with RMPs; Fe-out/SS-in. 83624-83629 have 10A on the control coils; 83630-83632 have 20A

Note that since only half the coils are energized, if comparing against Daisuke's results, the gain on all of these shots (in the sense of the quantity of flux pushed toward the plasma surface) is essentially half the gain he used. I believe he also had the shells more fully inserted, which will further reduce the effective gain-at-plasma-surface during my run, compared with his, although I may have bought back some of that by running at R0=92 rather than 93 or 94.

As an aside, Jeff and I discovered that the unused flux loop cables jump during TF-only shots. These cables are twisted pair connected to the flux loop box through a D-sub connector and isolated (fixed to an unplugged connector terminal rated to hundreds of volts) at the other end, since the inner set were never installed. These have been lashed down with cable ties to avoid the unlikely case of them jumping enough to contact a TF casing; they only jumped a couple of inches when we watched, however, and shouldn't be able to reach the uninsulated part of the TF casing. However, this is a bit of a puzzle, since there should be no current flowing. It may or may not be worth figuring out what exactly is happening there.

Monday March 31 2014 5:30 pm Stoafer 83633-83663 TS Alignment

The goal of the run was to align the collection optics to maximize TS signal using plasma shots. Overall, the day was successful with signal in channels B and C, which are not laser line channels, for all views. I was planning sending an image with the data, but had issues in the latter part of the day with the laser, so was unable to get an image of the signal.

In the beginning of the day, I spent time aligning the laser and fixing an issue with the laser triggering system. This all seems to working properly now, except I had a couple of shots at the end of the day where the laser did not fire. I will look for this issue in the following run days.

I was developing a shot the stayed relatively constant in major radius and got hotter through the shot (based on SXR emissivity data). Target shots are 79717 and 79755. I got reproducible shots towards the end of the day.

I will continue tomorrow with lens alignment. There is still a lot of room for improvement (I have not even tried adjusting fiber bundle focus location), showing we will have good Thomson data, at least for views 92cm through 100cm!

Tuesday April 1 2014 3:12 pm Stoafer 83663-83701 TS Alignment

Today was spent making adjustments to the collection system for Thomson scattering. The day went very smoothly, with reproducible plasmas, allowing rather efficient alignment of the system. The collection system is now aligned to gather decent signal in all views. A more optimized alignment will require the chamber to be filled with dry nitrogen to see Raman scattering to optimize the light in a non-stray light channel.

The following run days will be spent doing temperature (and relative density) scans for various shot styles. The current shot style is interesting and will be scanned first.

Wednesday April 2 2014 6:28 pm Stoafer 83702-83731 TS scan attempt

Today was spent debugging the laser triggering setup. The triggering worked fine for the previous run days, but became flaky towards the end of the day. Today we had intermittent problems all day. I plan to continue debugging before tomorrow's run, so hopefully tomorrow will be successful.

Thursday April 3 2014 6:27 pm Stoafer 83735-83785 TS scan

Today's run was rather successful for doing a time scan in looking at TS signal. The laser trigger now seems to work as expected; we had no problems at the end of the day. I was able to get reproducible plasmas, so was able to do a time scan over the shot for TS signal at 3 spatial locations. The major radius also varied during the shot, which is seen with the TS signal. I will complete a full analysis after the run campaign and send out a report. From preliminary analysis, when the plasma is centered around 92cm, I found the plasma temperature to be between 100 and 140 eV in the 92cm and 96cm views and a plasma temperature of 40 and 70 eV in the 100cm view.

In order to fix the laser triggering issues, I rewrote the arduino triggering program to make it more intuitive for myself. There were issues with laser triggering early in the run today, but I found that the issue was pickup along the cords that run from the control room to the laser room to be able to turn on and off the flashlamps. I changed the code to account for this pickup (just ignoring the signal from these buttons during a shot) and the triggering worked as expected for the rest of the day. It might be helpful to reduce the pickup along these cords in the future.

In the beginning of the day there were crate issues. When running crates_on we got an error for every crate. It ended up being a communication problem between Spitzer and the server room crate. Jeff helped sort out the issue. The issue was not fixed when rebooting Spitzer initially, but we tried rebooting the server room crate and rebooting Spitzer while that crate was on, which fixed the problem.

Tomorrow I am going to try a different shot style to perform a scan and, perhaps, change the 100cm view to a different location.

Friday April 4 2014 6:06 pm Stoafer 83786-83827 TS scan

The run today was designed to be able to gather TS signal before and after a minor disruption with a SXR crash. We have found that we were able to do this by starting with an outboard plasma around 95cm and bringing it inwards with a large plasma current ramp. The edge q approached 2 and the plasma ends up having a minor disruption at around 5ms. We also used the control coils to induce a disruption by applying a 2/1 RMP for 0.2ms at 4.5ms. Afterwards, we used the control coils to apply an additional vertical field, opposing the VF from the VF coil. This keeps the plasma outboard, prolonging the plasma, as well as allowing TS signal in all three views 92cm, 96cm, and 100cm. The plasmas were working out pretty well, but we had an issue towards the end of the day, which ended the run.

Late in the run, an unusual noise was heard near the machine, sounding like loose parts moving around. It was identified as a G10 insulating shim that had dislodged from its normal position between the TF and OH coils. Without this piece in place, there is less material supporting the weight of one of the TF magnets, and less insulation between the local TF magnet (at the Thomson scattering section of the machine) and the lower-outer-OH coil. This insulating shim should be repositioned before further running of the machine. Neighboring shims may also dislodge soon without remediation, and continuing runs before fixing this issue could pose a risk of arcing between the TF and OH in the future.

Wednesday April 9 2014 7:45 pm Hughes 83843-83862 Ion Gauge

Today was a repairs/upgrades testing day. Shots 83855 and 83856 were primarily to check for any more unfortunate noises during vacuum shots. The machine sounds okay, so it seems the problem has been fixed.

All shots including the above included a gas puff to test behavior of the digitized ion gauge signal. There are some very peculiar quirks, in addition to the 60Hz signal which we assume comes from the 60Hz temperature variation of the ion gauge filament.

We found that the effect of firing the banks on the ion gauge signal was much more complicated than the scalar signal reduction we had discussed in the meeting. The signal goes totally wild when any banks at all are energized (see Fig1). To get a base pressure (averaged over the gray area on all plots) and a fill pressure (averaged over the green area on all plots), we had to change the start time and basement_tf time to -240ms, and the puff start to -220ms (the black dotted line on all plots).

As visible in Fig1 comparing vacuum+puff shots 83852 (black) and 83853 (red), as anticipated from what we saw with the 60Hz signal, the SNR is roughly constant, meaning the higher the signal (i.e. the more gas puffed in) the more the noise is amplified. We can also see that 500us and 1000us puff times are pretty much equivalent, although 2000us and 3000us are markedly different.

An even more striking feature of Fig1 is seen in plasma shots such as 83854, 83855, and 83857, where the ion gauge signal is very strongly suppressed over a period far outside the lifetime of the plasma (the light blue area). No vacuum shot had this behavior. As far as I know, everything should look the same between a vacuum shot and plasma shot at t=200ms... but the ion gauge apparently does NOT. It also seems that--at least fir the first fraction of a second following the plasma--the machine actually pumps down FASTER with a plasma than without.

Fig2 shows shots 83854, 83855, and 83857 from above, as well as a puff-only shot, 83856, to compare the pumpdown characteristic. We should expect 83853 (green) to be higher and 83852 (red) to be lower than 83857 (blue), which should have about the same pumpdown as 83856 (black); however, the plasma shot is notably lower than all three non-plasma shots, whereas firing the coils doesn't seem to upset the trend.

We also took a few static-fill shots in order to compare the front-panel readout of the gauge against the digitized and processed signal. The results are somewhat puzzling. Although there is relatively good agreement between observed base pressures and analyzed base pressures ("p1" on each plot) to within 10-15%, the fill pressures do not agree nearly as well.

The following table compares the gauge readings and deuterium-corrected fill pressures, compared against the fill pressures calculated and displayed in Fig. 3

Shot #	p_gauge(uTorr)	p_gauge_D(uTorr)	p_calc(uTorr)
83858	50	143	100
83860	31	89	63
83861	27	77	54
83862	51	146	99

I'm not sure where this disagreement could be coming from.

In short: the good news is that we can get a base pressure and a fill pressure; the bad news is that there is a lot of stuff going on we don't entirely understand. I'll leave the tree changes in place while doing my FW+RMP runs, so we can go back and use the data retroactively if we want, but I won't be messing with this until that run series is finished.

Fig. 1 Fig. 2 Fig. 3

Thursday April 10 2014 6:29 pm Hughes 83863-83926 RMP-response FWM study (baseline)

Today's run objective was to get several shots with a range of control coil amplitudes, in order to find a response/amplitude curve for the Fe-retracted configuration. There isn't much analysis to offer at this point, since my RMP-response analysis code isn't working at all, yet.

However, plasmas were very reproducible, today, yielding at least 6 reasonably good RMP shots at each of six control coil currents (5A, 10A, 20A, 25A, 30A, and 40A). The only plasma that disrupted near 4ms were one or two which pushed up against the outboard limiter and a couple of 40A RMP shots on which the TF was lowered, pushing q* closer to 2.7.

Since these were only a couple of shots, it's hard to say what they mean, although it does point to the importance (which we already know about) of q* for the RMP-response effects. A more comprehensive study may have to dedicate additional time to enhancing the amplitude scan with q* scans.

Noteworthy shots

• 83864: Daily vacuum shot. NB: This shot alone had *no gas puff*, but the ion gauge signal sees a rise of ~100nTorr (Deuterium calibrated) around 400ms, peaking at about 650ms.
• 83875: Control coils and gas puff only.
• 83902,83903,83904,83905,83913,83914: Good 5A RMP shots
• 83883,83884,83885,83896,83897,83898: Good 10A RMP shots
• 83877,83879,83886,83888,83889,83899,83900: Good 20A RMP shots
• 83906,83907,83908,83910,83911,83912: Good 25A RMP shots
• 83880,83881,83882,83890,83892,83893,83894,83901: Good 30A RMP shots
• 83915,83916,83917,83918,83919,83924: Good 40A RMP shots
• 83920,83921: 40A RMP shots that disrupt near t=4ms.

I tried to take a few non-RMP shots, as well, at the end of the run (as well as one in the middle), but for whatever reason those shots were generally less well-behaved, with more wandering in R0 and q*.

Friday April 11 2014 7:17 pm Hughes 83927-83938 Attempted FWM+RMP run

Not much to say. Had breakdown problems all day, couldn't find the problem. The puff seems to be fine. The shells are in a configuration used recently without such severe breakdown problems. It's possible the e-gun has been moved, or that something else has changed, but unless the machine magically fixes itself by Monday morning, we'll have to poke around and see what could be going wrong. I would argue that even if the machine *does* magically fix itself by Monday morning, we should spend Wednesday at least on trying to track down the problem systematically.

Shot 83931 and shots 83935-83938 were used by Kyle and Qian to test some feedback code, but I can't say much about that.

Noteworthy shots:

• 83928: Daily vacuum shot
• 83929: Characteristic of the few plasmas formed today; mysteriously feeble in spite of increasing OHS and OHB settings. The very healthy plasma shots from yesterday exhibited a very low bias-phase current, so that isn't a completely clear indication of the problem
• 83930: Characteristic of a shot that failed to break down

Hopefully Monday will go a little better.

Tuesday April 15 2014 9:46 am Hughes/Byrne 83940-83984 Attempted FW+RMP, diagnosis, shaping coil

This would have been sent yesterday, but the wiki was down.

Not much to report. Plasmas were uncooperative all day, and eventually failed to break down. After some fiddling with the e-gun, a setting was found in which it reproduced plasmas more similar to those on Thursday, with similar bank settings. It seems to prefer a heating current (at least, in the "Fe-in/SS-out" shell configuration) that puts the bias right on the cusp of the sharp transition.

Targeting 83918 was difficult, because the plasmas had to pushed considerably harder (e.g., 10-12% higher bank settings) to produce similar equilibrium parameters, which seemingly led to greatly reduced reproducibility. Only three not-awful shots with RMPs were made before plasmas stopped breaking down and the focus shifted to diagnosis. For shots 83980-83984, I handed the remainder of the run day over to Pat for testing the shaping coil, so I can't really say much about those shots.

Noteworthy shots:

• 83942: Daily full vacuum shot
• 83958: One of the longest-lived plasmas seen recently
• 83969: Decently good plasma with an amplitude=1 (nominally 10A in the control coils) -3/1 RMP on odd-numbered (Fe) shells only

Below is a table of the e-gun heating currents and (active) bias voltages measured while diagnosing breakdown failure:

I_H [A]:	V_B [V]:
8.5	8
8.0	9.6
7.5	11.6
6.5	175
7.0	~25

Notably, where most of the bias voltages varied by a couple percent on the multimeter, the roughly 25V measured at the 7A setting varied continually by at least a few volts. It doesn't seem to be a matter of poor contact with the multimeter leads. My guess is that the system was at a cusp in the nonlinear relationship between heating current and bias voltage (about 230V with the e-gun turned off) and the voltage changed significantly with small variations in the current; but I don't know of a reason for the current to be unstable.

Tuesday April 15 2014 6:12 pm Hughes 83989-84020 FW+RMP w/ various problems

Today was another attempt at getting RMP response with the ferritic wall inserted, which was met with moderate success at best, due to problems with Caliban as well as breakdown problems. The Caliban issue was resolved with multiple reboots--apparently, Caliban sometimes boots up well, and sometimes it doesn't, and we don't really know why.

This is not dissimilar to the breakdown problems, although we have some guesses. Having found yesterday that the best breakdown performance seems to be found at a point on the V_B vs I_H curve where V_B is highly I_H-dependent, we tried increasing I_H (thereby lowering V_B) and increasing the bias power supply from 250V to 400V (thereby increasing V_B for a given I_H), hoping to get a nice V_B at a more stable part of the curve.

This met with at least some success, as we had no failures to get breakdown after this adjustment, but plasmas were still not as good as they had been on Thursday, whereas one of the settings we stumbled across yesterday while trying blindly seemed to produce a more Thursday-like plasma. While the Monday-esque plasmas aren't totally useless, they exhibit a roughly 10% lower plasma current (and thus, higher q*); correcting the plasma current with higher OHST tends to throw them out to a higher R0, and the behavior around t=1.3ms is extremely variable, shot-to-shot. Some plasmas jump in almost to 90cm. Others sit at 92 but then swing outward, driving q* down and away from the target profile.

Part of the difficulty is that I_H changes on its own; after setting it to 7.5A around midday, by the end of the day it was hovering around 7.9A, and visibly wavering by perhaps 100mA. Nick has suggested this might be solved by replacing the supply's variac. We may spend a part of Wednesday exploring the parameter space of the e-gun at long last, in order to make a more deliberate and less ad-hoc choice of settings. We are also discussing how we might record the bias voltage and/or emission current, to be better able to diagnose future breakdown issues.

Problems aside, I did get a few reasonably good shots with 0.5 and 1.5 helical amplitudes (-3/1 structures on at 3.0ms, flip at 3.5ms, off at 4.0ms). I'll probably need more to make for really meaningful analysis, but I also need a working analysis code, so I can't say too much about it just yet. Although some shots flew in or out and died young, it's interesting to note that at least a couple of relatively well-centered plasmas died around 3.8-4.2ms, but only (as far as I recall) on the A=1.5 series.

Notable shots:

• 83986: Daily full vacuum shot of the day. The TF timing is a bit off, and the VFS power supply wasn't turned on, hence the peculiarities
• 84012: Best A=0.5 RMP shot of the day
• 84019: Best A=1.5 RMP shot of the day
• 84014,84017: Plasmas that died near 4ms despite being relatively well-centered

Thursday April 17 2014 6:36 pm Hughes 84035-84089 FW+RMP

Today was a very mixed bag. After adjusting the e-gun settings yesterday afternoon, plasmas were very strong with considerably higher start IP for the same bank settings, and relatively reproducible after a bit of shot development/cleanup. However, before long, the e-gun's filament heating current began wandering erratically by significant fractions of an amp.

Although we now operate in a *less* I_H-sensitive part of the e-gun's V_B/I_H curve, we are still at the edge of that rollover, so fairly small changes in I_H can change bias voltage by more than 10%, changing the breakdown characteristics. Between shots, I_H changed by as much as 2.25A in one extreme case; since it can only be checked by entering the tokamak room, it's impossible to say how much it actually changes regularly, and impossible to correct it before firing the banks.

While the plasmas were cooperating, I was able to get 2-3 good RMP shots each with an amplitude setting of A=0.5, 1.0, and 1.5... I also attempted to get shots at A=2.5, but these disrupted around or before 4.0ms or had undesirable R0 and q* traces. However, at least one A=0.5 shot also disrupted at almost the same time as the early-dying A=2.5 shots, so it isn't clear that this had anything to do with the RMP.

Noteworthy shots:

• 84036: Daily full vacuum shot
• 84053: Longest shot I've seen in months, disrupting around t=9ms. Also a very good A=1.0 RMP shot
• 84049-84051: Excellent A=1.5 RMP shots
• 84079,84082: Very good A=0.5 RMP shots

Hopefully things will be more settled during the day tomorrow.

Friday April 18 2014 6:43 pm Hughes 84090-84138 FW+RMP

Today was a more successful study of RMP response with the ferritic shells inserted and non-ferritic shells retracted. Plasmas exhibited quite a bit of shot-to-shot variation; only 1 in every 2-3 plasmas was useful. Although the e-gun is now set to parameters such that we don't seem to have breakdown problems, the changes in bias voltage due to the wandering heating current seems to drive significant changes in the breakdown characteristics, changing the startup plasma current by +/-10%, roughly.

As Jeff pointed out today, it is possible that configurations with the ferritic wall inserted and the plasma well-centered at breakdown (with a very early VFST time) are more responsive to these variations than far-outboard plasmas with non-ferritic walls inserted.

All RMPs were -3/1 helicity, turning on at 3ms, flipping at 3.5ms, and turning off at 4ms. Analysis will have to wait until next week, when I have some code up and running, although interestingly, I did often see rotating modes suddenly appear at the phase flip, especially at higher amplitudes. I'm not yet certain whether this is caused by the RMP, or an intermittent feature of the shot style.

Noteworthy shots from today:

• 84091: Daily full vacuum shot
• 84106,84124,84136: Good A=2.5 RMP shots
• 84117,84119,84121: Good A=1.5 RMP shots
• 84113,84116: Good A=1.0 RMP shots
• 84135: A=2.5 RMP shot that disrupted at about 3.7ms, shortly after the phase flip despite being well-centered

I believe the next run day is Chris'. I'll be reversing the configuration to ferritic walls fully retracted (less 5mm padding) and non-ferritic walls fully inserted. I'm also working on a --showgas flag for the bigplot.py code to replace the big current shot # display with puff/base pressure/fill pressure information (since there's really nowhere else to put it). Anyone who wants to take a look at it and give feedback on alternatives for displaying this information can find and run it in "home/peh2109/Scripts/hbt/bin/". I won't send it on to the repository until I have some approval, though, so the regular bigplot.py we use will remain untouched for now.

Monday April 21 2014 5:58 pm Stoafer 84139-84146 TS scan

Today I was attempting to do a TS scan on a plasma that disrupts around 4.5ms. However, I immediately had problems with breakdown with 2 attempted plasma shots. It was found that the puff valve was stuck open, filling the chamber with deuterium gas. The small deuterium gas bottle was noted to have around 400psi towards the beginning of the run day was empty.

It was found early that the cryos were frosting up, so the gate valves were closed, causing no damage to the cryos. They are being re-genned.

We were able to pump out the deuterium from the chamber safely by bleeding in dry nitrogen while slowly pumping on the chamber with a roughing pump. The chamber is now roughed out and valved off while the cryos re-gen.

We have another bottle of deuterium, but tomorrow we need to fix the problem with the puff valve.

While the chamber was a relatively high pressure, I performed alignment of the collection system using Raman light.

Monday April 28 2014 5:27 pm Byrne/Stoafer 84192-84229 TS/Shot development

OH lead broke around 3pm, stopping the run. The cause has been isolated, and the leads have been removed and stripped and are ready for installation of a replacement. I will be stopping at an Electric supply wholesaler tomorrow to have spare parts first thing in the morning. We should be running again by the afternoon. No damage to the machine, other than the broken lead was detected.

From Chris:

Thomson scattering measurements were taken throughout the day. Plasma center temperatures ranged from 30eV to 60eV throughout the day with higher temperatures being reached as more cleanup shots were taken. There were a series of shots taken where the TS measurement was taken when the plasma was centered at 96cm. In these cases, the temperature was read to be around 45eV in the 96cm view and 30eV in both the 92cm and 100cm views.

After switching off with Chris at lunch time, shots 84206-84226 were fired with the aim of developing a longer lived shot and cleaning the chamber.

Base pressure from shot-to-shot was roughly 35-40nTorr, which is rather high for a clean chamber. Currently the chamber is at 11nTorr, and has not had gas puffed in in 2 hours. At the start of the day, the pressure was ~9nTorr. We'll see if there has been any improvement.

Ferritic shells are still pulled back, and are not being cleaned. They can be inserted if we think the represent a virtual leak as is, or if not, it should be held in mind that they will require cleanup on the next ferritic experiment.

Wednesday April 30 2014 6:08 pm Byrne/Stoafer 84230-84273 Shot development/cleanup

Lost the morning when it was discovered that the high voltage power supply providing the e-gun bias had died. We currently have no idea what happened, but as a stopgap, eleven 9V batteries were daisy chained together to provide a 103V DC power source. With the e-gun at full heating current, the e-gun has a 25V drop, and a 0.75mA output. The batteries are rated to 350mA hours, so we should have plenty of time to work out a new solution, as long as we remember to turn off the e-gun when not in use.

Day was ended when the Shaping coil stopped discharging. on shots 84271 and 84272, the bank was seen to fully charge, but only released around 10% of the expected current. A shaping only vacuum shot, 84273, with 1/3 the previous voltage (200VSt, 50VCb) did not seem to discharge at all. The batteries on the pulser for the SCR switch were inspected, and stood at 9.21 and 8.9V. The 8.9V battery is a bit low, but not really in the range where we would expect to see its performance change. It's not been ruled out yet that the SCR is undamaged. Let's hope that's not the case. Bank was dumped (still works) and shorted.

Otherwise, the day went rather well. Lots of shots were taken, almost all shaped. There was some trouble getting MR to behave the way we want - to rise then fall *gradually*, rather than crashing in, but we seemed to be making good progress until the problems discharging arose. Chamber seemed to clean up as the run progressed - this is only the second half-day of cleanup since the leak in the gas-puff, so the shots are still quite variable. Running the shaping coil decreases plasma lifetime by around 1ms across the board, but nevertheless, it seems that if we were to run a shot optimized for high IP and/or long life we would be able to hit star shots like 70000 and 71000.

I'll start work on the Shaping bank first thing tomorrow, and hopefully we'll be able to run.

Thomson Scattering seems to be working well too and the data is saved in the tree if anyone wants to look at it. Node is Sensors.Thomson.

Pat

Thursday May 1 2014 5:57 pm Byrne/Stoafer 84274-84299 Troubleshooting/RMP Shaping

Issue with the shaping bank turned out to be the trigger pulser. Two 9V batteries run it, one was at 8.97V. This was apparently enough to cause the SCR to not trip. Checked the pulse shape once I replaced the low battery, and the pulse is only 2V peak into an Oscilloscope. According to the SCR's datasheet, it should require 3V to turn on, so we may be marginal and very sensitive to the batteries drooping.

Shaping bank ran well all day. Began with the control coils running as pseudo-VF coil. This actually worked quite well, and by 84292 we had a q* trace very similar to shot 69642, from Daisuke's RMP run, used as reference, in the time of interest.

Began having crate troubles by the end of the day. Shot 84293 carried the first 3-1 RMP, overlaid on top of the VF pulse. 10A of VF, 10A of RMP. Concerns over the 50% shell coverage led us to boost the current to 10A VF and 30A RMP for subsequent shots. No indication of blown fuses. 84295 and 97, the TF did not digitize. This was seen once yesterday. Paul had apparently encountered this in his campaign as well. q* can not be reconstructed using tree data in this case, though we can assume the toroidal field is very similar from shot to shot and use a reference without introducing too much error.

82498 and 99, the RMP would not fire. No clue as to why yet. Will look into it tomorrow. Run day was closed down by the 4:10 class Chris and I had to attend.

Star shots are 82492 to compare our base shaped equilibrium with reference 69642, 84293-97, to look at 3/1 response in a shaped plasma

Pat

Friday May 2 2014 7:25 pm Byrne/Stoafer 84300-84346 Shaping + RMP

Tested Caliban with 2 crates only shots first thing, no problems with getting the control coils to fire. Control coils worked fine all day afterwards. Not sure why it would have failed at the end of the day Thursday

shots 84318 - 339 represent the compass scan. The python function I called to get a random angle seems a little busted, as the values tended to cluster together for several shots, before a big jump to a new cluster zone. Last couple shots were chosen deliberately to fill in some of the larger gaps. Still, we covered the compass pretty well. Lost a fuse in one of the RMP coils at 319 (made worse since we are only pulsing half of our coils) and did not notice until 33 but the 3-1 component of the RMP should still be dominant.

Shots 340-46 represent the amplitude scan. Phase is set to zero, amplitude is varied from 3-.5 (or 30-5A in the coils, unsure the Gauss strength of the 3-1 field). All fuses were fine for all of these shots.

All shots were very close to Daisuke's shot 69642 in q* during the time of interest. The 'VF' pulse on the RMP coils was extended .5ms before and after the 3-1 RMP so that we have time to examine the effects of the one in isolation from the other. q* is a pretty rough parameter considering all the issues with the shaped geometry, but it's the best we have so far. Turn-on,flip and turn off times are visible in the stripey plots, so there should at least be something worth looking at.

Computer problems:

Take shot script hung at one point early this morning. Banks failed to auto-fire, and Take Shot script needed to be force quit. Command line control did not return even then, until a Ctrl-C was given, at which point a segfault was reported. Chris said he saw something similar before the gas puff failed, so Athena was rebooted just to be safe. The gas puff was not triggered at any time up to this error, and no irregularities were seen after the reboot.

The TF probe is often failing to digitize. This was seen yesterday, and it seems to occur more often around the end of the day. Heating in the rack shouldn't be an issue, but I can't think of why else. Jeff has suggested that the extra-long time base of the TF digitization may mean it is still digitizing when the store command comes through, causing an error. Suggestion is to disable the TF phidget, and fire manually, hitting pulse, store and analyze separately. This is impractical for my run, but may be worth trying later.

This may not be the case for anyone else, but running with Athena can be a pain. Despite running as much code as I can on Spitzer or Caliban, I still need to switch between many programs to analyze and prepare each shot using an interface run on a computer that has trouble supporting it's own load. If a reboot is necessary, that's another 10 minutes wasted. I feel a lot more could be gained from a runday if computer latency weren't a time-limiting factor. Athena is a pretty central part of any experiment we do, so if it can be done without messing things up too badly, the next time we have the opportunity and means, we should probably see about replacing Athena with a computer with at least an average amount of oomph.

Monday May 5 2014 6:07 pm Stoafer 84347-84411 TS scan minor disruption

The goal of the run was to do a TS scan of a shot with a reproducible disruption to see the Te profile before and after the disruption. I am still working on shot development to reproduce shot 83817. Shots are getting caught with an early 3-1 mode, which prevent them from getting to lower q. I try to get the plasmas to approach q=2 and use an RMP to induce a disruption at 4.5ms.

Changing plasma and machine conditions have slowed shot development, but I am starting to get shots I can use. Since my previous run the gas puff valve has been changed, making the back pressure different.

We had breakdown issues in the morning, which was found to be problems with the gas puff valve. We increased the back pressure on the valve from 9PSI to 10PSI and increased the gas puff time from 75us to 1000us, which resolved the problem. This dramatic increase in the gas puff timing shows that the valves can have different characteristics week to week.

Tuesday May 6 2014 6:26 pm Stoafer 84412-84474 TS scan minor disruption

Today's run continued to perform a TS scan of minor disruptions. Towards the end of the day we were consistently getting a minor disruption at around 4.5ms or 5.5ms (shots typically deviated to either on of those). The TS laser timing was varied to determine Te and relative density distributions throughout the shot.

Plasmas were not reproducible in the morning and it was found that the neutral gas pressure was varying. The cause ended up being the battery for the gas valve circuit was not charged. The battery was had a full voltage reading yesterday, but had apparently not been charged over the weekend (it was outside of the gas valve circuit box and near the charger on Monday morning, but not connected to the charger).

The low battery voltage made, what seemed to be sporadic behavior is the gas puff, with changing fill pressures from shot to shot. Once the battery was replaced, the fill pressures leveled out, but required a longer gas puff time (1500us instead of 1000us) to get to the same fill pressure.

Thursday May 8 2014 7:05 pm Byrne 84475-84533 Shaping + RMP

Star shots are 84506,84527, 84529, 84530, 84531, 84533. They are shaped shots, with RMP, that persist for at least as long as the RMP. 84533 has 30A of RMP current, the others have only 16. Had to reverse the VF RMP polarity, to push the plasma in rather than pull it out. shot 84506 is from before that switch happened, so it is comparable to the last run for comparing 3/1 to -3/1, but probably not much else.

Unfortunately that's about all I got. Was unable to match the shots from last Friday well for most of the day. Took a lot of development to bring the plasma around, but we do have a reasonably repeatable shot. Hopefully we'll still have it tomorrow morning.

Pat

Friday May 9 2014 4:25 pm Byrne 84535-84588 Shaping + RMP

Spent a lot of this morning trying to improve on the equilibrium from yesterday. Ended up settling on the original equilibrium, with the tweak of only imposing the VF cc pulse early in the shot, at the time the early mode causes MR crashes. This did not eliminate the occurrence of early modes, but did help reduce the frequency. Star shots are 84577, 84574,73, and 71. These shots all had q and MR close to the model shot of 70000, Shaping that started at 1.2ms, and RMP phase flips with 30A of current from 2-3ms.

Tried to get an unshaped equilibrium, but it was too late in the day to do anything useful. This will have to come later.

Pat

Friday May 9 2014 5:11 pm Byrne/Levesque 84589-84594 Unshaped + RMP

Added a few more shots to the end of the run day. Was able to get a couple of unshaped RMP shots that are similar enough for direct comparison to the shaped shots from today. Additional good shots are: 84589 and 84591.

Monday May 12 2014 7:18 pm Hughes 84595-84644 FW+RMP extension

The objective of today's run was to fill in the high-amplitude gap in the ferritic wall part of the response-vs-RMP amplitude plots I showed in the last couple of meetings. I started by taking some non-ferritic RMP shots to make sure the response is comparable to my previous shots before casually inserting new data. However, on reversing configuration to insert the ferritic walls, the plasma became much less cooperative, displaying erratic breakdown problems and intense major radius jogs sometimes over 1cm, at around 1.3ms, as well as being extremely sensitive to bank settings and varying widely from shot to shot.

More shots will be required to say anything definitive, although a majority of shots that weren't rejected for excessive deviation in major radius and q* instead disrupted between 3.4ms and 4.2ms. All of these were at RMP code amplitude of 4 (est. about 50 Amps); I may try A=3 or A=3.5 RMPs tomorrow, instead of pushing for A=4.0 as I did today. Thomspon Scattering was running today, but Chris can say better if it saw anything noteworthy.

Due to breakdown problems shortly after switching to the ferritic configuration, the e-gun settings were changed. The e-gun is now set to a supply voltage of 151V; with the heating current at 8A, this yields about 35V filament bias. Plasmas broke down much more willingly after this change, but also displayed the large major radius jumps (possibly due to early mode activity) mentioned above.

Noteworthy shots:

• 84595: Daily vacuum shot
• 84610: Star A=4.0 non-Fe shot
• 84618: Star A=1.0 non-Fe shot
• 84637: Star A=4.0 Fe shot
• 84635,84638,84640,84641,84644: Disrupt during or very soon after the end of the RMP

Analysis will follow when there's more data to look at and I'm not busy running, hopefully in time for the Wednesday meeting.

Tuesday May 13 2014 6:10 pm Hughes 84645-84681 FW+RMP extension

Although plasmas continued to be somewhat uncooperative in the same manner as Monday afternoon, several shots with RMP amplitudes of 4 and 3.25 in code were taken. The resulting calculated response show considerable shot-to-shot variation, with some data points agreeing with the slope of the A=0.5-2.5 series and others suggesting a saturated behavior.

The interpretation of the new data is therefore not obvious, and will probably require more detailed shot-by-shot analysis and careful reasoning to decide whether a particular shot is "bad"--i.e. having an equilibrium that changes radically, making subtraction difficult, or with an equilibrium different enough from the other shots to experience very different MHD damping/amplification and therefore exhibiting different saturation behavior for the same RMP.

This seems like highly subjective territory, so my inclination for the present is to set the extended data aside as Not Yet Understood, and focus on the prior set of data which is at least relatively consistent with itself.

Noteworthy shots:

• 84646: Daily vacuum shot
• 84653,84658,84659,84660,84666,84667,84679: A=4.0 shots that disrupted between 3.5 and 4.5ms
• 84670,84671: A=3.25 shots that disrupted between 3.5 and 4.0ms
• 84678: Best A=4.0 shot of the day (preliminary analysis suggests saturation)
• 84672: Best A=3.25 shot of the day (preliminary analysis suggests saturation)

One interesting observation is that the rate of disruptions between the time of the phase flip and a short time after the RMP ends seems to be MUCH higher for high-amplitude RMP shots.

Monday May 19 2014 5:46 pm Qian 84684~84732

The run was for experiment data to apply subspace identification method. Most of today's run are spent on shot development. Shots are not very reproducible during the day. Ferritic walls are in and perturbation is applied on those shells. The perturbation doesn't seems to disrupt the plasma so the data should be able to be obtained from this method. More shots will be tested tomorrow.

Tuesday May 20 2014 5:54 pm Qian 84733~84798

all the shells are put in today and the plasma becomes more reproducible. plasma with randomized peturbation are performed and data is obtained to test the subspace identification method.

Wednesday May 21 2014 5:15 pm Stoafer 84799-84852 TS scan of 92cm plasma

The goal of today's run to do a TS measurement scan of plasmas that stay at 92cm throughout the shot. These are the shots that Qian has been using for preparing his state space feedback algorithm. For these shots both the ferritic and stainless shells are inserted. The stainless shells are inserted 13mm back from their nominal position to match the location of the ferritic shells.

We saw high temperatures throughout the shot with a relatively low density compared to other shots with TS with the new alignment. This makes sense because the soft x-ray signals are low compared to other shot styles. Temperatures in the 92cm and 96cm views ranged from 100 to 200eV based on the measurements.

I have been looking at the analysis I use for determining Te from TS measurements and have been working to improve it to be more reliable. Ideally, we would be able to do some Raman scattering shots to get a good averaged shot that I can fit each plasma TS measurement to to reduce the affect of noise in the measurements.

For now, I believe the measurements might be high, but I do believe the plasmas get to be at least 150eV towards the center based on the raw signals from TS.

Thursday May 22 2014 4:26 pm Qian 84853~94912 feedforward shots

Perturbation with fixed rotation frequency has been applied today. These data, along with the randomized frequency obtained on Tuesday, should be able to give an estimation for the response matrix for the system. More data analysis will be done later.

Shots are quite reproducible with current wall configuration. Plasma sometimes disrupts suddenly around 4~6 ms. This is not correlated to the perturbation that this applied.

Tuesday May 27 2014 3:03 pm Stoafer 84914-84955 TS scan higher density

The purpose of the run was to perform a TS scan of a similar shots to the previous scan (shots 84799-84852), but with higher density. The fill pressure was increased from 74uTorr to 135uTorr, which more than doubled the density. I was able to develop a shot similar to the low density shots.

The temperatures were lower than the low density shots, but I believe the density is raised enough to have a net increase in pressure.

I will continue with taking data tomorrow and will perform analysis of the two shot styles.

Tuesday May 27 2014 8:17 pm Levesque 84956-84976 TS scan, normal density

Ran more shots to extend the run day and collect TS data with a different discharge evolution. Developed a shot that starts with R₀ near 96cm and q_* near 2.5, then gradually moves inward as the current rises and q_* stays roughly constant. Good overlaying discharges of this type were 84959, 84962, and 84963. The fill pressure and density were typical of our recent shots, being lower than Chris' run earlier today. The discharge evolution was not too reproducible, but it did provide reasonable TS data as the laser timing was changed. Temperatures were typically around or below 125eV throughout the shot, and were generally lower than data taken during recent runs. I attribute this lower temperature to the lower starting current that was required to maintain q_* above 2 with a small minor radius.

At the end of the day, I started developing a discharge similar to the high-current plasmas from June 2013 (reference shot 79771), but ran out of time before getting a good shot.

Wednesday May 28 2014 4:31 pm Stoafer 84977-85032 TS scan of 92cm steady plasmas

I finished the TS scan of the higher density plasma shots with a steady 92cm major radius. The density measurements were significantly higher, but the temperature measurements were lower. I will perform a quick analysis looking at the temperatures, densities( arbitrary), and pressures (arbitrary) between the shot styles to see if we can develop similar plasmas with different pressures.

Thursday May 29 2014 4:44 pm Hughes 85033-85048 Extended timebase ion gauge measurements

Today's (brief) run was aimed at studying the behavior of the ion gauge data on longer timescales than the normal t=1s limit (~120k samples x 10us/sample) by increasing the clock divide from 10 to 100. This provided a ~12s long basement A14 timebase.

We have long observed that plasma shots take longer to pump down and establish a higher base pressure than shots in which gas is puffed but no plasma forms. Since digitizing the ion gauge earlier this year, we have also observed that, contrary to the long-time (~10+ seconds) behavior, the pressure following a plasma shot drops below the pressure seen at the same time in a puff-only or non-plasma puff+banks shot.

With the extended timebase, we can see that (given a plasma shot and a puff-only shot with similar fill pressures) there is in fact a crossing point, generally between t=1.5 and t=3.0 seconds. The crossing time seems to be determined by fill pressure and vacuum condition at the time of each shot. Before the crossing point, even after all significant electromagnetic perturbation of the ion gauge dies down (around 1.0-1.2 seconds), the plasma shot exhibits a lower pressure than the non-plasma shot with puff; after the crossing point, the plasma shot's pressure is higher.

The best theory I can put forward is this:
When a healthy plasma forms, it sputters impurities off the walls, making the walls very clean; gas is pumped out fairly quickly by the cryopumps, but due to this wall cleaning, the walls help to accelerate the pumping until they saturate on a timescale of 1-3 seconds. The material knocked off the walls resists cryopumping much more than deuterium, suggesting helium as a top candidate. As a monatomic gas, helium should be more able to soak into the outer layers of the wall's lattice, which might help to explain its persistent abundance in the wall; its presence in the vessel would also be cloaked by the identical mass of the deuterium.

There may be other possibilities, and there are some puzzling aspects about this theory--particularly the persistent abundance of helium after every shot, months after it was last introduced during glow discharge cleaning. If anyone has clever ideas, it would be interesting to hear them either by email or in our next meeting.

One no-puff, no-banks shot was also taken in which the e-gun was switched on after hitting PULSE ON: this shot shows the time characteristic of the pressure rise we observe whenever we switch on the e-gun, revealing a surprising amount of temporal structure with a large-scale risetime of about 2 seconds, but unexplained perturbations with ~0.5s timescales.

Noteworthy shots:

• 85033: Daily full vacuum shot
• 85040: Puff-only shot, 1500us puff
• 85044: Plasma shot, 1500us puff
• 85045: Puff+banks, no e-gun, 1500us puff
• 85047: No puff, no banks, e-gun on after PULSE ON

I'll be setting the timebase back to its usual settings to avoid the risk of breaking code that doesn't anticipate alternative timing arrangements. For future reference, extending the basement A14 timebase requires changing CLOCK_DIVIDE in the basement A14_04 node, as well as modifying the trigger settings in .devices.basement:J221_09>"Setup Device">"01" from [5000000] (5,000,000us) to at least 13,000,000us (I used 15,000,000 for simplicity). I'll be sending around plots of the data discussed here.

Friday June 13 2014 5:53 pm Stoafer 85049-85085 Bias probe at floating potential

Today we ran with the bias probe after it was re-installed on the machine. The plasmas cleaned up very quickly, with the bias probe retracted and the gate valve open, it only took 4 shots to get back to normal-operation plasmas.

This bias probe was disconnected from the biasing electronics, so it would remain at the floating potential, with a voltage divider to measure the voltage. The bias probe was fully inserted, with the front of the probe tip at 103cm. After about 10 shots, the plasmas looked similar to before the probe was inserted.

The reference shot used was 84941. 85073, 85076, 85077, and 85078 were good shots with the probe at 103cm. 85080 and 85081 were good shots with the probe tip at 107cm. 85085 was a good shot for the probe tip at 106cm.

During the run, we noticed that the voltage on the probe was reading positive, which did not make sense. We checked the electronics and found that there is a sign inversion associated with the A14. Bryan found this as well, as noted in his run report on May 11, 2011. The inversion is now accounted for in the tree (sensors.bias_probe), which also accounts for the voltage divider.

On shot 85060 we noticed large voltage measurements on the probe, which were caused by ground loops. Jeff found that the grounding wire of the BNC cable inside of the south rack was very close to the rack casing, which might have been causing the ground loop. The grounding wire was pulled back from the rack casing to resolve the problem and we did not see the issue again.

In all of the plasma shots today, we saw additional noise in the major radius measurement than usual. Jeff suspected a ground loop either associated with the bias probe or other work done in the area and worked to resolve it, but the issue is still not fixed. This is not a very bad issue, but we will continue to try to resolve it. Note the cosine rogowski is in the west rack, which is the same ground as the chamber. The bias probe circuit uses the chamber ground, but its voltage is measured in the south rack, which uses a different ground. The plasma current measurement is also in the west rack, but the noise is not seen. This might be due to different RC circuits for making the measurements.

The run will continue on Monday to see how the floating bias probe affects the plasma conditions. TS measurements will be taken. We also plan to run the bias probe in a grounded configuration to measure how that affects the plasma. Tests are planned to measure the crown amp output through the transformer to figure out how to properly run with biasing the probe.

Monday June 16 2014 5:39 pm Stoafer 85086-85106 Bias probe floating and grounded

Today we ran in three configurations: bias probe retracted (shots 85087-85091), bias probe inserted to 103cm and floating (shots 85092-85096), and bias probe grounded at same location (shots 85097-85106).

There was no extreme noticeable change in running with the probe retracted versus floating, perhaps a decrease in natural mode amplitude, but this needs to be analyzed.

The natural modes seemed to slow down and become larger in amplitude, as expected, when the probe is grounded compared to when the probe is floating. This is exciting and will be looked at further.

We stopped the run to install a more appropriate Pearson for measuring the current to the probe, which was done at shot 85106 and works well. We also plan to change the voltage divider for measuring the voltage on the probe.

Towards the end of the day we started setting up the Crown amp to be run with a function generator triggered from the south rack. This setup will be completed for tomorrow's run, where we will control mode rotation by biasing the probe in various ways.

Star shots include:

• Probe retracted: 85091
• Probe floating: 85095
• Probe grounded: 85106

Tuesday June 17 2014 6:09 pm Stoafer 85107-85128 Bias probe with amplifier

Today's run goal was to run with a bias probe connected to the crown amplifier to affect mode rotation. Shot 85127 and 85128 were plasma shots with the bias probe driven at +18V from 2ms until 6ms, which confirmed that we are biasing the probe as expected. More shots with different probe voltages are planned.

Most of the day was spent troubleshooting issues with the bias probe. This included Jorway triggering and using a function generator as input to the Crown amp. These seem to be resolved currently, so we were able to take a few plasma shots.

Tuesday June 24 2014 5:40 pm Stoafer 85130-85144 Bias probe

Qian and I setup the bias probe to be controlled by the same system as the control coils. This allows us to easily change the signal we apply to the probe and do feedback with the bias probe in the future.

A short run was performed after the setup was complete with a triangle wave applied to the bias probe. For shots 85140 and 85141 a triangle wave was applied from 2ms to 6ms with a period of 1ms ramping from 0V to about 45V. We saw mode amplification (possibly slowing) at the peaks of the triangle wave, which is expected. Shots 85143 and 85144 had a negative triangle wave with the same parameters and the mode did seem to speed up, but not at the same time the the triangle wave peaked negative.

A run with similar ramps will continue tomorrow.

Monday June 30 2014 5:36 pm Byrne 85151-85177 Cleanup/conditioning

Started around Noon, as the shells needed to be repositioned to begin the runday.

Plasmas were very short lived and SXR light was very high for most of the day, suggesting a high amount of impurities. A High, steady outboard limited plasma was developed and run several times. Following that a few shots were taken with the same settings as the target shot (82575).

Ip was lower, LV was higher, lifetime was shorter, and SXR was marginally higher. Plasmas seemed to have trouble getting below q=3. the final shot of the day, 85177, is a good example. Whether it's a coincidence of Ip and MR development or not, it holds almost constant at ~3.25 before the plasma begins an inward crash ending in disruption.

Plasma is developing, and despite the heat, things seem to be improving. Hope to have good data by the end of Tuesday.

Tuesday July 1 2014 6:09 pm Byrne 85178-85236 Shot development

Progress, however slow. Shots have finally begun approaching a lifetime of 5-6ms, roughly the standard, as well as reaching high current, and straight current ramps, instead of a 'slumpy-saddle backed' trace (see early shots,85182 for example)

Still having great difficulty getting q below 3. A very high OHramp (OHE at 340 out of 350V) shot was able to push q down, but IP was obviously too high for suitability for shaping. (shots 82523,24,25) dropping OHE, in parallel with VFE, and TF to attempt to reduce IP, while keeping the ramp and MR/q trace intact has not worked yet, though only a few shots have been taken and development is still in progress.

Hope is to scale down the IP trace, while maintaining the plasma's transit through the chamber and it's q, to allow us to ensure diversion when we shape. The few shots taken with shaping have disrupted too early for useful data to be gathered. There will probably be extra development needed once an unshaped equilibrium is arrived at.

Wednesday July 2 2014 6:22 pm Byrne 85236-85275 Shot development

Am still unable to find a shot that reliably goes below q =3. Any suggestions are welcome.

Have been experimenting with a 'VF' type control coil pulse, to push the plasma quickly outboard, dropping q below 3. Have had some success, however, I have not successfully managed the actual VF bank such that the plasma is not immediately pushed back, and q sent above 3 again.

85245 is a good example. passes through q =3 twice (passes back out again due to q =2 disruptions) but even at late times, has trouble getting through q =3. My theory before today was that there was something about the breakdown (hollow IP profile, strong eddies in the shells/chamber...)that was causing the plasma to exhibit this behavior. Since we see it at times after 4ms, I think it might a permanent feature, at least with the shot settings I am using.

Have heard that this is due to high OH settings, but that was not seen to be the case in a previous run. Have been steadily dropping OHSt and OHE over the course of today and have yet to see the behavior significantly affected.

Ran a study of moving the VFstart later and later to force q through 3 via large MR/low a, (shots 85245-49) and it seems like there are two very different operating regimes on either side of the q = 3 line. A plasma that is finally pushed through q of 3 is suddenly very positionally unstable, and flies out until q~2 causes a large excursion in MR back to a more stable position. If that position is above q of 3, there will be more hitching. If the plasma can then pass through a second time, the same thing will happen

If anyone has run in the last few weeks/months, with q below 3 and MR above 92, please let me know

Also, the control coils in chamber 1, top shell, midplane were found to have shorted to the shell. Last time they worked was shot 84890. There is pickup due to the OH bias firing seen in the shunt, which suggests that the wires are not broken, but that the insulation has worn away somewhere, providing a second ground path to the coil. Repair work may be possible in the future, or it should be possible to remove the large coil from the circuit and use the medium or small coils as well.

Thursday July 3 2014 6:25 pm Byrne/Levesque 85278-85340 Shaping

Quick heads up, the leak rate of the silicon oil seems to be going up. Not sure why, but the reservoir was completely filled on Monday, and by Thursday morning, the level had fallen below the 'T'. As a head's up to anyone else who may be operating the machine in the near future, the new automatic pump fills the reservoir much faster than it empties into the system. I got a bit of an oil shower this morning waiting when the reservoir filled before the line had even filled to the yellow rope. Best to turn it on for 5-10 seconds, then proceed with the startup checklist.

Early part of the day was spent cleaning up the machine. At about 1pm, the 'hitching' behavior of the plasma at a q of 3 suddenly stopped. Since the day and week were ending, rather than develop the shot further we took what we could get and ran the shot as is, along with a shaped equilibrium that was very similar.

q goes below 3 early, drops to about 2.5, then rises to 3 or above after ~2ms as the plasma falls in. It may be difficult to BD these plasmas, as they move a lot in the chamber, but they are very repeatable, and we were able to overlay shaped shots on them quite well. Star shots would be anything from 85318 to 85328 for unshaped, and 85330 to 85340 for shaped. in particular 38 was run with 7kA of shaping current, and had good q behavior, so there is a very good chance it is diverted.

Monday July 7 2014 7:29 pm Byrne 85341-85408 Shaping

Got many more star shots, in that they were repeatable in the main, and went below 3 and back up again, with some time after before disrupting.

Shots are still highly variable in their actual MR/q, but that was an inescapable feature of this campaign. The shot database is much larger than for others, however, so that should help us determine what is an actual feature of shaped MHD and what is unrepeatable 'noise'

Took two shots at EOD with 3-1, 20A RMP on the 10 inserted shells. the shell with the broken coil is not among them. Have not analyzed, but figured it would help to plan the next campaign. As always, I would not read to carefully into any differences with a sample size of one, but if you care to look, they are shots 85403(unshaped) and 85404 (shaped, disrupts right as RMP ends. Did not see this at all without RMPs) and 85408 (shaped, does NOT disrupt despite similar q evolution.)

Total Star Shot (as in comparable data) database is:

From today:

Unshaped: 85368,70,72,93,96,97,99,401,402

Shaped: 85378,83,84,85,87,88,89.

From yesterday:

Unshaped: 85318-28

Shaped: 85311,13,30-40

There is a lot of variation within these shots, due to the large amount of early mode activity, (which antedates the shaping, if any) and causes unpredicatble behavior as q=3 is crossed. should still be plenty to work with

Pat

Tuesday July 8 2014 6:26 pm Qian 85409-85459 feedback shots

The purpose is to test the improved feedback algorithms with phase dependancy.

All shells are inserted for the current configuration. non-ferritic shells are backed by 13mm from fully insertion. Early today, the ferritic shells were fully inserted and the plasma was behaving quite unstable: minor disruptions happens on the frequency of 1/0.3ms and the major radius behaves like piece-wise step function. Example shot is 85421. These could be interesting shots to examine later but are not suitable for the current purpose.

Later it was found out that ferritic shells should also be backed by 5mm and that got rid of the strange behavior of the plasma.

desired shot style was reached later today and feedback shots are tested. There was a bug in the code found during testing and took some time to fix it. So we didn't have much time in fully test the performance of the algorithm.

Some feedback shots without phase dependency was performed to reproduce the previous result. It seems that non-feedback shots have a tendency to disrupt when q reaches around 2.2. With feedback at a given phaseshift, this disruption is avoided. and plasma lives longer. feedback shots: 85450,85456 non-feedback shots: 85451~85455 which all disrupts when q reaches certain point.

Further tests with phase dependent algorithms will be tested tomorrow.

Wednesday July 9 2014 6:39 pm Peng 85460-85505 feedback shots

The them of today's run is to test the feedback algorithms.

The shot style has a roughly constant MR around 92cm. q drops across 3 around 2.5ms and reaches 2.2 around 4ms. When q reaches this low value. a strong m=2 mode appears and grows, usually leading to disruption. The growth of this mode is nearly exponential and can be used to calculate a growth rate.

feedback coil are turned on at 2.7 ms and turned off at 5.5 ms. It can be seen that with proper value chosen, the growth rate of the m=2 mode is reduced. It still disrupts finally, but q stays at the critical value around 2.2 for longer time. star shots for comparison are:

non-fb shot: 85502, 85505

fb shot: 85503,85504

the phase dependent algorithm is used in the above feedback shots. But it seems that the phase shift and gain parameters affects the performance more. even with the reversed phase dependence, with the proper phaseshift and gain, the growth rate is still reduced to an extent: 85501

more tests will be done tomorrow.

Thursday July 10 2014 6:03 pm Peng 85506~85584 feedback shots

today's theme is feedback test. Plasma current is lower compared to yesterday's run with the same settings. Increasing the puff time or OH start or OH elect does not have much effect on boosting the plasma current to yesterday's level. This resulted in a different q trace and yesterday's shot styles could not be reproduced.

For today's shots. The early activity still occurs frequently and I think this is related to the instability introduced by ferritic walls. MR is chosen around 93 mm where this effect is sort of mitigated. q traces are higher today compared with yesterday's shots because of lower plasma current. Feedback effect seems to be observed when q drops across 3 and stays around 2.4~2.7. with MR around 93mm. Suppression can be seen when the plasma parameters are in this region.

star shots are 85581 (nonfeedback) ,85582 (feedback). the control coils are turned on after 3.5 ms.

The plasma is not very reproducible and shots with different parameter traces are hard to compare. more tests would be done tomorrow.

Friday July 11 2014 4:47 pm Peng 85584~85640 feedback shots

Shots similar to Wednesday was performed today. Shots disrupts with a exponentially growing m=2 mode as q reaches near 2.2.

When feedback is on, such growth rate can be reduced.(not fully stop the exponential growth, but the time is extended.) More analysis will be done over the weekend to show a more clear result.

Monday July 14 2014 5:36 pm Stoafer 85641-85684 Exploratory Bias Probe

Today's run goal was to explore how the bias probe interacts with the plasma. I had recently tested the crown amplifier on a resistive load to determine optimal settings and limits on the input voltage. With this information, I setup triangle waves and square waves to see how the bias probe interacts with the plasma. We also explored medium density and low density plasmas by varying the puff time.

For triangle wave plasma, we saw expected IV curves, which will be sent in a separate email. We did not see the probe clearly get into the electron saturation regime.

With the square waves, we applied a positive voltage from 2 to 3 ms, then a negative voltage from 3 to 4 ms. We also reversed the voltage with negative from 2 to 3 and positive from 3 to 4. We saw the plasma mode reverse direction, as well as lock, for the maximum amount of positive voltage applied.

The following are lists of shots taken and their conditions:

• Medium density triangle waves were shots 85659 and 85660
• Medium density pos to neg square waves was shot 85662
• Medium density meg to pos square waves were shots 85663 and 85664
• Low density triangle waves were shots 85675, 85677, and 85679 (which is a star shot)
• Low density pos to neg square wave were shots 85680 and 85681
• Low density neg to pos square wave was shot 85684

You can find bias probe data saved in the tree under sensors.bias_probe:voltage and sensors.bias_probe:current. You can also find a useful jScope configuration under bias_probe_new.jscp

Tuesday July 15 2014 5:09 pm Stoafer 85685-85726 Bias Probe feedback

Today's experiments with the bias probe focused on getting feedback working with Qian's help. We were able to get a simple algorithm to feedback on the bias probe voltage to maintain a certain mode frequency. For a first try, this worked very well, but it will be tricky to optimize it because the bias probe suppresses mode amplitudes making it difficult to determine the mode frequency.

A scan of bias voltages using step functions was also run, which will be used as a reference for the feedback algorithm.

Star shots for feedback:

• 85705
• 85706

The following are star shots for the scan of input voltages into the Crown amp:

• 0V: 85708
• 1V: 85709
• 2V: 85711
• 3V: 85715, 85716
• 4V: 85717, 85718
• 5V: 85720
• -1V: 85721
• -2V: 85723
• -3V: 85724
• -4V: 85726

Thursday July 17 2014 4:39 pm Peng 85728~85771 feedback shots

Today's shot is for feedback tests. There was some problem with turning on the crates this morning. It might be caused by the thermal issue in the control room. Shutting spitzer down, cooling it and then rebooting fixed the problem. Plasma shots are not very reproducible today. We still got some shots that is similar to the target shot and did a brief phase scan. More tests will be done tomorrow.

Friday July 18 2014 3:02 pm Peng 85772~85808 feedback shots

more feedback shots are performed today. Plasma shots are not very reproducible. The MR radius swings between the target style (outboard) or falls inboard. This is largely due to the early mode activity which cannot be controlled by now.

a brief phase scan is performed and the results will be analyzed.

Monday July 21 2014 9:27 pm Levesque/Stoafer 85809-85842 Low q* shot development

In the beginning of the day we performed rotation feedback with the bias probe. The target rotation was 9kHz, using a positive bias to drive faster modes. The feedback worked well with the simple feedback algorithm and plasmas seemed to be more stable, however only a small number of shots were taken with this focus. Following the brief feedback run, shots were developed which have a centered major radius and q_* decreasing to near 2 at 4ms. These shots will be used for later magnetic feedback attempts.

Starting at shot 85824, the focus changed to developing plasmas that have the initial q_* in the range of 1.5 < q_* < 2 at the breakdown. We had moderate success, and were able to reproducibly get plasmas with initial q_* down to around 1.5. These plasmas were very unstable, with the major radius always crashing in quickly and/or plasma current falling quickly to move q_* above 2. When q_* rose above 2, the plasma current tended to rise as programmed unless R₀ had decreased too far (and led to a major disruption before the plasma could recover). Strong 2/1 modes were seen while q_* was below 2. This q_* < 2 time window lasted less than ~0.5ms, and the major radius was always changing quickly. 3/2 modes were also seen in the magnetic sensors for these shots. Further shot development is needed to see if it's possible for us to maintain these discharges for longer times; not much exploration has been done yet.

The method for getting to low initial q_* today was to have high I_p at breakdown and use a much lower TF than normal. Typically we run the TF bank at 6.1-6.5kV, while today we ran it at 3.6-4.0kV. We ran the OH Start bank at our maximum comfortable limit, which is still well under the rating of the bank. We will soon clean and evaluate the OHST bank to see if we are willing to raise our present maximum charge limit.

Noteworthy discharges are:

• 85828: One of the best shots with q_* below 2 for a long time. Note that the q_* measurement has not been evaluated for accuracy under these conditions.
• 85830: SXR fan array appears to show a "snake mode" around 1.2ms. Other shots may have this feature too, but this was the clearest.
• 85832: Bursts of 2/1 activity throughout, and definite 3/2 activity at 1.5ms.

Tuesday July 22 2014 4:41 pm Stoafer/Levesque 85843-85876 q<2 shot development

We continued with shot development for plasmas with q* less than 2. A good base shot for the early development of the shot style is 85866. The plasma moves outward to about 94cm with a large starting plasma current, bringing the q* below 2 quickly in the beginning.

However, there is not enough OH ramp to keep the plasma outboard, so it crashes inwards. We were able to adjust the settings to get this good starting behavior, but are having trouble in keeping the plasma going. We used a high OH bias of 5000V and set the OHE at its max of 350V to get a good current ramp. We also raised the limit of the OHS bank from 10kV to 12kV with Nick's approval after inspecting the bank. For this shot to be developed more, we would need to raise the limit of the OHS or OHE settings, which we are not interested in doing at this time.

We can still explore regimes in reducing the TF for lower q shots. In these shots, the TF bank is set to 4700V, about 2/3 the normal setting. We have seen that we can lower it even further and still get breakdown.

No specific plans for continuing this shot development are in place.

Tuesday July 29 2014 4:48 pm Peng/Stoafer 85877~85920

Shot developments are done today to make a shot style with q starts below 3 and reaches 2 around 4ms. The shots are not very reproducible and gets worse towards the end of the day. But progress is made and more tests will be made tomorrow.

Monday August 4 2014 5:50 pm Peng 81921~81975

Feedback tests was done today. q was maintained roughly stable with a low q rather than a decreasing case. shots have less tendency to disrupt with this q trace. When feedback is on, the modes frequency can be changed but the amplitude is not affected much. maybe the mode amplitude is saturated and is maintained at a certain level.

Wednesday August 6 2014 6:07 pm Peng 81975~81925

More shots developments are done today. We can let q reaches below 3 quickly and rise up approaching q=3. The major radius is roughly maintained constant. The shot style is reproducible at a rate about 3/4. So we should be able to do more feedback tests on such shot style.

Wednesday August 6 2014 6:30 pm peng 85975-86025

This is a correction to previous run report. the shot numbers are 85975~86025

Thursday August 7 2014 5:42 pm Peng 86025~86065

Shots today are targeted to have q stays below 3 and have a kink mode. Feedback control is tested on such modes and suppression effect can be seen.

star shots are 86060 and 86061.

Thursday August 14 2014 6:01 pm Peng 86066~86115 plasma shot

most of the time today was spent on getting the same shot styles as last Thursday. Should be able to perform feedback tests tomorrow.

The broken CC06_S1 control coil continues to give the spike around 1ms even after the fuse in the loop has been taken out. It suggests that the signal is coming from the digitizer side.

Friday August 15 2014 5:47 pm Peng 86116~86160 Plasma shot

Shots today are still not very reproducible and much time is spent on tuning the parameters. Some comparison is finally obtained in the end though.

86155 (fb shot) compared 86157 (ref) and 86156 (fb) compared with 86160 (ref) shows the feedback suppression effect. Feedback is turned on at 3.8 ms.

Tuesday August 19 2014 6:52 pm Peng 86162~86109

feedback experiments continued today to see the effect of turning it off early. The targeted shots are not easy to reproduce. The comparison pair is 86190(ref) and 86205(fb) where feedback is turned on between 2.5ms and 4.2 ms. Suppression can be seen during this window.

Thursday August 21 2014 5:26 pm Peng 86211~86258

fbshot tests are done today. And the targeted shot style was obtained and tested. a good example group are: 86229, 86232,86240,86242. where 86232 and 86242 are reference shots. and 86229and 86240 are shots with feedback turned on at 2.5ms and turned off at 4.5ms.

Friday August 22 2014 5:27 pm Peng/Stoafer 86259~86309

Rowgowski coils are tested today. signals can be measured by the oscilloscope. But the noise level is big in the circuit and has to be further reduced to provide reliable measurement.

Biased prob is put in and used to drive a slowed mode. This certainly makes the plasma more stable and led to early disruption events. We are still working on this to obtain a target shot style.

Monday August 25 2014 5:53 pm Peng/Stoafer 86309~86347

The code to control bias probe and feedback control coils together are tested today and working. Plasma becomes very unstable when the bias probe is turned on to slow the mode. feedback might delay the disruption a little bit but is not conclusive.

Tuesday August 26 2014 5:47 pm Peng 86348~86390

feedback shots with bias probe inserted in were tested today. With the bias probe inserted in, plasma behave quite differently. It is find out that with bias probe maintained at -40 volt (floating potential) the plasma could last pretty long sometimes. Most of the shots have a roughly constant MR around 93 cm and q around 2.4 The plasma shows a strong rotating 2,1 mode. When feedback is tried on such modes, no suppression effect can be seen. This is consistent with the previous observation that a tearing mode cannot be suppressed. A kink mode style is targeted later in the day but can hardly be achieved.

Wednesday August 27 2014 4:56 pm Peng 86390~86429

More shot developments are done today but the targeted shot style are hard to obtain.

Friday August 29 2014 3:29 pm Hughes 86430-86452 Shot development for bias probe

Today's run was meant to retrieve the 83918-style equilibrium with bias probe slowed modes. For most of these shots, the bias probe voltages were -2 and +4 at the amplified input (corresponding to about -40 and +90 real Volts at the probe), the negative voltage being the floating potential. For most shots, the bias ramped from 2.5-3.0ms or 1.5-2.0ms later in the series. All shots showed significant variability in R0 traces, and significant disruptivity, with only a few relatively good slowed plasmas with steady R0. Nearly 50% of plasmas with slowed modes disrupted by 5ms, as compared with all plasmas with negative or zero bias lasting longer than 5ms; this is merely an interesting observation, however, as the sample sizes are very small for both sets. With that cavet, since the ferritic wall is inserted, the higher disruptivity at slowed rotations MAY indicate increased mode amplification as rotation drops to 1kHz and slower, as various simulations and tests (Mike's simulation, the freq. rolloff on the mu tests, the amplification of ~1kHz phase flips but relatively unchanged natural mode amplitudes) have suggested. More shots are needed, however, and I would strongly suggest once we're satisfied with the shot programming, we should start swapping between ferritic and stainless walls again, even without RMPs.

Shots of interest:

• 86430: daily vacuum shot
• 86439: shot with a nice R0
• 86447: disruption at 3.5ms, exceptionally clear 50kHz(!!) mode at 4.7-4.9ms. Hard to gauge m number since the plasma is far inboard...
• 86449:rather than slowing, bias goes from -40V to -60V (speeds up to ~10kHz), messing up R0, but despite dropping close to q=2, plasma survives until 7.75ms, longest shot of the day.

One problem of note is that the mode analysis built into the bigplot script (including Chris's version that plots the bias probe traces) doesn't pick up slow modes well at all. At some point, I want to overhaul it to make a smarter polynomial smoothing code that knows to skip the disruption. Might do that early next week, or over the weekend if I can get a stable connection to the lab server.

Tuesday September 2 2014 2:01 pm Hughes 86453-86467 Shot development for bias probe

More shot development with bias probe. Mostly saw plasmas with highly reproducible IP, R0, and q traces but disrupting early (mostly between 3.5ms and 5ms) at bias voltage ~80V. Turned slowing bias voltage down by half, and saw mean disruption time of 5 shots move back from ~4.6ms to 5.2ms, suggesting relationship between bias voltage and disruptivity, but this could be due to the radial current or the bias itself or the rotation or something else entirely.

Handing the run over to Chris at 2pm.

Wednesday September 3 2014 11:05 am Peng 86468~86476

this is a run report for yesterday's run after Paul. A brief test was done on feedback using the shot style developed early in the day. Major radius a roughly constant around 92cm. q drops gradually over time, acrossing 3 around 2.5ms. Bias probe was initially floating at -40V and then quickly ramp up to 40V around 2ms, then maintained at 40V.

suppression can be seen with 3000gain at 270 deg. The gain is higher than previous cases without bias probe (2000 gain at 330 deg). But the current on the control coil is around 20mA which is at the similar magnitude. Then phase shift is also changed from 330 deg to 270deg. The effect is not much at 330 deg.

refshot. 86470,86475. fbshot: 86473, 86474.

Thursday September 4 2014 5:48 pm Byrne/Levesque 86486-86515 Shaping vacuum shots, bias-slowed plasmas

Took vacuum shots to measure pickup from shaping coil on sensors for a variety of shaping currents. Also took some control-coil-only shots with a 3/1 applied field to check for pickup on sensors. Note that all shells are at the same radii as for the previous few weeks (all SS portions of shells are at the same radius, slightly retracted from the SS full insertion location).

Also took plasma shots with the bias probe set near floating or set near ground. These expand on the shots taken earlier this week at higher voltages. Only a few shots were taken (86509-86515) before running out of run time.

Monday September 8 2014 6:33 pm Hughes 86517-86560 Fe wall bias w/ long RMP

Today's run was meant to gather a series of shots with long RMPs and positive (slowing) bias to begin a Fe/SS comparison. For this run, the Fe shells were inserted (back 5mm from maximum insertion) and the SS shells retracted fully. Unfortunately, plasmas were not very cooperative, and most disrupted between 4ms and 5ms, and often even earlier. As a result, the usual 3ms-4ms phase flip wasn't suitable. A new shot style that crosses q=3 at about 2.5ms rather than 3.0ms was developed, although the major radius is considerably less steady in this shot style. The bias probe begins ramping from floating potential (-40V) at 1.5ms and reaches full bias (+80V) at 2.0ms, holding until 8.0ms; the RMP amplitude is set to 2.5 (roughly 25 amps), turning on at 2.5ms, flipping at 3.4ms, and turning off at 2.5ms, giving 900us of soak-through time. Analysis will be performed later, but as a preliminary observation, the disruptivity and general fussiness of the plasmas seemed greater today than on other days.

Noteworthy shots:

• 86517: Daily vacuum shot.
• 86542,86558, 86560: Best strongly slowing-biased plasmas without RMP.
• 86547, 86549, 86552, 86553: Best strongly slowing-biased plasmas with A=2.5 RMP.
Tomorrow, I plan to continue with the Fe-in configuration, and hopefully the plasmas will be nicer.

Tuesday September 9 2014 6:33 pm Hughes 86561-86588 Fe wall w/ bias probe and long RMP

Attempted to continue the campaign from yesterday, but preliminary analysis showed very low response to the RMP in some cases; in other cases, the plasma seemed to lock as though responding to an RMP when none was applied. Some of this may be due to difficulties in getting a good equilibrium subtraction on today's and yesterday's shots, but, regardless, the plasma response seems to be much less clear than in shots prior to the bias probe's insertion with similar control coil fields on the plasma. Plasmas today have been more variable than usual, including a series of shots with radically different equilibrium evolutions in spite of no changes to any settings. A series of shots with very high RMP amplitude (A=4 or around 40-50 amps peak) were applied; some kind of response is visible, but on some shots, it seems as though the locking/response begins to appear before the RMP is applied, so the actual role of the RMP is unclear. Almost all shots with very high RMPs disrupted between 3.47ms and 3.86ms, while the RMP turned on at 2.5, flipped at 3.4, and turned off at 3.5ms; the reason for the seemingly huge disruptivity isn't obvious, since looking at the raw signals, I didn't see any obvious indication of the cause. A number of non-RMP slowed and unslowed shots were taken, as well, to look at natural modes, but these have not been analyzed yet. Qian pointed out that in his feedback experiment, he had a lower bias (80V above floating, vs 120V above floating), and the very high bias may interfere somehow with the RMP. The mechanism for this is not clear.

Shots of interest:

• 86561: Daily vacuum shot.
• 86574: Not good for intended purpose, but q* drops well below 2 before disrupting.
• 86580, 86581: All settings identical, completely different equilibrium evolution.
• 86569: Good shot with slowed rotation and very strong RMP.
• 86564, 86575, 86586, 86587: Usable shots with slowed rotation and no RMP.
• 86577, 86585: Usable shots with bias probe floating and no RMP.

Wednesday September 10 2014 4:44 pm Rivera 86589-86597 rogowski calibration and testing

Shot 86589 with TF only and bank charged to 2.5KV. Rogowski was wrapped one turn around the TR magnet closest to ground. Noise from krytrons created large offsets. 86590 with all but TF Krytons off was better but the TF crowbar Krytron still created an offset. We plugged the amplifier into the UPS and that cured that problem.

• 86592 & 86593 was done at 3.5KV
• 86594 at 4.5KV became obvious that the peak of the rogowski was leading the peak of the TF probe
• 86595 5.5KV
• 86596 6.5KV normal operating voltage.
• 86597 Rog. moved to VF wrapped once around outer coil which is four turns. Signal weak with low frequency oscillation.
• 86598 Rog. wrapped 3 times around outer VF. Still showing low frequency oscillation.
• Gain of amp is currently at 1000. I'll try increasing to 4000 for calibrating, maybe?

Thursday September 11 2014 6:25 pm Hughes 86599-86619 Unbiased (floating probe) Fe Shells

The intention today was to collect some more shots with the bias probe set to the floating potential, with and without RMPs, as these shots have been uncooperative on previous days. Unfortunately, the plasmas were even more recalcitrant than usual, today, along with the rest of the machine. After resolving a series of minor but distracting problems (e.g. the TF shorting stick falling apart when picked up, ion gauge cable to the basement rack coming unplugged, etc), serious plasma operation wasn't achieved until about 2pm; they were generally poorly behaved throughout the day, tending to bifurcate strongly (R0 falling in or out from shot to shot with the same settings), showing little or no change in IP with changes to OH start, etc. This resulted in plasmas with a different R0, IP, and q* evolution than the target--enough to make it difficult to compare sensible with RMP shots.

Shots of note:

• 86600: Daily vacuum shot.
• 86606: The only actual good shot of the day, with floating bias probe and no RMP.
• 86607: An interesting shot. The plasma falls outward to R0=93, and the bias probe can be seen to go from nearly zero current to a small but measurable current (~5A). Since the probe is set to a particular floating potential, which should become more negative as T_e increases, this was a nice sanity check in a day of otherwise disappointing plasma misbehavior.
Hopefully tomorrow will be a day of a more auspicious beginning and more cooperative plasmas. I'm adding to the maintenance list that we should check the VFE and OHE shorting sticks in particular to make sure they're constructed more solidly, since the problem seems to have been a screw that isn't quite long enough for the job and falls out if it loosens slightly.

Friday September 12 2014 6:16 pm Hughes 86620-86650 Fe wall bias w/ long RMP

Today's objective was to get a few shots with strong phase-flip RMP (~40A on at 2.5ms, flip at 3.4ms, off at 3.5ms) for comparison against the slowed case(s). The fussiness of the plasmas persisted, however, resulting in only two good non-RMP shots (although one, shot 86627, did have a very nice major radius) and a single shot with RMP that simultaneously matched the IP, R0, and (therefore, obviously) q* traces nicely. A few other shots were of interest because there was a visible, generally negative probe bias at the end of the shot; this indicates a floating potential nearer to 0V than the -40V set point, suggesting a lower electron temperature. What's interesting is that if the R0 measurement and probe insertion position are both trustworthy, this low-Te plasma seems to be centimeters outside the limiter-defined plasma edge, suggesting a much thicker and more conducting scrape-off layer than we usually discuss. Some more deliberately targeted studies in this direction might yield interesting results, possibly even something worth mentioning in the proposal.

Shots of interest:

• 86620: Daily vacuum shot.
• 86623, 86627: No RMP, floating bias shots. 86627 had a particularly nice R0 for this shot series.
• 86648: High amplitude (A = 4 or 40-50 Amps) RMP with bias probe at floating potential.
The run ended at about 4:30 since everyone else had left. Analysis will follow at some point, but the code I used to do the response analysis previously has developed a somewhat mysterious fault (something to do with timing, since it works fine with older 3.0-4.0 shots, but on the recent 2.5-3.5 shots, it confuses itself with empty data vectors) so I'll have to fix that before I can say anything even preliminary about results. Also, one high-RMP floating shot does not statistics make, and we observed considerable spread in the response, so results may not be very meaningful. However, it is interesting that 86648 lasted until nearly 6ms, in contrast to most of the high-amplitude, slowing-bias shots dying before 4ms. What this means, though, really won't be clear (or, at least, won't be clear to me) without a lot more analysis, and probably some run time dedicated to understanding exactly how the bias probe actually affects the plasma.

Monday September 15 2014 6:50 pm Hughes 86651-86698 Fe wall w/ bias probe and long RMP

The mission of today's run was to gather more data on shots with a floating (-40V) or moderately-slowing (V_float+80V=+40V) bias probe. The first half of the day was not very fruitful, producing no really useful shots until 86673, but most of the objectives of the run were accomplished. This was mainly achieved by relaxing the standards of IP, R0, and q* matching. Even so, one or two actually good shots were taken. Looking at the raw sensors, some shots do exhibit fairly strong RMP responses on the order of 1-2G which don't seem to show up on non-RMP shots, or at least not on all RMP shots, but no real analysis has been done, just some eyeballing. If there was a difference in response between unslowed and moderately-slowed plasmas, the difference is more subtle than I could pick out this way.

Shots of interest:

• 86652: Daily vacuum shot
• 86651: First attempt at vacuum shot, but OHB heat lamp was off. It should be noted that the resulting OH ringing showed up somewhat on the SXR midplane and was seen very clearly on the bias probe voltage.
• 86673,86674: Floating probe, high-amplitude RMP shots. 86674 may be the best shot of the day, at least for R0.
• 86677,86690,86692,86693: Moderately-slowed probe (80V), non-RMP shots. 86677 is best of the set.
• 86686,86695,86696,86698: Moderately-slowed probe (80V), high-amplitude RMP shots. 86686 is okay; 86698 is perhaps the best or second-best shot of the day.

There are still no bias probe shots in the SS-in/Fe-out configuration, so I expect this will be the next step, unless there are other shots we need in this configuration or the ones from today prove unsuitable.

Tuesday September 16 2014 4:11 pm Rivera 86700-86703 TF Rogowski Calibration

Took the new TF rogowski off of the TF and wrapped it around the outer VF coil and fired only the VF electrolytic bank. On shots prior to 86700, it is wrapped around a TF magnet.

For shot 86700, it was wrapped only one time around. For subsequent shots, it was wrapped three times.

Pat's analysis code shows that the TF field from shot 86697 taken yesterday, for example, is 17 to 18 percent higher than the TF probe shows. This is close to the results Paul had in 2012.

Pat's code is VF_cal.py It's in my home directory and his.

Wednesday September 17 2014 6:20 pm Hughes 86705-86744 SS wall w/ bias probe and long RMP

At the beginning of the day, the shell configuration was reversed to stainless-in, ferritic-out, so all ferritic shells are at their maximum safe retraction (inserted 5mm from maximum physically possibly retraction) and stainless shells are retracted 13mm from their maximum insertion. It took some shot development to get good plasmas, but shots with floating bias probe were much, much nicer and more reproducible, today, than in the ferritic configuration. However, the strongly positive-biased (80V) shots tended to disrupt early (before 4ms) and have wild jag in R0, so I don't have any really good slowed shots, yet. It may need some more shot development, or it might have just been bad luck at the end of the run day.

Noteworthy shots:

• 86705: Daily vacuum shot.
• 86727, 86728, 86734: Floating bias probe, no RMP.
• 86730, 86732, 86733: Floating bias probe, 40-amp RMP.
Tomorrow will feature shot development to get some good slowed shots, and then perhaps a return to the floating potential shots to get more data, depending on how the day goes.

Thursday September 18 2014 6:52 pm Hughes 86745-86783 SS wall w/ bias probe

Today's run targeted getting more slowed and unslowed plasma shots with the stainless walls inserted (13mm back from full insertion) and the ferritic walls retracted (5mm in from full retraction). Plasmas remained frustratingly disruptive, and continued to exhibit a confusing lack of RMP response on slowed modes. Since the purpose of this run campaign is ferritic wall research, however, it was decided to set the RMPs aside in light of their unclear results, and acquire data with a strongly positively biased (V_float+120V) and floating probe. Although the bias+RMP campaign promises to be interesting, it would probably be more productive if it follows a more in-depth study of the bias probe's influence on plasma performance, as there are a number of features we simply don't understand even about our biased plasmas without RMPs, and leaping straight to bias + RMP + Fe wall seems to be getting ahead of ourselves. Among the peculiarities observed, there are a number of cases in which the bias probe current drops from, say, 40A to 20A for 0.5-1.0ms before returning to near its prior value; no other standard diagnostic (IP, R0, D_Alpha, either SXR, or loop voltage) shows any sort of matching disturbance that I've been able to see. There is some evidence that this dip might correlate to a locked mode, but as Chris pointed out, the polynomial equlibrium subtraction which indicated this can be misleading. Also, shots with RMP turned on have been observed to have rises in bias probe current at the time of the RMP turning on, where non-RMP shots remain steady or even drop in probe current; however, these same RMP shots do not necessarily show any clear response on the magnetic sensors, or correlations in other equilibrium parameters (e.g. R0, IP, etc).

Shots of interest:

• 86745: Daily vacuum shot
• 86746-86747: Two calibration shots for Nick's Rogowski
• 86758: Only decent RMP shot of the day. Bias probe moderately positive (40V=V_float+80V).
• 86763, 86765, 86767, 86769: Bias probe moderately positive (40V=V_float+80V). No RMP.
• 86773, 86775, 86777, 86779, 86780: Bias probe strongly positive (80V=V_float+120V). No RMP.
• 86781, 86583: Bias probe set to floating potential (-40V).
Although the major radius doesn't fit perfectly between the SS and Fe series of shots (the Fe shots dip inward more), they will hopefully be close enough for comparison of mode amplitudes, although this may require some amount of cleverness. It may be possible to normalize against the signal picked up by the outboard-most PA sensors on the two arrays, since PA1 seems to be unaffected by the ferritic material on the section 3 shells.

Monday September 22 2014 7:09 pm Byrne 86784-86853 Shaping shot development

Run was focused on creating two equilibria to increase our database of shots for studies on natural modes in a shaped plasma.

Some delays due to changes in the machine since the last run. Bias probe was inserted, which was limiting the plasma and causing early disruptions (early in the sense of q*, disruptions characteristic of q* ~2 were occurring around 2.5)

It was assumed that the ferritic shells were the cause, and they were fully retracted. When they were being re-inserted it was noticed that even at the extra 5mm retraction, the q* disruptions were happening at higher q* than with them out. Fe shells were retracted one chamber at a time. Chamber 5 seems to have the furthest-in ferritics, as q* at disruptions changed only on that chamber.

Ip at breakdown was low, and this was found to be a function of the gas puff. Settings were the same as the reference shot, but fill pressure was ~85% of what it was. Increasing Gas Puff timing from ~2ms to 2.75ms fixed the problem.

Shots showed the characteristic q=3 'hitching' behavior all day. Several rapid shots were taken to see the effect of base pressure. As base pressure rose, the hitching stopped, but a low base pressure shot afterwards had the same behavior.

As the q=3 hitching subsided, q* would drop through to q* ~2 and disrupt. rest of the day was spent dialing the settings in, increasing VFE and taking several shots until the hitching subsided.

Shaped shots were taken, but showed the same behavior, at higher values of q* (as read from the tree). TokaMac simulations of the shots showed a small mr plasma, which is consistent with a lower q* than what is in the tree. Simulations also showed the plasma to be diverted, with an X-point poloidal limiter distance of ~5mm on average.

Work will continue tomorrow.

Pat Byrne

Tuesday September 23 2014 6:33 pm Byrne 86854-86925 Shot development

Shots displayed the usual bifurcation. 86922 is a good example of 'q =3' hitching, and 86923 is a good example of a 'q=2' crash. These shots have the same settings.

A few shots bordered on acceptable, for instance 86892,90,86,85, but were tinkered with in an attempt to match the reference shots from the last run campaign, 85402. These shots had a reasonable q, but at the cost of rapid IP-ramp and MR fall-in. It was desired to have slowly developing MR to help with analysis and reduction of eddy current effects.

Only a few shaped shots have been taken, as unshaped equilibria are usually less of a challenge. 86884 is one such shot, with a similar equilibrium to the unshaped ones, and diverted.

Any advice will be welcome at the meeting. End of today was a return to the settings of 85402, and an exploration of the settings nearby in parameter space. A shot without the two modes of disruption was not found.

Unable to say whether good shots will be able to be gathered by end of day Wednesday. This has been characteristic of shaping campaigns, but disappointing all the same

Byrne

Wednesday September 24 2014 6:04 pm Byrne 86926-86986 Shot Development

Shots still highly variable, and difficult to match the reference shots.

Several different configurations of VF and OH banks were used, but IP at medium times (2-4ms) was generally unresponsive. A few shots were taken with q going below, and remaining below 3. These shots had little early mode activity. It should be possible to investigate them, but they bear little resemblance to the reference shot. For those interested, those shots are 86955, 86980, and 86966. 86935 has similar q evolution to the reference shot, but very different Ip and MR.

End of day involved playing with gas puff settings. Settings for this campaign had to be significantly higher to get the same fill pressures as the reference shot. Intention was to see what effect increasing the gas puff from the bare minimum would have. A puff time of 1.75ms was necessary for a breakdown at 58uTorr. 1.5ms, which provided a fill of 54uTorr, did not break down. Only one shot was taken after this, with 1.85ms, and 62uTorr, but it suggests that the evolution of the plasma (especially initial conditions) are highly sensitive to gas puff in this area. Tomorrow will be spent quantifying this effect, with an eye towards returning to the reference equilibrium.

Shots similar to the star shots of this run will also be taken. In the event that the reference equilibrium is unreproduceable in this campaign, it will at least give us some data to look at.

Byrne

Thursday September 25 2014 8:14 pm Byrne 86987-87051 Shaping/Shot Development

First real progress of the campaign.

Remain unable to replicate the Ip trace of the reference shot (85402) but have developed an unshaped equilibrium that is slightly further out, keeping the q* trace roughly the same, and falls in slower, making BD analysis valid over longer subsections of the shot. Star shots of this equilibrium are 87000,01,04,19,&20.

Once a repeatable equilibrium had been found, work began on replicating it as faithfully as possible in the presence of shaping. Have not developed such a shot yet, but shots 87047,48&51 are probably useable, and development continues.

Will focus on increasing the lifetime of the plasmas (ideas for morning - more OHST to reduce early modes near q* =3 & more OHE/less VFE to reduce the rate at which the plasma falls in. Probably once an hour, I will undertake to get a few more unshaped shots. Original plan was to get ~10 more of each type of equilibrium to improve the statistics on our plasmas.

As mentioned in my email to the group, this morning I was unable to get breakdown. I had ended the day Wednesday by dropping GP duration below the breakdown threshold, and then raised it. The lowest pressure at which plasma broke down was 58uTorr. The last shot of Wednesday was 62uTorr. The first shot of Thursday was also 62uTorr in fill pressure, but did not break down. This happened twice in a row, at which point the puff duration was increased radically (from 1.85ms to 3ms) to check that the system was not broken. Experimenting on the Gas Puff and HBT-EP's breakdown physics was not the purpose of this campaign, so nothing more was done. However, it does seem as though the history of the chamber can affect the plasma in non-obvious, but significant ways. This may be due to background pressure (less than 1% of the fill pressure) playing a larger role, or possibly more (or less) impurities in the chamber at the beginning of the day.

Pat

Friday September 26 2014 5:23 pm Byrne 87052-87104 Shaping

equilibria were highly variable. However a database of about 10 shaped shots have been taken. Though they crash quickly, and have fast fluctuations in MR, there should be enough to draw some inferences from their bulk behavior. All star shots ar outboard-limited, with q* dropping below 3 at early times, and rising to above 3 before disruption.

Reconstructions will need to be undertaken to determine exactly what the true q* of the shaped shots was, but it seems a safe bet that these shots, with a strong IP ramp, falling MR, and thus rising q*, should be susceptible to external kink modes. At least according to the theory we use for our circular plasmas

Star shots are:87075,76,77,82,84,87,89,91,92,94,95. No analysis has been done as yet.

Gas puff fill pressure was much higher at beginning of day today than end of day yesterday. Battery was NOT charged overnight. Pressures were in the 100uTorr range, while 77uTorr was the fill pressure the day before. Reducing the pressure by reducing puff time had no noticeable effect on the plasmas.

At one point near end of day, a breaker near the south door (door 1) tripped, gating off the cryo. This was eventually noticed by Jim, but had the effect of raising the base pressure from 80nTorr to around 120nTorr when the problem was discovered. Again, no noticeable effects either from the elevated base pressure, or the lower base pressure once power was restored.

Pat

Monday September 29 2014 11:42 am Byrne 87105-87107 Shaping

Intention was to grow the shaped-shot natural-mode m=3 kink database.

During vacuum shots, the TF charging water resistor was seen giving off a plume of steam. HV was shut down and the resistor inspected. The electrode had completely eroded away. Work is being done currently (noon on Monday, 9/29) on a replacement.

Banks were inspected and cleaned. Very little water had splashed on the TF caps, fortunately. Most was on the resistor itself, or the floor.

Pat Byrne

Tuesday September 30 2014 5:45 pm Byrne 87108-87168 Shaping

Shots became good around 3:30pm. 6 shaped star shots gathered, defined as Ip, and MR similar to the unshaped reference 87020.

This campaign was marked by plasmas significantly less repeatable than the previous one that straddled the 4th of July. Even after cleanup, the same settings gave very different plasmas from those seen in the previous campaign. During rundays, equilibria were seen to be highly variable, and required bank tune-ups every 10-15 shots to ensure useable shots were taken. Even still, perhaps 1 in 3 shots was useable, shaped or un-.

MR tended to fluctuate quickly (but with small amplitude) during shots, but generally fell in slower than the previous campaign, which was one of the main goals

Nevertheless, many useable shots have been gathered. This should be plenty to come to conclusions about the natural mode content of shaped plasmas vs unshaped.

Star shots are:

• Unshaped: 86873, 86890, 86892, 86941, 86955, 86955, 86956, 86960, 86966, 86980, 87001, 87004, 87012, 87013, 87014, 87015, 87017, 87018, 87019, 87020.
• Shaped: 87047, 87048, 87051, 87075, 87076, 87077, 87082, 87084, 87087, 87088, 87089, 87091, 87092, 87094, 87095, 87130, 87146, 87147, 87156, 87157, 87159, 87161, 87165
Byrne

Thursday October 2 2014 5:54 pm Peng 87210~87259 Plasma shot

Shots development continued today. The plasma major radius is more stable compared to yesterday. But the q value varies quite a bit form shot to shot and the mode amplitude is very sensitive to the q value. It can be seen feedback seems to have effect with q around 2.8, but reference shots are not able to be obtained. A better target shot style shot be tried to get more reliable parameter traces.

Wednesday October 8 2014 5:55 pm peng 87260~87305

Shot development is tried today. All stainless shells are in, All ferritic shells are out.

The mode activity is very small in the targeted shot style, the examples can be seen 87301~87305 where the major radius is already below 92cm. The mode amplitudes are also small when the major radius is above 92cm.

The mode becomes bigger when q drops below 2.6

the 2-1 mode becomes dominate when q drops below 2.2 but does not leads to disruption. The q calculation may not be very accurate with this configuration.

Friday October 10 2014 5:46 pm Stoafer 87308-87325 Mach probe setup and test

Today's run was spent setting up and testing the Mach probe (five-tipped probe). The electronics have been setup to have each of the outer tips measure ion saturation current, completing the circuit through the center tip. Current on each of the four outer tips is measured with Pearson rogowskis, so the entire circuit is floating. A 150V battery was used to bias the tips into the ion saturation region. This voltage is relative to the center tip. A capacitor is used in parallel to the battery supply enough charge through the shot to mitigate droop.

We were able to measure current for each tip; the rogowskis are currently hooked up to a scope. The voltage reading is a little over 50mV, which corresponds to about 25mA, which is in the range that I expected, but is a little low. The wire from the probe tips is each wrapped around a corresponding rogowski 20 times to increase the signal. We hope to increase the signal further by either wrapping the wire around more times (and confirming that there is no inductive effect in the regime we are measuring) or using an amplifier before sending the signal to the digitzers.

We also saw that the tips would arc to the center tip. At least one would arc in every shot. One shot was taken with a reduced voltage of 45V to try to prevent arcing. One of the probe tips did seem to arc, but the signal did not change as severely as the case with the higher voltage. This will be further studied in future runs. It is possible that the probe design might need to change by putting an insulating barrier between each of the outer probe tips and the center tip in order to prevent arcing.

Further setup of the system include setting up the rogowski signals to be digitized and using voltage dividers to measure the voltage on the probe tips.

Wednesday October 15 2014 4:40 pm Stoafer 87326-87343 Mach probe tests

Tests continued today to troubleshoot problems with the 5-tip probe. The voltage was set to 45V for the run and the shells at the Mach probe were retracted. Only one shot showed a large negative current for one of the probe tips, so the rate of the occurrence has decreased, but we do not know the cause of it yet. Tests will continue tomorrow.

Tuesday December 2 2014 5:27 pm Stoafer 87357-87403 Clean up / Mach probe

In today's run, clean up shots were taken and troubleshooting the Mach probe continued. Shots were almost back to normal by the end of the day. There was no evidence of arcing with the Mach probe. The bias voltage for the probe was 140 V. The shells at the Mach probe were retracted at the beginning of the run, but were inserted in the middle of the day to test if the shell position could lead to the arcing behavior. Tests will continue tomorrow with changing the radial position of the probe. We will also try to get the probe signal digitized, but we worry about spikes in measurement voltage due to the arcs we saw previously.

Thursday December 4 2014 6:13 pm Stoafer 87404-87440 Mach probe setup

Now that the signals from the Mach probe tips are being digitized, we began testing different configurations with the probe. We started with the probe at 103cm, then on shot 87419 we moved the probe out to 104cm. We also tested various orientations of the, as described below.

The signal from the Mach probe seem reasonable and usable. I use probe tips labeled C and D. When the rotatable flange of the probe is at 0 deg, tip C is CCW and tip D is CW when looking down on HBT. At 90 deg tip C is up and D is down in the z-direction.

sensor nodes in the tree have been created starting on shot 87414 at .sensors.mach_probe:tipC. The signals are raw voltages from the Pearsons, divide by 2 to get current. A ratio of the signals is related to rotation.

Shots were taken with different orientations of the Mach probe:

• 0 deg: 87414-87426 and 87436-87440
• 180 deg: 87427-87432
• 90 deg: 87433-87435
Run will continue tomorrow with applied torques from bias probe.

Friday December 5 2014 6:09 pm Stoafer 87441-87478 Rotation studies

The goal of today's run was to control rotation with the bias probe and measure the effects with the Mach probe. This was done successfully with promising qualitative, and perhaps quantitative results with more analysis, results. The Mach probe measured an increase in plasma rotation when the bias probe was biased positively. This was determined by taking the logarithm of the ratio of two Mach probe tips as a value the corresponds to plasma rotation. More details analysis of these signals in required to associate a measurement of ion flow or plasma rotation frequency.

Below are a list of notable shots:<

• -2V bias, 0 deg: 87451, 87452
• 0V bias, 0 deg: 87453
• -5V bias, 0 deg: 87454, 55, 56, 57, 58
• 2V bias, 0 deg: 87459
• 5V bias, 0 deg: 87460-69
• 5V bias, 180 deg: 87470, 1
• 5V bias, 90 deg: 87472, 4
• 5V bias, 270 deg: 87475, 6
• 5V bias, Mach probe behind shells: 87477, 8
The west cryo needs to be regenned, which we will do next week in conjunction with further rotation study experiments.

Monday December 8 2014 5:47 pm Stoafer 87479-87515 Rotation studies

Today's run was focused on shot development with a target shot of 85721. In the run, the stainless shells were fully inserted and the ferritic shells were retracted. Tomorrow, shot development will continue with the same target shot, but with all shells inserted. Stainless 1.5cm back from nominal; ferritic 0.5cm back from nominal.

Tuesday December 9 2014 7:10 pm Stoafer 87516-87558 Rotation studies

The goal of the run was shot development with all the shells inserted. I aimed for a shot where q starts below 3 and rises with a target shot similar to 86061, however, I hope to have more current ramp. Shots 87549 and 87558 have good starting conditions, but need a smaller current ramp so that q rises back up to 3. Work on this will continue tomorrow. Once a shot style is developed, scans of different bias probe voltages will be performed while measuring with the Mach probe.

Thursday December 11 2014 7:07 pm Stoafer 87560-87597 Mach probe sweeps

For today's run, one of the Mach probe tips was connected to the setup used to bias the bias probe. Therefore, we could apply arbitrary bias functions to the probe tip and read the current and voltage. The was done successfully with the Mach probe at 104cm. Attempts were made to take these measurements at other radii (105 and 106cm), but towards the end of the day the plasmas behaved differently. This will be continued tomorrow.

Star shots:

• Probe at 104cm: 87586, 84, 83, 81, 80
• Probe at 105cm: 87592, 91 90
• Probe at 106cm: 87595
The IV characteristics looked reasonable and as expected, but with more variation than the bias probe. A report with this information will be drafted.

Friday December 12 2014 6:15 pm Stoafer 87598-87599 Mach probe sweeps

Today the Crown amp used for applying a voltage to the bias probe (today hooked up to the Mach probe) overheated and started smoking. I noticed after the second shot of the day and turned it off as soon as I noticed. The rest of the day was spent troubleshooting what happened and preventing the issue in the future. Here is what we know:

When I noticed the amp was smoking, the lights that indicate the amp is amplifying a signal were on and it seemed like it was constantly outputting a signal.

On the first shot of the, 87598, I programmed for a triangle wave, but starting with a voltage of -80 (input voltage of -3) from 0ms until 2.5ms, at which time the triangle wave should have been applied. The input seemed to stuck at -3V. Also, the program used to send the input to the Crown amp froze for this shot. We have seen this behavior before, mostly for the first shot of the day, but it never caused an issue before because the system is designed to decay to an input signal of 0V. Qian mitigated the problem with the software freezing by conditionally checking if the program is not receiving signal information, where it might be getting stuck. He also added a termination function to set the voltage into the Crown amp to be 0V if anything goes wrong with the program.

There is also a large signal in the voltage measured on the probe around 20ms. This can be checked with the tree node .sensors.bias_probe:voltage. This voltage is measured on the secondary windings of a transformer, on which the Crown amp biases. This is a 2:1 step down transformer.

For preventative measures, the software changes made by Qian should help if the fault occurred because erroneous inputs to the Crown amp were generated. I will also write up a procedure for using the AWG controller when starting up to prevent this behavior on any of the Crown amps. Nick also found a circuit we can use as a gate to set a certain window that allows signals to be sent to the input of the bias probe Crown amp that we can trigger with a Jorway. This will be a good hardware addition to prevent this from happening if there were signals being sent to the amp after the shot was taken.

For now, we plan to run without the amplifier and focus on shot development. It is possible that we can fix the amp; I found two wire in which the insulation was melted and looked like it could have been the smoke. No other parts are obviously damaged, but that is still possible. We plan to replace these wires and test the Crown amp on the bench. In the meantime, based on a suggestion by Jeff, I re-wired the control coil input and output for CC 6B, which shared an amplifier with CC6T, which is shorted to the chamber and not usable. The wires have been routed to the amplifier that was used by CC 1T, which has the same problem. Since the inputs and outputs are still associated with each other no software mapping has to be done for this change. I also made notes about the changes and taped them to the control coil rack, in addition I added tape to label the changes.

We will not run with the Crown amp on the bias probe or Mach probe until sufficient preventative measures have been implemented for this problem.

Monday December 15 2014 7:14 pm Stoafer 87602-87644 Mach probe sweeps

Today, I continued shot development and Mach probe sweeps with the Mach probe at different radial locations. The run was relatively successful with good shots with the probe at 104, 105, and 106cm sweeping the voltage on the center tip. The run will continue tomorrow with more locations.

Below are star shots:

• 104cm: 87640
• 105cm: 87632
• 106cm: 87634
An older generation Crown amp was used to amplify the voltage applied to the probe tip, which worked well, given the setup does not require a lot of current or precise frequency response.

A few safety checks were put in place to prevent damage to the amplifier. A fuse with a smaller current limit was put in the output circuit of the amplifier; similar to those for the control coils. Qian implemented some software checks to prevent the input into the Crown amp from getting stuck at a voltage when the script is improperly terminated. We also don't turn on any Crown amp until a crates only shot is taken in the beginning of the day and the do_awg script runs properly on caliban. We ran with no problems today.

Tuesday December 16 2014 6:59 pm Stoafer 87645-87693 Mach probe sweeps

The run was continued today to measure voltage sweeps of one of the Mach probe tips as a Langmuir probe. The plasmas were not as reproducible today, so only a couple of shots were used. We now have a course spatial measurement (102 - 106cm in 1cm increments) of probe positions. The campaign will continue with finer spatial increments at the edge.

Star shots:

• probe at 103cm: 87671
• probe at 102cm: 87685
• probe at 105.2cm: 87688

Wednesday December 17 2014 6:24 pm Byrne 87694-87729 Shaping Vacuum Pickup/Vertical displacement

Took 15 shaping-only vacuum shots at 3 different current levels to determine a good response function subtraction, and the dependence of the response function on bank settings.

Second part of run, determining plasma motion during disruption in presence and absence of shaping has been hindered by the failure to digitize a sin_theta signal.

The integrator box was found this morning at the end of a BNC cable at the bottom of the west rack. This rack had flooded in the past, and the integrator was in the water. There was significant corrosion, so the integrator was rebuilt and replaced.

As there are only 3 AM502 amplifiers in the west rack, the cos_3 rogowski was disconnected, and the amp settings were changed to be the same as the cos_1 rogowski, for lack of any guidance as to the correct settings. During plasma shots, however, the signal in the sin_1 CPCI channel - that is the amplified, partially integrated output of the rogowski, was very near zero, with no clear signal, and the properly processed signal showed no motion, up or down, with respect to shaping.

The Amplifiers for the cos_1 and sin_1 were switched, and the CPCI digitized signal of both was seen to change. Since it was the same signal, going through the same integrator, going through an amp with the same settings, this suggested a problem with either the amp or the digitizer channel

The sin_1 rogowski was then digitzed into the sin_3 rogowski's channel, and the signal was seen to be the same. It suggests a problem with the sin_3 AM502. At this point, determining plasma vertical position with the sin_1 rogowski was abandoned and the AM502 had the sin_3 connections and amp settings restored

15 plasma shots were taken, 5 unshaped, and 5 apiece with bank settings at 500/125 and 700/175 SHST/SHCB. The PA sensors were on and digitized, and they will be used to determine the vertical position of the plasma as it disrupts in the presence or absence of shaping.

Byrne

Thursday December 18 2014 5:58 pm Stoafer 87735-87779 Mach probe sweeps

The voltage sweeps of a Mach probe tip continued today, focusing on getting good spatial resolution at the edge.

Star shots are as follows:

• 106.0cm: 87751
• 106.2cm: 87754, 64
• 106.4cm: 87766
• 106.6cm: 87776
• 106.8cm: 87777
Run will continue tomorrow.

Friday December 19 2014 4:51 pm Stoafer 87781-87827 Mach probe sweeps

The run was completed today for looking at probe voltage sweeps at different edge radial locations.

Star shots are listed below:

• 107.0cm: 87786
• 106.0cm: 87790
• 105.8cm: 87792
• 105.6cm: 87796
• 105.4cm: 87807
• 105.2cm: 87810
• 105.0cm: 87811
• 104.5cm: 87812
• 104.0cm: 87822

Monday January 5 2015 5:15 pm Stoafer 87838-87877 Rotation studies

Today's run was spent reconfiguring the five-tipped probe to act as a Mach probe and triple probe by measuring probe tip currents and voltages. Once all the signals were being properly digitized I began shot development. This will continue with the run tomorrow.

Tuesday January 6 2015 6:08 pm Peng_and_Stoafer 87878~87917 plasma shot

Shot development is done today. The shotstyle with q approaching from below is very difficult to obtain and given up in the late afternoon. New target shot has a flat major radius and q goes across 3 from above. The reproducibility is much higher and an increase in mode amplitude is observed when q goes below 3. The star shot is 87914. Plasma is slowed with bias probe for the last few shots and increase in the mode amplitude is seen.

Wednesday January 7 2015 6:26 pm Stoafer/Peng 87920-87952 Rotation studies

In today's run, the bias probe was used to torque the plasma while measuring the response with the Mach probe and magnetic sensors. The shot style with a constant major radius and decreasing through q=3 slowly was used.

We see an affect on the measured Mach number at different bias voltages, but it is difficult to see a transition in the measurements when applying a step in bias voltage change. This will be investigated further. Tomorrow the run will continue with a radial scan using the Mach probe.

Star shots include:

The following have the Mach probe at 104.0 cm

Floating bias probe: 87921, 87929, 87948

Bias probe square wave -3V -> 0V: 87922-27

Bias probe voltage -2 -> 0V step: 87936

Bias probe voltage -2 -> 3V step: 87940

Bias probe voltage -2 -> 5V step: 87942-44

The following have the Mach probe at 105.0 cm

Floating bias probe: 87948

Bias probe voltage -2 -> 3V step: 87946

Friday January 9 2015 10:43 am Stoafer 87953-87968 Probe troubleshooting

There was an issue sending out the run report yesterday, so I am sending it again.

Today was spent troubleshooting the five-tipped probe. Voltage dividers had been setup to read the voltage on the center tip, as well as the two probe tips that are floating.

With the probe at 104.0cm (where the center probe tip is at 104.0cm and the side tips are at approx. 104.3cm) the center probe measured around -30V (this probe tip was also biased to collect electron current to source the ion saturation collection probe tips). Probe tip A (below the center tip) measured +10V and probe tip E (above the center tip) measured -20V.

When the probe was rotated by 180 degrees, the center probe tip measured -20V, probe tip A (now above the center tip) measured -20V, and probe tip E (now below the center tip) measured +15V.

All voltages measured relative to chamber ground using voltage dividers.

I got similar results when I disconnected the battery used for collecting ion saturation on other probe tips.

This indicates relatively large electric fields in the plasma, which doesn't seem physical, or the probe is behaving in an unexpected manner. We are still trying to figure out what is going on.

Further troubleshooting includes taking shots at more angles and setting up voltage dividers for the other probe tips.

Tests were taken with the bias probe at 103cm, but floating (not connected to amplifier or transformer).

We saw no significant voltage with a vacuum shot or when the probe was pulled back outside of the plasma.

My current theory is that the measured voltages are being affected by both the ion and electron fluid flows, changing the measured voltage for upstream and downstream probe .

Friday January 9 2015 5:21 pm Peng/Stoafer 87969~88000 plasma shot

feedback with slowed plasma is tested today. Bias probe is set to 40V to slow the plasma down to around 5kHz. Plasma is more unstable when slowed by the bias probe. The feedback suppression can be observed.

The star shots are:

ref: 87974 87975 suppression with 285degree: 87990, 87991

The five tipped probe was instrumented to measure floating potential on all five probe tips using voltage dividers. An angle scan of the probe was performed at 103.0 cm and 105.0cm with consistent results.

Monday January 12 2015 5:15 pm Stoafer 88002-88036 Floating five tip probe

This was used to test the five tip probe with each probe tip separately reading the floating voltage. Shots were taken with different bias voltages on the bias probe and a rotation scan of the probe orientations was begun. The scan will continue tomorrow.

Tuesday January 13 2015 5:58 pm Stoafer/Peng 88037~88072 plasma shot

A scan of radial positions was performed with the five-tipped probe while all probe tips were floating.

later in the afternoon, feedback tests are performed with bias probe at voltage 0V ( floating potential is -40V). Only a few shots are taken and more data on the phase scan is needed.

Thursday January 15 2015 6:07 pm Peng 88073~88103 plasma shot

feedback test is taken today to with bias probe voltage setting at 0V (floating potential is -40V). Plasma is slowed and feedback suppression is observed. good comparison: 88102(ref) 88091(fb with 3000 gain at 300 degree)

Friday January 16 2015 4:17 pm Stoafer 88104-88130 Rotation studies

The five-tipped probe was connected today as a Mach probe and triple probe, using both of the probe tips above and below the center tip to measure floating potential. These measurements will be used to determine the floating potential at the center tip for triple probe measurements.

Shots were taken with stepping the bias probe up from floating potential to measure changes in mode rotation and plasma rotation. Shots 88120, 88121, and 88122 show this really well. Continued run plans are to collect shots similar to these for comparisons across many shots.

Tuesday January 20 2015 5:28 pm Byrne 88131-88155 Shaping Hardware

Today was spent investigating the noise measured on the Shaping coil Rogowski. Since we are attempting to use Fourier Transforms to remove pickup, the presence of high-level, wideband noise, is causing significant problems in the analysis.

Findings were that the amplifier array in the basement rack has seen its noise level increase in recent months to ~double its previous level, to 4mV rms, and also that the so-called 'krytron noise' that occurs on most diagnostics when a bank fires is related to the j221 firing, not the krytron or ignitrons. Details below.

A series of crates-only shots were taken. Shots were taken with the shaping rogowski on different channels of the amp used to amplify OH current rogowski and TF probe signals.

Shots were also taken on those channels with a 50 Ohm terminator in place of the rogowski.

These zero-signal shots were compared to shot 85402, which had no shaping current. The RMS noise level was less than half that of the current shots (1.6mV vs 4.0mV), as well as those of the Response function run before the Winter Break.

This noise was much lower when the output of the amp was replaced with a 50 Ohm terminator, essentially jumpering the digitizer bank inputs with a resistor. The noise level was basically the same as the older shot, 85402.

This all suggests that the amp in the basement rack has been outputting significantly more noise in the last few months, from a level low enough that it was indistinguishable from the digitzer's bitnoise

Once this was done, the next step was to turn off the krytrons, to see the effect of the noise on the fourier transform of the shaping current. It was noticed that turning off the krytrons did NOT remove the noise spikes. Disabling the J221s that fired those krytrons in the tree, however, did! I am not sure if anything can be done about this, but it definitely seems easier to isolate the pickup now that we know where it is coming from.

Pat

Wednesday January 21 2015 7:02 pm Stoafer 88156-88189 Mach probe compass scan

Today was the beginning of a campaign to better understand the characteristics of the five-tipped probe. The probe was instrumented with 3 tips acting as a floating Mach probe (the center tip and two side tips), while reading the voltage on the other two tips. Shots were taken at different probe angles ranging from 0 to 360 in increments of 30. The scan will continue tomorrow to complete this course scan and also look in increments of 10 from 0 to 90.

Thursday January 22 2015 7:24 pm Stoafer 88190-88228 Mach probe compass scan

The compass scan of the Mach probe continued today. There was interesting behavior when the probe was rotated to between 220 deg and 330 deg, where one probe tip had twice as much current as the other. Because of this behavior, a fine scan was performed around those rotation angles.

I was unable to start the scan with all floating tips, so that run will come later.

Also note that Jeff and I added two columns to the SQL database with bias probe position and Mach probe position, both given in centimeters.

Wednesday January 28 2015 6:12 pm Peng 88252~88285 plasma shot

Feedback test with ferritic wall in stainless wall out continued today and bias probe is out. It took a while to get a good development and the major radius is still not completely stable. a few feedback shots was done later in the day with some parameters. Gain is increased from 3000 to 5000 to obtain some feedback effect. More tests will be needed tomorrow to get detailed information about the feedback performance with this wall configuration.

Thursday January 29 2015 6:35 pm Peng 88286~88323 plasma shot

More feedback tests were done today.

The water coolant for the GPU card is found to be low again today. It causes inaccuracy in the computation errors in driving the control coil. The tank is refilled and performance is restored. It was refilled around a year ago last time. Details about the refilling procedure can be found in the wiki: nvidia_qpu

Feedback tests resume later in the day and a rough scan with gain 5000 was performed. feedback suppression can be observed. A few shot with gain 4000 is test in the end and feedback effect is not very strong. More tests will be done tomorrow.

Friday January 30 2015 5:50 pm Peng 88324~88358 feedback test

More feedback test is done today with gain at 5000,4000,and 3000. It is found that with ferritic walls in and stainless walls out, the feedback suppression need a higher gain( above 4000) to have sensible effect. The overall suppression is not as good as all walls in configuration.

Tuesday February 3 2015 7:26 pm Stoafer 88359-88397 Triple probe scans

The run today used the five-tipped probe connected as a triple probe. Two opposing outer tips were connected to the other two outer tips as a double probe and the center probe measured floating potential. Since the measured floating potential varies with the rotation angle of the probe, an angle scan of probe orientation was performed. This most reasonable measurements, based on previous probe sweep results, was found around a probe rotation of 110 degrees. This probe orientation was used for a radial scan.

A radial scan of probe position was performed from 101cm - 110cm. From 101 - 106cm, the temperature was measured to be constant around 20eV with a decrease in density radially outwards. Probe signals were measured all the way out to 110cm (although small) with the plasma limiting at 107cm, suggesting a large scrape off layer.

Validation of the accuracy of these measurements (of Te and Ne) still needs to be done based on how the probe is connected and its orientation to the magnetic field.

Wednesday February 4 2015 6:14 pm Peng 88398~88432 feedback shot development

feedback shot development is done today. The ferritic walls are in and stainless walls are out.

The major radius is mostly maintained between 92 and 90 cm and q slowly goes down across 3. Mode activity becomes much stronger after q goes across 3. All shots before 88427 are without feedback with major radius not exactly the same. The mode is quiet at first and grows very fast to a saturation point quickly. A typical example is 88425. This implies a very big growth rate.

some feedback test with 5000 gain was done later but the major radius goes out quickly and makes it not comparable with the reference cases.

Thursday February 5 2015 6:13 pm Peng 88433~88471 feedback shot

feedback shots with ferritic walls in and stainless walls out continued today. The shots have a roughly constant major radius bewteen 91 and 92cm. a phase scan with gain at 5000 was performed. feedback effect is less effective compared with the all walls in case which is consistent with the theory that less wall coupling effect would decrease the feedback performance.

Monday February 9 2015 6:20 pm peng 88472~88513 shot development

Shot development is done today. Stainless walls are in and ferritic walls are out. The major radius still falls in with a OH elect at 325 and VFE at 95, which suggests keeping the major radius constant with such wall configuration might be very difficult.

Tuesday February 10 2015 6:03 pm Peng 88514~88549 feedback shot

feedback shots on stainless walls in ferritic walls out continued today. Most of the time was spent on shot development to get shots with strong modes activity.

shot style similar to ferritic case seems to give little mode activity. and a different shot style was reached later in the day. Major radius is roughly constant around 93cm before 3 ms and then falls in, drops below 90cm after 5ms. Mode activity is stronger in such case with a natural rotation frequency around 5 kHz.

Some feedback shots was performed at 4000 gain with phase shift at 300, 330, 350 deg and some acceleration, suppression, deceleration was observed.

Wednesday February 11 2015 6:50 pm Peng 88550~88592 plasma shot

feedback shots were taken with a better shotsyle targeting at 83909. Exact traces were not obtained but a fair reproducible shot style were obtained. A typical one would be 88586.

A phase scan with gain at 4000 was performed to collect enough data for the contour plot.

Friday February 13 2015 4:57 pm Rivera/Byrne 88593-88605 Rogowski Calibration

At end of day, a breaker was found to have been tripped, and the roughing pumps had shut down. One of the cryos was warm, and pressure was in the 10-6 range. Jim caught it early enough that it was not a problem. He suggested the system had not been like that for more than 2 hours or so.

The problem was that the cryo gate valve interlock for leaks was turned off. If you finish running **please ensure that you follow the shut down checklist** If Jim had not caught this problem, it's likely that the machine would have sat unpumped over the weekend, potentially fouling the vacuum for weeks.

Run was taken to test the new TF rogowski integrator and amp circuit. SH rogowski was piggy backed on this. Purpose was to ensure that the large noise on the shaping coil trace is due to the amplifier currently being used. It is. As the SH rogowski is digitized in the North Rack, and the amp it is currently using is noisy, and in the basement, it seems like a good idea to re-route the signal to a new amp in the north rack, both to eliminate noise and the potential for ground loops.

The TF rogowski was used to measure a set of VF and TF only shots. Analysis to follow.

Pat

Monday February 16 2015 6:18 pm Rivera/Byrne 88606-88616 TF Rogowski/SH Rogowski

Digitized the old TF probe through the AM502 instead of the old homemade digitizer. Signal level was seen to rise by ~6%. This represents half of the disagreement in peak signal seen by the Probe and Rogowski. Time-shifts in peak location remain unexplained, however

A few shots were taken with the SH Rogowski amplified through the AM502. The noise level was seen to be roughly the same (although slightly smaller). Nick suggests this may have to do with grounding of the signal.

Terminating the signal chain later and later, it seems like if there is a problem with grounding it is downstream of the integrator, as that is where the noise began to be pronounced. I'll check that tomorrow

Pat

Tuesday February 17 2015 5:57 pm Stoafer 88617-88645 Vector Mach probe

The five-tipped probe has been instrumented to read the current on every probe tip. In today's run each of the outer probe tips were biased negatively to collect ion saturation current with the center probe collecting electrons to source that current. This allows a "Mach" measurement in 2 directions. A rotation scan was performed to check Mach measurements and directions at different probe orientations. Analysis has not yet been completed on the measurements.

Thursday February 19 2015 6:27 pm Stoafer 88646-88693 Mach probe calibration

In the run today, the scan of probe rotations of the five-tipped probe connected as a 2-dim Mach probe was continued. The goal was to determine calibration coefficients for the ion saturation current for each probe tip. Most of the necessary shots were taken; the run will continue tomorrow.

There was evidence of arcing of probe tipD in shots 88688-88692 where the probe was rotated to 90deg (so probe tipD is along a field line with probe tips E and B). The current on probe tipD was in the opposite direction of ion saturation and the voltage on the center probe tip (tip E) was much higher (and positive) than usual. I will continue to investigate what is happening.

The gating circuit was installed and tested for the bias probe, which will help protect the crown amp.

Monday February 23 2015 5:41 pm Stoafer 88694-88738 Mach and Bias probes

The rotation scan of the five-tip probe position was completed today to determine calibration coefficients between the probe tips.

The bias probe was also used while measuring ion flow with the five-tipped probe as a 2-dim Mach probe. A scan of bias voltages and radial locations of the Mach probe is planned in this run campaign.

Tuesday February 24 2015 6:20 pm Stoafer 88740-88776 Radial scan with Mach probe with bias probe active

The Mach probe was used to measure ion flow at various radial locations ranging from 103cm to 106cm. The bias probe remained at 103cm and various biases were applied to control plasma and mode rotation. Initial results show minimal effect on plasma rotation until maximum positive bias (~90V), whereas mode rotation is significantly changed. More analysis of Mach probe measurements will be done to understand the results.

Wednesday February 25 2015 6:24 pm Stoafer 88777-88808 Rotation studies

The radial scan of the Mach probe with and without biasing the bias probe was completed today. Analysis to come with calibrated probe tips.

At the end of the day I applied an n=1 field with the control coils with different phases while applying a maximum bias on the bias probe to look for unlocking through ion flow. There were no obvious changes of unlocking, but there is evidence that it affects mode rotation. Analysis to come.

Thursday February 26 2015 6:07 pm Peng 88809~88853 shotdevelopment

Plasma shots are developed with all walls out configuration. The major radius is maintained around 91cm to 92cm. q slowly goes across 3 from above. The m=3 mode was not observed much. m=2 mode begin to showup after q goes below 2.5.

Friday February 27 2015 5:53 pm Peng 88854~88899 feedback shot

More shots with the same shot style like previous day are taken today. Mode activity is usually small until 4ms when q goes below 2.7 Comparing the TA sensors shows that the mode is indeed smaller than the walls partial in case. Some feedback shots are taken today with gains as high as 5500. The feedback effect is not very strong. More shots will be taken Monday.

Monday March 2 2015 7:16 pm Peng 88900~88986 feedback shot

feedback shots with all walls out configuration is performed today.

Good mode activity is observed when q is below 2.5(m=3 mode) and MR below 92cm. The Plamsa was not very reproducible but a phase scan is still performed. Feedback have effect on the plasma by slowing it down and speed it up. The suppression effect is not very strong. More analysis will be done.

Monday March 16 2015 10:09 pm Levesque 89006-89046 Cleanup shots, Ferritic wall configuration

Moved the walls to the "ferritic-inserted" configuration, with odd-numbered shells outward 5mm from full insertion, and even shells 5mm inward from full retraction. Attempted to reproduce target shots of 84046 and 83909, but plasmas would not last past 4ms regardless of the wide range of settings used. Could not get a reasonable positive current ramp. Diagnostics suggest than impurities are too high, and that cleanup is required. Discharge evolution throughout the day was also characteristic of needing cleanup shots and overnight pumping. SXR levels were much higher than the target shots with similar settings, but this does not appear to be from a high temperature. Later shots were targeted at putting in high OH power and trying to clean up the walls/limiters. Cleanup and shot development will continue tomorrow.

Control coil FB03_C2 is presently broken. At the beginning of the day, the respective CC amp channel light was green (bad) upon power-up, but the fuse is intact. Using the coil as a sensor did not produce good signal like the other coils. The current in this coil was good on the last shot of the previous run, but was it wasn't working from the outset today -- moving the shell configuration may have jostled the coil/connections a little, in case it has an intermittent failure. Troubleshooting will be needed for this coil before its next usage.

Tuesday March 17 2015 9:52 pm Levesque 89047-89085 Cleanup attempts, Ferritic wall configuration

Continued shot development and cleanup from yesterday. Shots did not show much improvement. Tried a wide range of bank and puff settings. "Cleanup"-style shots that went outboard were tried for parts of the day, but probably not enough or at high enough power to make much of a difference. Shots continued to terminate by 4ms, and only weak current ramps were achievable, even while maxing out our safety margin on the OHEL bank. This did not appear to be strictly caused by mode activity, which is otherwise common in the case of high OHEL voltages when plasmas are well-behaved. The longest-lived shots with reasonable parameters were those that remained up/down limited for the longest. There were no "good" shots for the day.

The HBT room is much hotter than for normal operating conditions, and impurities on the RGA before starting the run in the morning were higher than in other cases of normal run days that I found on the logsheet. Although the SXR signals were not as high today as yesterday for similar shot settings, I would still surmise that vacuum conditions (e.g. high outgassing due to high room temperature) are the culprit for the poor machine operation.

Shells were in the same configuration as yesterday for most of today. Since conditions didn't appear to improve, the even-numbered (SS-only) shells were fully inserted toward the end of the day with the intention of improving stability. Several shots were taken in this configuration, but still without significant improvement.

The problem that appeared in control coil FB03_C2 was investigated before running. I found a bad connection at the MS connector plugged in to the vacuum feedthrough, which is intermittent depending on the local cable bend. The black cable has clearly been stressed and worn over time, likely due to its small amount of slack and ease of stepping on it. Propping up the cable to reduce the tight bend at the connector fixed the problem for now. This also explains why the coil failed as a sensor after the shell+connection was moved. The "fix" is only temporary; a permanent fix of the cable end should be done before running the control coils again as actuators.

Wednesday March 18 2015 5:06 pm Levesque 89086-89116 Cleanup shots

Continued cleanup shots with all plasmas outboard limited. No improvement was seen throughout the day. Base pressure and RGA impurities were a little better today than yesterday, but still there seemed to be more N2 and O2 than should be expected in the morning following a run day. Ended the run early since there were many CAMAC errors that were getting worse, including failure to record the TF and fill pressure.

Yesterday I noticed a problem with another A14 that records control coil currents. Now a third A14 for the CCs has an intermittent bad timebase, for coils FB04_C{1-4} and FB05_C{3-4}, meaning that 18 of the 40 control coils have faulty timebase recordings. The problem with that A14 was not as prominent today.

Thursday March 19 2015 9:20 pm Levesque 89117-89154 Ferritic wall configuration, natural modes

Before running, did static fills (Shots 89117-89121) with the RGA on and found that the puff line was contaminated with nitrogen. Using a short puff of 200us to avoid swamping the RGA, we found that ~25% of the fill gas from the puff line was nitrogen (8.1e-8 torr) with the balance as mostly D2 (2.3e-7 torr), assuming that the relative calibration between N2/CO and D2/He is correct. After purging/diluting the puff line with D2 and roughing it out ~10-15 times, the static fill puff test was repeated. This gave nitrogen partial pressures well below 1% of the injected D2. This shows the importance of purging the puff line with the working gas several times after roughing it out. Note that this puff contamination was a result of testing the puff valve for leaks at the end of last week, and not purging the line a few times before starting to run again.

Shots from the past 3 days were consistent with the polluted puff line: loop voltages were higher than normal for similar bank settings; SXR signals were either very high or very low, possibly indicating bifurcated plasma equilibria above or below the nitrogen radiation barrier; radial evolution suggested a lack of plasma thermal pressure, etc. Normal plasmas were achieved after only a few shots following the D2 purge. The wall configuration started out the same as yesterday, then was changed to the "ferritic configuration" of 3/16/15 for shot 89136 onward. Plasmas that were reasonably close to the target of 84046 were achieved. 3/1 mode growth rates can be calculated for at least 8 of these shots. The run will continue tomorrow, along with growth-rate analysis in conjunction if there's time.

Friday March 20 2015 10:14 pm Levesque/Hughes 89155-89209 Ferritic-inserted and ferritic-withdrawn, natural modes

Continued campaign for comparing natural 3/1 modes in the ferritic-inserted and ferritic-withdrawn wall configurations. Produced around 5 usable ferritic-in shots before changing the wall to the "ferritic-withdrawn" configuration for shot 89170 onward (SS walls 13mm back from full insertion, Fe walls retracted to the Fe-safety-stoppers). Around 20 of the shots from the Fe-withdrawn config are usable for growth rate analysis with equilibria matching those of the Fe-in config. Overall the Fe-inserted configuration produced faster growth rates on average, consistent with results presented at the APS meeting and the forthcoming PoP paper. Detailed analysis of modes will be done next week.

Tuesday March 24 2015 6:24 pm Stoafer 89215-89241 Mach probe calibration

Shots today were taken to confirm the calibration settings of the Mach probe by testing a few probe rotation orientations.

There was a problem with a crates in the afternoon that caused the TF, OHE, and OHS banks not to fire. If this is a recurring problem, we will look into it.

Wednesday March 25 2015 6:40 pm Stoafer 89243-89271 Rotation studies

A radial scan of the Mach probe, measuring toroidal and poloidal flow, was completed with the bias probe floating. This helps complete a radial and bias scan to measure plasma flow.

There were issues with firing the banks in the beginning of the day. Jeff found a disconnected cable from the Krytron to the OH start bank and the lead was close to a power strip, possibly in contact. We weren't sure if the leads were suppose to be connected, so we moved the cable away from metal contacts. The run proceeded normally for about 10 shots then there were more bank firing issues. I talked to Nick and we decided to attach the cable to where it was originally attached (the Krytron is labeled as not working). This seemed to resolve the issue.

Tuesday March 31 2015 2:10 pm Stoafer 89281-89288 Rotation

We had issues with the crates today. Initially the crates wouldn't initialize with the take shot script. Issue was resolved by running original take shot.

Time base was off for all signals later in the day. Machine was shutdown at this point.

Thursday April 2 2015 6:06 pm Byrne 89296-89343 Shaping Shot Development

Run was delayed until after lunch today, as there were triggering issues. Problem was solved when the basement rack J221-02 and J222-01 were removed from the CAMAC crate, and had their contacts scoured with a steel brush, rubbed with a pencil eraser and wiped down with IPA. Notes from previous times we had this problem suggest that switching around the j221's, which worked, was really just re-seating the contacts.

Otherwise, run went well. Star shots are 89315 -17, -18 (unshaped) and 89331, -32, -39, -41, -42, -43(shaped) Main focus was replicating shot 89200.

This shot had a very steady MR, allowing for good use of the BD to isolate mode activity, was well centered, allowing for good BD discrimination of mode shape on the inboard sensors, and a decent current ramp, sending q through 3 (by the old, slightly off measure of q).

Shots are still under development, but I was able to use the SH coil as a pseudo-VF. The attraction to the shaping coil causes the plasma to fall inwards, and by boosting OHE and starting the SH coil earlier, and with lower SH Start, I was able to ramp the shaping current in such a way as to hold an outwardly unstable plasma roughly still at roughly 92cm.

This isn't perfect, but represents by the standard of shaped shots thus far the most well-centered, most steady, and at with a SH start time of 1.25ms, and a disruption time of 4-5ms, the longest lived shaped plasmas to date.

Plasmas seem to disrupt due to MHD. cylindrical q* is often not near a resonance, but mode activity is seen in SXR. Excited to model these, will take star shot 89341 and run it through BD, TokaMac and DCON to see what we get. Will forward results by end of day today.

Will not be running until tomorrow Afternoon, around 1pm. I have a class that runs 9-noon. This would be a good time for anyone with work to do in the cap room or machine room claim.

Byrne

Tuesday April 7 2015 5:50 pm Stoafer 89345-89363 Rotation studies

A new circuit for the bias probe was constructed using 2 transformers and two amplifiers. This setup was tested, then used with plasma shots. We saw a large rotation for shot 89356 with the additional power as compared to the previous system. The setup will continued to be tested this evening and the run will continue tomorrow.

Wednesday April 8 2015 7:20 pm Stoafer 89364-89404 Rotation

Today's run began with troubleshooting outputs from the AWG controller. The output for one of the bias probe amplifiers was found to be bad, so it was switched to another output. I planned on troubleshooting what caused the failure at the end of the run.

With the additional power to the bias probe many shots showed a transition from slow rotation to fast rotation states. This is exciting and will be analyzed to see what other changes we see with the measurements. There seems to be a large shear in poloidal flow in the fast rotation state based on a radial scan using the Mach probe.

Towards the end of the day the fuse blew for one of the amplifiers. During troubleshooting, I found that the AWG output went bad, so we stopped running to not cause anymore damage. Tests also show that the crown amp is not functioning properly. Troubleshooting and figuring out what to do next will continue tonight and tomorrow.

Monday April 13 2015 6:27 pm Hughes 89409-89457 All shells "in" growth rates

Today's run targeted shots 89140 (Fe) and 89203 (SS) from Jeff's Fe/SS growth rate campaign. Plasmas have been somewhat dirty, starting with either very high or very low SXR signals, and retaining somewhat irregular SXR and consistently high loop voltage signals throughout the day. With few exceptions, the trend has been shots with gross plasma parameters that follow the target shots relatively well, but exhibiting radically different perturbations. These seem to include beta collapses that radically alter the equilibrium, typically just before the plasma crosses downward through q=3 according to our current q calculation. Plasmas do not generally show clear exponential growth periods as in the example target shots, but extremely bursty mode amplitudes, most often two sharp spikes near the q=3 crossing time. This apparent mode activity may be a failure of equilibrium subtraction attempting to smooth over the beta collapse and producing the illusion of a large helical mode. At the beginning of the day, the loop voltage monitor was malfunctioning, but was repaired before 89414. Shots of note: 89409: Daily vacuum shot. 89436,89455: May have exponential growth periods that could be fit to. 89427,89429,89430,89431,89432,89436: Most steady shots with abrupt peaking suspected to be beta collapses. 89432,89451: Disrupt later than 9ms. Running will continue tomorrow. The RGA will be connected and left on over night to check impurity levels in the morning. If impurities are the problem, the only solution may be more cleanup shots.

Tuesday April 14 2015 6:30 pm Hughes 89459-89513 All shells "in" growth rates

Today was a continuation of the all shells "in" series expanding on the effects of Fe/SS wall configurations on mode growth rates. Plasmas were slightly better behaved than yesterday, but continued to exhibit the same trend of minor disruptions/beta collapses near the q*=3 crossing, making later-time behavior extremely unpredictable, and casting some doubt on how well these shots' equilibria can compare to those of the Fe-in and SS-in cases. Preliminary analysis of yesterday's data suggests reason growth rate fits can be extracted from perhaps 30-50% of these shots using cos(n=1) fitting rather than BD, since (at least in my experience) BD mode-mixing often pollutes the helical modes with signal actually belonging to large, abrupt equilibrium shifts. More analysis to come.
Noteworthy shots:
89460: Daily vacuum shot.
89491,89492,89505: Especially steady R0, gentlest pressure collapses, and long lived.
89463,89467,89470,89474,89476,89479,89480,89482,89485,89495,89496,89498,89499,89501,89503,89507,89513: Moderately steady shots. Not as good, but should be analyzed individually for clear growth rates and matching equilibria.

Wednesday April 15 2015 6:27 pm Hughes 49514-49540 All shells "out" growth rates

Today's run was the beginning of the all shells "out" (i.e. ferritic walls retracted to new safety stoppers, stainless shells retracted to 5mm in from chamber safety stoppers) segment of the growth rate study of wall configuration effects. Today was essentially only a half-day for running purposes, and the main objective was shot development for the all "out" wall configuration, using 89140(Fe) and 89203(SS) as target shots. Many plasmas exhibit lower SXR signals early in the shot, as well as cusps in both SXR and and R0 around 2.2-2.4ms, when q is in the neighborhood of 3.5 in most shots. This has made for highly unreproducible, unpredictable plasmas, so far, generally with very jumpy R0 traces. Noteworthy shots:
89514: Daily vacuum shot.
89535, 89536, 89539, 89540: May still be usable for analysis, although not very good matches for the target shots.


Thursday April 16 2015 6:26 pm Hughes 89541-89585 All shells "out" growth rates

Today was a continuation of the all-walls-out branch of the study of ferritic/stainless wall configuration effects on growth rate. Plasmas have continued to be less than cooperative, exhibiting large jumps in major radius around 2.3ms, although a few plasmas have had equilibria more similar to the target shots, 89140 (Fe-in) and 89203 (SS-in) that Jeff acquired.

Noteworthy shots:
89541: Daily vacuum shot.
89565,89566: Shots with various CAMAC failures.
89567-89569: Krytron testing
89544: Very strange I_P trace seen in several shots today, where plasma seems to disrupt but then recover. Other examples of this behavior include 89549,89574,89581.
89557,89559,89585: Relative promising shots, close to target.
89545,89546,89547,89550,89553,89558,89575,89578,89580,89582,89583: Not very close to target, but may still be useful. Most have very jumpy R0, especially around 2.2-2.4ms, and it is not entirely clear how these events (which seem to coincide with a loss of energy) will affect later growth rates around 3-4ms.


It should be noted that using the four top and bottom chamber-mounted PA sensors to measure vertical position, SS-in shots typically sit about 1cm below the midplane, while Fe-in and all-in shots sit about 2mm higher and all-out shots sit about 2.5mm lower.

PS: On an operational note, the day was also plagued by CAMAC difficulties that caused timing failures and bank trigger failures for a few shots. Power-cycling all crates and the server room AC (which sometimes fails to restart its compressor and has to be manually restarted) seemed to fix the problem, in this instance. I have noticed that the auto-fire has intermittently been slow, sometimes firing in under 1 second, sometimes taking 2-3 seconds before firing; I don't know if this is new but normal, or abnormal behavior. The problem may be related to overheating in the server room, but this is only a guess.

PPS: There have also been some minor take_shot_hbt.py problems, mostly warnings thrown up in the terminal and one or two crashes. Since I'm not sure what goes on in the code, I'm including the error messages here, just in case they are harbingers of a problem that can't be solved just by restarting the program:

The crash (Dispatch Control closed itself, but jDispatchMonitor stayed open) was accompanied by the error:
"The program 'take_shot_hbt_revised.py' received an X Window System error.
This probably reflects a bug in the program.
The error was 'RenderBadPicture (invalid Picture parameter)'."


The warning didn't accompany a crash, and simply appeared at some point in the terminal window:
/opt/hbt/bin/take_shot_hbt_revised.py:482: GtkWarning: GtkTextView - did not receive focus-out-event. If you connect a handler to this signal, it must return FALSE so the text view gets the event as well
gtk.main()

Friday April 17 2015 6:41 pm Hughes 89586-89632 All shells "out" growth rates

Today was the end of the all walls "out" run campaign. Plasma performance was similar to Thursday, with several possibly-good plasmas to study. Additionally, a few extra shots were taken with very different current ramps for use in testing code to compare SXR profiles shapes, which may help in verifying similarity of plasma equilibria.

Noteworthy shots:

• 89586: Daily vacuum shot
• 89588,89602,89608,89622,89625: Most promising plasmas for analysis.
• 89589,89592,89593,89595,89599,89600,89603,89607,89609,89613,89616,89619: Marginally good prospective plasmas for analysis.
• 89626,89628,89630,89632: Plasma with a range of positive, level, and negative I_P ramps for testing SXR profile shape comparisons. 89628 is noteworthy as it retains plasma current until 10ms.

Hardware Status Notes: The section 9 and 10 control coil amps lit green on shutdown (Thursday) and startup (today). They all had blown fuses which have now been replaced, and the amplifiers light red; however, the control coils should of course be tested before use to make sure all is well. It isn't clear when this happened, since it was between startup and shutdown on Thursday, but the CAMAC failure is the most obvious culprit. Also, the screen room crate did not turn off when the power button was pressed; Chris and Pat plan to take some crates-only shots, now, so I will leave it for them to take care of.

Monday April 20 2015 5:48 pm Stoafer/Byrne 89640-89699 High bias / shaping

The run today started with rotation studies using the new circuit for the bias probe. The circuit worked well all day with no issues and we saw transitions into the fast rotation state. An attempt was made to apply an m/n=1/1 error field correction with the control coils to see if the transition happened more easily if canceling an error field. This has not yet been successful, but runs will continue on Thursday, 4/23.

There were crates problems in the beginning of the day that ended up being caused by the take_shot script crashing too many times and blocking connection ports to the CPCIs. This was resolved by rebooting Spitzer.

During the run today one of the Krytrons for the VFstart bank stopped working. We are using the thytrotron in its place while Nick installs a new Krystron.

Byrne's run was shots 89681-99.

Intention was low-IP equilibria.

Got a good low IP, MR steady unshaped equilibrium (requires more refinement but the main thrust was shaping the equilbrium).

When I tried to shape it, the plasma disrupted early. Unfortunately, the MR code is set to only trip on discharges where IP exceeds 10kA. These shots did not reach that current before disrupting, and so MR was not plotted. This made it difficult to adjust bank settings.

Ended the day rather than continuing to fly blind. Last shot had low Shaping and higher OHE, just to ensure all was well. Got an outboard disrupting plasma, as expected, and MR digitized, as IP exceeded 10kA pre disruption.

Will edit the MR code to accept any plasma with current higher than 2kA, and pick up where I left off tomorrow.

Tuesday April 21 2015 6:35 pm Byrne 89700-89750 Low-IP shaping

Encountered difficulties developing the shots. Eventually got star shots 89750 and 89749. These shots eventually drifted outward and disrupted, but there is a period, lasting until about 3ms, when the plasma is well centered in the chamber. Shaping requires 0.8ms to turn on, and is turned on at 1ms. This gives us a millisecond of plasma that is well centered, low-IP, and under the influence of strong shaping ~5kA/turn.

A quick Chi-squared analysis of the TokaMac equilibria is being run, and the results will be shown at tomorrow's meeting. Roughly speaking though, the plasma seems to be low-pressure, very small (so little wall coupling, no chance of outboard limiting, and q95 is lower than q*), and diverted by about 1cm past the poloidal array limiter. There seem to be very active modes during the period in which the plasma is centered. I will send that analysis as well, and we can talk about it in the meeting.

Of interest also, shots 89715, 25, and 26 all displayed sawtoothing. These shots had large MR excursions, out to about 96cm, and so despite being shaped, their LCFS is likely close to circular, and they are almost certainly outboard limited. These shots showed large, repeated fluctuations in their SXR traces, and the fan array showed core cooling with edge heating, suggesting sawtoothing. These fluctuations were correlated with inward motion of the plasma major radius.

Wednesday April 22 2015 7:02 pm Byrne 89751-89825 Low-IP shaping development

Was unable to significantly improve the plasma performance from yesterday. Was able to greatly increase the database of shots similar to 89749 and 89750, however.

In addition, shots 89760, 89777 and 89802 and 803 are very well centered to over 4ms. These are the run's star shots. The rest of the run turned up a large number of shots with steady MR until about 3ms, at which point MR diverged outwards. Examples include -772, -773, -789, -790, -792, and -794.

Will analyze tomorrow, but if the mode shape that we saw with 749 and 750 persist, there should be more than enough shots to prove beyond a doubt.

Also took a few RMP shots, to help guide the planning for the next runday. Shots 89813-25 had -3/1 RMP's. Phase was chosen to get CC01_3 close to a node, as it is dead, and the central coils should couple most strongly to the plasma.

30A RMP's tended to disrupt the plasma, moving the plasma major radius around in the chamber. However, shots -816 and -817 may be useable. RMP amplitude was toned down, and shot -822 has a very stable MR during a 10A RMP. 20A RMP's were attempted, but there was not enough time to run to get good shots. -824 and -825 have MR roughly stable till about 3ms, which is when the phase flip ends.

Pat

Thursday April 23 2015 8:06 pm Stoafer 89826-89886 Error field studies with high bias

The goal of today's run was to see if we could use the control coils to cancel out error field by applying a m/n=1/1 structure at various phases and amplitudes. To test if the error field was reduced we used the bias probe a a large positive bias and measured the torque from the bias probe required to enter the fast rotation state. We also ran with Thomson scattering to measure the profile during different plasma states.

Overall, results are fairly inconclusive about whether we were able to cancel out some error field, but there is still some data exploration and analysis to understand the dynamics of the shots and how the plasma enters the fast rotation state.

Phase scans and amplitude scans of the applied field were conducted. It seems that a phase around 270deg and input voltage around 0.5-1.0 (5 - 10 amps in the CCs) might produce the best error field cancellation, but this still needs to be confirmed by exploring the shots.

The following are interesting shots:
phase, amplitude, shot
280, 1.0, 89881
280, 0.5, 89882
280, 0.75, 89883
0, 0.75, 89884
90, 0.75, 89885
180, 0.75, 89886

Friday April 24 2015 7:42 pm Stoafer 89887-89946 High Bias EF and Bias radial scan

The beginning of today's run was a continuation of looking for error field cancellation using a high bias on the bias probe. Two phase scans were completed at different amplitudes (7.5 amp and 15 amp maximum CC current). There is nothing obvious the signifies an error field cancellation based on the torque required to enter the fast rotation state. A more detailed analysis will be completed to see if there is anything I missed. The scans with shot numbers are listed below:

7.5A scan: shots 89889-89906
15A scan: shots 89907-89923

Later in the run day, scans were performed to see how the bias probe affects plasma conditions. The bias probe was floating and positioned at different radii. The five-tipped probe was instrumented as a triple probe. With each bias radius, the triple probe was scanned between 102 and 106cm to get a radial profile. This was done for the bias probe at 103, 104, 105, and 106cm. This will be continued in the next run day with the bias probe out of the plasma. Thomson scattering was also active for each shot with the laser firing at 3ms.

Monday April 27 2015 8:09 pm Stoafer 89947-90012 Error field / bias probe

Today's run began with looking at error field cancellation using the high bias torque method from the bias probe. Field errors were applied using the bottom mid-plane array of control coils with an n=1 perturbation. The phase and amplitude were adjusted to reduce the amount of bias probe current required to cause the plasma to transition into the fast rotation state. A phase of 0 degrees and amplitude of 10 A were found to reduce the required torque the most so far, however more scans are needed.

Towards the end of the day, the triple probe scan with the bias probe removed from the plasma was completed to determine how the bias probe affects plasma conditions. Thomson scattering was also active during this time to measure the core.

Key shots include:

• vacuum shot: 89948
• CC off: 89979
• Amp = 15A:
• phase = 0deg: 89974
• phase = 90deg: 89975
• phase = 180deg: 89976
• phase = 270deg: 89977
• Amp = 10A:
• phase = 0deg: 90000
• phase = 90deg: 89998
• phase = 180deg: 89999
• phase = 270deg: 90001

Friday April 28 2015 7:42 pm Stoafer 90013-90067 The scan of applying different field errors with the control coils continued today using the bottom midplane array of CCs at various n=1 phases and amplitudes.
Plasmas and max torques seemed to vary throughout the day, so I need to look at the data carefully before looking for optimal error field cancellation.

Key shots include:

• vacuum shot: 90013
• CC off shots: 90044-90055
• 6A scan: 90024-90042
• 10A scan: 90057-90065.

Monday April 29 2015 6:56 pm Byrne 90068-90112 Chris was feeling under the weather, so I took over the runday.

Intention ws to compare shaped plasmas to unshaped plasmas, in terms of their temperature profile via Thomson Scattering and measurements of edge location via insertion of the mach probe.


Initial results suggest for a low-ip shaped plasma centered at 92cm, a minor radius of ~13cm. Center tip of the 5 tip probe, at floating potential, was seen to go to zero when the plasma was at or near 92cm. This has been correlated in the past with the probe being outside the plasma LCFS.

Thomson data has not yet been analyzed in anyway.

Star shots:

• 90078 - This plasma is unshaped. It has very similar plasma conditions to the shaped star shots from last run, 89802,89777,and 89760
• 90083,84 - unshaped shots, similar.
• 90095,99,105,110,112 all shaped shots. Had Mach probe at a variety of radii. results suggest edge of shaped plasma with MR at 92cm has edge at 105-105.5cm. Thomson data was taken at 3cm. In all cases plasmas are well centered and not moving major radially quickly.

Friday May 1 2015 6:39 pm byrne 90149-90175 Finding Shaped Edge

Used the downtime today to run with the Mach probe inserted floating at 105cm.

Expect that floating potential should reach zero or go positive when probe tip leaves plasma edge. Treating MR as a good number for plasma centroid, subtracting the MR value when the floating potential goes positive gives an edge location.

In all cases, this was seen with the plasma between 90.2 and 92. With a circular plasma, 105cm should be inside the LCFS throughout this region. This seems like solid confirmation that the LCFS is NOT limited on the outboard limiter.

Depending on the minor radius given, and on the assumption that the plasma is roughly centered to ~1cm above the midplane (TokaMac's prediction) we also find that it is unlikely that the plasma is top/bottom limited.

All TokaMac equilibria (and my own fixed-filament code) predicts the inboard edge of the plasma to be severely pushed in by the counter-ip coils, so inboard limitation is almost certainly out.

I will begin looking at SXR data, to see if the profiles are lower for shaped plasmas at the outer chords than for circular plasmas next, but these results are very encouraging.

A plot of the data will follow this email.

Pat

Monday May 4 2015 5:52 pm Stoafer 90176-90216 Rotation studies

Today's run started with shot development for low-Ip ramp, stable plasmas. This is for an experiment to apply phase flip RMPs to stable plasma with various bias voltages, controlling rotation. Some initial shots were taken with RMPs. Note, the RMPs are 1 ms earlier than desired because of timing inconsistency with prep_awg. This was fixed in later shots.

Vacuum shot: 90176
RMP phase flip from 1.5 to 2.0 to 2.5 ms:
shot, bias v, CC amp
90188, 0, 1.0
90189, 0, 0.5
90190, -40, 1.0
90192, -40, 0.5
90193, +40, 1.0

Shots 90193-90197 had bias probe at +40V with RMP from 2.5 to 3.0 to 3.5 ms. Shot 90199 had same RMP with bias probe near floating potential.

Later in the day, the high bias torque scans continued by looking at shots with a large error field applied compared to shots with no applied error field. Error fields had m/n=1/1 configuration. The following shots are part of the scan:

shot, phase, amp
90201, off, off
90202, 180, 2.0
90203, 180, 2.5
90204, 180, 2.5
90206, 180, 2.5
90207, off, off
90208, off, off
90210, 180, 2.5
90211, 180, 2.5
90212, 180, 2.5
90216, off, off

Tuesday May 5 2015 6:27 pm Stoafer 90217-90273 Rotation/error fields

Today's run focused on two experiments: (1) RMPs on stable plasmas with rotation control and (2) Error field studies with bias probe torque into the fast rotation state.

(1) The beginning of the run day focused on shot development of a stable plasma, on which we can apply RMPs. A target shot was developed, 90243, which will be used in future runs.

(2) In the afternoon, studies were continued with applied error fields and high bias torque. Shots with high amplitude error field with m/n=1/1 configuration were taken at various phases to compare to shots without an applied error field. The following shots will be used for analysis:

shot, phase, amp
90247, off, off
90249, 180, 2.5
90253, 180, 2.5
90254, 270, 2.5
90260, 270, 2.5
90262, 0, 2.5
90269, 270, 2.5
90270, 270, 2.5
90272, off, off
90273, off, off

Wednesday May 6 2015 8:07 pm Stoafer 90274-90339 Rotation/error fields

Today's run was similar to yesterday's: RMPs with bias probe in the beginning of the day and applied error fields with high bias torque towards the end of the day. Results from both of the experiments are promising.

There were a couple of successful RMP shots with the bias probe at -40V (shot 90292, near floating) and at +40V (shots 90297, 90298, 90300, and 90301, slowing the plasma). Plasma response has not yet been analyzed. It is interesting to note that the Mach probe sees plasma flow in both the toroidal and poloidal directions during an RMP.

A phase scan was performed with an m/n=1/1 error field testing for maximum bias probe torque. With many shots, there is a clear peak at 135/180 degrees (requiring more torque) with less torque required at 315 to 45 degrees.

The following shot numbers were added to the scan:
shot, phase, amp
90306, off, off
90307, 270, 2.5
90308, 0, 2.5
90309, 180, 2.5
90310, 45, 2.5
90311, 315, 2.5
90312, 225, 2.5
90313, off, off
90314, 225, 2.5
90317, 0, 2.5
90319, 180, 2.5
90325, 45, 2.5
90326, 315, 2.5
90327, 90, 2.5
90329, 315, 2.5
90330, 225, 2.5
90331, 45, 2.5
90332, 135, 2.5
90333, 135, 2.5
90335, 315, 2.5
90336, 135, 2.5
90337, 135, 2.5


Thursday May 7 2015 8:19 pm Stoafer 90340-90410 Fast rotation state / error field scan

Today's run started with shot development to have an outboard plasma with major radius greater than 94 or 95cm in the fast rotation state so that the SXR fan array can view the plasma edge. Shot development was relatively successful with some minor adjustments needed for a longer lived plasma. Target shot from Jeff's work is 81519 and today's most successful shot was 90359.

Later in the day, another error field phase scan was performed, but with lower amplitude (10 amps). Results are as expected as compared to the previous scan at 25 amps. The shots in the scan are:

shot, phase, amp
90360, off, off
90361, off, off
90362, off, off
90363, off, off
90364, off, off
90366, 45, 1.0
90367, 135, 1.0
90369, 225, 1.0
90370, 315, 1.0
90371, 0, 1.0
90372, 90, 1.0
90374, 180, 1.0
90375, 270, 1.0
90376, 45, 1.0
90378, 135, 1.0
90381, 225, 1.0
90382, 315, 1.0
90383, 0, 1.0
90384, 90, 1.0
90386, 180, 1.0
90392, 270, 1.0
90393, 45, 1.0
90395, 135, 1.0
90397, 225, 1.0
90400, 315, 1.0
90402, 0, 1.0
90403, 90, 1.0
90405, 180, 1.0
90407, 135, 1.0
90410, 45, 1.0

Friday May 8 2015 5:46 pm Stoafer 90411-90463 Fast rotation / error field

Today's run again started with outboard plasmas in the high rotation state to measure the plasma with the SXR fan array. Shot 90428 is a good shot with fast rotation and shot 90424 is a similar shot with the bias probe floating.

Later in the day, an amplitude scan of applying an error field with m/n=1/1 at 315deg and 135deg to suppress and amplify the error field. Shots in the scan as follows:

shot, phase, amp
90432, 315, 0.5
90434, 315, 0.5
90436, 315, 1.5
90437, 315, 1.5
90439, 315, 2.0
90440, 315, 2.0
90442, 315, 1.0
90443, 315, 1.0
90446, 135, 1.0
90447, 135, 1.0
90448, 135, 1.0
90449, 135, 0.5
90450, 135, 0.5
90451, 135, 0.5
90452, 135, 1.5
90453, 135, 1.5
90455, 135, 2.0
90457, 135, 2.0


Afterwards, shots were taken with the bias probe floating and conditions that show the most error field cancellation, phase=315 and amplitude=10A. We are looking at MHD mode activity to see if there is reduced mode locking in mode rotation. These were shots 90459-90463.

Monday May 11 2015 6:29 pm Byrne 90464-90477 SXR Tomography

Attempted to route more of the SXR tomography spool piece channels into the North Rack CPCI. Current channel is #204, just above the midplane (horizontal view). Added 205, just below the midplane horiz, 118, the far outboard edge vert, 201, edge top horiz, 208, edge bottom horiz, and 121, edge x-point, diagonal.

found 205 to be working, but none of the other channels gave appreciable signal. Channel 121 has strong signals, but they begin before the shaping occurs, and are the same regardless of whether the shaping is present.

Tried moving the soft x-ray channels to the next ones further in. (plots will be sent around). Found the same behavior. The two midplane horizontal channels worked fine, all other channels did not. Interestingly, channel 122 had the same odd behavior as channel 121. No reason why this should be so.

Will continue to work on this tomorrow. Will check up/down symmetry of 204 and 205 since they are working, in the presence of shaped and unshaped plasmas. Will also attempt to bring more channels online, as a secondary goal.

Pat

Tuesday May 12 2015 8:33 pm Byrne 90480-90533 SXR

Worked further on the SXR box. As it turns out, several channels were working, but recieving such low signal that it was hard to tell. Also found that some SXR channels have loose connections. Moving the wires around seems to effect the signal gathered, especially on the X-point viewing chord.

However, the X-point viewing chord has a very large pickup at the disruption of a shaped plasma. Similar pickup is *not* seen during the disruption of an unshaped plasma, but a vertical inboard viewing angle was installed and did not show such a large signal. The question remains open whether it is an artifact.

Since a large dataset was inadvertently taken yesterday with shaped plasmas, and several more shots besides today, unshaped shots were developed.

Unshaped plasmas that fell inwards had their outboard edge signal fall to shaped levels when they were significantly inboard. Have not quantified how far in yet. However, SXR signal at channel 117, which is 5cm from the outboard limiter is generally negligible in shaped plasmas, while in unshaped plasmas, especially as the plasma heats up, it can be clearly measured, if not large.

Also intend to look at the ratio of above to below midplane SXR signal. Expect it to be higher in shaped plasmas.

Wednesday May 13 2015 7:44 pm Stoafer/Byrne 90551-90591 High torque with EF / shaping

Pat ran in the beginning of today's run with shaped and unshaped shots and looking for SXR tomography channels to use for viewing the plasma edge and X-point.

Towards the end of the day, I continued an amplitude scan of applied error fields in measuring the torque required to get into the fast rotation state. Included shots are as follows:

shot, phase, amplitude
90566, 315, 2.0
90568, 315, 1.5
90570, 315, 1.5
90571, 315, 1.0
90572, 315, 1.0
90575, 315, 0.5
90576, 315, 0.5
90577, 135, 0.5
90578, 135, 0.5
90580, 135, 0.5
90581, 135, 1.5
90582, 135, 1.5
90583, 135, 1.5
90585, 135, 2.0
90586, 135, 2.0
90587, 135, 2.0
90591, 135, 2.0

Thursday May 14 2015 6:30 pm Stoafer 90592-90644 RMPs with rotation control

Today's run focused on RMPS with rotation control using the bias probe. Both 3/1 and 4/1 RMPs were applied in separate shots to measure the response with different bias probe voltages. RMPs were applied from 3 to 4 ms with a flip at 3.5. The following are key shots from the run for analysis:

shot, bias voltage, RMP config, RMP amplitude
90606, +80, off, off
90609, +80, 4/1, 2.5
90614, -60, 4/1, 2.5
90622, +60, 4/1, 2.5
90627, +60, 4/1, 3.5
90638, -60, 4/1, 3.5
90639, +150, off
90643, +150, 4/1, 3.5
90644, +150, 4/1, 3.5

Friday May 15 2015 5:46 pm Byrne 90645 - 90691 Shaped RMPs

An opportunity to run presented itself late in the morning, so I did some exploratory running with RMPs on a shaped plasma.

Lost some time early in the day, as I was misreading the output of the bigplot code. On correcting that, it was realized that several shaped shots similar to the target had already been taken.

Star shots are 90655 and 90657, as they are good matches to the model, and unperturbed. 90680-84, 90686, 90688, 90690, 90691 are shaped shots with RMPs. RMP amplitude varied from 10 to 20 to 30 Amps. -3/1 helicity, -54 degree phase shift to have nodes near the dead sensors.

Have not done any analysis, but Mach probe was in to 106cm, and during RMP significant changes in signal were seen, implying the plasmas LCFS is significantly displaced by the perturbed mode. Or, perhaps a loss of confinement increasing the electron density in the scrapeoff layer

Patrick Byrne

Monday May 18 2015 7:58 pm Byrne 90692-90739 Shaped RMPs

A cryo was regenned over the weekend, but there were no issues with startup. Matched target shot almost immediately. Shots are all similar to the standard shaped shot, Ip around 10kA, steady MR at around 93cm, disupting around 4ms.

First plasma shot of the day saw TA sensor 9_s2P go bad. The signal was very high and rising, seemingly over integrated, and looking at the raws did not provide a clear explanation, other than a much lower, and slower decaying initial pulse at the breakdown of the plasma. Worked again fine the next shot, but there are indications other sensors may not be working well either. Chris had a TA sensor fail recently.

Star shots are 90700-702, 90712, and 90736. These were the initial shaped plasma shots, and two shots taken during the RMP run to ensure the base plasma was the same. These are well representative of the normal plasma, and will be helpful to look at unperturbed modes.

A phase scan was taken, as the Mach Probe was inserted to 106cm, and was floating. The voltage seen on the center tip has been seen to correlate with plasma edge location. When the RMP is active, we see the tip voltage vary. The phase scan is intended to see if the edge is being distorted by the RMP's. It seems likely that is the case, as the voltage goes up or down depending on RMP phase. All RMP's -3/1.

Star shots are 90715 and 17 (base phase 85, amplitude 10A), 90718 and 20 (base phase 265, amplitude 10A), 90722 and 25 (175, 10A), 90726 and 27 (355, 10A). An amplitude scan was taken with a phase of 85 degrees (seen to zero out the signal that would go through the two dead coils best...). Shot 90728 (5A) 90732 and 33 (15A) 90738 (20A).

No analysis yet.

Pat

Tuesday May 19 2015 8:19 pm Stoafer 90740-90806 RMP with slow rotation and high torque with error field scan

Today's run began with RMPs of slowed, stable plasmas. The bias probe was set to a ~100V to slow the plasma to ~0kHz for 3/1 and 4/1 RMPs applied during stable plasmas. Analysis similar to Daisuke's method is being developed to look at these slow plasmas compared to more quickly rotating plasmas in both the poloidal and radial field response. Notable shots include:

Shot, Bias voltage, RMP, RMP amp
90748, +100, off
90751, +100, 4/1, 3.5
90753, +100, 4/1, 3.5
90759, +100, 4/1, 2.5
90762, +100, 3/1, 3.5
90766, +100, 3/1, 2.5


Later in the day, the high bias torques with applied error fields experiment was continued with applied m/n=2/1 EFs. Initial results show significant effects of different EF phases on the required bias probe torque to enter the fast rotation state; more complete analysis to come. Shots from the scan include:

Shot, phase, amp
90774, 0, 1.5
90775, 90, 1.5
90776, 180, 1.5
90778, 270, 1.5
90779, 0, 1.5
90781, 90, 1.5
90783, 180, 1.5
90785, 270, 1.5
90789, 45, 1.5
90791, 135, 1.5
90792, 135, 1.5
90793, 225, 1.5
90794, 225, 1.5
90795, 315, 1.5
90796, 315, 1.5
90797, 45, 1.5
90799, 45, 1.5
90800, 270, 1.5
90801, 0, 1.5
90802, 90, 1.5
90804, 180, 1.5
90806, 90, 1.5

Thursday May 21 2015 7:59 pm Byrne 90889-90943 Shaping/RMP/phase scan

Forgot to issue a run summary for yesterday, will include both days.

Intention of run was to scan the plasma response to RMP's across a variety of helicities. n =1 m = {0, -0.5, -1, -1.5, -2, -2.5, -4} have been taken. In addition, many m=-3 shots have been taken at various other times, and will be included in the data base. It remains to complete m = -3.5, and possibly -4.5, as was the original plan.

Plasmas being RMP'ed with a new helicity tended to require shot development before good shots could be taken. All completed helicities have 3 shot in which the plasma is roughly centered around 93cm, and disrupts after 4ms. The phase flip occurs from 2.5-3.5 ms, and the shaping coil fires at 1ms, peaking by 1.8ms.

Lost several shots today (Thursday) when the shaping coil bank did not provide as large a current as was expected. Signs pointed to the dump, but it was intermittent and we were unable to solve the problem. During shutdown, I checked out the SH coil dump, tugged on the leads and the other wires in the dump, and found that the lead from the solenoid to the terminals to which mains power is directed was loose. The wire pulled out of its lug with almost no force at all. This happened around 7:20pm, so the decision was made to shut down and work will begin on the dump tomorrow. Should be very simple, unscrewing the dump from the bank, attaching a new lug, and reconnecting the solenoid.

Pat

Tuesday May 26 2015 7:15 pm Stoafer 90982-91048 RMP / error field with rotation control

Today's run began with an RMP scan with controlled rotation. The plasmas were developed to have a more steady major radius than the previous RMP scans and included lower RMP amplitudes.

Key shots include:
90982, vacuum shot with RMP
90987, -60, off
90988, -60, 4/1, 1.5, 90
90990, -60, 4/1, 2.5, 90
90992, +80, off
90993, +80, 4/1, 1.5, 90
90994, +80, 4/1, 2.5, 90
90998, +80, 4/1, 2.5, 90
91003, +100, off
91005, +100, 4/1, 1.5, 90
91009, +100, 4/1, 2.5, 90
91010, +40, off
91013, +40, 4/1, 1.5, 90


Later in the day, the high torque experiments were continued for scan with the 2/1 error fields. A phase scan was completed and amplitude scan has been started.

Wednesday May 27 2015 8:11 pm Stoafer 91049-91112 RMP of stable modes

Today's run was spent developing a stable plasma to do m=0, n=1 and n=2 RMPs to measure the response of a stable mode. Key shots include:

91093: Vacuum shot with m=0, n=1 RMP amp=4.0
91097: Vacuum shot with m=0, n=2 RMP amp=4.0
91090: RMP m=0, n=1, amp=4.0, from 3.5 to 4.5
91096: RMP m=0, n=1, amp=4.0, from 3.5 to 4.5; more outboard and stable MR
91099: RMP m=0, n=2, amp=4.0, from 3.5 to 4.5


Thursday May 28 2015 8:18 pm Stoafer 91113-91191 Rotation / applied field studies

In the beginning of today's run the bias probe was used to measure how it affects the plasma. Triangle waves were applied to the bias probe and the mode and plasma rotation were measured. Error fields were applied based on the results from the high torque experiments to see if the MHD mode activity changes with constant applied error fields. The following are key shots from these experiments:

91123, 1.5kHz triangle wave, no EF
91126, 1.0kHz triangle wave, no EF
91129, 2.0kHz triangle wave, no EF
91130, 1.0kHz triangle wave, 2/1 EF amp=1.5 phase=135
91131, 1.0kHz triangle wave, 2/1 EF amp=1.5 phase=315
91140, bias probe floating, 2/1 EF amp=1.5 phase=315

In the middle of the day, non-resonant error fields were applied to drive a stable mode. The following are key shots:

91146, m=-1 n=2 amp=40
91149, vacuum shot, m=-1 n=2 amp=40
91153, m=-1/2 n=2 amp=40
91154, vacuum shot, m=-1/2 n=2 amp=40


Towards the end of the day, experiments were performed with high bias probe torque applying 3/1 error fields with a phase scan. The following shots were used in the scan:

91168, off
91169, 270, 1.0
91170, 0, 1.0
91171, 90, 1.0
91172, 180, 1.0
91173, 45, 1.0
91174, 135, 1.0
91175, 225, 1.0
91176, 225, 1.0
91178, 315, 1.0
91180, 0, 1.0
91181, 0, 1.0
91182, 90, 1.0
91183, 90, 1.0
91184, 180, 1.0
91185, 180, 1.0
91189, 45, 1.0
91190, 135, 1.0


Friday May 29 2015 5:30 pm Stoafer 91192- Fast rotation state studies

In the beginning of today's run, RMPs were applied to the plasma in the fast rotation state. The plasma seems to transition out of the fast rotation state when a moderate RMP is applied (4/1 amplitude 15A and above). Some RMPs were taken with lower amplitudes to measure the plasma response while it in the fast rotation state without transitioning out. The following are key shots:

91204,

Friday May 29 2015 6:41 pm Stoafer 91192-91265 Fast rotation state studies

The previous run report submission was a mistake.

In the beginning of today's run, RMPs were applied to the plasma in the fast rotation state. The plasma seems to transition out of the fast rotation state when a moderate RMP is applied (4/1 amplitude 15A and above). Some RMPs were taken with lower amplitudes to measure the plasma response while it in the fast rotation state without transitioning out. The following are key shots:

91204, no RMP, target shot
91215, 4/1 RMP amp=15
91217, 4/1 RMP amp=15
91219, 4/1 RMP amp=10
91221, 4/1 RMP amp=10


Vacuum shots were taken throughout the run for the RMPs:
91212, vacuum, 4/1, amp=15
91214, vacuum, 4/1, amp=25
91216, vacuum, 4/1, amp=35
91220, vacuum, 4/1, amp=10


Later in the day the high torque error field scans were completed. Shots began with completing a phase scan of 3/1 EF:

91223, 270, 1.0
91229, 0, 1.0
91230, 90, 1.0
91231, 135, 1.0
91234, 315, 1.0
91239, 270, 1.0
91241, 180, 1.0


The amplitude scan of the 2/1 EF was also completed from previous runs:

91244, 315, 0.5
91245, 315, 1.0
91247, 315, 1.5
91248, 315, 2.0
91249, 315, 0.5
91250, 315, 1.0
91251, 315, 1.5
91253, 315, 2.0


Towards the end of the day, a 2-D Mach probe radial scan was performed for the fast rotation state:

91254, 106
91255, 105
91256, 105
91259, 104
91264, 103
91265, 102


Monday June 1 2015 6:24 pm Byrne 91266-91319 Low-IP, Circular-RMPs

Today's run focused on developing an unshaped shot with a similar equilibrium to target shot 90657

Star shots are 91283,91287,91288,91289,91291,91296,91297, 91298, 91300,91305. 91298 and 91305 were the best shots of the development cycle, and are a very good fit to the shaped plasma, in terms of MR and IP, though SXR is a bit lower.

Once these plasmas were developed, RMP's began. 0/0 shots were taken. Star shots are 91306-91310 and 91313. -1/0 shots were also taken, shot 91314 is a good match to the unshaped equilibrium. Shots were somewhat bifurcatory, either falling out or in. Roughly one in 3 was acceptable. Will continue to develop the helical scan, and by extension, the unperturbed database during the morning startup, over the course of the week.

Day ended when crates began to not digitize. As it was late, and this failure mode usually requires a half hour or so to recover from, it was decided to shut down instead

Also, the oil leak seems to be accelerating. Filled the oil line to above the yellow line, and perhaps just to the reservoir. Found the oil level down to the 'T' juction at end of day (5:30pm). Chris has seen something similar. Will begin checking oil level at 2pm to ensure safe conditions.

Pat

Tuesday June 2 2015 6:50 pm Byrne 91323-91386 Unshaped, Low IP helicity Scan, Amplitude Scan

Rebooted Spitzer at the beginning of day as crates were still not initializing. That fixed things.

Started with m=-1 n = 1 RMP's, and proceeded through m=-2, m=-3, m=-4, m= -0.5, and m = -1.5, all with 30A of control coil current.

At the end of the day, took a few m = -3 shots with differing amplitudes. 2 shots apiece were taken with 5A, 10A, 15A, and 20A.

This will be compared to similar shots from shaped equilibria. Preliminary analysis shows that the unshaped shots respond *less* strongly to almost all helicities of RMP. A measurement of response vs driving current/field has yet to be made.

Star Shots: no rmp - 91313, 29, 43,65, 68, 57, 50, 51 m = 0 - 91306, 07, 08, 09, 10

m = -1 - 91314, 31, 34

m = -2 - 91339, 40, 42

m = -3 - 91347, 49, 53, 54, 73, 74 (30A)

m = -4 - 91355, 56, 58, 59

m = -0.5 - 91360, 61, 63, 65

m = -1.5 - 91356, 67, 71, 72

m = -3, A = 5

91375, 91376

m = -3, A = 10

91379, 91380

m = -3, A = 15

91383, 91384

m = -3, A = 20

91385, 91386

Wednesday June 3 2015 10:30 pm Byrne 91387-91463 Vacuum RMP response/Unshaped plasma RMP response

Morning was spent taking vacuum shots with different helicities. n =1, m = 0 to -4.5 (by half integers) all had 2 shots taken. 3.5V, and 0 phase These are shots 931404-931424

Following that, the unshaped plasma run continued. RMPs were taken at -2.5 and -3.5.

Star shots are:

no RMP - 91428,91437,91449,91454

-2.5/1 - 91429,91438,91440,91446,91450

-3.5/1 - 91455,91457,91458,91460,91463

Also the oil leak is definitely in the oil pump. Twice during the runday the oil was checked and found to have drained from the reservoir to the t-junction. This is significantly more than drains throughout the entire night or even weekend, when the pump is not on.

Thursday June 4 2015 7:11 pm Byrne 91464-91536 Sausage Modes/RMPS

Day was spent developing a shaped, low-ip plasma RMPed with sausage modes

star shots:

91486 & 817 - shaped, no RMP

91494 & 91505 - shaped, -0.5/2 RMP 0 phase, 3.5V

91507 & 91510 - shaped, -1/2 RMP 0 phase, 3.5V

91514-17 - Unshaped -4.5/1 RMP 0 phase, 3.5V

91518, 19, 21 - unshaped, 0/1 RMP 0 phase, 1.5V

91522, 23, 25 - unshaped, -1/1 RMP 0 phase, 1.5V

91526, 27, 28, 30-33 - unshaped -2/1 RMP 0 phase, 1.5V

The low amlitude shots were taken to get a low level response, since the high amplitude shots may have driven the plasma into saturation. The response will be compared to the high amplitude shots to make sure that the lack of resonant response is a feature of the RMP, not it's amplitude.

Friday June 5 2015 6:01 pm Stoafer 91537-91601 Triple probe scan with bias

The goal of today's run was to measure density and temperature fluctuations of MHD mode activity with different bias voltages and to measure the change in temp/dens profiles with different biases.

The following are the key shots from the experiments with the triple probe location and input voltage for the bias probe.

102cm:
-1.2V: 91601
0V: 91599
+1.0V: 91597
+2V: 91598

103cm:
-1.2V: 91595
0V: 91594
+1.0V: 91592
+2V: 91590, 91

104cm:
-1.2V: 91578, 80, 81
0V: 91562
+1.0V: 91566
+1.5V: 91568
+2V: 91569, 70, 72
+2.3V: 91574, 77

105cm:
-1.2V: 91538, 39
0V: 91560, 61
+1.0V: 91555, 57, 59
+1.5V: 91548
+2V: 91540, 41
+2.3V: 91545, 46, 47

106cm:
-1.2V: 91582, 83, 84
0V: 91585
+1.0V: 91586
+2V: 91587


Wednesday June 10 2015 8:42 pm Byrne 91610-91706 Sausage Modes/RMPS

Run Summary is for previous two days.

Tuesday's run was ended when the shaping bank again was not discharging its designated current. This has happened before, and last time a loose wire was found in the dump circuit. It was supposed this was causing the dump to intermittently lose power and close, dropping the bank voltage. The dump was repaired, but the problem rearose. A blown fuse was found on the start bank, but all components in the chain were checked and found to be in working order. No issue was found, and this morning the bank worked fine. The problem was run-up to by errors in the phidget. The phidget seems to get stuck in bad states when the banks dump. Why this should be so, and why the initialization of the interface program, which should reset all relay states in the phidget is not known. There is not time left to investigate/repair the problem. It is something we will have to be aware of and look out for in the next week and a half.

Several crates errors of the type seen monday occurred as well. Started with the J221 in the West rack, as Monday, then the failures spread to all A-14s in the Server Room, North, South and Basement racks. Rebooting spitzer and turning off all crates for the reboot fixed the issue.

Also, the control coil currents in coils FB02_c3 and c4 are clipping, at low levels of current. 91679 is a good example of this. Whether this is a digitizer issue, or something worse (a short to chamber diverting the current from the amp circuit?) I can't say. Problem did not get worse since it was noticed this morning

Several shots were taken, with the intention of creating a measure of m=0, n=1 and m=0,n=2 RMPs. DCON predicts a strongly m=0, n=1 and !50/50 m=0/m=1 n=2 mode. Since our coils cover 90 degrees of poloidal angle, a certain m=1 amplitude can't be avoided, so it was hoped that the n=2 mode would couple better the the coils

Vacuum shots for the n=1 mode are 91465,,91622. Good plasma shots are 90824, 90826, 90827.

Vacuum shots for the n=2 mode are: 91657,91658,91659,91660 Good plasma shots are: 91679,91682,91683,91693,91704

Thursday June 11 2015 6:49 pm Byrne 91707-91745 Non-resonant perturbations

Run went well, despite the late start. Took high amplitude m=0 n=1 RMP shots.

Vacuum shots (all banks included) are 91716-91719

Plasma shots 91727,91733,34,39,41

Un perturbed shots, with good equilibria 91721,26

Preliminary analysis shows significant n=1 response in the TA, but oddly, a phase shift. Not sure if that's correct with a stationary perturbation. These shots have had their phase shifted by 50 degrees to align better with the PA sensors.

Friday June 12 2015 7:46 pm Byrne 91747-91809 Re-phased sausage modes

Found that there is a phase shift between the perturbation and the response. Re-ran shots with new phasing to maximize the signal at the PA sensors.

Shaped plasma shots were difficult to get. Allowing the base pressure to drop into the 20ntorr range helped, but only a bit. Since phasing could only be determined after looking at a few good shaped, rmp'ed shots, this meant settling on a phase took most of the day.

A few crate failures, but they did not persist. Seem to be the same as yesterday, but more sporadic. Also, Jeff noticed that certain PA sensors seem to have RC time constants that are off. All sensors feed into in one amp box http://sites.apam.columbia.edu/HBT-EP/wiki/lib/exe/fetch.php/magnetics:hd_amps_annotate.jpg BOT 3.

m=0,n=1 shots had a phasing of +110 degrees in prep_awg. Good shots:

Vacuum:91761,91762

Plasma:91758,59,60,91777,91790,793,794,795

m=0, n= 2 shots had a phasing of + 90 degrees in prep_awg. Good Shots:

Vacuum:91802,91803,91804,91809

Plasma:91805,91806,91808

Un-RMP'ed shots:91755,91756,91776

Monday June 15 2015 6:23 pm Stoafer 91811-91867 RMPs with bias probe

Today's run focused on getting more shots for a scan of RMP amplitude and bias probe voltage. The run was rather successful with the following shots used for the scan:

shot, bias voltage, RMP amp
91820, -60, 15
91821, -60, 15
91824, -60, 35
91826, 0, 15
91828, 0, 25
91829, 0, 25
91830, 0, 35
91833, 0, 35
91835, 0, 15
91838, +40, 15
91839, +40, 15
91841, +40, 35
91842, +40, 35
91843, +80, 15
91845, +80, 15
91846, +80, 25
91848, +80, 25
91849, +80, 35
91853, +100, 15
91854, +100, 15
91857, +60, 15
91858, +70, 15
91859, +70, 25
91861, +70, 25
91862, +70, 35
91864, +70, 35
91865, +70, 15
91866, +70, 15
91867, +70, 15


Tuesday June 16 2015 8:01 pm Byrne 91868-91945 Sausage Modes/RMPS

Had issues with Caliban early in the morning. Despite running the correct order of startup commands (init/prep/do), when sending the pulseon command, ./do_awg would not trigger. After several reboots of Caliban, this was eventually fixed by a combination of restarting take_shot, rebooting Caliban, and running crates_off followed by crates_on. Crates_off/on was the only novel step taken, so it's likely, but not 100% that that is what fixed things.

The oil barrel is also very low. Had to fill it this morning, which was difficult alone. We should probably replace the barrel with a fresh one, or come up with a more simple way of transferring reclaimed oil to the barrel.

Star shots are: Vacuum shot + RMP: 91915, 91940

natural shaped plasmas: 91890,91939,91937

shaped RMPed(0/1 3.5V, 110 phase) plasmas: 91905, 91912,91914, 91945

shaped RMPed(-4/1 3.5V 110 phase) plasmas: 91917

Shots were very badly behaved. It was noticed earlier this morning that many shots taken so far have large MHD activity around 3-3.5ms. While this doesn't cause a disruption, it still prevents an equilibrium fit across the RMP. This activity is seen in non-RMPed shots frequently as well.

Also, a few shots were taken at the end of the day with extended 1ms RMPs, with no flip. The hope being that a BD on a stable equilibrium, with a stable perturbation, should allow us to separate the constant response from the eddies that have so far kept us from getting a clear idea of the structure of the mode on the plasma.

These shots are:

91936 - vacuum

91928, 91932, 91933, 91934, 91935 - with plasma.

Wednesday June 17 2015 8:27 pm Byrne 91946-92005 Sausage Modes/RMPS

On the final shot of the day, the OH bias bank and the OH start bank failed to fire. It's not clear if the OH start bank's failure to fire is due to the start bank's failure. Shot 92004 was burned, in order to turn on the gas for a plasma shot. Shots 92002 and 92003 were vacuum shots, and show all 3 OH banks firing, with no indication of what could have caused 92005's failure. As Chris and I were the only two in the lab at the time, the decision was made to shut down rather than attempt to diagnose the problem further. In the basement during shutdown, there were no obvious signs of damage, and the heat lamps were all on. It's possible the charging resistor has evaporated, though this morning the level seemed too high to evaporate that much.

Star shots of the day:

Natural shaped Plasmas: 91949, 91954, 91955, 91964, 92001

driven plasmas: m=0 n=2: 91969, 91974, 91980, 91981

driven plasmas: m=0 n=1: 91987, 91990, 91991, 91993, 91996

driven plasmas: m=0 n=1, with large minor disruptions, may be worth studying: 91994,91995,91998

Vacuum Shots + 0/1 RMPs: 91950,2002,92003

Vacuum Shots + 0/2 RMPs: 91952

Thursday June 18 2015 8:44 pm Byrne 92006-92075 Amplitude Scan

Problem with the OH bank was indeed the charging resistor. It was refilled with water, and the first vacuum shot worked fine.

Shots took a while to develop. Looking at the VF and OH traces, with the same bank settings as the target shot, 91375, the current traces were slightly different. This seems due to a higher loop voltage on these shots, which was the only clear difference.

At any rate, modifying the bank settings to get close bank currents, shots began being stable and lasting until or after 4ms. At least two shots were taken at -3/1 ampiltudes 0.5V, 1.5V, 2.5V, 3.5V

Finally shots were attempted with no-flipped RMPs. These shots were much more disruptive, suggesting that eddies induced by the flip reduce the growth of the mode, in addition to rotating it 180 degrees. Only a few of these shots were taken, at high amplitude. If time permits tomorrow, a few may be taken at the end of Friday

Star shots:

92012, 92041, 92043, 92069 - unperturbed shaped plasma

92050, 92052 - 0.5V, phase 0

92054, 92055 - 1.5V, phase 54 (this is the phase the other -3/1 RMPs were taken at. the 0 phase for 0.5 was an oversight)

92057,92060 - 2.5V phase 54

92062,92063,92064 - 3.5V phase 54

Friday June 19 2015 6:26 pm Byrne 92076-92103 OH charging resistor replacement/Natural Mode measurement/'paper rip test'

On first vacuum shot of the day, a bad smell was noticed, and the OH bank did not fully discharge. Going downstairs, the charging resistor was very brown and cloudy on top, and a deep turquoise at the bottom. The glass at the bottom was room temperature. The glass near the brown cloudy liquid was extremely hot.

It seems like the ionized water was denser than the non-ionized water. This has been noticed before, with crystalized sulphur crystals floating above an invisible density layer. The water was filled to the top of the resistor yesterday, and was high the day before. What seems likely is the doped waster was concentrating in the bottom, with de-ionized water resting on top. This led to the voltage drop being mostly across a narrow layer of deionized water, depositing a large amount of heat. Since the hot water was on top, there was no convection. This fed back until the top layer evaporated to below the electrode.

The residue from the boil-over was cleaned, the residue was removed, and the bank was hooked up to the replacement heating element resistor, already installed. Test shots are 92080 and 92081. Banks charged and fired fine, and the resistor did not feel hot to the touch, although only two shots had been taken when it was felt.

Next a set of shots were taken with the control coils active, but set to zero volts. The purpose is to ensure that the control coils are driven as sensors by the naturally existing mode. star shots are 92092, 92095, and 92096.

Following this, based on a suggestion from Prof. Boozer during the thesis proposal for this research, paper rulers were printed out. They were wrapped around the top and central bundles of the shaping coil and different circumferences, as measured on the rulers. The circumference was increased until the natural snap-out of the coil stopped tearing the paper.

Paper was slid to beneath the TF case, midway between the fixed shaping coil holder, and the hanging positioner. Toroidal variance was found, unsurprisingly. Chamebrs 4, 2 and 7 were measured. Chamber 4 failed to rip at 26.5cm. Chamber 7 failed to rip at 27.5cm, and chamber 2 failed to rip at 28cm. Resolution of measurement was 5mm. That is, after a rip, a 5mm longer loop would replace that one that ripped. The minimum loop length between the two bundles was about 25cm, in that both ends barely connected when the ruler was that length. As the rulers are loops, the 5mm length increases allow 2.5mm of extra jump-out. Work will be done with simulations to determine how the fields are affected by these separations. It is not expected that the coils will spread this much during the plasma lifetime, but it will serve to put a ceiling on the error fields.

Tuesday July 7 2015 7:14 pm Levesque 92129-92149 Checking mach probe I_sat measurements after reinstalling probe

Ran plasmas to collect I_sat and mach probe data after the removal and reinstallation of the 5-tip probe on the machine. The main purpose of today was to see if the presently-recessed tips have been recessed for a "long time" relative to the data collection period over the past few months, or if they became recessed when the probe was pulled out of the chamber a couple weeks ago. Comparing the I_sat levels measured before and after the recent venting gives insight into when the problem occurred -- we did not change the probe tips after observing the recession, so if the measurements from today yield similar currents as were seen in the past couple of months, this implies that the tips had been recessed prior to the venting. After running cleanup shots, a plasma reasonably close to the target (89256, from March) was achieved. When comparing star shots 92146 and 92147 to the targeted 89256, the probe currents from today do look very close to what they were in March. Both old and new shots have the mach probe at 105cm and 45degree orientation, though 89256 had the bias probe inserted and floating. The reduced effective areas of the recessed probe tips are likely responsible for the much-lower-than-expected plasma density that has been measured by the probe (while assuming the nominal probe area).

The 5-tip probe is severely misaligned within the bellows. While inserting the probe, I noticed that the bellows was catching near the probe tip, and there was a significant kink in the bellows wherever the probe tip was. Rotating the probe rotates the bellows kink, and by pushing on the bellows it's clear that the probe is far off-center. Only Chris or Jim would be able to tell if this catching/alignment is worse than before the venting. Nick and I slowly inserted the probe to avoid having the central tip hit anything, though a component along the shaft was catching in multiple places. Since the misalignment is so bad, I retracted the shells in that section before fully inserting the probe and kept them out for the plasmas. This was to prevent the probe from possibly hitting a shell upon insertion.

The TF oil was very low prior to running. While refilling and running the pump, I found a large white flake in the system flowing above the magnet on the right half of Section 8. This likely flowed out of the magnet, or dislodged from a local barbed fitting, given where I saw it. Nick thinks it's insulting material that was in the magnets when we got them, and that it's fine to run the system as is. The flake caught and stuck on another local barbed fitting while flowing.

PA2_S14P is broken. This sensor had problems before, but now has new problems as of this run. We "fixed" or modified the amp for this sensor during the recent RC checks (I think the one was missing a filtering capacitor), but a different problem is now present. We will troubleshoot it when the is a convenient time.

Several shots have interesting SXR fan array features during mode activity, including 92132 and 92138.

Thursday July 23 2015 5:45 pm Hughes 92194-92214 Vacuum shots to measure FBp pickup of VF from ferritic wall

Today's run was a series of vacuum shots to examine the amount of poloidal field measured on the midplane feeback poloidal sensors during TF+VF-only shots as a function of toroidal field. Peak VF for all shots is approximately 110G, whereas toroidal field varies from 0 to about 2800G at the sensors, assuming a 1/R gradient. Results of this run include finding that the TF bank would fire when set as low as 300V, but would not fire at 200V. There may have been some leakage due to the charge time of the VFST bank, however, so the true TF ignitron breakdown voltage may actually be somewhat lower than the setting on the control panel.

Thursday September 3 2015 4:57 pm Hughes 92232-92264 VF+TF vacuum shots, continued

Due to an arcing problem discovered yesterday during the hipot, the run was delayed until today. A wire (possibly from an old diagnostic) was found far inboard, and removing it seemed to resolve the problem. The suggestion was put forward that the TF-dependence of VF amplification by the ferritic wall might be a pitch angle effect rather than a saturation effect. To test this, two sets of vacuum shots were taken at varying TF settings, one with very high VF (the VF banks were specially cleaned for bank safety), 33% higher than the previous series, and another at low VF, 50% of the previous series. Little of note for normal machine operation purposes, except that 250V is the lowest recent TF bank setting recently observed to fire successfully. This represents a TF of about 45 Gauss. A TF bank setting of 200V failed to fire as of late July.

Wednesday September 16 2015 10:56 pm Levesque/Donald 92275-92297 Cleanup shots, Thomson scattering

Ran cleanup shots and started taking three-point Thomson scattering data, following the recent Rayleigh calibration. The calibration has not yet been implemented, but data can be back-calibrated following collection. Reasonable plasmas were achieved quickly, with shots lasting 6ms by the end of the day. Thomson data were collected at 3ms for the last two shots of the day.

During high-potting, a flash/arc was seen between Sections 1 and 2, and a crackling was heard near Section 8. These were likely due to overcharging the TF magnets to higher voltage then normal (around 11 or 12 kV, instead of 10kV). No problems were found in those areas, and a follow-up high-pot to normal voltage passed.

Outside of cleanup/development, much of the run was spent troubleshooting. An unusual popping sound, thought to be from around the tokamak, was heard for a couple of shots. After one of the shots, there was a mild smoke smell in the tokamak room. Inspections of the machine and capacitor room didn't turn up any significant problems. One of the roughing lines was found touching TF magnet and was moved, but this shouldn't have caused problems. Nothing unusual was seen in the current traces from the banks. Subsequent checks for arcs with vacuum shots didn't show any problems. After checks around the machine and banks, the run continued with no abnormalities.

Nodes were added to the tree for recording data from the laser energy monitor, but the data is not yet storing properly; this will be fixed tomorrow. The change to the tree caused a bizarre error in bank timings after one of the instances when make_tree was run, where trigger times were shuffled around to the wrong banks. This may have been caused by running make_tree without closing and reopening the take_shot script. The end result was that both the VF and OH banks rang for shot 92294, and the VF crowbar bank fired after the first return swing. Bank currents were checked with a follow-up vacuum shot after rerunning make_tree with take_shot closed, and everything functioned normally.

Wednesday September 23 2015 6:46 pm Levesque 92298-92319 Cleanup shots, Thomson scattering

Ran cleanup shots and Thomson scattering, following the rebuilding and reintroduction of the 5-tip probe to the chamber two days ago. Thoroughly checked around and above the TF magnets before running to make sure everything was normal, following the unusual sounds last week -- no problems were found. Started the run with a low-power TF shot before going to full power. The machine behaved normally all day. Was able to get current ramps and plasma lifetimes near the target shot (89501) by the end of the day, but loop voltage and x-rays are still high.

TS data were collected for shots 92300-92311, except 92308 in which the TS system didn't trigger. The TS computer stopped functioning at shot 92312; it now fails to boots up. The Thomson system cannot be run until this computer is fixed.

Monday October 5 2015 10:47 pm Levesque 92320-92340 Cleanup shots, Thomson scattering

I was able to revive the TS computer last week. Ran TS system during cleanup shots today. Increased resolution on the TS scopes to improve the data collection and analysis, but the storage code was not correctly accounting for this changed scale; the resulting density for shot 92326 is wrong. George modified the code to accommodate the change, as long as the code is restarted after changing the scale. The digitized scope traces do not have the correct offset from the scopes, but the signal amplitudes appear to be correct, so the analysis should work fine. A timing problem intermittently occurred for signals from polychromator #4, causing a small shift in the data which prevents the analysis code from working. This happened for about 4 shots today. The code could be made more robust to handle this if it remains frequent. The vacuum control system was switched from the PLC to the PAC last week, and it's been operating as expected. Cables for controlling the mini turbo will be run this week.

Last week the bias probe circuit was reconfigured to use the single Crown XLS 5000 amplifier, which is necessary for the bias feedback system to work. Part way through today's run, I inserted the probe to 103cm and set the hardware switch to float the probe. The measured floating voltage is as expected, and no problems were encountered. The presence of the probe had little effect on the magnetically-measured equilibrium early in the discharges, but may have had an effect on the Thomson data -- I didn't get enough data to see statistical significance.

Inserted the 5-tip probe to 105cm with an orientation of 45 degrees at the end of the day. The battery for biasing the Isat probes was dead (at 1V instead of near 12V), so I couldn't proceed with data collection. Started charging the battery at the end of the run. After 30 minutes of charging, the battery appeared to be holding a charge.

The shells in Section 2 were fully retracted to insert the 5-tip probe without the possible shell obstruction, then reinserted to their previous (13mm retracted) position after inserting the probe. After this work, I noticed that the "Fault" lights on the Crown amp for CCs in Section 2B were active. I haven't seen this condition before, but it may indicate a problem with the CC hardware for that Section. After turning off the amp, waiting 20 seconds, and turning it back on, the amp appeared normal. The Fault condition returned again within 10 minutes while there had been no activity around the machine.

Wednesday October 7 2015 9:40 pm Levesque/Hughes/Donald 92341-92358 Testing 5-tip probe

Started the run with the 5-tip probe at 105cm in the dual-mach probe configuration. Saw large, wide (~0.5ms) spikes in current and voltage measured from the mach probe tips. The voltages measured at the CPCI inputs got up to 2.8 volts, which should not be possible given the protection diodes in the circuits (these should limit the voltage to ~0.6V). Pulled the probe out to 115cm and progressively moved it into the plasma to look for the onset of current/voltage spikes. Spikes were not present at 107cm (nominal plasma edge for outboard limiting), but were present at 106.5cm. We have not yet been able to explain these spikes. I think the spikes with short duration (~6us) were similar to what Chris occasionally saw when the probe was at certain orientations, but they appeared at all orientations tested today. The voltage on central probe tip E was always positive today, while it had been generally negative for this probe configuration in the past. Note that the correct polarity was verified for the tip biasing.

Probe orientations between 45 and 225 degrees were tested in 45-degree increments while the probe was at 106.5cm. The initial 45 degree orientation produced the fewest spikes.

The control coil amplifier for Section 2B showed no problems today. Currents measured in the relevant control coils were normal for the cases checked. TS data were collected during the run.

Thursday October 8 2015 5:37 pm Hughes 92359-92386 Cleanup, mach probe diagnosis

The purpose of today's run was cleanup shots, as well as diagnosing the peculiar behavior of the mach probe. A 45-degree-step orientation scan begun yesterday (10/7) was completed, covering angles 233 through 0 to 53 degrees on the current alignment gauge. Isat current continues to be generally much lower than expected (~5mA), but with large spikes and occasionally with very large bursts (~1A for 1ms), and tipE V remains small and positive (~5V) for insertions from 106.5 to 106cm (just inside the limiter). Diagnosis will continue.

Shots of Note:
92359: Daily vacuum shot
92362,92363: Higher positive tip E voltage (~+30V) at 278 degrees and 106.5cm... but this is not seen again at the same orientation and 106cm (shots 92375-92386).
92370: Probe insertion changed from 106.5cm to 106cm.
92375: Large current and voltage burst of the characteristic type at ~5ms.


Thursday October 15 2015 4:47 pm Hughes 92387-92402 Testing floating 5-tip probe, cleanup shots

Today's run was intended to test the 5-tip probe with all probe tips floating, as well as taking some cleanup shots. Shots 92387-92392 are low-power OH-only shots to check for ground loops; unfortunately, large spikes (3-5V) are seen on all probe channels even at half-normal OH settings. The problem appears to be somewhere between the input leads of the voltage divider box and the CPCI, since the noise was still present even when the input leads were grounded to the probe stand (92391). All resistances measured seem to be consistent and appropriate. After shot 92391, the cable from the voltage divider box to the North Rack was unplugged at the box, and noise spikes dropped to 10s of mV. Smaller spikes had been appearing on other, unused CPCI channels, and these also dropped by an order of magnitude or more when the cable was unplugged (92392). Shot 92393 was the first full vacuum shot of the day, and following shots were all typical cleanup shots targeting 89501. The 5-tip probe was inserted to 105cm for these shots to take advantage of the opportunity for cleaning the probe.

Monday October 19 2015 9:37 pm Levesque/Hughes 92403-92413 5-tip floating probe testing

Made plasmas with the 5-tip probe measuring floating potential. The probe was positioned at 105cm, 103cm, and 101cm while attempting to measure expected floating potentials around -50V. Measured potentials were generally in the range of -20V to +20V for the shots at 105cm and 103cm, strongly disagreeing with the floating bias probe at 103cm, which measured potentials around -60V. The floating bias probe was then retracted in case it was limiting the plasma, but this caused no change to the 5-tip probe signals. At 101cm, the floating potentials for all tips were fluctuating in the -30V to 0V range, still significantly above what was expected. A reference shot 88014 saw potentials in the range of -60V to -40V for tip E. Shots today generally were not as far outboard as the reference shot, but should have been close enough for measurements to agree. For shots today, all tips saw close to the same values as each other. The exception is probe tip B, which saw much lower signal around 0V all day, and might not actually be connected. For one shot, tip E had a near-zero signal similar to tip B. After moving the probe for the following shot, tip E voltage returned to looking similar to tips A/C/D. This suggests that tip B has not been connected (today at least), and the connection for tip E is intermittent. Only 2 orientations of the probe were tested.

Later in the day, we visually inspected the probe tip by looking through the port at Section 10 with a theodolite. There are no visual abnormalities of the tips. A couple of the angled tips may be sticking out < 0.5mm from the insulator instead of being flush, but we didn't inspect it carefully enough to quantify this. The other angled tips appear flush. The tip precesses by about 5mm as the probe is rotated 360 degrees, which is less motion than before the probe was repaired.

Friday October 23 2015 4:31 pm John/Qian/Josh/Melissa 92414~92415 crates_only shot

Crates_only tests for the bias_probe phase dependent feedback configuration is tested today. The output of AO32 is connected to an oscilloscope instead of the amplifier to moniter if the output is correct.

A brief introduction about the bias probe circuit is given during the test for new graduate students. The GPU is found to be of low coolant again and the reservior is refilled. The modified code had a bug (calling cos() instead of cosf() ) causing the system to freeze and is fixed. The parameters for the output waveform are tested with different values and the observation on the oscilloscope matches expectation.

Tuesday October 27 2015 4:52 pm Hughes 92416-92428 5-tip probe repair testing

Today's objective was to test Jim's repairs to the 5-tip probe MS connector. The good news is that most of the signals are much more reasonably, using Chris's shot 88014 as a reference. Because the probe stand was moved, probe positions are estimated from other measurements and diagrams, but this should be officially calibrated, soon. Shots were taken at approximately 103cm (92418-92424) and at 101cm (92425-92428), at orientations of 8 degrees (92418-92422,92428) and 188 degrees (92423-92427).

Preliminary examination of the raw voltages suggests that two or more probe tips may have been swapped to different MS connector pins, relative to the information on record, but this is based on comparison to data from shot 88014. Tip B appears to be a broken connection, as the noise level and signal on that shot resemble a vacuum shot more than a plasma shot.

As described below, on certain shots, some kind of rather dramatic event takes place which drops the floating potential as seen by all the probe tips and is followed by a rise and sometimes a flat-top of the soft x-ray midplane sensor trace. No obvious correlation appears in any other sensor on the standard hbtplot.py display (the fan array doesn't seem to show the same flat-top that the midplane sensor does), and it's not obvious what would cause this, although there are sometimes large hiccups in the loop voltage at a similar time, which we have seen before. Sometimes the loss of floating potential is gradual, and sometimes very abrupt.

Noteworthy shots:
92417: Daily vacuum shot. 92422, 92426, 92427, 92428: Around 3-4ms, the probe voltages all drop from -40 to -60V to near zero for 1-2ms, before returning to their prior voltages shortly before the disruption. On shots 92426 and 92428 in particular, the SXR-midplane trace flat-tops about 1ms after this loss of floating potential.

Tomorrow, we intend to inspect the probe tip again for signs of melting or burning, before conducting any more testing along these lines.

Friday November 6 2015 5:53 pm Brooks/Qian 92429 92469

Executive Summary:

The feedback control system was modified to drive the bias probe at the same frequency of the dominant mode in the system. Today's experiments proved that the system is working to that effect, a compass scan was performed to assist in determining the effect bias probe feedback system, and the data will be analyzed next week.

Goals:

1) Show that the feedback control system was setup correctly to provide mode-phase information to drive the bias probe with a appropriate oscillatory response.

2) Determine if the mode locked oscillatory response of the bias probe had an effect and if so, what kind.

3) Provide me (John) with experience operating the system and analyzing the data.

Tests performed:

1. The first sets of shots were to determine the correct voltage settings for the coils to provide the ideal system response. While the system was calibrated so that dominant 3/1 mode consistently observed, there was some variation in the length and timing of these modes.

2. Next, the bias probe was tested with a non-phase locked cosine wave to demonstrate that the code was working. After this, the bias probe was setup to provide a constant -40 and -80 volt signal. The goal of these tests was to further verify that the code was working (which it was) and to see if the system responded to the bias probe (which it seemed to do). Further analysis is required.

3. Next, the phase-feedback feature of the control system was turned on, and the bias probe appeared to generate the correct oscillatory response and have some effect on the system. The bias probe offset was -80V with +-20V oscillatory swings.

4. Lastly, a compass scan was performed with 90deg spacing and with three measurements taken at each location. As before, the bias probe offset was -80V with +-20V oscillatory swings. As a control, three additional measurements were taken at -80V without any oscillatory swings.

Results:

No official results can be provided because data analysis will not be performed until next week. However, I can make a few qualitative comments. First, the bias-probe feedback system appears to be working, and the system appears to have been influenced by it. Second, the mode amplitudes and durations were somewhat inconsistent so it's possible that the data may be less conclusive than ideal. Lastly, I (John) gained quite a bit of experience working with the machine and feel much more comfortable concerning its operation, setup, and measurements.

Wednesday November 11 2015 6:06 pm Donald/Hughes/Levesque 92470-92530 TS Data Collection

The purpose of the run was to characterize plasma temperatures and densities across a range of time periods using Thomson Scattering. Target shots were 84046 and 89199.Reasonable data points were recorded for 7 seperate time intervals with 4, 5 or 6 data sets being taken for each interval. Time intervals ranged from 1.5ms to 6ms.

It was interesting to note that a strange type of mode locking was observed for a number of shots - examples of this can be seen in shot numbers 92514 and 92528. Typically the plasma locks at ~2.5-3.5ms, followed by a strong rotating mode (on some occasions the mode briefly locks a second time). Not sure what has caused this as no RMP or biases were applied.

Friday December 11 2015 5:38 pm Hughes/Levesque 925331-92562 Cleanup shots + spectrometer testing

Today's run was mostly cleanup shots following the brief up-to-air for installing the new port in section 5, as well as ~3 weeks downtime. As in typical cleanup runs, shots started very weak at the beginning of the day and grew out to nearly 6ms by the end, although the plasma current ramp remained very weak despite using nearly the maximum OHE.

The new spectrometer was also tested for the first time looking at plasma light, and seems to be working well, showing strong D-alpha and D-beta, but also some strong carbon lines, as is typical following up-to-air. The spectrometer was able to capture 2-3 spectra per shot, depending on shot length. An animated gif of intensity vs. wavelength and time can be found here. In principle, a normal, healthy plasma could be expected to have 3-5 spectra recorded.

Cleanup shots will continue early next week. We intend to calibrate the actual integration window limits of the spectrometer with a xenon lamp in the near future, as well. Once this is done, the timing of the spectrometer triggers may be changed, although 2.0ms seems to be the lower boundary on triggering interval.

Noteworthy shots:
92539: Daily vacuum shot.

Monday December 14 2015 3:22 pm Donald/Hughes 92563-92588 Cleanup shots

Today was the second day of cleanup shots since resuming post-APS, post-up-to-air operations. Although plasmas were well behaved and relatively good, including some disrupting later than 6ms, most were shorter-lived, and many were slow to heat up, perhaps due to the unusually high base pressure (~4.5x10^-8 on the gauge). The lab was exceptionally hot, reaching 85+F in the control room, and hotter in the server room, screen room, and machine hall. The TS spectrometers reached 99F, and the more heavily-loaded cryopump effectively stopped pumping, along with a string of camac errors including the familiar "x=0" heat-related errors in the North rack, and repeated triggering failures on the Flame spectrometer. The run day ended early in order to regen the loaded cryo pump, hoping for better vacuum conditions (and less heating problems) tomorrow.

Noteworthy shots:
92563: Daily vacuum shot
92585: Typical shot with anomalously low SXR-midplane signal up to ~2ms.


Wednesday December 16 2015 5:38 pm Hughes/Donald 92589-92605 Cleanup shots + spectrometer

Today's run was a continuation of cleanup shots and spectrometer recording. The lab was very hot again in the morning, so the run only began after the end of the weekly meeting at 2pm, after things had cooled significantly, hence only a few plasmas. The spectrometer triggered for nearly all plasma shots in this run, but there were some peculiar triggering problems including skipped trigger pulses and stretched trigger intervals. The bias probe was inserted, as well, to begin cleaning it following the up-to-air and extended down time. Plasmas generally developed higher plasma currents and stronger plasma current ramps for the same bank settings than on Monday and Friday, suggesting the cleanup campaign is making progress.

Noteworthy shots:
92589: Daily vacuum shot.
92596: Longest-lived plasma of the day.
92601,92603: Plasmas with anomalously low SXR midplane signals, lasting long into the shot. These are also short-lived plasmas and have slightly weaker plasma currents. It's not obvious what distinguishes them, yet.

Thursday December 17 2015 6:47 pm Hughes 92606-92635 Cleanup shots + spectrometer

More cleanup shots. Although only two shots failed to trigger the spectrometer at all, several had individual failed triggers during the shot. The cause of this behavior has not yet been identified. The carbon line does appear to be lower than it was on Friday (Dec. 11th), as expected. SXR-midplane traces were generally close to the pre-APS target shot, although a number of shots (most notably 92629) have very low SXR-midplane signals for much or all of the plasma lifetime; these plasmas tend to be shorter-lived, and most also show lower TS temperatures, although noise issues have been plaguing the precision of TS measurements. An animation of the run day's spectrometer data is available here.

A problem with the 96cm TS line was resolved by correcting an erroneous scope trigger threshold, which has been causing a timing error and pegging the temperature and density readings to zero. The 96cm line seems to be working as of 10 shots after the fix.

Noteworthy shots:
92606: Daily vacuum shot.
92629: Extremely low SXR midplane levels, as well as very low TS temperature at 2ms.

Friday December 18 2015 4:58 pm Hughes 92636-92665 Cleanup shots + spectrometer

Continuation of the cleanup shot series targeting pre-APS shot 92519. There were a number of "x=0" failures in the screen room A14, which plagued the run day. There was also a relatively large number of low SXR-midplane shots, which may or may not be related to low fill pressures, as they seemed to become more frequent later in the day, and the puff pulse battery was at the lower end of acceptable voltages (12.8V) at the beginning of the day. In addition to the bias probe, the 5-tip probe has been inserted for cleanup, and both are showing reasonable floating potentials, higher on the bias probe (~-60V) than on the 5-tip probe center tip (~-40V). The 5-tip probe's insertion gauge needs to be recalibrated, however, so they may not be at the same depth. Flame spectrometer data was recorded, but could not be analyzed yet, since it currently needs to be transferred manually by thumb drive.

Noteworthy shots:
92636: Daily vacuum shot.
92643,92645,92655,92658,92659,92660,92665: Screen room A14 failures. 92642,92653,92657,92662,92663,92664: Low SXR-midplane shots. Typically also have lower TS temperature, but loop voltages very similar to target shots.

Wednesday December 23 2015 1:41 pm Hughes 92666-92680 Spectrometer testing

Today's run was crates-only to study the integration window of the Flame spectrometer using the xenon flash lamp. The flash lamp was triggered from the screen room crate J221 off of input 7. Three sets of crates-only shots were taken, with a range of Dt values, where Dt=(t_flash - t_spect) in microseconds.

For the first set (92666-92668), Dt=+-750,+-500,+-250,0.
For the second set (92669-92674), Dt=-1750,-1500,-1250,-1000,-750,-500,-250,0,+250,+500.
For the third set (92678-92680), Dt=-200,-150,-100,-50,0,+50,+100,+150,+200.

The result is that while light from a flash triggered at Dt=0 is seen as expected, although signal strength reduces somewhat as Dt approaches 0 from below; however, a flash triggered at Dt=+50us or more is seen on the next spectrometer trigger, with little or no loss of signal strength. This suggests the CCD dumps and resets in less than 50us, and integrates more or less evenly over very nearly the entire time between trigger pulses.

Some confusion remains regarding the internal spectrometer timebase, which can record trigger times up to +-300us off from the J221 trigger time settings, or sometimes skips 1-2 triggers. However, when the spectrometer's internal timebase claims to have triggered late or early (e.g. 18.3ms rather than 18.0ms), the resulting spectrum still shows the expected peaks, as though the trigger had been on time; on the other hand, when the internal timebase registers skipped triggers (e.g. 20ms to 34ms, rather than 20ms to 24ms), the spectra observed seem to make sense only if the internal timebase is correct.

This may partly be on the software end, since the software currently in use is known to be extremely flakey, including a history of failed triggers, failed recording, and regular crashing. For standard operation, it will definitely be necessary to get or create new interface software.

Wednesday January 6 2016 4:54 pm Hughes 92681-92710 Cleanup and shot development

Today's run was intended to be cleanup and shot development, but the behavior of the plasmas suggests that cleanup needs to be the priority, right now. Comparison to older shots from the Ferritic Wall study show that both loop voltage and SXR midplane traces are much higher now than they were previously, and a considerably higher OHE setting is necessary to replicate the plasma current trace, all suggesting impurities remain quite high. This may be partly due to issues with the west cryo pump, which has been observed at an elevated temperature (~20K this morning) and may be pumping less effectively. However, it should be noted that the SXR and loop voltage traces are similar to those seen in the week prior to APS (e.g. 92519). Cleanup and shot development as necessary will continue tomorrow.

Noteworthy shots:
92681: Daily vacuum shot.
92708: Best shot of the day, relative to target shot 92519 (prior to APS), a more recent proxy for ferritic wall study shots 89140 (Fe) and 89203 (SS).


Thursday January 7 2016 4:59 pm Hughes 92711-92741 Cleanup and shot development

Today continued the cleanup/shot-development run from yesterday, targeting 89140 and 89203, using 92519 (from just before APS) as a benchmark for recent performance. Similar problems persist, where higher bank settings are needed to produce similar plasma parameters, although this has improved somewhat since yesterday. SXR-midplane traces also remain high (e.g. 92713), although many exhibit an intermediate behavior where the SXR traces is low until around 2ms and then rises (e.g. 92725); a rare few have SXR traces similar to the target shots (92740), but it has prove difficult to get this behavior from good plasmas, and these seems to be no obvious correlation to bank settings, plasma current, or major radius. Loop voltages remain considerably higher on all recent shots, including those with SXR traces close to the target shots.

We performed a TF-only listen-check for the first time in months (shots 92714 and 92715), just in case this peculiar behavior is caused by loose ferritic material in-vessel, but no untoward mechanical sounds (rattling, banging, or scraping) were heard.

Noteworthy Shots:
92711: Daily vacuum shot.
92725: Good (close to target) early SXR which rises later.
92740: SXR trace close to target shots.


Friday January 8 2016 6:50 pm Hughes 92742-92789 Cleanup shots

Today's purpose was to see if cleanup shots with more aggressive settings could bring plasma performance closer to the 89140 (Fe) and 89203 (SS) style shots from the ferritic wall growth rate study in March of 2015. Plasmas with start plasma current around 15kA and either flat-topped or rising to about 18kA were developed using OH Bias up to 4kV (normally run at 2kV), which may also be of interest as the Thomson scattering system becomes more mature (i.e. fully calibrated, more viewing lines in use). It isn't clear yet that this approach to cleanup has been very different from more "regular" cleanup shots, but this may be judged best by plasma performance on Monday.

Toward the end of the run day, the DI water in the TS laser power supply had to be refilled; the laser safety interlocks had stopped the laser as intended. Additionally, there appears to be an open-circuit fault in the 2T-midplane control coil. The fuse was checked and appears to be all right, suggesting a problem in the coil or cable. We don't know yet if this problem is in-vessel. The west cryo pump is currently at 30K, notably higher than it should be, but the vacuum quality has been typical of a rigorous cleanup day. Flame spectrometer data was taken for most of the shots, today, and there were a few impurity lines which (normalized against D-alpha) were higher than had been seen in more standard shot styles.

Noteworthy shots:
92749-92784: High-power shots. 92780 is an especially good example. Note that the SXR midplane amplifier is consistently saturated (but not the digitizer).
92787,92788: Two shots with the same settings and exhibiting the different SXR midplane behaviors. Note that loop voltage is very similar even as SXR deviates.

Monday January 11 2016 4:51 pm Hughes 92790-92805 Shot development

This run was meant for shot development with the bias probe and 5-tip probe (in dual Mach configuration) inserted, as well as investigating changes in performance following last week's aggressive cleanup day. However, the 5-tip probe showed unexpected and anomalous signals, including some very large currents (about 1.6 in unspecified units with a calibration factor of 2.0 in the tree) on Tip E while the Tip D current is approximately 1/4 of Tip E, and the Tip B current resembling the negative of the Tip A current for the first half of the shot but resembling positive Tip A for the second half. After seeing these seemingly very large currents (more than an order of magnitude larger than what seems to be expected judging by the preset axes in hbtplot.py), it was decided to retract the 5-tip probe until its wiring can be examined in more detail.

It was also tested whether the bias probe or the ferritic wall might be responsible for the elevated loop voltage and soft x-ray signals, perhaps due to either introducing additional impurities to the plasma, or changing its limiting characteristics, but retracting each separately seemed to have little or no effect on the traces in question. After retracting the ferritic wall, it became difficult to maintain a steady major radius with the bank settings used previously, but only a few shots were taken. Further attempts to target 89140 and 89203 will continue on the next run day.

Noteworthy shots:
92795: Shot with very high Tip E and Tip D currents, as well as anomalous Tip A and Tip B relationship.

Thursday January 14 2016 5:25 pm Hughes 92808-92833 Cleanup & shot development

Today's run was to test whether plasmas useful for research could be produced at bank settings similar to those in the March series of ferritic wall studies. It seems clear that plasmas with similar parameters require much higher bank settings, which accounts for about 2/3 of the difference in loop voltage, particularly from the much higher OH electrolytic bank voltage (>320 as compared with ~240, in March). Floating potential on the bias probe seems similar to what it generally has been. The trend persists of high SXR midplane signals on most plasmas, with occasional plasmas exhibiting SXR signals closer to the targets but with relatively poor plasma performance, generally falling inward despite similar plasma currents.

Noteworthy Shots:
92810: A plasma with bank settings similar to the target shots. SXR midplane is higher, and plasma falls in rapidly; starting plasma current is somewhat higher, but with almost no I_P ramp. Note that the loop voltage is also considerably higher than on the target shots, despite nearly identical bank settings.
92832: The OHE power supply failed to charge all the way to the set point before the TF finished charging and the phidget auto-fired; even so, OH current and plasma current are higher on this shot. The OHE power supply voltage was turned up after this shot, but it's possible the knob had been bumped.
92833: SXR midplane matches targets, but plasma falls inward, as is usual on recent shots.

Friday January 15 2016 5:11 pm Hughes 92834-92840 5-tip probe testing

The purpose of today's run was to test performance of the 5-tip probe after an overhaul. The results are relatively similar to previous 5-tip probe tests, in that the interpretation of the data is very unclear. Tips A and B continue to show mirrored vacuum pickup, but almost identical signals, otherwise. With the probe bias supply off or at low voltage (~70V), these signals are at reasonably low levels (in relation to the plot axes preset by Chris when he was using the probe), with modest oscillations; with the bias supply at higher voltage (~150V), the signals become far more erratic. There may be remaining ground problems in the cabling of the probe electronics, or other as-yet undiscovered issues. The calibration is also unclear, displaying numbers with no units (I presume mA, here, but this is only a guess, until we verify the calibration).

Noteworthy shots:
92838: Probe inserted to 103cm, bias supply turned off. Tip E voltage looks like floating potential, small I_sat current (I assume this is the sum of currents from tips A-D).
92839: Probe bias supply turned on to about 70V. Tip E voltage still resembles floating potential, but I_sat current is about three times higher.
92840: Probe bias supply turned on to about 150V. Both tip E voltage and I_sat exhibit wild swings; individual probe tip currents show the strange plateau and broad spike behavior from before.

Wednesday January 20 2016 5:16 pm Hughes 92841-92861 5-tip probe testing

Today's run was to test whether the peculiar behavior of the 5-tip probe in Isat configuration was orientation-dependent. Although the individual tip behavior did change, the general trend remained the same, including very large bursts of current on tip D, tips A and B, and/or tip E (possibly arcs). Tip C remained at low signal levels on all shots. Tips A and B still exhibit identical or nearly identical real currents and mirrored magnetic pickup for reasons currently unknown. There were also delays due to some digitizer failures which were rectified by power cycling the crates. The 5-tip probe is current retracted (as of shot 92855) completely from the vacuum vessel due to its ongoing bad behavior.

A few additional shots were taken (92855-92861) to see if the probe's presence was a large factor in the poor quality of recent plasmas, but the current ramp remains anomalously weak and soft x-rays and loop voltage both high relative to target shots 89140 (Fe) and 89203 (SS).

Wednesday January 27 2016 4:37 pm Hughes 92862-92883 Bias probe testing

Today's run was geared toward testing the bias probe, since it hasn't been operated in some time. There were some initial problems with the setup routine and hardware issues, but the bias probe does seem to be performing as expected, and is apparently ready for tomorrow's bias probe feedback run.

Today was also the first run day since the ground loop corrections from Friday (1/22), so all the Mirnov sensors, control coils, and other standard diagnostic traces were checked to make sure there were no newly anomalous signals created, and none have been found. Additionally, the puff line was purged due to a concern that it might have been contaminated with air, and plasma performance on the few plasma shots today was considerably better than in other recent run days, achieving similar plasma currents to the target shots (89140 and 89203) with an OHE setting of 270, rather than the 320-340 required the last few weeks (compare about 240 on those target shots). Loop voltage and SXR midplane traces are also closer to target but not quite there. Another puff line purge may be in order, since the line was at one point left under vacuum.

Relatedly, the ion gauge digitization has failed, possibly due to a number of wiring projects going on in recent days, including Friday's ground loop work. This will hopefully be fixed either tomorrow or Friday.

Noteworthy Shots:
92883: Perhaps the nearest-to-target shot of the last few weeks, with strong plasma current, fairly steady R₀, and less-elevated SXR midplane and loop voltage.

Thursday January 28 2016 4:57 pm Brooks/Hughes 92885-92901 Bias probe feedback

Today, Paul and I setup and performed phase-locked bias probe feedback. The main goal was to familiarize John and Josh with setup/shutdown as well as shot development. In addition, John tested his updated bias probe feedback code. The code was debugged with the exception of a "stair-step" like ramp which requires further work. Using this code and after Paul performed shot development, several feedback tests were performed. Oscillations of the bias probe current were observed, but an insufficient number of tests were conducted to allow for detailed analysis. Further testing will be conducted next week.

The most relevant feedback shots were 92900 (0 phase offset) and 92901 (for a 0 amplitude feedback).

Monday February 1 2016 5:08 pm Hughes 92904-92926 Shot development

Today's run was intended to develop a shot style similar to shots 89140 and 89203 from last March's ferritic wall growth rate campaign, in order to study the effect of rotation on stainless and ferritic configuration mode dynamics. The shot style was achieved relatively quickly, with nearly identical SXR, I_P and q*, steady R₀ between 91 and 92cm, and only slightly higher bank settings and loop voltage. A range of bias probe voltages (-2, 0, +1, +2V at the amp input, roughly corresponding to -60, 0, +30, +60V on the probe tip) were used in order to study the effect of the bias probe on mode rotation in the context of this shot style. Analysis will follow, although strong bias does seem to interfere with mode growth.

Noteworthy Shots:
92917-92919: Strong positive bias (+2V at amp input, about +60V on probe).
92912-92914,92925-92926: Bias probe floating (-2V at amp input, about -60V on probe).

Thursday February 4 2016 5:32 pm Brooks/Hughes 92927-92954 Bias probe feedback

The goal of today's shots were to continue to train John in the operation of HBTEP in addition to John's bias probe feedback research. Also, Josh used several shots during the day to diagnose the 5-tip probe.

The results from John's tests were that the Bias Probe feedback continues to be able to phase lock with the dominant modes and produce an output. Initially, the results appear inconsistent, but further data analysis will be conducted to study the bias probes ability to speed up/down the plasma in addition to mode excitation/dampening. Shot numbers of note were: 92939 (feedback on with 0 deg phase offset), 92951 (270 deg phase offset), 92937 (flat bias probe signal), 92948 (90 deg phase offset).

A few crates-off shots were performed at the end of the day, and the bias probe ramp issue was fixed. The code was performing integer divisions instead of the appropriate float divisions.

Tuesday February 9 2016 4:51 pm Hughes 92955-92986 Shot dev. & slowed FWM study

Today's run was oriented toward shot development for the slowed FWM study, but plasmas were well enough behaved early in the day to take preliminary data for the "stainless configuration" (Fe walls retracted to stoppers, SS walls retracted 13mm from full insertion). Plasmas continue to show the behavior seen previously, where modes would be relatively quiescent (1-3 G on TAp) around 2-4ms, during which time mode frequency was strongly dependent on bias probe voltage, but a large mode (~10G) would grow and simultaneously spin up at around 4ms, reaching typical rotation speeds in both floating-potential (-60V) and slowed (+45V) configurations. However, the spontaneous rotation in the "slowed" state did seem to generally be slower (~5kHz) than the rotation observed in the "floating" state (~8kHz), but the effect was far less dramatic than the difference in rotation during the 2-4ms quiescent period. "Slowed" plasmas showed significant weaker modes (1G or less) during this period, however, than "floating" plasmas.

Noteworthy shots:
92965: Star "slowed" plasma. Other best-quality shots in this set are 92966, 92968, and 92984.
92963: Star "floating" plasma. Other best-quality shots in this set are 92962 and 92977.

Wednesday February 10 2016 4:49 pm Hughes 92987-93020 Shot dev. & slowed FWM study

Today's run focused on the "ferritic" configuration with the bias probe slowing (+45V) versus floating (-60V). Fortunately, relatively little shot development was needed to recover good plasmas after swapping shell configurations, between shots 92994 and 92995. Notably, SXR midplane signals remain similar to the "stainless" configuration, but D-alpha signal is considerably higher, when all other plasma parameters are similar. Shots seem to disrupt early more often, but have also been slightly less cooperative than they were this morning, prior to changing shell configuration. Differences in mode rotation between the shot types is somewhat less clear than in the stainless configuration, but mode onset seems at first glance to be considerably earlier in time (at high q*).

Noteworthy shots:
92996: Disrupted early, following several dramatic events (possibly internal) from about 2.5ms to 4.5ms. 93010: Example of a good "slowed" type (probe bias +45V) shot. Others include 93009, 93013, and 93019. 93016 Example of a good "floating" type (probe bias -60V) shot. Others include 93003, 93005, 93006, and 93014.


Thursday February 11 2016 4:48 pm Hughes 93021-93042 Slowed FWM study

Today's run was a continuing of yesterday's bias probe campaign comparing "slowed" and "floating" plasma with the ferritic wall inserted and stainless walls retracted. The target shot for the run was 92925. Similar behavior was observed, with "slowed" plasmas generally exhibiting mode growth at earlier times and higher q* values. Six good shots (judged on the basis of plasma parameters closely matching the target shot) of each type were taken, as well as a few less consistent shots that drifted in or out in major radius.

Noteworthy shots:
93039: Example good "floating" shot with the bias probe voltage at -60V. Others include 93022, 93026, 93027, 93032, and 93033
92041: Example good "slowed" shot with the bias probe voltage at +45V. Others include 93029, 93030, 93031, 93040, and 93042.

Tuesday February 23 2016 1:02 pm Hughes 93043-93047 Fe wall w/ slowed modes

Today's run began late and ended early, and produced very few shots. The intention was to begin looking at lower fill pressures, to see if good plasmas could be made, and whether it would be easier to slow mode rotation with the bias probe in such plasmas. The prior runs were at a typical fill pressure of about 75uTorr; today, fairly good plasmas down to 65uTorr were made. One plasma was made at 59uTorr fill pressure, but the breakdown was very poor, with about 30% lower start I_P and much, much lower SXR signal for similar bank settings. There was not enough time to develop a shot properly, however, or to test the effectiveness of the bias probe, which was at the nominal floating potential during each shot.

Noteworthy shots:
93044: A shot at 71uTorr in which large (30-60V) swings can be seen on the bias probe voltage as the plasma falls radially inward.

Wednesday February 24 2016 4:55 pm Hughes 93049-93069 Slowed plasma exploration

Today's run was intended to seek methods of improving the bias probe's control of plasma rotation. Three main categories of plasmas were explored, albeit briefly: low fill pressure (hoping for reduced neutral drag making the plasma more responsive to probe bias), high probe bias (up to about +105V), and the high-pressure, high-current style from Chris' thesis. Although some of these are very interesting plasmas that would be good to investigate in more detail, all were relatively similar in terms of bias probe effectiveness: the bias probe was moderately or very effective early in the shot when the modes are strongly suppressed in positively biased cases, but shortly after crossing q*=3 from above, the plasma would spin up to 6-8kHz regardless of bias probe voltage.

For the record, plasmas with fill pressures ranging from 60uTorr to 118uTorr were made today. Although not obviously relevant to the intended research pursuits, it is worth noting that one major difference between high and low fill pressure was that the soft x-ray trace rose much earlier with a concave-down curve in the high-pressure case, and much more slowly with a concave-up curve in the low-pressure case, although (in long-lived shots, at least) they saturated to similar levels. The D-alpha light looks very similar everywhere except at breakdown, where a lower fill pressure produces less breakdown light.

Thursday March 3 2016 5:34 pm Hughes/Levesque/Cabrera 93070-93092 5-tip probe repair testing

Today's run was focused on testing a repair to the 5-tip probe current measurements' voltage clamping diodes. Diodes are installed in BNC boxes, each with 5x higher diode clamping voltage (both positive and negative) than the old boxes, and a 50-ohm resistor (up from direct short) on the probe side of the diodes.

Switching to the new boxes for tips E and D did change the character of large spikes and saturated high level (~2V) signals somewhat, but did not prevent them completely. Moving the BNC boxes to connect directly to the Pearsons rather than being installed inside the rack seemed to improve performance and remove the anomalous signals--but moving the boxes back again did not produce those signals, so the box position does not appear to be the sole determining factor at this point.

Some shot development for the upcoming/ongoing slowed FWM study was also done. The Chris-style shot (targeting 85708) has some good properties, but q* becomes low later in time (below 2.5), leading to an internal 2-1 rather than an external 3-1; attempts to develop a shot with lower peak IP and higher q* (~2.7?) to study a growing external mode have not yet settled on a good shot style.

Noteworthy shots:
93077,93078: 5-tip probe unchanged, in I_sat configuration. Big spikes show up, at breakdown or at disruption. Tip E box is at rack, tip D box is at Pearson.
93079,93089: New BNC boxes, tip E at rack, tip D at Pearson.
93080, 93086: New BNC boxes, tip E at and tip D at Pearsons.
93083: Old BNC boxes, tip E at and tip D at Pearsons.


Monday March 7 2016 4:56 pm Hughes 93093-93122 Slowed FWM study

Today's run was intended to target plasmas with a strong current ramp, steady major radius near 92cm but not outboard limited, and growth of an external 3-1 mode at q* below 3 without large MHD bursts at the q*=3 crossing. Shots were taken with the bias probe forced-floating (driven to the measured typical floating potential) and biased at V_float+120V. The shot style (currently targeting 93096, although a steadier major radius nearer 92 is preferable) proved relatively reproducible, although some analysis will be needed to gauge whether there is a significant difference in mode rotation between the two cases, using this shot style.

Noteworthy shots:
93096: Current target shot. Sets lower bound for good major radius for this shot style.
93096,93102,93103,93111,93118,93119: Best shots for "floating" probe.
93104,93106,93110,93120,93121: Best shots for "biased" probe. 93107 and 93108 may also be OK, but scrape outboard limiter.

Tuesday March 8 2016 5:10 pm Hughes 93123-93147 Slowed FWM study

Today's run continued the slowed FWM study campaign from yesterday, expanding the set of floating and biased plasmas to include more strongly biased (V_float+150V) plasmas. Shots were also taken with stronger negative bias (V_float-45V) but this had very little if any effect, and the bias probe was only able to drive about 10% of the current drive in the positive bias configurations. Plasmas have been well-behaved and quite reproducible in terms of equilibrium parameters, although mode growth and frequency remain unpredictable.

Noteworthy shots:
93124,92125,92126,93136,93137,93145,93147: Biased to V_float.
93127,93130,93132,93133: Biased to V_float-45V. Does not seem to have much effect on mode frequency, relative to floating case.
93138,93139,93140,93141,93142,94143: Biased to V_float+150V. Mode frequencies seem generally slower than V_float+120V shots from yesterday.


Wednesday March 9 2016 5:02 pm Hughes 93156-93205 Slowed FWM study

Today's run continued the slowed FWM study series with the ferritic walls retracted (to Fe retraction stoppers) and stainless walls inserted (back 13mm from insertion stoppers). Plasmas were not very cooperative, requiring considerably higher OHE settings (~300V up from 270V) to achieve similar plasma current traces; as a result, the plasma current ramp changes more in today's shots than in the previous shots of this campaign. Early soft x-rays are also slightly lower, and (due to the higher OH settings) loop voltage is higher. Natural rotation appears to be slower with the stainless walls in than with the ferritic walls in, as well. Bias probe voltages of -45V to +145V relative to floating potential were applied Although q* traces do not match as well as those on the previous run days, the best shots of today (in terms of equilibrium parameters) are listed below..

Noteworthy shots:
93183,93185,93187,93195,93196,93202,93203: V_bias = V_float.
93188,93189,93190,93204: V_bias = V_float+120V.
93192: V_bias = V_float+145V.
93197,93199,93200,93201: V_bias = V_float-45V.


Friday March 11 2016 4:41 pm Brooks 93215-93228 Bias Probe feedback

Tested new bias probe feedback code (which appears to work) and ran a number of tests using the code. A large portion of today was calibrating the feedback parameters to trigger in the presence of a suitable mode. Particularly, these parameters governed how large a mode and how long lived a mode should be to trigger the feedback.

Spent some time hunting for a regime where plasma instabilities are large and consistent but had trouble finding such a regime. Large mode amplitudes consistently developed when the major radius exceeded 92cm, but I'm not sure if this a good plasma to use for the study.

The new feedback code, once initiated, didn't appear to show any immediate response in the plasma.

Run 93220 was an interesting run in that mode amplitudes appeared to increase after the feedback was turned on. Thanks to Paul for his help today.

Tuesday March 15 2016 5:25 pm Brooks 93229-93242

The goal of today's run was to reproduce one of Qian's previous runs, 87980, which was characterized by consistent mode instabilities and a major radius consistently near 92cm. The secondary goal was to continue testing the bias probe feedback code.

I spent most of the day learning shot-development but never reproduced the instabilities in the target shot number. Part of this was due to my inexperience and possibly due to the poor chamber conditions.

As an aside, I'm struggling with python library dependencies. This needs to be resolved in order for me to run hbtshot.py and a few other tools.

Thanks to Paul and Jeff for teaching shot development today.

Tuesday March 22 2016 5:12 pm Abler 93288-93300 Rogowski Tests

Ran "low" and "normal" bank levels for VF only, OH only, and TF only, duplicating each shot. Also ran one shot with the mid plane (short lead) Rogowski disconnected from section 4 to check for pickup, which appeared to be minimal. A polarity test of the Rogowskis still needs to be conducted - hopefully tomorrow afternoon. Further analysis will be forthcoming in the next day or two. Huge thanks to Paul and Nick for all the help today! -Mel

Thursday March 24 2016 5:06 pm Brooks 93312-93351 BP Feedback

The goals today were to continue training shot development in addition to reproducing one Qian's shots. I reproduced the shot to the best of my ability, but we decided to pursue a more exploratory shot development to find the desired mode types. After a bit of searching, Shot numbers 93348 and 93351 both show fairly repeatable mode growth with a major radius near 92cm. The downside is that the modes appear to be 2-1 instead of 3-1.

Also of note is that Jeff disconnected a portion of the PA1 sensor array and replaced it with Mel's Regowskis. More specifically, Jeff disconnected "BOT3" HD magnetic sensors at Section 3 in the north amplifier box. They were disconnected before shot 93329 and have not yet been reconnected.

Tuesday March 29 2016 5:00 pm Abler 93388-93421 Rogowski Tests

The goal of the run was to look at currents across the 3/4 quartz break now that the Rogowskis are better shielded. "Low," "normal," and "high" VF only and OH only shots were taken (in triplicate), along with "normal" TF only shots (also in triplicate). 7 plasma shots were also taken: 3 pseudo-clean-up (93415-93417), one trying to crash the plasma inboard (93418), one trying to crash the plasma outboard (93419), one "high power" (93420), and one low power (93421). I'll send an email once the revised Rogowski calibration is done (ideally tomorrow night) and has been applied to these shots. Many thanks to Paul for the guidance on bank settings. ~Mel

Thursday March 31 2016 5:07 pm Brooks 93422-93464 BP Feedback

Summary:

I found shot style that was characterized by coherent modes, relatively flat major radius near 92cm, fairly repeatable shots, and a mix of 3-1 and 2-1 modes. After discovery, I performed a phase sweep using the bias probe feedback over a number of phase offsets and with feedback off. I managed 4 duplicates of every phase setting. Preliminary results suggests that I may find something, but I won't begin analysis until next week.

Goal:

The goal of today's run was to discover an ideal shot style and perform a phase sweep using the bias probe / GPU feedback system. The ideal shots would be characterized by coherent modes lasting an appreciable amount of time (>3 ms would be nice), a relatively flat major radius near 92cm, and being repeatable.

Repeatability:

The data recorded today consisted of very similar safety factors and mode growth. However, the major radii often drifted and was periodically characterized by an intermittent and rapid major radius decay. Care will have to be taken when comparing shots.

Startup issue:

The ignitron water cooling interlock didn't trip correctly, and the water flow needed to be turned up.

Tuesday April 5 2016 5:22 pm Abler 93508-93537 Rogowski Tests

The run day started with a quick crates-only 5ms "square pulse" test to use for time domain analysis (93508-93510). The rest of the run was split between vacuum shots with both, either, or neither strap connected across the 3/4 quartz break (93514-93525) and plasma shots with both, either, or neither strap connected across the 3/4 quartz break (93526-93537). Cursory examination of shots shows at least some similar-looking plasmas (similar Ip, major radius, and mode activity) with different strap attachment configurations, which should allow for a good comparison between those cases. In the (likely) event that there are not enough comparable shots to allow comparison of all four strap combinations, I'll run again Thursday to try to round out the data set. Calibrated results will be emailed out and put on the Wiki ASAP.

Thursday June 2 2016 2:14 pm Abler 93667-93679 RMP/Halo Currents

Today we attempted to continue RMP work with all shells in. During initial shot development, some shots failed to break down (93673 and 93678), and those that did make plasmas died very quickly (under 4 ms), forming inboard (around or inboard of 90cm) and crashing further inboard right away. Initial attempts at correcting the problem by charging the e-gun and puff batteries again, altering puff timing, and altering e-gun current all failed to improve performance. The puff battery is fully charged, and the e-gun battery will be left charging overnight. The hope is that this will remedy the situation, since Mel, Pat, and Jeff ran out of ideas for other (quick) fixes.

Tuesday June 7 2016 4:37 pm Hughes 93708-93711 Fe wall limiting tests - interrupted

It was discovered on the first plasma shot of today's run (93709) that there was a small but substantial arc occurring on the OHEL bank, on the left-hand half of the bank, in the 12th row up and 10th row right. Apparently, a connection from the cap to the rail was loose. Besides burn marks on insulation, the only substantial damage caused was to weld the bolt to the rail, which Jim has already repaired. Reviewing security camera footage (available back to Wednesday) indicates that this arc has been present since at least the beginning of the run day on Thursday. However, based on shot 93709, it does not appear to have affected plasma performance in an obvious way, and if it has any effect on the OH traces as viewed in jScope, the effect is very small. The damage from this minor arc is already repaired. However, there is evidence on the cameras of some other, smaller arcs, as well--possibly broken/damaged wires making resistive contact--and other lose connections were noticed, suggesting that we should invest some time in checking at least the OHEL bank for other loose or broken connections. Checking over the OHEL bank and conducting repairs consumed the bulk of this run day, so the run will be rescheduled for another block of time.

Thursday June 9 2016 3:17 pm Abler 93712-93754 Rogowski Calibration and Bank Checks

Shots 93712-93752 were used to re-calibrate the Rogowski coils after discovering that power to the amp board was shoddy during the previous calibration. Shot 93753 was an all banks vacuum shot - the OHE and OHS banks were both observed to light up on this shot in at least one location. Mel and Jeff investigated, but didn't see any signs of charring and all fuses seemed intact. After some deliberation, shot 93754 was taken to see if the problems were repeatable; both banks lit up again. The decision was made to shut down for the day, with plans to clean all cap leads on the OHE and OHS banks tomorrow. When Mel entered the cap room after shot 93754, it smelled intensely like burning plastic - the source (based on smell) seemed to be the OHE bank, but again, no charring could be found. Further investigation will be required tomorrow.

Wednesday June 15 2016 4:52 pm Brooks 93762-93763 OH Bank testing and Cleaning shots

Repaired the OH banks and ran a series to test to confirm that everything was working. After a little more tweaking, no more sparks/arching was observed.

Next, I ran a series of cleaning shots and played with bank settings in preparation for my run tomorrow. After successfully firing three plasmas, the 4th and 5th shot each blew one of the 3 phase fuses associated with the OHE bank. Some possible considerations include:

1.When repairing the OHE bank, we effectively added 46 capacitors to the bank when we replaced broken fuses. Perhaps more capacitance and charging time changed something?

2. Also, the current knob on the OHE power supply was bumped at some point, and no one seems sure where it should be set. We tried 9 and then 8 amps limit for the 4th and 5th shots, respectively.

Jim ordered more fuses, and we're waiting until tomorrow to decide on how to proceed.

Tuesday July 12 2016 5:12 pm Brooks 93808 93871

Today was largely a shot cleanup day due to our month-long bank repair. I also used the opportunity to work on developing my shot style for my upcoming run day, next week hopefully. My desired shot style is a relatively flat major radius profile along 90cm. Shots 93863 and 93864 were my best attempts.

Friday July 15 2016 4:46 pm Hughes 93974-94024 Cleanup shots

Today's run was focused on cleanup shots for experimental operation to resume next week, with some shot development on the side. The machine was generally well-behaved, although base pressure rose considerably during the day (likely due to vacuum condition and the lab being hotter than usual) and shot reproducibility was middling, partly due to what appear to be strong early modes on many shots. As a cleanup shot run, there was no target shot in particular, besides long-lived, high-current plasmas.

All walls were "in" for this run, meaning stainless shells are back 13mm from maximum insertion, and ferritic shells are back 5mm from the FW stoppers. The OHE bank was successfully run at 300V (up from 260V) for much of the day.

Despite not being a research run, there were a few interesting plasmas, described below.

Noteworthy shots:
93975 swings outboard, then falls in, giving it an interesting q* evolution in time.
93990 and 94005 both disrupt (i.e. have a current spike) while "well-centered" (i.e. R₀ in the 90cm-92cm range).
93998 has an interesting period of strong mode activity (3-4ms) between quiescent periods (2-3ms, 4-5ms).
93999 has a very slowly rotating (~1kHz?) mode that grows from 2-3ms, as well as an interesting loop voltage evolution.

Monday July 18 2016 5:42 pm Brooks 94025 94075

Ran Bias Probe SOL tests today. The BP was at 105 cm for all shots today. The probe will be placed at 106 and 107cm tomorrow. The goal of today's experiments were to study the effect of the BP voltage and current in the SOL and study any correlation between the BP current and the Rogowski currents.

Todays tests can be broken up into three sections.

1. First, I measured the floating potential of the BP in both the plasma and in the SOL. The floating potential in the SOL was about 0 to 5V. 94045 and 94046 are good examples of the probe drifting between the plasma and SOL and measuring floating potentials in both.

2. Second, I ran with BP at a constant potential of ~45 V from 2.5ms onwards. 94061 and 94066 are good examples of the probe in the SOL, and 94065 is a good example of the probe in the plasma. Preliminary analysis shows that both BP current and Rogowski current demonstrated the same frequency in current fluctuations, but causation/correlation is difficult to determine. The current calibration for the Rogowskis have not yet been conducted so nothing can be said about the amplitudes.

3. Lastly, I ran the BP with a constant potential of ~45 V but inserted square pulses that dropped the voltage to 0 V sometime during the run. 94073 and 94075 are good examples of the probe inside the SOL, and 94072 and 94074 are good examples of the probe inside the plasma. While the BP current is clearly seen to drop low when the BP voltage is set to 0, there is no clear correlation with the Rogowskis.

Tuesday July 19 2016 4:26 pm Brooks 94076-94119 BP-SOL Current Tests

Today was a repeat of yesterday's experiments but with the BP placed at 106 cm and 107 cm instead of 105 cm.

The 106 cm case showed some results. Shot 94080 shows the BP floating voltage while the major radius moves in and out of the SOL and plasma. Some oscillations are observed as the probe moves from the SOL into the plasma. Further tests were performed with a fixed BP voltage and a BP voltage that pulses between +45V and 0V. Preliminary results are inconclusive.

The BP was moved to 107cm, and the plasmas within HBTEP were incredibly inconsistent. Very little usable data was recorded during this time.

Monday August 1 2016 6:17 pm Hughes 94132-94157 Cleanup and shot development for Fe Wall Asymmetry

Today's run was plagued by a number of technical issues, followed by poor reproducibility. Vacuum condition seems to be not-terrible, although the east cryo is again at 30K on its gauge, and base pressures have been a bit high all day.

The main issues have been around the puff and e-gun. The puff circuit power supply was previously repaired and was reinstalled, today; however, the supply was set to 495V after repair. I erroneously remembered it should be set for 450V, and set it to that value (as of shot 94136); continued misbehavior (very high fill pressures) led us to correct the setting to 250V as it should be (as of shot 94143). Documentation to this effect has been added to the wiki. The puff seems to have been working well since (based on the metric of good fill pressures), so it doesn't seem to have been damaged.

Shortly afterward, plasmas stopped breaking down (94145). The problem was traced to probably being the e-gun, and the bias was adjusted from -25V on (-150V off) to -29V on (-180V off), which seemed to solve the problem insofar as getting breakdown successfully (94148). However, the remaining plasmas of the day were highly irreproducible, changing behavior significantly with no changes to bank settings.

The only shots of any special note are those from the beginning of the day when plasmas (e.g. 94135) with very high fill pressures (~200uTorr) were produced. They were generally very weak, although it does at least demonstrate that we can still generate plasmas in that fill pressure regime.

Tuesday August 2 2016 5:51 pm Hughes 94158-94183 Cleanup and shot development for Fe Wall Asymmetry

Today's run was again aiming for cleanup shots and shot development for the Fe wall asymmetry run. Plasmas continued to have relatively poor reproducibility. Some time was also dedicated to studying the problem of the North rack A14 timebases, although the results remain highly variable and confusing; it appears that the timebase errors change depending on whether the banks are fired, and also differ between vacuum and plasma shots.

A continuous sine wave was applied to several channels to study stretching of the timebase. On at least one shot (94169), several channels can be seen to change their apparent frequency spontaneously, and neither frequency agrees with the other channels.

Shot 94182 exhibited a good, steady R0, but at the cost of a peculiar early plasma current trace. Some other shots (e.g. 94172, 94174, 94176) had reasonable but not ideal R0 traces, and if this style proves more reproducible, it may be acceptable.

Wednesday August 3 2016 4:44 pm Hughes 94184-94216 Shot development and Fe wall asymmetry

Today's run was meant to begin the Fe wall asymmetry study. Early plasmas still showed poor reproducibility. The fill pressure was somewhat erratic in relation to puff time setting, increasing irreproducibility and preventing breakdown on one shot (94194). The puff time (and therefore fill pressure) was increased, and reproducibility improved enough to take a few good shots (94200, 94203, 94204, 94207, 94208). However, fill pressure began to decline again (around 94210). Puff time was increased again, but shortly after the CPCIs began throwing "status 0 on SPITZER %SS-W-NOMSG" errors, failing the Prepare stage. Power cycling the racks and restarting take_shot_hbt.py did not resolve the issue. The next possibility is to reboot Spitzer, but given the late time of day and the time needed for Spitzer to reboot, this marks the end of the run day.

If everything is working tomorrow morning, the walls will be swapped into a strongly n=1 asymmetry with only Fe shells in sections 1 and 3 inserted, while all stainless shells remain inserted and all other Fe shells are retracted to their stoppers.

Thursday August 4 2016 6:04 pm Hughes 94217-94270 Fe wall asymmetry study

Following up on yesterday's shot development and initial run (all-in, symmetric), the target shot was 94203. Today covered the Fe wall asymmetry study, examining four cases:

Case 1 (phi_Fe ~ 72 deg): All stainless walls in the "in" position (back 13mm from full insertion). Fe walls in sections 9, 1, and 3 retracted fully; those in sections 5 and 7 in the "in" position (back 5mm from Fe wall stoppers).
Case 2 (Fe out, symmetric): All stainless walls in the "in" position (back 13mm from full insertion). All Fe walls retracted fully.
Case 3 (phi_SS ~ 108 deg): All Fe walls retracted fully. Stainless walls in sections 2, 4, and 10 in the "in" position (back 13mm from full insertion); those in sections 6 and 8 back 5mm from full insertion.
Case 4 (phi_Fe ~ -144 deg): All stainless walls in the "in" position (back 13mm from full insertion). Fe walls in sections 3, 5, and 7 retracted fully; those in sections 1 and 9 in the "in" position (back 5mm from Fe wall stoppers).

Case 1 vs. Case 2 (and "Case 0" of the mostly-symmetric configuration from yesterday) is the essential comparison between configurations for this study. However, comparing with Case 4 makes it possible to look for changes in preferred mode phase, and including Case 3 compares effects of Fe material vs. effects of close-fitting conducting structures. Results will require analysis, of course.

Machine performance was generally well-behaved, compared with the rest of this week, and several good shots were taken in each configuration. The failures at prepare from Wednesday afternoon did not return; it seems that either rebooting Spitzer fixed the problem, or it fixed itself overnight.

The erratic fill pressures have also returned to normal behavior, steadily dropping during the day (at roughly 1uTorr/10 shots), easily compensated with small changes in puff time. Notably, last night, I recharged the puff battery with the dumb charger (which often charges above 13.5V) rather than either of the smart chargers (which seem safer but also charge less aggressively to ~12.7), but this correlation is only two data points.

Wednesday August 17 2016 4:22 pm Brooks 94284-94312 BP2 initial testing

Today, I finished connecting the new bias probe, designated BP2, to North Rack CPCI channel's 96 and 89. After some debugging, I inserted BP2 into the plasma and took a few shots in both "floating" and "grounded" configurations. 94310 and 94311 show both cases, respectively. Tomorrow, further calibration (probe radial location, voltage, current) will be conducted as well as setting up for a feedback run. /home/brooks/jScope/configurations/northRackCh89And96.jscp is the jScope config file for easy viewing of the above data.

Tuesday September 13 2016 11:54 am Brooks 94586-94580 New

The run reports discussed here cover my runs on both 9/9/16 and 9/12/16.

The goals of these runs were several fold. Jeff requested that I pursue 3-1 and 4-1 modes that were outwardly limited and assist him in studying the phase relationship between the dominant mode amplitude and the current measured at both bias probes. BP1 was placed at 106cm and BP2 was placed at 106.5cm in order to place them on the boundary between the plasma and SOL. Both were in their "grounded" configuration in order to measure plasma current. Jeff appears to have found a good correlation between the phases of the two.

The second goal was to finalize the setup and calibration of the BP2. BP2 appears to be working as expected, but the its major radius location doesn't quite agree with similar measurements conducted with BP1.

Third, BP1 and 2 were placed at different major radii, and their current measurements were studied to look for signs of shadowing (the radial location of one probe affecting the other). No signs of this were found.

Fourth, the first triple probe was instrumented and initial tests have begun using it located well outside of the plasma. There appears to be a number of bugs plaguing both wiring and the instrument box. Work is underway to address these issues.

Fifth, Mel's Rogowskis were placed at the 3-4 midplane and later at both the 3-4 midplane and 9-10 midplane. Relationships between the BPs' and Rogowskis' currents were studied but no obvious correlation was observed. Additional study is required.

Thursday September 15 2016 9:07 am Brooks 94681-94722 TP1

9-13 and 9-14 run reports. Spent two days working on TP1, and it appears to be close to working correctly. Shots 94721 and 94722 shows TP1 at 105.2cm. /home/brooks/jScope/configurations/TP1.jscp shows the details.

In addition, ran a number of shots for Jeff. BP1 was as 106cm, BP2 at 106.5cm, and both were in their grounded configurations. Rog B was at the 3-4 midplane, and Rog A was at the 9-10 midplane. Shots 94698-94717 were fairly good data points for Jeff where 98, 99, 01, 03, 04, 07, 08, 09, 10, 11, 15, 17 were the better shots.

Thursday September 15 2016 3:55 pm Hughes 94724-94744 Fe tile mu test

Today's run used only the TF magnets and a hall sensor to measure the field between two Hiperco 50 tiles, and the toroidal field in the same location in absence of the tiles. The TF was scanned from 500V to 6500V in both cases. Comparison of field measured by the hall sensor indicate a fixed peak permeability of 8. The permeability appears to droop at higher fields (>2kG); however, this may also be due to nonlinearity of the hall sensor above 1 Tesla. The results are plotted on the appropriate supporting experiment discussion page on the wiki, which will have more to say, later. Digitizer timings in the west rack were changed for this run, and will be returned to normal settings.

Tuesday September 20 2016 4:36 pm Hughes 94753-94779 Toroidal asymmetry of ferromagnetic walls

The purpose of today's run was shot development and cleanup for a continued study of the effect of toroidal asymmetry of ferromagnetic walls on edge MHD dynamics, particularly mode rotation, locking, and phase-preference, following up on this experimental campaign by adding a 72-degree phase comparison. Today's plasmas were not reproducible enough for the study, but by the end of the day, they became much better-behaved with loop voltages and SXR midplane traces much lower and closer to the target shot (94203). The most likely cause is the recent work installing new probes in the machine leading to increased impurities and a need for additional cleanup shots. The best shots from today were 94775, 94778, and especially 94779. This run is planned to continue tomorrow.

Wednesday September 21 2016 6:06 pm Hughes 94780-94834 Fe wall toroidal asymmetry study

Today's run targeted shots 94203 and 94779 to study the effect of toroidally asymmetric ferromagetic walls on mode dynamics, particularly rotation. Shots were taken with SS walls only inserted (back 13mm from full insertion), with S9+S7 Fe walls inserted (back 5mm from full insertion), with S5+S7 Fe walls inserted, and with only 9B inserted and only 7B inserted. This is meant to show if there is any change in preferred n=1 mode phase from 72-degree changes in Fe wall location phase. Results will follow.

It appears shell 9B was already inserted from some point in the past, although it isn't known when. This means good shots 94775, 94778, and 94779 belong in today's "SS + 9B only" category.

Good shots (chosen for similar I_P and steady R₀ between 91 and 92.5cm for best coupling to Fe wall):
SS only: 94825,94827,94829,94833,94834
SS + S9+S7: 94800,94802,94804,94805,94806
SS + S7+S5: 94807,94811,94814,94815,94817
SS + 9B only only: 94784,94787,94788,94790
SS + 7B only: 94818,94819,94820,91821,94822


Wednesday September 28 2016 5:13 pm brooks 94845-94870 Triple Probe testing

Spent the day diagnosing issues associated with the triple probe at section 2. I fixed the issues with the reversed signal polarities and another issue associated with naming conventions on jScope. Finally, I had all three probes tips in their floating configurations and placed the plasma at 105 and 104 cm. Tip C provided results (particularly the offset) that appeared similar to the floating Bias Probes inside the plasma. Tip B, however, did not work correctly. While it showed some of the oscillations observed on Tip C, they were smaller, and the offset was pretty much zero. I checked many times over the hardware/wiring and could not discover any explanation. Shot 94867 best shows this. Use /home/brooks/jScope/configurations/TP1.jscp to view the results.

Monday October 10 2016 10:46 pm Abler -94964 SoL Currents as a function of nearby shell location

This is part 1 of 2, conducted Friday 10/7/2016. Wednesday was lost to classes/hardware struggles. On a positive note, this lead to the solution of an optional extension being added to Rogowski B to allow it to span either the 3/4 quartz break, or the slightly wider 5/6 quartz break. Photos will be uploaded to the wiki tomorrow. Thursday was also lost to a combination of ion gauge and bank firing problems - after some troubleshooting, we did not have a solution, and tried the "turn it all off and see if it works tomorrow" approach, which seems to have panned out. On Friday, Rogowski B was attached across the 5/6 quartz break midplane, and Rogowski C was attached across the 10/1 quartz break midplane. The goal is to get highly outboard limited shots (centered at 94 cm or more) with consistent mode activity and relatively consistent major radius for a few ms - target shot was 81915. Under these conditions, the "upstream" (lower phi) walls from the jumper are retracted varying amounts, and vessel borne currents through the jumpers are compared. Notably, the currents at the 5/6 break are consistently larger than those at the 10/1 break (by a factor of about 5). To confirm this was not an equipment malfunction on the part of Rogowski C, Rogowski A was switched in across the 10/1 break for shots 94947-94951; current amplitude was similar during these shots, so this seems to be a real phenomenon. As a result, we choose to focus on the 5/6 break by moving sections 4 and 5, rather than the original plan of focusing on the 10/1 break and moving section 10. Shot 94942 revealed itself as a good update to the original target shot. Later in the day, after beginning to move the shells, both maintaining an outboard plasma and repeatability became difficult and a resumed/continued run is planned for Monday 10/10/2016.

Tuesday October 11 2016 10:25 am Abler 94965-95008 SoL Currents as a Function of Nearby Shell Location

This is part 2/2 of this run campaign (the first having been Friday 10/7/2016).

As before, the goal is to have plasmas centered at or beyond 94 cm major radius, then to vary the location of shells at sections 4 and 5 to study the changes in halo currents through Rogowski B across the 5/6 break (Rogowski C is across the 10/1 break). Shot 94942 was the loose target shot for the day. Below, find a list of shots which meet the desired major radius profile, sorted by shell location:

fully inserted: 94979, 94983

0.5 cm back: 95000

1.0 cm back: 94991, 94995

1.5 cm back: 95002, 95008

2.5 cm back: 94988, 94989, 94990

Fully inserted is following the previously established convention (5 mm back from stoppers for ferritic shells, 13mm back from stoppers for stainless shells). All distances are measured from that "fully inserted" position. I thought I had two good shots at 0.5 cm, but apparently don't.

In terms of physics results, a dropoff in currents through Rogowski B can be observed as shells 4 and 5 move outwards. I don't have the analysis code built up yet to separate out effects of mode amplitude, but will be working on that over the next few days. Graphs of the results I do have will be presented at group meeting.

Tuesday October 11 2016 4:55 pm Brooks 95009-95021 BP SOL testing

I managed a new shot profile, 95018, but had trouble getting Caliban to run. Caliban is the PC used for feedback. It ran out of internal cooling water last week, and I pulled it out to replace the water. The current issue may be related to my removal of the computer, or it may have been something that was related to it being unused for several months. Hopefully the reboot of Spitzer tonight may help.

Thursday October 13 2016 9:42 am Brooks 95022-95082 BP SOL tests

The focus of today's experiments were several fold.

The first set of experiments placed the section 9 BP in the SOL and drove the probe with an oscillatory voltage of 3kHz and 7kHz. Quartz break rogowskis were also present. Data analysis will start tomorrow and will focus on developing phase relationships between the BP current and B-tilde amplitudes. The rogowskis will likewise be studied.

The second set of experiments placed both section 5 and section 9 BPs in the SOL, and they were connected in a double probe configuration, connected with a wire that ran externally along the west side of HBT. Several shots were taken with the BP amplifier disconnected and several with it connected. BP-S5 voltage was picking up a lot of "bank" signal and may be unusable.

Tuesday November 15 2016 4:35 pm Hughes 95141-95147 e-gun digitization testing

This was a series of crates-only shots to test e-gun digitization. Although heating current is recorded well, there are some peculiarities in the other two parameters recorded. The emission current measurement is inconsistent between settings, which may partly be explained by low-level signals and digitizer offset. The supply voltage reading has unexplained ~5kHz signal about the same strength as the DC signal; this may be due to the amplifiers, although the problem was believed to have been resolved on the bench.

The digitization board will go back for more bench-top testing to track down the noise problems, although some of the noise may originate in the e-gun stand which is known to be electrically noisy. Once the board has been repaired and retested, it will be calibrated in-situ.

Tuesday November 22 2016 5:32 pm Brooks/Hughes 95154-95180 Commissioning all banks, e-gun digitization testing

Today's run was directly mainly to verifying that all banks and critical systems are running properly, in particular following the installation of the new OHE power supply. In parallel, the e-gun digitization/isolation circuit was tested further. The noise problems remain, but the only saturating noise spikes are 2-3 time points (20-30us) wide. Shots 95154-95170 were vacuum shots. 95171 was puff-only, and 95172 was the first plasma of the day.

Shots 95173-95180 were cleanup shots, which improved over time as usual. More cleanup shots will probably be required.

Monday November 28 2016 5:30 pm Hughes 95218-95248 Shot development for Dov's upcoming run

Today's run was oriented toward developing a shot style optimized for studying effects of the ferromagnetic wall on the poloidal mode spectrum. The plan was to target q~2.5 with relatively well-centered plasmas (R₀~92-93cm, compared with Jeff's q~2.5 shots where R₀~95cm). This requires higher current, and the resulting plasmas have unpredictable early (t~1.0-1.4ms) behavior. Modes observed appear to be predominantly slowly-growing but large m=2 structures, suggesting tearing modes in spite of the fairly strong (up to ~2.5MA/s) current ramp.

hbtplot.py has been updated to use e-gun heating current data to display an alert next to the shot number if the current is below a minimum threshold (e.g. the heating current wasn't turned on).

As a note, on most of today's shots, the security camera feed has cut out for 1-2 seconds at the time when the banks fire and the ion gauge switches momentarily to bypass. This appears to be a new behavior. The monitor does not switch off, and the video is stored normally and can be reviewed afterward, suggesting the interruption is in the signal to the monitor.

Noteworthy Shots: 95248: Nearest shot to today's target shot style. Relatively long-lived at q~2.4-2.5, eventually suffers minor disruption from what appears to be a large 2/1 mode. 95238,95244: Shots with large excursions in R₀, but very steady q~2.2 from about 3-5ms.

Tuesday November 29 2016 3:35 pm Brooks 95181-95215 Triple probe testing

The goal of Wednesday's (11/23/16) run was to continue testing the Triple Probe (TP) in Section 2. After testing the banks and power supplies after swapping around supplies and an extended break from running the machine while at APS DPP, I first started at developing a shot style. I based my run style on a previous shot style, but discrepancies (likely caused by the increased capacitance in the OH bank after we added caps during the summer) required that I develop a new one.

Once developed, I ran several shots with the TP at various major radii. Vacuum shot = 95208, 109cm=95207, 108cm=95210, 107cm=95211, 106cm=95212,95213,95215. The amount of response at each probe tip certainly increases as the probes are pushed farther in. In addition, shot 95213 at 106cm suggests that the probe tips moved between the plasma and the SOL. Data to follow this message. TipC is the floating tip.

A few other observations. The vacuum shots show a fair amount of noise picked up from the banks. Jeff and I spent a few shots last week moving wires around (induction loops) to try and change the pickup but with no success.

Because only 1 of the three shots at 106 cm shows the floating potential move between positive and negative regions, I'm beginning to speculate that my radial probe location may not be correctly calibrated.

Tuesday November 29 2016 4:18 pm Abler 95249-95257 RMPs + Rogowskis

Today was a first attempt at applying RMPs to outboard-limited plasmas. Shot 95252 has become a new target shot, as it rode very precisely at 93 cm for more than 2 ms. Storage failures in the basement, north, and south rack A14s beginning on shot 95255 could not be resolved by power cycling or taking a cooling break (all the racks felt quite cool when we were investigating). The run was ended after we ran out of ideas for how to fix the storage failures. Hopefully the machine will be more cooperative tomorrow.

Tuesday December 6 2016 9:47 am Brooks 95316-95345 Triple Probe Section 2 and 7 testing

Today's (12/5/16) tests concerned the testing of Triple Probes at section 2 and 7. TPS2 has been an ongoing project, and TPS7 was only inserted for the first time today.

I discovered a broken BNC connector on Tip A (negative tip) of TPS2 and repaired it. TPS2 is now measuring current but now suffers from current spikes (95333 and 34). Diodes were installed to mitigate these spikes around 2.5V (95335) and appear to prevent saturation.

Another issue with TPS2 is that there appears to be a voltage different between tips B (positive tip) and tip C (floating tip) (95339). Swapping tips B and C at the instrumentation box causes the voltage difference to swap to negative (95341), implying the issue is with the probe tip itself or at least the wiring leading up the tip. Perhaps there is a bad electrical connection adding resistance?

TPS7 was inserted at the end of the day in a floating configuration at 110 cm, 108cm, and 105cm (95343, 44 and 45 respectively). It looks like only 1 tip is providing reasonable results . Further testing is required.

Looks in my /home/jScope/configurations/ directory for the TPS2.jscp and TPS7.jscp files to view the triple probe data. A few images will follow this report.

I still am not on this mailing list.

Wednesday December 7 2016 10:39 am Abler 95346-95376 Rogowskis & RMPs

Today was continued work on shot development and some initial data for the RMP/Rogowski experiments. However, set-up involved a number of notes which will be addressed first.

During hi-pot testing, sparks were repeatedly heard and/or seen near section 5, typically when the voltage was at or above 7 kV. After a series of inspections by Paul, Jim, and Jeff, no source could be found, and hi-potting was re-attempted. This cycle repeated several times with increasing worry - eventually, some insulation near section 5 was moved (Jim will be able to speak to this more). After group meeting we hi-potted again, and although there was one spark near section 5 when it was held around 10 kV, the spark did not repeat after more than 30 seconds, and it was deemed safe to run. Operators should be alert to any changes, especially strange sounds coming from the machine room and extra care should be taken when hi-potting until we know more.

It was observed on the second RMP shot of the day (95351) that some of the control coil currents were clipping, and investigation by Jeff revealed that this was because the dials on the amplifiers for sections 2B, 3T, 6B, and 7T were not in their normal positions (5 clicks down from max). This was corrected for shot 95352.

As far as the actual run is concerned, progress was made! I now have 2 shots with no RMP (95252, 95304), 0 phase (95359, 95364), and pi phase (95365, 95369). Graphs/analysis will be sent out ASAP. On a machine operation note, the plasmas I was aiming for (steady at 93 cm major radius over the RMP or longer) are very sensitive to base pressure - 26 nTorr was the sweet spot, and even a plus/minus 2 nTorr variation was enough to either move the plasma inboard or give a minor disruption during the RMP. I will run again on Thursday and Friday to try to fill in the rest of the phase space.

Monday December 12 2016 8:11 am Abler 95401-95427 RMPs + Rogowskis

Further attempts to get a complete set of RMP shots. 95422 was the closest to a workable shot, but I had a lot of problem with disruptions during the phase flip. Extreme sensitivity to base pressure remains.

Monday December 12 2016 8:13 am Abler 95444-95481 RMPs + Rogowskis

More RMP/Rogowski shots. I now have data for a pi/2 phase (95459, 95465) and 3pi/2 phase (95468, 95481). Any runtime I can get my hands on next week will be use to fill in the rest of the phase space.

Monday December 12 2016 11:33 am Hughes 95429-95443 e-gun isolation circuit calibration

The purpose of this run was to test the calibration of the recorded e-gun parameters. The channel carrying emission current is currently not working, so only heating current (I_H) and bias voltage (V_B) were calibrated. Calibration is between the digitizer with expected calibration for the isolation circuit, versus voltage measured with a DVM or current measured on the display of the heating current transformer.

Calibration factors for V_B fall in the range 0.943-0.103, averaging 0.99+/-0.02 relative to the circuit-based expectation. Interestingly, when the heating current is on, the calibration factor increases significantly (meaning that digitizer-measured bias is lower than DVM-measured bias) if gas is puffed into the chamber. At V_B=35V, and I_H=9.2A:

Tpuff	Pfill	Vexp/Vmeas
None	30nTorr	0.986
1ms	45μTorr	1.386
2ms	83μTorr	1.608
3ms	116μTorr	1.675

Calibration factors for I_H were more variable with heating current, ranging from 0.978-1.194, averaging 1.08+/-0.09 relative to an empirical calibration constant. The circuit tends systematically to undervalue lower currents.

Monday December 12 2016 6:14 pm Hughes 95482-95513 SS vs Fe m-spectrum study

Today's run was the beginning of a study to compare the m-spectra of modes in the stainless and ferritic configurations, with the change of keeping the S8 shells (and therefore PA2) inserted in the latter configuration. The target shot selected was 93188, a stainless-configuration shot from March.

Since today's run used the stainless configuration (which is not changed from previous SS-vs-Fe runs), the results were expected to be similar, but plasmas were somewhat fussy with this shot style, possibly due to the relatively high fill pressure, and tended to drift in or out rather than sitting steady. Fourier-decomposed mode activity was also markedly different from the target shot; most shots had very weak m=2 components except for the last ~1ms before disruption, rather than the long-lived m=2 activity seen in shot 93188.

Noteworthy shots: For m-spectrum analysis, shots 95499, 95501, 95503, 95508, and 95512, which follow the target shot reasonably well. Shots 95510, 95511, and 95513 do not follow the target shot as well, but having lower q* and reasonably good wall coupling, may still be usable. Shot 95487 was also interesting, due to an uncommonly steady spin-up of the mode rotation in the 2.8ms to 4.0ms window.

Hardware notes: Despite using the same bank voltage settings as those from the morning of the target shot, the VF start current was ~15% lower on today's vacuum shot. A VF-only shot however matched a same-settings VF-only shot from later that month (93395). The reason for the lower VF start current is not presently known (the VFS bank may have a higher charging current or leakage current than it used to), and was compensated for with slightly higher bank settings, but this should be noted in case of trying to replicate older shot styles.

Wednesday December 14 2016 4:28 pm Brooks 95514-95537 Triple probe testing

The goal of today's test was to test TP-Section 8 (the new probe) against TP-Section 2 (the existing probe). After checking wiring, I inserted both probes to 105cm and began testing. Because I can only measure 2 of the 3 probe tips during an given shot, there are a total of 6 probe to channel configurations. I next ran all 6 configurations on both probes simultaneously. Both probes provided similar results but the signals often look inverted. Still thinking through this.

Also of note, the puff valve failed on shot 95531. It dumped in enough DT into the chamber that the ion gauge reset and closed all of the cyro gate valves. Jim was present and reset everything. Further testing didn't have any additional valve failures.

Tuesday December 20 2016 12:39 pm Brooks 95539-95573 Triple Probe Testing

The goal of today's (12/19/16) testing was to check the results of triple probe section 2 (TPS2) against TPS8. The results of the powered tests are remarkably similar for certain probe tip configurations and less so for others. I'm not sure why the probe configuration seems to matter so much. Also, TPS8 is pulling almost double the current as TPS2 at the same major radius.

Also of note, shots 95552-55 resulted in no plasma. Paul, Jeff, and I investigated, and Paul sent out an email last night on the results. Our best guess is that there was some hiccup with the egun that went away by shot number 95556.

Tuesday December 20 2016 5:37 pm Hughes 95574-95601 SS vs Fe m-spectrum study

Today's run was shot development for an experiment series associated with Dov's modeling work that predicts a change in poloidal mode spectrum between Fe and SS configurations. We attempted to target a modified version of shot 93188 with higher I_P and therefore lower q_*, as well as a more standard p_fill. These higher-current shots were uncooperative, however, exhibiting strong, persistent mode activity after breakdown, which consistently produced a jumpy major radius in the 1-2ms time range. Plasmas tended to fall in, and none with increased current sat in the target 91-92cm range.

The purpose for this altered shot style was to look at m=2/m=3 proportions with q_*~<3, but it may be just as effective to look at m=3/m=4 proportions at higher q_*, which may be a better-behaved shot style. There were also issues with Caliban, as init_rtm-t and do_awg threw errors unexpectedly. We will try to sort this out tomorrow morning.

The one really noteworthy shot was 95596, when I forgot to switch on the e-gun until ~1s before the shot fired. On the e-gun trace, I_H (as noted, actually measured as a voltage) can be see rising from 7A at -240ms to 9A at 900ms; V_B remains near 215V, only drooping to 195V at 900ms. Notably, the plasma is feeble, compared to another shot with the same bank settings on which the e-gun was turned on several seconds earlier. Although emission current (I_E) isn't presently measured, it can be extrapolated from the loading of V_B; given the ~218V supply setting and the 87kOhm supply resistor, I_E evolved from ~30uA (typical for a cold filament) to only ~0.3mA, in contrast to I_E~2.1mA, typical of a hot, biased filament.

Wednesday December 21 2016 6:11 pm Hughes 95602-95641 SS vs Fe m-spectrum study

Today's run was dedicated to generating plasmas with steady R₀ and q_* in the 3-4 range for m-spectrum studies examining relative m=3 and m=4 components in RMP response. Applied RMPs include combined m=3+m=4, m=3 only, and m=4 only. For this run, all Fe (odd-numbered) shells were back 5mm from full insertion, and all SS (even-numbered) shells were back fully except for S8, which was back 13mm from full insertion so that PA2 will be unchanged between Fe and SS configurations.

Plasmas were much more cooperative than the previous run day, due to targeting shot 93188 more closely. After several good shots, the new (more recent) target shot was set to 95623. Calculation of q_* was based on the use of the --correct-tf (increases value from the TF probe by 15%) and --use-vp (corrects minor radius with a vertical position estimate) flags in hbtplot.py, and will appear different if these corrections are not applied. Note that with --use-vp, q_* does depend on sensor RC times being accurate, and there are known RC time errors in PA2. Plans are underway to correct this retroactively in the tree. The run was ended as plasmas became less cooperative, possibly due to high base pressure (~40nTorr on the control room gauge).

Noteworthy shots:
95612,95613,95616,95618,95619: RMP amplitude 1.5(m=3) + 1.5(m=4).
95620,9562195622,95623,95624: RMP amplitude 3.0(m=3 only).
95627,95628,95630: RMP amplitude 3.0(m=4 only).
95634,95636,95638: No RMP applied. Shots 95632, 95633, and 95640 have less well-behaved R₀, but may still be usable for a no-RMP baseline.

Thursday December 22 2016 5:07 pm Hughes 95642-95686 SS vs Fe m-spectrum study

Today's run was the stainless configuration series in the study of the ferromagnetic wall's effect on poloidal mode structure. Plasmas targeted shot 95623 from yesterday, and equilibrium parameters were generally roughly similar. However, R₀ drifted much more, and a higher OHE was needed to get similar plasma current, leading to I_P traces that tended to be slightly higher around 1-3ms, and then drop off more quickly after 4ms. RMPs were applied, with combined 3/1+4/1, 3/1 only, and 4/1 helicities. Quality of shots is based on similarity of R₀ and q_* traces in the 2-5ms window.

Soft X-ray midplane signals were consistently low at breakdown (1-2ms) and highly erratic afterward, sometimes more than 50% below the target shot's trace for extended (1ms+) periods, as though the plasma was unable to heat up. The resulting variations in major radius in particular greatly outweighed effects of bank settings. It may be advisable to flush the puff line, since strange SXR signals have previously been correlated with puff line contamination.

During the run, two changes were made to machine operation between shots 95655 and 95656: 1) Jeff changed the PA2 RC times in the model shot to be more accurate, and 2) the optical fiber for the USB spectrometer was moved to have a better view of plasma light, in case the low SXR behavior correlated to differences in visible light spectra.

Noteworthy shots:
95658,95660,95661,95662: Best m=3+4 shots. 95664 may also be usable.
95676,95677,95678: Best m=3 only shots. 95679 may also be usable.
95681,95682: Best m=4 only shots. 95684 and 95686 may also be usable.
95667,95669,95671,95673: Best No RMP shots. 95665, 95670, and 95674 may also be usable.
95666: Example low-SXR midplane shot. No obvious trends stand out in visible light spectrum, comparing low-SXR shots and normal shots, although some low-SXR shots do see a slightly higher line (~463nm, suspected to be C-III).

Friday December 23 2016 11:58 pm Levesque 95688-95703 SS vs Fe m-spectrum campaign

Continued yesterday's run with the same wall configuration. Was not able to get any shots with equilibria similar enough to the target (95623) for direct comparison of mode structure. No RMPs were applied.
Before the beginning of the run, Jim purged the puff line several times due to concerns over possible puff line contamination. Subsequent plasmas behaved differently compared to shots from yesterday and before. Without changing the puff time, the calculated fill pressure was much higher than earlier this week -- yesterday a puff of 2100us duration gave fill pressures of 82uT, while today the same time gave fill pressures of 101uT. Reducing the puff time to 1830us produced the target calculated fill pressure of 82uT. Note that the raw signal recorded for the chamber pressure had different features than yesterday, and I'm not familiar enough with the calculation to know if this would produce erroneous calculations of the fill pressure.
Throughout the day, breakdowns were more variable than usual, especially during the OHB stage of the discharge. They seemed characteristic of when we've had puff problems in the past, but the chamber pressure recording for these shots didn't reveal anything different for each shot. Of the 10 shots with banks firing and a puff, 6 had breakdowns with reasonable variation, 3 had breakdown currents that were lower than expected by around 10%, and one didn't form a plasma. While shutting down, I found that the e-gun bias was probably off all day (the switch was off when I checked it, but I don't remember turning it off beforehand) -- this would easily cause differences in breakdown, but it was surprising that so many plasmas were successful. The digitized e-gun information also suggests that the bias was off all day, if I'm interpretting the raw data correctly.
More-significant problems started occurring at shot 95700, leading to shutting down for the day soon afterward. Problems occurred as follows:
95700: Attempted a normal shot. The shot was auto-triggered when the TF was fully charged, but the banks didn't fire. The North rack CPCI alone failed to store -- this is unusual since one CPCI failing is usually accompanied by all of them failing. The puff appeared to blow in far too much gas, as if the valve was stuck open for much longer than usual. The chamber pressure got high enough for long enough that the ion gauge turned off and cryo gate valves closed. After the shot, turning the ion gauge back on showed the chamber pressure steady at 1.6e-5 torr with the cryo gate valves closed. The valves had to be opened by bypassing the ion gauge interlock while the chamber ion gauge still displayed around 1.6e-5 torr. The fill pressure calculation read much higher at 2.6mT, suggesting there may have been a problem with the ion gauge and its output instead of there actually being high chamber pressure. The digitized ion gauge signal abruptly rose much faster than normal and clipped at the time when the TF was supposed to trigger.
95701: Crates and puff only, with the e-gun off. The puff had the same problem as the previous shot, where it appeared to puff in too much gas, and the gates valves closed. No further puff attempts were attempted in order to protect the valve. The calculated fill pressure was 0uT despite there being lots of gas. The basement A14 also did not record correctly -- it stopped early. The A14 problem persisted for the rest of the shots. The digitized chamber base-pressure offset was now different, despite the base pressure being the same. This change in base-pressure offset continued for the rest of the shots. Shots 95702 and 95703 were crates-only without puff for testing the puff reading and e-gun diagnostic. The planned run on Wednesday next week should start with some puff-testing in case problems persist.

Thursday December 29 2016 11:28 am Brooks 95704-95725 Triple probe double puff tests

Yesterday's (12-28-16) run focused on double puff testing with the triple probes. Only the TPS2 probe was used (S8 was offline for some reason). A second D puff was injected from 2.0-2.2 ms, and a noticeable drop in Temperature was observed. There was also a noticeable shift in pressure in the chamber pressure data. Shots 95718 to 25 were the relevant shots (odd numbers = double puff, even = single puff). I emailed out a few figures (plasma major radius and plasma temperature) for people to look over. Density calculations forthcoming.

Egun failure occurred on 95710. No plasma breakdown. Only occurrence that day.

Puff and bank triggering worked fine compared to the failures last Friday.

The shells were accidentally inserted (all the way) while the triple probes were inserted. No damage occurred, and the probes seem to fit. The only comment was that the lower probe tip of TPS2 was resting on the shell when it was in and may have been slightly deflected. Hard to tell because the lower probe tip was already partially deflected toward the upper two probe tips.

SS shells 13mm back, FE shells 5mm back.

Friday December 30 2016 2:13 pm Brooks 95726-95764 Triple probe radial scans

The goal of yesterday's tests (12_29_16) were to perform radial scans using the two triple probes. After experiencing a lot of arching on one probe while not on the other (even though they were at the same radius), we thought it best to double check the radial calibrations. As it turned out, the section 8 probe (TPS8) was almost 2 cm further into the chamber than previously suspected. Further tests at the same radius now show that the current profiles between the two probes now agree where previously the section 8 probe was always much higher. However, the probe tip configuration for TPS8 no longer provided voltage measurements that matched TPS2. Please remember here that I swapped around the tip configuration on TPS8 until it matched the voltage output of TPS2 and that the two different configurations didn't match but were required to be that way to provide the same voltage profiles. After re calibrating TPS8, I had to find a new configuration to provide similar voltages as TPS2. Not sure what this means.

After experiencing repeated arching, Jeff used the telescope to view the TPS8 tips. A picture will be sent out via email to accompany this report. As you will see, there is a slight sign of charring (black) around the middle probe tip (which was the negative tip around -140V).

Most of the run day was done by the time the probes were recalibrated so the radial scans were not completed. Preliminary results suggest that the electron temperature is decreasing as the probe is pulled closer to the chamber wall.

SS walls 13mm back, FE walls 5 mm back.

Smooth run day with the machine performing as expected.

Thursday January 5 2017 10:16 am Brooks 95765-95818 Triple probe radial scan

The goal of yesterday's (1-4-17) test was to perform a radial scan of the plasma using the triple probes. In addition, we were checking for shadowing. The section 2 probe (TPS2) was held fixed at 103cm, and TPS8 was moved radially from 103cm to 107cm in 0.5 cm steps. Preliminary data analysis suggests that temperature falls off roughly as expected.

To check for shadowing, TPS2 was moved to 106 cm, and a few inner radial points were remeasured with TPS8. Results forthcoming.

Another set of tests that we performed was to insert TPS8 into the probe until arching occurred and record it with a lime lapsed camera. This was done, and arching is clearly visible with TPS8 at 102 cm. Arching, also observable on the probe current traces, will henceforth be avoided to protect the probes.

On shot 95765, the first shot of the day, only the TF bank fired, and the North Rack failed to fire. No obvious explanation was provided.

One of the control coil amps, a crown xls5000, keeps throwing its breaker. Tests on it will be performed tomorrow to determine the cause. Specifically, this unit is located on the left rack, 4th from top, and controls the 2B coil.

SS walls 13mm back, FE walls 5 mm back.

Monday January 9 2017 3:23 pm Hughes/Levesque 95819-95821 Crates only

These shots were just taken to test the updates to the tree following the addition of nodes for the triple probes, bias probes, and external rogowskis, as well as a new operational diagnostics top node "oper_diag" for secondary data collection, e.g. ion gauge and e-gun parameters. With some bug-fixing, everything works fine, now--except that .oper_diag.e_gun:I_emis and .oper_diag.e_gun:I_emis:I_proxy both are mysteriously given the unit "volt" rather than "ampere" in spite of the correct units being named in the TDI language call.

Thursday January 12 2017 4:10 pm Hughes 95822-95857 e-gun optimization

Today's run focused on the influence of e-gun settings on plasma breakdown. Shots were taken with standard V_supply=220V and I_H=9.5A settings (95825-95829), with heating current raised (65830-95832) and lowered (95833,95836,95837), and also at below-threshold heating currents (95839,95840). With V_supply raised to 320V, shots were taken with I_H=9.5A (95841-95843), I_H=11A (95844-95846), I_H=7A (95848,95850,95851), and I_H=6A (95847). With V_supply dropped to 113V, shots were taken with I_H=9.5A (95852-95854) and I_H=11A (95855-95857).

It does appear that breakdown is nearly impossible below I_H~7A due to insufficient emission current (there is a steep transition in I_E and consequently V_B between 6 and 7 amps). Away from this transition region in I_H, the strongest influence on I_E is V_supply. Analysis of any more subtle results will follow.

Friday January 20 2017 10:37 am Brooks 95858-95904 TPS2 angular scan

The goal of yesterday's experiment (1/19/17) was to perform an angular scan of the section 2 triple probe (TPS2). In total three sub-sets of experiment were completed. For all of the following shots, TPS2 and TPS8 were both at 103cm.

1. TPS2 and TPS8 were both set to floating. TPS2 was rotated in 45 deg increments (0,-45,-90,-135, and -180), and 2-3 shots were taken per location. Results suggest the previously observed odd floating voltage measurements followed the probe tip and not the probe tip location.

2. At both extremes (0 and -180 deg), the floating TPS2 probe had all 6 permutations of its wires changed around, and a shot was taken at each location. This data is still being studied.

3. TPS2 and TPS8 were both set to powered, and task 1 was repeated. This data has not be analyzed yet.

FE walls back 5mm. SS walls back 13mm. Section 2 and 8 shells were completed retracted.

Of note, the rotating mechanism on TPS2 is leaking at certain angular positions (-135 and -180 deg) but not at others (-90, -45, and 0 deg). For reference, the base pressure when leaking was around 2e-7, and it was 3e-8 when not leaking.

Tuesday January 24 2017 11:33 pm Levesque/Brooks/DeSanto 95905-95929 Testing prototype EUV upgrade TIA board

[Run summary for January 20-23.] Tested prototype EUV upgrade transimpedance amplifier (TIA) board in a temporary setup on the machine, connected to the diode array that is presently used for the SXR Fan diagnostic. The first shots (up to 95911) were plasma shots using the standard SXR Fan array setup in order to establish a baseline for the later tests of the new amp board. Shots after 95911 were crates-only for testing the prototype board. A high-level high-frequency oscillation (MHz range) was seen on the recorded outputs of the TIA board. This oscillation has not yet been corrected. No banks-active or plasma shots were taken with the prototype board connected due to the high-level oscillation. SXR fan array hardware was set back to the normal configuration at the end of testing.

Wednesday January 25 2017 9:36 am Levesque/Brooks 95930-95944 Preparation for experiment JPL09 with VF only shots

Set up the machine to prepare for run plan JPL09. External Rogowski D is temporarily connected around the North rack grounding strap; positive currents in this diagnostic are a measure of current from the rack case (the intended single-point ground location for chamber Sections 1-3) to the grounding bus.

Initially connected Ext Rogowski A across quartz spool 9.5 top, and the biasing hardware across spool 9.5 midplane, in order to measure voltages and currents for VF-only pulses with the biasing hardware switch set to 1) "floating" and 2) "grounded" (i.e. shorted together at the diagnostic box). No jumpers were installed across quartz spool 0.5. Most of the day was spent troubleshooting the unexpected signals that were seen during this test. At low VF settings, the voltage measured in the floating configuration was reasonable and low. Higher-than-expected currents flowed through the external Rogowski diagnostics -- this has not yet been resolved. At higher VF settings that were used for recent plasma shots, the voltage measurement was unreasonably high and saturated the digitizer. Several VF settings were tested between the high and low VF currents above, and a threshold is seen where the apparent measurement suddenly jumps from ~-0.2V to clipping in the A14 channel at -5V. This is suggestive of a large common-mode voltage at the digitizer that could be introduced by an inappropriate ground reference, as induced voltages in the circuit should be too low to allow an arc. A Jensen isolation transformer was installed between the voltage divider and the digitizer, but this didn't resolve the problem. No significant current was seen in the biasing hardware during the floating tests, as expected.

Using a VF-only pulse in the "grounded" switch configuration (shot 95944) produced current in the biasing hardware, as expected from the pickup loop established around spool 9.5. A corresponding current appeared in the Rogowski A jumper that completes the pickup loop, but it competed with a similar magnitude opposite current that was present for all VF tests.

Currents appeared in all ext Rogowskis for all VF shots today, including those where no current was expected. Rogowski D showed ~50-100A flowing through the North rack grounding strap, suggesting that a large ground loop exists and includes the grounding bus in the loop. Troubleshooting will continue today. Shell configuration is presently: Sections 2, 8, and 10 fully retracted, other sections are in the "all-in" configuration of the post-Fe-wall standard layout. At the beginning of the day, gains were corrected for the new External Rogowski nodes that were generated on January 9th.

Wednesday January 25 2017 6:06 pm Levesque/Brooks 95945-95967 VF-only shots and troubleshooting Chamber 10 biasing XP voltage/current measurements

Continued troubleshooting the intended hardware setup for the Chamber 10 biasing experiment. Problems with saturation in measured voltage appeared to go away at the beginning of the run, but I later found that I had accidentally disconnected the terminal blocks at the Jensen isolation transformer input. Repairing this connection showed that the voltage spikes had not disappeared. The source of this problem is still unknown.

The e-gun setup is producing a ~3VAC (60Hz) difference between chambers sections 9 and 10, measured with a multimeter, when section 10 is set to floating. This difference changes with the e-gun setup, and goes away when the e-gun leads are disconnected, but not when the e-gun circuit power is simply disconnected. Paul, Pat, and I found and investigated this a couple of months ago, but were unable to resolve it or determine if it was really a problem for the experiments at the time.

The external Rogowskis located on the machine were found to have significant magnetic pickup when there is no enclosed current. This can be seen by comparing VF-only shots 95953 and 95954, where the only difference was that I rotated each Rogowski ~180B0 in place about its jumper axis. The apparent current seen by Rogowski A during the VF startup for example changed from +30A to -30A upon rotating (not flipping) the Rogowski, while the jumper was actually disconnected and would not allow enclosed current. Rogowski D is far from the equilibrium coils and still saw ~100A flowing through the North rack ground strap, suggesting that this current is real. Careful pickup subtraction may need to be used when interpreting the external Rogowski signals, depending on the experiment. Further pickup tests should be done to quantify the potential impact on previous results, especially during disruptions.

Saturday January 28 2017 10:13 am Brooks 95968-96086 TP - FE wall test

To goal of Thursday's (1/26/17) to Friday's (1/27/17) experiments were to test if the presence of the FE changed the boundary of the plasma. TPS2 was radially scanned with the FE wall "in" (5mm back) and out. The SS walls were "in" (13mm back) the entire time. Preliminary results were emailed out Friday evening, but no strong change in the plasma boundary was observed between the two cases. Of note, pulling the FE walls out between cases caused q to shift down and the plasma current was shift up by quite a bit. Therefore, the above findings come with the caveat that the plasmas in the two different cases weren't identical. ShotNo 96034 shows an example shot style with the FE walls "in", and ShotNo 96085 shows an example shot style with the FE walls out.

Thursday run day experienced two bank triggering failures on ShotNos 95990 and 96001. In both cases, several of the banks didn't fire (but different banks each shot).

Thursday February 2 2017 9:45 am Brooks 96097-96102 TP and BP setup and testing

The goals of yesterday's tests (2/1/17) were to vet TPS8 and determine its optimal configuration for data acquisition. In addition, BPS5 and BPS9 were to be reinstalled and testing using their new enclosures. Unfortunately, we had KMAC issues, and the banks never fired. We finally setup a wireless webcam to view the KMAC in the basement to help us diagnose future issues (today).

Thursday February 2 2017 4:56 pm Brooks 96103-96116 TP and BP setup

The goal of today's run day (2/2/17) was to finalize the setup of TPS8 as well as BPS5 and BPS9. TPS8 appears to be working just fine, but I am using a different probe voltage layout that TPS2. The reason for this discrepancy isn't know. BPS5 and BPS9 both appear to be working in their floating configuration. Jeff helped me solve a auto-DC removal process in the tree that was affecting some of the data. I still need to recheck their radial calibration as well as try both in a grounded and driven mode.

KMAC issues this morning prevented me from running. Later in the afternoon, the KMAC in the basement rack decided that it wanted to work after all, and I managed a limited run day this afternoon.

Thursday February 9 2017 11:09 pm Levesque/DeSanto 96117-96126 EUV prototype board testing

Crates only plus a vacuum shot and 2 plasma discharges (96125, 96126) for testing the EUV prototype board. Up to 5 diodes from the SXR fan array were connected to the new TIA board and digitizer. Recently-observed bad features of the signals were not present for these tests. It's possible that problems in the most recent previous tests were caused by the leads from one of the unused diode channels being shorted together in the amplifier box, which was found and fixed before shot 96119. Signal levels from EUV TIA board outputs during the plasma shots matched expectations. Detailed analysis of the signals still needs to be done, including noise analysis and measurements from mode activity.

The base pressure was also observed to be high (up to 1.0e-7 torr) since the beginning of the week, independent of this run. Cryo pumps were both significantly warmer than usual (above 20K rather than around 15K), and the RGA showed almost entirely deuterium. The puff line was pumped out and refreshed several times to test for a leak across the valve (no evidence found for a leak to the chamber), and also since a puff test suggested that some air was in the line in addition to the D2. The large pump stand cryo was regened; this brought the base pressure back down to a normal range when only one cryo is active (~1.8e-8 torr).

Friday February 10 2017 3:35 pm Brooks 96127-96137 Bias Probe testing

The goal of today's run was to continue testing the Bias Probes. A few shots were taken today, but KMAC issues plagued both the onset of the day and returned later to force the experiments to stop early.

Shots 96127 and 37 both suffered from KMAC failures. 96128 gave NOX issues from the North Rack. From shot 96137 to the end of the day, the KMAC system wouldn't work, and the run day was abandoned after some waiting, resetting, etc.

BPS5 and BPS9 both appear to be working although the current for BPS9 is inverted and should be flipped in make tree. Also, BPS5 is showing a fair amount of magnetic pickup from the banks. The last time this much pickup was observed on a bias probe, it turned out to be a grounding loop issue. BPS5 and S9 still need to be tested for in their driven mode of operation.

Even shells were in. Odd shells were out.

Tuesday February 14 2017 5:40 pm brooks 96142-96157 BP testing

Shots today suffered from more CAMAC issues. After trying several things, we again pulled CTX out of the Jorway system, and everything seemed to working again. This matches our observations that our most recent CAMAC issues started about the same time CTX was put back in.

The goal of today was to test the BPs in sections 5 and 9. Both appear to be fulling functional. BPS9 was successfully driven in a feed forward configuration using a sig gen and an amplifier. When trying to do the same with Caliban (the GPU feedback computer), an odd error occurred. I'll poke at it some more, but I may need to bug Quien about it.

Tuesday February 14 2017 5:41 pm brooks 96142-96157 BP testing

Shots today suffered from more CAMAC issues. After trying several things, we again pulled CTX out of the Jorway system, and everything seemed to working again. This matches our observations that our most recent CAMAC issues started about the same time CTX was put back in.

The goal of today was to test the BPs in sections 5 and 9. Both appear to be fulling functional. BPS9 was successfully driven in a feed forward configuration using a sig gen and an amplifier. When trying to do the same with Caliban (the GPU feedback computer), an odd error occurred. I'll poke at it some more, but I may need to bug Quien about it.

Forgot to include Jeff's comment: Brief external Rogowski magnetic Pickup tests were done during the run. Rogowski jumpers A and B were configured as they were for the shot 94664 series used in the recent IAEA Nuclear Fusion journal submission. The two Rogowskis were rotated by 180 degrees part way through the run to look for a difference in apparent measured jumper current during current quenches. Two appropriate shots to compare are 96153 and 96155, where both Rogowskis were carefully rotated by 180 +- 5 degrees between shots. Upon quick inspection, there is no significant change in the measured jumper currents during the disruption (i.e. no strong external pickup) -- a strong toroidal jumper current is still observed predominantly in the co-I_p direction.

Wednesday February 22 2017 10:28 am Brooks 96164-96185 TPs radial scan with biased BP

Yesterday (2/21/17), I set out to do several TP radial scans with the BP at several bias voltages. After about 20 shots, the previous CAMAC issues started reappearing. At first, lights 4 and 6 on the basement slot 1 didnt' light up, and the VFE and OHE banks didn't fire. After additional testing and giving time to "sit", only lights 1, 2 and 7 worked. Run ended. Of note, shot 96185 showed the failure associated with lights 4 and 6 off and the VFE and OHE banks not firing.

Wednesday February 22 2017 6:45 pm Brooks 96190-96235 TPs radial scan with biased BP

First, I'm pleased to report that no CAMAC issues happened today. Second, I developed a reasonable shot style and managed to take most of the data I needed for this experiment. I biased the BPS9 at +80V, -80V, and floating and inserted it at 104.5cm. Next, I performed a radial scan with both TPS2 and TPS8 at 103, 103.5, 104, 104.5, 105, 105.5, 106, 107 and 108cm for all three BP voltages. I have not looked at the data in detail and cannot report any findings just yet.

SS walls "in" and FE walls out.

Thursday March 2 2017 10:36 pm Levesque/Brooks 96236-96273 Rogowski jumper pickup and ground loop tests

Ran plasma and vacuum shots to measure direct magnetic pickup on External Rogowskis A and B, jumpered across the 9/10 and 3/10 midplane quartz breaks respectively (the same as for the shot 94666 campaign). Both bias probes were inserted to roughly the same radii as for 94666 campaign shots, and both triple probes were out. Rogowski leads were carefully disconnected for pickup tests, while keeping Rogowskis themselves positioned and oriented the same before and after changing jumper connections. Rogowskis A and B were previously seen to have some direct pickup during vacuum shots which depended on their orientation, but direct pickup from the plasma equilibrium fields had not been quantified. In today's run, Rogowski A was found to have vacuum pickup that is different from vacuum shot 94746 (i.e. the reference vacuum shot for 94666 campaign), but has no pickup from the plasma current itself. Rogowski B has pickup at similar levels from both vacuum fields and from the plasma current. The best discharges for comparison are:

• 96256, 96257: Plasma shots, Jumper A and B leads connected across quartz breaks.
• 96261, 96262: Plasma shots, Jumper A and B leads disconnected on +phi side.
• 96258: Vacuum shot with Jumper A and B leads connected across quartz breaks.
• 96259: Vacuum shot with Jumper A and B leads disconnected on +phi side.
• 96263-96266: VF-only an OH-only shots for measuring pickup. Standard and low power.

At the beginning of the day, a Pearson 6062 (0.01V/A) was inserted around the West rack ground strap, connected to the local CPCI channel 96, to measure current flowing from the rack to the grounding bus. Rogowski D measured current from the North rack to the grounding bus (as it has been configured since shot 95930). Variations in these ground currents during the later times in plasma shots agreed well between the two current measurements to within a small offset (~4A); when fluctuating current leaves the plasma/chamber via the north rack, it does return through the west rack as expected when assuming those are the only two grounds. Early in the plasma shots however, the offset or amplitude between ground currents are much larger (~75A difference at startup) -- the difference between them decays in time. At shot 96269, another Pearson 6062 was inserted around the North rack ground strap to replace Rogowski D.

Shots 96267-96273 were used to measure the levels of currents that arise when ground loops are intentionally generated by jumpering across additional quartz pieces. With Rogowski Jumper C additionally connected across the 10/1 quartz break, a peak of ~230A flowed from Chamber 9 to Chamber 1 via the jumpers surrounding Chamber 10 at the start phase of a normal OH-only shot. Measured currents exiting/entering the racks via the grounding straps were: +100A at North, -150A at West. Source and sink of the difference in ground/jumper currents were not investigated. The same OH setting with no jumpers shorting across quartz pieces produced -70A at the North Rack and +20A at the West rack, indicative of ground loops even without the jumpers. When Rogowski jumpers were connected across all 4 breaks for a low-power OH shot, the resulting currents saturated the jumper diagnostics, and could not be calculated, as could be expected for having the full toroidal loop. Lower power tests could be used to measure the current, but were not attempted due to time constraints. At the end of the run, Rogowski jumpers were returned to just having Rog A and Rog B across the 9/10 and 3/10 midplane quartz breaks respectively.

Thursday April 13 2017 12:53 pm Levesque/Brooks 96274-96290 Ip Rogowski pickup test

This run report covers the run on Monday 3/6/2017. A series of shots were taken to measure the possibility of direct pickup on the continuous plasma current Rogowski coil due to current in the nearby (~1 inch away) toroidal jumper across the quartz piece at Section 9.5. Shots were taken with this jumper disconnected, with it directly connected across the break at the midplane (as we have configured it for previous tests), and with it connected via a radial extension running about 8 inches away from the Ip Rogowski. When connected, currents up to ~500 Amps were measured in the jumper during disruptions. Comparing the direct jumper connection to the extended jumper connection cases, there is no evidence for significant direct pickup on the plasma current Rogowski measurement due to current in the nearby toroidal jumper -- fluctuating signals in the Ip Rogowksi did not significantly change with the location of the jumper.

Friday April 14 2017 8:57 am Brooks 96292-96347 TP dual radial scan

The goal of yesterday's run (4/13/17) was to perform a dual triple probe scan to characterize the phase relationship between the MHD and the plasma properties as measured by the TPs. Unfortunately, shot development proved difficult and was never completed. SS walls were completely in, and FE walls were all out.

In addition to the TP scan, the HS camera was resurrected by Jake Lee and used yesterday to record HS videos of the TPS8 at 106.0 cm and 101.5 cm. The 106.0 cm case showed little of interest, but the 101.5 cm case showed several ablation events of the negatively biased probe tip. This ablation corresponded with arching events as recorded by the TP data.

Wednesday April 19 2017 10:34 am Brooks 96348-96426 TP dual radial scan

This run focused on developing the shot style for the TP dual radial scan test where I'll be measuring the phase relation between B_1_poloidal with both triple probes. The shot style, characterized by 96408, is mostly developed, but I damaged TPS2 on insertion. It needs to be removed and repaired.

While diagnosing the triple probe issue, I discovered that TPS8 V_{-} doesn't seem to be recording (which is optional anyway). In addition, the new TP nodes have their offset subtracted, and this makes their measurements questionable.

SS walls completely in (against stopper) and FE walls all out.

Friday April 21 2017 10:54 am Brooks 96428-96450 BP-GPU testing

The goal yesterday's run (4/20/17) was to test that the GPU control system is working and can control the bias probe. The GPU system is in fact working, and it is mode locking to the BP. The next step is to place the BP in the plasma and SOL, perform a phase scan, and study mode growth for the different phases.

Friday April 21 2017 4:11 pm Stewart/Brooks 96451-96461 Bias Probe Mode Rotation Study

With the FE walls all out and the SS walls in against the stopper, the first 7 shots of the series were made in order to calibrate the plasma major radius to an appropriate size below 93cm. The preliminary shots were also used to ensure the bias probe was working properly with the desired waveform outputs. The goal of the last 4 shots was to verify results from Chris Stoafer's thesis, namely the relationship between mode rotation frequency and bias probe voltage detailed in Figure 5.5 of his thesis. The mode frequency decreased between applied bias probe voltages of -10V to -2V (shots 96458-96461) with an amplifier gain of ~90%, in agreement with previous work in the thesis. It will be the topic of future runs to verify mode rotation frequency behavior for positive voltages on the bias probe. As an important technical aside, the new AETechron 7224 model amplifier was able to output approximately -73V on the bias probe from the input -10V.

Thursday April 27 2017 1:00 pm Brooks 96462-96532 BP in plasma, control phase scan

To goal of this run day (4/25/17) was to perform a controlled BP phase scan with the probe in the plasma. The BP was placed at 105cm, and the AETechron amplifer, operating at 100% gain, was driven with the control system at 5 +- 5 V sin waves. The phase of the sin wave was controlled by Qian's FB-Cuda code, and I performed a scan with the phase offset. The phase scan consisted of mostly 90deg steps with a few 45 deg steps. Data analysis is ongoing.

FE walls all out. SS walls all in against stopper.

96475 was an interesting shot with very steady mode amplitude and rotation.

Wednesday May 10 2017 6:43 pm Stewart,_Brooks,_&_Levesque_ 96533-96542 Searching for Arcing

In shots 96533 and 96534 (vacuum shots) with the bias probe inserted at 105cm of major radius, an anomalous noise was noticed and a spike in the OH coil current profile was also found in both shots. Subsequently, shots 96535-96542 were devised in order to determine the source of the current spike and noise with the hypothesis that there was arcing on the upper or lower portions of the OH stack. After a thorough investigation of video footage and visual inspection of the machine, no arcing could be found originating from the magnetic field coils or the vacuum vessel. From the subsequent runs, it was found that the appearance of the current spikes on the OH coil reading coincides with presence of the almost in-audible noise noticed by the several of the members of the HBT group. A possible resolution of this problem can be found in the run report associated with shot 96557, where a loose screw in the OHE capacitor bank (second section, eleventh row from the bottom, leftmost capacitor) was found to create a discharge.

Wednesday May 10 2017 7:12 pm Stewart 96552-96579 Bias Probe Mode Rotation Study

The purpose of this run campaign was to verify the results found in Chris Stoafer's thesis regarding the relationship between mode rotation and bias probe voltage. However, shots 96552-96553 were intended to investigate a prior hypothesis of arcing on the OH coil and consisted of a vacuum shot at half bank power and a vacuum shot at full bank power respectively. With no anomalous noise present or spike in the OH coil current profile, the campaign was continued with plasma shots 96554-96556. In these shots, both the nearly inaudible noise and current spikes in the OH profile were present, with the noise now localized to a lower portion of HBT or the basement capacitor banks. An analysis of footage during run 96556 showed that there was an electrical discharge from the OHE capacitor bank. The ensuing investigation revealed that a capacitor in section 2 (eleventh row from the bottom, first capacitor on the left) had a loose screw that may have been causing the discharge. After tightening the screw, no anomalous noise was heard and no discharge was seen upon investigation of basement footage at that location during the subsequent run campaign for shots 96558-96579. Runs 96558-96566 were designated for shot development with the bias probe inserted at 105cm with floating potential voltage (SS walls were against the stopper and FE walls were all out). The bias probe position was recalibrated before shot 96565 and the probe placed at 103cm of major radius (probe tip location), as it was in Chris's thesis. The input bias probe voltage for shots 96567-96579 was varied between -4v and +4V with an output bias probe voltage ranging from ~-76V to ~+32V. As was found in runs 96458 to 96461, the mode frequency was higher with the more negative bias probe voltages. A more thorough analysis is required to verify this analytically and to compare with the results from Chris's thesis.

Friday May 26 2017 5:49 pm Stewart 96590 96604

The focus of this series of shots was to evaluate possible cross-talk on the EUV prototype board. The setup consisted of 2 diodes from the SXR Fan array attached to channels 6 and 8 on the EUV prototype board and an external diode array with 1 of its diode elements attached to channel 7, again on the EUV prototype board - i.e. adjacent to the SXR Fan channels. If crosstalk were to occur, it would likely show up on this channel (This experimental setup was discussed with Jeff Levesque prior to this run). For shots 96590-96602, no cross-talk was observed for the three adjacent channels (channel 6-8 on the EUV prototype board). It is however interesting to note that using vacuum shot 96600 as a baseline, it appears that there could have been x-rays detected by the diode in the box situated on top of the fan array amplifier box during a time corresponding to a disruption. The ensuing shots (96603-96604) were intended as an extension to the campaign of clean-up shots that were also used to test for cross-talk.

Friday June 9 2017 11:19 am Stewart 96605-96637 Bias Probe-Mode Rotation Study

The goal of this study was to fix the spectrometer system and to confirm results from previous work on bias probe induced mode rotation. Shots 96606-96607 were used to debug the issues with the spectrometer, which was giving an unusual noise-like signal output. For shot 96608, the BNC cable connecting the spectrometer to the screen room crate was replaced. The spectrometer functioned normally for the next 29 shots of the experiment campaign, thus verifying the problem as a fault in the connection wire. Shots 96623-96630 were designed to test the operational range of the new amplifier with inputs of up to +10V and -10V. It was found that an input voltage of +10V produced an output voltage of +90V, whereas an input voltage of -10V was capable of producing an output of -175V. This leaves the high positive regime of +90V -> +110V inaccessible using this amplifier setup. It was found that the high negative bias voltages induced fast rotation in the plasma in the direction of natural mode rotation as was found before. The mode frequencies were found to hover around 0 for the positive voltages, denoting a requirement for higher positive bias for mode rotation in the opposite direction. It is also interesting to note that the shots with high negative biases lasted consistently longer than unbiased plasmas with the same coil settings. The plasma currents also persisted much longer for the negatively biased cases than for the positively biased cases, which showed almost no mode rotation and appeared to be more susceptible to what could be locked modes. Further analysis is need to verify this phenomenon.

Wednesday June 14 2017 5:53 pm Stewart 96638-96646 Shaping Coil Testing

The goal of this series of runs was to verify that the shaping coil system was still operable and able to create a single null diverted plasma configuration. The first shot, 96638, was intended as a test to determine if the shaping coil would trigger correctly. The bank charged independently of a signal from the control system for charging the banks, indicating that the wiring to the phidget board may have been altered. The output current for the shaping coil from this shot was ~2000A. The shape of the current profile also indicates that the SH start bank did not fire. The subsequent shots were used to debug the issues with the shaping coils, with the connection BNC cables for the SH start and SH CB between the banks and the phidget board having been found by shot 96642. It remains to be resolved if the logic circuit connecting the SH CB power supply is in the correct configuration and why the SH start is not firing during the shot.

Wednesday June 14 2017 6:51 pm Brooks 96647-96672 TP Locked mode detection

Interesting run day today. After a long setup, I finally started running around noon. A few notable things happened.

1. TPS2 (triple probe section 2) was reintegrated. The probe started arcing quite a bit, but it did less and less as the runs progressed. Jim suggested that "dirty" tips from his recent reinstall may be the culprit. Also, the negative probe tip measurements on TPS8 are still weird. I think it's picking up noise from something. Fortunately, I trust the positive and floating tip measurements and the power source. Therefore, that particular measurement is redundant and therefore ignored.

2. Jim repaired the stuck shell motor on section 8 top. Thanks, Jim.

3. I started my run by using Ian's developed shot style from yesterday. He had detected what he thought were locked modes, and my primary goal of today's run was to use the TPs to detect locked modes. Ian's shot style also consisted of using BPS9 (biap probe section 9) at +50V at 103cm. While trying to reproduce this shot style, I accidentally programmed the BP to start ramping up at 0.4 ms for 1 ms and hold steady until 6ms. Ian started his probe a full ms later. Of note, plasma breakdown on HBTEP occurs at 1.0 ms. This resulted in a very fast negative rotating mode (approx -20kHz) around 1.5ms. The frequency slowed down to 0 (around 3.5 to 4ms) where a locked mode appears to occur briefly. Then the mode rotation switched to positive frequencies, the mode amplitude started growing, and the plasma crashed inwards without an obvious disruption. 96670 is a good example of this behavior.

4. I performed a radial scan with both TPS2 and TPS8 from 103cm to 108cm in 1cm increments. Both probes were always at the same location. Of note, the probes show interesting behavior at about the same time that the locked modes were observed in the stripy and frequency plots. More specifically, the probes often (but not always) showed a deviation in temperature and density measurements at these times. Also, TPS2 frequently showed a large negative dip in floating voltage. I'll try and send out sample plots tonight.

Final notes: SS walls all in to stopper. FE walls all out.

Thursday June 15 2017 5:02 pm Brooks 96673-96740 TP radial scan

Today's run day focused on a radial TP scan from 103 cm to 108 cm covering both the plasma and SOL. The goal was compare the phasing of the triple probe data with the measured B oscillations as a function of radius. It is expected that a phase difference between the two should be a function of radius. Data is being processed.

At the end of the run day, Ian and I successfully repeated some of yesterday's results. We also pulled the plasma off of the outer limiter (76740) and did NOT observe the large negative rotation that dominated the plasmas against the outer limiter.

SS walls were all in against stopper. FE walls were all out.

Wednesday June 21 2017 10:04 am Stewart 96742-96784 Radial and Temporal Bias Probe Scan

The goal of this set of runs was to determine the effect of timing of the bias probe ramped voltage and the radial location of the bias probe on mode rotation. Shots 96742 to 96752 were devoted to shot development, where the VFS and VFS bank voltages were altered to achieve a plasma centered at approximately 91cm of major radius. The next 30 shots (96753-96782) were devoted to determining the effects of bias probe timing on mode rotation with previous bias probe work by Brooks indicating that an early start time for the bias probe voltage (with the end of the ramp at ~1.5ms) induced faster negative mode rotation than previous runs with the end of the ramp set to ~2-2.5 ms. This fast negative mode rotation state was observed for shots 96767 and 96763, which were set to ramp intervals of [0.0-1.0 ms] and [0.5-1.5 ms] respectively, with the onset of negative mode rotation corresponding to the end times of their respective voltage ramps at ~1.2 and ~1.5 ms. The peak negative mode rotation was at ~ -24 kHz for shot 96767 and ~ -23kHz for shot 96763. Shots that had later bias probe ramps such as 96780 [1.0-2.0 ms] and 96782 [1.5-2.5 ms] showed lower peak negative rotation frequencies of ~ -11.5kHz and ~ -13 kHz at times corresponding to the end of each ramp at 2.0 and 2.5 ms. For the shots with the faster negative mode rotation (96763 and 96767), there was a marked decrease in the current measured flowing through the bias probe and an increase in the voltage during the fast mode rotation interval, perhaps delineating the formation of a transport barrier or other mechanism impeding current flow to the bias probe. From the temporal scan it is also possible that density plays a role in the magnitude of the mode rotation frequency, where early shots show approximately double the mode frequency induced by the bias probe in comparison to the later shots (formation of the plasma is at ~1ms and the highest rotation states were found to be induced at this time interval). The radial scan of the bias probe from 103cm to 105cm of major radius during the last set of shots showed a decrease in induced mode rotation frequency as the bias probe was moved radially outward. The shots that were taken farther from 103cm also showed a decrease in current running through the bias probe. This may indicate that the optimum position (to collect the most current and induce fast mode rotation) for the bias probe is indeed at 103cm.

Wednesday June 28 2017 10:09 pm Stewart 96796-96845 Two Bias Probes: Calibration

The goal of this series of runs was to verify that the two bias probe setups at sections 5 and 9 were working properly and were able to operate independently of one another. From the first set of shots (96797 - 96806) that used different combinations of both bias probes, programmed to execute voltage ramps at 1.5ms, it was apparent that the waveform output on BPS5 was being truncated both in terms of voltage and in time. The remaining shots were devoted to debugging the issues with the section 5 bias probe and it was later found around ~96840 (a crates only shot) that the input to the Jensen needed to be grounded in order for the voltage ramp waveform to be completely passed to the amplifier and sent out to the probe. The bias probe at section 9 worked as it has in previous runs with an output voltage of approximately ~110V with an input of +10V. It was also determined that the CE4000 amplifier had a 1 ms delay that was not present in the AETechron amp. This time delay between input pulse and output pulse was corrected for by programming the probe input to begin the ramp 1 ms earlier. The waveforms between the two bias probes matched up in the later shots of the debugging section (e.g. shot 96845). It is however interesting to note that in the shots where both the probes were in amplifier mode and only one probe executed a voltage ramp, there was a current observed in the non-ramping probe with the opposite sign of the current flowing into the ramping probe. This may indicate that a current is flowing through the plasma between the two probes.

Wednesday July 5 2017 8:23 am Brooks 96846-96891 TP test and other misc.

A number of tasks were completed on 6/29/17.

1. A thorough hardware and electrical connection test was conducted on the two triple probes (TPS2 and TPS8). Both worked as expected with the exception that TPS8 had a fairly noisy signal even with crates only shots. In addition, the zero offset feature still needs to be removed from the tree nodes.

2. Next, both TPs were tested against one another in plasma shots. Both were setup in their default setup and at 103 and 105 cms. Both provided nearly identical measurements. Both probes are therefore expected to be working correctly.

3. Next, BPS5 and BPS9 were used independently to attempt to lock the MHD mode to the chamber. Unlike previous tests, BPS9 only tests were unable to lock the mode to the same location despite repeated tests. BPS5 also was unable to lock the modes to the same location.

4. Next, step-up/step-down ratio on the ACME transformers were changed from the typical 1:1 (240:240) configuration. For background explanation, the AETechron amplifier appeared to be saturating at 50 amps. Ian noticed that the manual listed 50 amps as the maximum output. In these tests, changing the transformer to 2:1 (240:120) provided a small increase in total power delivered (500 to 1000 W).

5. Finally, I attempted to continue my TP radial scan from the previous week but the modes (magnetic sensors) were either too weak or didn't strongly couple to the TP data. I have some older data that I'm going to look through.

6. Ian finished the day with a few double BP tests.

SS shells were in against bumper. FE shells were all out.

Wednesday July 5 2017 1:22 pm Stewart 96893-96918 Two Bias Probes: Study

The goal of this series of experiments was to determine the effect of using two separate bias probes on sections 5 and 9 on the plasma mode rotation. Shots 96893-96898 were devoted to testing the double bias probe setup with independent timing for each probe (the AEtechron and CE4000 amplifiers were used in this study). The section 9 bias probe waveform was held constant with a voltage ramp from [0V +10V] of input while the section 5 probe waveform timing and final voltage were modified. It can be seen from shots 96899 (BPS5 ramp [.5ms 1.5 ms]), 96908 (BPS5 ramp [1.0ms 2.0ms], 96912 [2.0ms 2.5ms], and 96915 [2.0ms 3.0ms] that the negative mode rotation initiated by BPS9 is sustained for successively longer periods of time with negative rotation spikes occurring when the second probe completes the voltage ramp.

In particular, shot 96912 has a negative mode rotation state lasting approximately 3ms. The current in both probes also drops three times throughout the shot, which is perhaps indicative of transport barrier formation, however the drop occurring at ~1.5ms is the largest. This drop occurs when the current ramp from BPS9 has been completed. At coinciding times, the fast reverse mode rotation peaks below -20kHz. This behavior is also clearly seen in shot 96908 with the same -20kHz threshold.

For shots 96910 and 96913, when the plasma was moving radially away from the two probes, there was an increase in voltage in BPS9 and a decrease in voltage for BPS5 periodically (with a period of roughly 0.2ms, which corresponds to the period of the n=1 mode observed during the probe voltage oscillation).

Wednesday July 12 2017 10:29 am Stewart 97125-97160 Two Bias Probes: Study

The goal of this series of experiments was to perform checks on the previous series of runs and to test the QSC amplifier. For shot 97154, the section 9 bias probe was set to ramp to +10V using the AE Techron amplifier, while the bias probe at section 5 was set to float. For a majority of the shot, the floating potential of BPS5 was ~ -23V but during the period of high reversed mode rotation the probe read a potential of +75V (at 1.6 ms). This may indicate a dramatic change in the floating potential during the reversed rotation state. As a side note, during testing, shot 97153 saw negative mode rotation for almost the entire length of the shot (~1 ms to 5 ms).

The last 5 shots were taken using the QSC amplifier as the source for BPS5. The peak power output for this amp can be seen (briefly) in shot 97160 (170V and 50A) 8500 watts at the 2ms mark. The waveform, however, sags during the course of the shot. Further testing is required to diagnose this issue.

Thursday July 13 2017 11:29 am Brooks 96924-97124 BP-GPU testing

The run report here covers three run days, 7-5-17 to 7-7-17. The goal of these three run days was to try and determine if the BP-GPU feedback system has any influence on the MHD modes in the plasma.

Unfortunately, the developing a reliable and consistent shot style proved very difficult. Much variation occurred from shot to shot and made developing a large, repeatable shot database difficult. In addition, the shot styles seemed to shift throughout the day, and it was not always possible to change the bank settings to cause the shifting shot style to return. Finally, finding a shot style that was both outward limited and with strong mode activity proved to be an extra complication.

On the 7th, I finally managed to decent shot style. In this case, BPS9 was placed at 106.5cm (near the SOL). Shot 97087 marks the beginning of my feedback shots for the day. I completed several full phase scans at a constant amplitude and a few amplitude scans at constant phase. Despite the repeatability and decent mode activity, I was not able to observe any obvious effect of the various feedback parameters on the plasma. More analysis is necessary.

SS shells were all in to stopper, and FE shells were out.

Thursday September 21 2017 11:10 pm Levesque/Stewart 97239-97248 Firing OH coil to look for dropped tools, bake problem

Pulsed OH-only shots to look for tools that were dropped around the tokamak during the up-to-air period over the past 2 months, per our standard tool-seeking procedure. There are 3 known offending objects being sought: a wrench near the small pump stand, a small hex key on the south side of the machine (fell down a tube -- probably trapped), and a strong permanent magnet near the edge of the base pad at Section 2. At high OH settings, Jim and I (Jeff) were unable to hear movement of any objects near the machine. We will continue with cleanup around the machine and searching for tools before further pulsing of the magnets, especially the TF. This venting period accomplished removal of the ferritic wall and installation of LFS scrape-off layer tile diagnostics in Sections #1 (8 tiles), #4 (4 tiles), and #8 (8 tiles).

The full CAMAC fiber optic highway was resurrected over the past couple days, with all HBT racks and CTX in the loop. We found that the south rack U-port adapter is very sensitive to insertion depth/focusing of the output fiber -- screwing the SMA connector on all the way prevented enough light from getting to the next crate, while having the fiber barely inserted gave enough light on the other side of the fiber optic cable to relieve sync problems.

Today's run started with the common CAMAC problem where pulse signals were not sent to the krytrons for most banks, but this seemed to heal itself after some unrelated troubleshooting of bank dump relays. The OHE bank dump initially would not open for the bank to charge -- this was fixed upon finding a disconnected banana plug in the back of the control panel.

Since OH was the only coil to be used, and all bake circuit connections come from one location with no intentional direct electrical connections to the machine, I figured it would be safe to pulse the OH with the bake still operating. However, pulsing the OH appears to have broken the Arduino controller for the bake power relays. This has ended the bake for now, until the controller can be repaired/replaced. The bake was active continuously for 6 days, following a 2-hour glow with D2 and 1-hour glow with He. The base pressure was around 3e-7 torr before the bake faltered, dominated by water, with D2/He as the next largest partial pressure. We noticed a problem with the bake because the chamber pressure was quickly dropping, which suggested that the bake had turned off and the chamber was cooling. While we were checking around the bake controller and power relays, Jim noticed smoke coming from the heater pad on the Thomson scattering section, so we turned off the bake power. This excessive blanket heating might not fully be caused by malfunction of the bake controller; the power connector on the chamber side for this blanket was replaced this week since it was too melted to maintain a good connection. This suggests that the TS section has had problems in the past, and might normally draw too much power under certain conditions. The resistance of this blanket is as expected from other blankets.

Saturday September 30 2017 12:06 am Levesque/Stewart 97294-97326 Diagnostic upgrade and vacuum shots.

This week, ex-vessel signal routing was completed for the LFS SOL diagnostic tiles. Measured currents are in the MDSplus tree under the nodes .SENSORS.SOL:LFS##s#" where the first ## is section number, and the last # is local sensor number, following the convention of the FB sensors. All 20 SOL sensors are functioning for tests so far, though some have significant drift in inferred current (such as LFS08_s4).

Thursday and Friday we attempted to take vacuum shots and make plasmas, but the CAMAC triggering was not cooperative. With several reboots and hardware unplugging/replugging, we were able to get enough partial-system triggers to fire OH shots (97300 was high power OH-only) to check for tools again, and to ramp up the TF to the usual 6.1kV setting (shot 97307 was 6.1kV). There were two successful full vacuum shots with TF, VF, and OH (97320 and 97324). Every attempt to include a gas puff resulted in the puff valve opening for too long due to a faulty trigger, and the the ion gauge tripping off as the chamber pressure rose too high. The North rack CPCI failed to store for quite a few shots. There were also triggering problems for the screen room J221 module, seen from the indicator lights on the J221 and failure of the USB spectrometer to trigger properly for most or all shots.

Successful shots revealed fairly strong pickup on the SOL tile current measurements from the OH and TF. During a typical plasma lifetime period, the worst sensors read up to 27amps calculated during the OH-only shot, and up to 20amps from the TF-only shot. VF pickup is also significant, but hasn't been separately measured. This pickup will have to be subtracted based on vacuum shot coupling. There is likely to be strong direct pickup from the control coils, though this hasn't been tested yet.

Regarding standards of machine operation, all ferritic-wall-related stoppers for shell insertion/retraction have been removed for storage, following removal of the ferritic wall. Shells were fully inserted for this run. From now on, unless otherwise indicated, "all-in" shell positions will have the shells inserted to the stoppers that were installed in 2010, and likewise for "all-out" shell positions (except for Section 6B, which has an extra retraction stopper from the installation of the TS viewing dump in September 2012). This is in contrast to the standardized Ferritic-related wall configurations established by Paul Hughes in 2014. The updated "all-in" and "all-out" positions now match those from before the 2014 Fe wall install.

Friday November 3 2017 11:56 pm Stewart 97401-97412 Clean-up/SOL testing

This was a series of clean-up shots taken after making the following modifications to allow puff-valve triggering: 1.) We swapped the J222 module from the screen room crate and placed it in basement crate, connected to the J221 module in slot 9. The triggers were set as: Output 3: Gas Puff and Output 5&6: TF Start (all in make tree). 2 .) The J221 module in the screen room was disabled since it no longer had a J222 (only in traverser) 3.) Outputs from a J222 in the Northrack were routed to the screen room to trigger both the Dalpha spectrometer and USB spectrometer (the outputs were only modified in traverser), since the J222 from the screen room was removed. This is a temporary solution as Jim is currently working on fixing the custom J222-equivalent circuitry from the basement rack. Reason for Modifications: The last working basement Jorway 221/222 trigger outputs sourced from slot 9 stopped working while we were trouble shooting puff valve problems. This reduced the number of basement triggers below the minimum number needed for plasma shots, which require the puff valve and TF start triggers. Shots 97403-97412 had sustained plasma currents of ~2 ms and major radii that varied greatly over the duration of the shot. These shots, however, indicate an error in the timing of the USB spectrometer and the Dalpha spectrometer most likely due to our work around using the triggers from the north rack. Of note is shot 97402 that had a sustained (low) plasma current for ~6 ms. This was the first plasma shot of the day and featured wildly fluctuating major radius and calculated q. These were also some of the first shots to feature data from the new SOL tile Rogowski measurements.

Monday November 6 2017 10:46 pm Levesque 97413-97433 Cleanup shots, Collecting LFS SOL tile data

Ran cleanup shots. There were no CAMAC triggering problems today, other than the continuing D_alpha spectrometer and USB spectrometer timings. Based on the time reported by the USB spectrometer, and measuring the trigger for the D_alpha spectrometer, the noise spike from TF_START firing seems to be generating a spurious trigger for both diagnostics. The result is that D_alpha data is recorded long before the plasma, and the main plasma spectrum from the USB spectrometer appears one spectral trace later than it should (presently showing up in spectrum_04 instead of spectrum_03). Improving the temporary trigger cables for these diagnostics could resolve this problem.

While setting up, I noticed that shells 2B and 4T were retracted most of the way, while all other shells were fully inserted. This was probably the configuration for the most recent run day(s), and could be checked by examining the FB sensor data for those two shells if shots from the recent couple of runs are analyzed. Shells were fully inserted for this run.

Plasmas greatly improved throughout the day. By the end of the day, plasmas lasted for up to 4ms, including the current quench. Better discharge programming could have lengthened the shots. Coherent modes were apparent for many shots. LFS SOL tile data showed correlation between current fluctuations at the tiles and rotating 3/1 modes. A good example of this is shot 97418 around 2ms, with vacuum shot 97421 having the same bank settings -- tile LFS04_S3 has peak-to-peak oscillations of about 10A, while nearby magnetic sensors (FB03_S3P, FB04_S2P, and FB05_S3P) have peak-to-peak oscillations of about 8G.

The radial location of the wall and associated tile does make a difference on the measured tile current oscillations as expected. This is seen by comparing the LFS04_S3 and LFS04_S4 tile signals from the Friday November 3rd run to their signals from today's run, and considering fluctuation levels in other tiles. These two tiles were unintentionally retracted by 3-4cm for Friday's run, while today they were fully inserted. Fluctuations were small/negligible when the tiles were retracted, while they were strong today. A scan of LFS SOL tile signal versus wall radius is planned for upcoming experiments.

FB sensors in two sections need to be checked or properly connected per the following. Section 8 top and bottom FB sensor cables were repaired after the pump-down, but still need to be checked and reinstalled (i.e. FB08_s** are not connected right now). Cabling/polarity needs to be checked for FB sensors in Section 1 -- the polarity has apparently flipped for sensors 3R, 1P, 2P, and 3P. Without knowing the cable pinouts offhand, this problem might be from top/bottom cables being swapped, and/or it could be from an actual change in sensor polarity that may have been introduced when repairing connections on the vacuum side of the feedthroughs (possibly damaged/repaired during installation of LFS SOL tile feedthroughs). If this is determined to be due to a polarity flip in the vacuum, then polarities in the tree will be fixed for shots that follow the recent up-to-air, including today's shots. FB01_S1R appears broken; it was working or at least different before the up-to-air.

Wednesday November 8 2017 8:54 pm Stewart 97434-97459 Clean-up Shots

The goal of this series of runs was to continue the clean up process. Unfortunately, the CAMAC system again had the pervasive issue, where the banks (with the exception of the OH bias) do not trigger and the gas puff does not receive the second trigger to close the puff valve. After restarting all of the crates and remaking the tree, there was a single successful plasma shot (shot 97438) that lasted approximately 3 ms. The rest of the run day (shots 97439-97459) was spent trouble shooting the issues with the CAMAC system. The solution of remaking the tree and restarting the crates was attempted multiple times but the system remained in a state where ~30% of the "pulse on" tests would trigger everything correctly and the other ~70% failed to trigger the banks and the puff valve closure. A full reboot of Spitzer also did not alleviate the issue.

Sunday November 12 2017 7:39 pm Stewart 97460-97482 Clean-up Shots

The series of runs on 11/09 were designed to continue the clean up process, as well as continue testing the CAMAC system. Before starting up, the tree was remade using make_tree and the crates were initialized. Following a successful test (shot 97460) of the triggering system, a vacuum shot was taken (shot 97461). While all of the triggers for the banks successfully showed up on the J222 unit in the basement, the VF start bank did not fire during the shot. After trouble shooting, it was found that the amplifier in the basement rack for the VF start was no longer operable. I then switched the VF start to the amplifier in slot A0 in the basement and conducted a crates only trigger test. After the swap, the VF start bank fired normally during the shot (see shots 97475 and higher).

Before the swap and the realization that the VF start was not firing correctly, 3 successful plasma shots (97462-97464) were taken with puff times of 2 ms and 1.7 ms. All three shots lasted approximately 4.5 ms with unusual dips in the plasma current about 1.5 ms into the shot. Additionally, the shots featured erratic major radius position behavior and relatively small amounts of soft X-rays (perhaps due to the lower Ohmic heating from the lower plasma current). As per Jeff and Spence's fix for the spectrometer timing coming out of the North Rack, the Dalpha spectrometer is now triggering at the correct time but the traces look like they have lost information between some of the data points, yielding large vertical jumps.

After swapping the amplifiers for the VF start, the CAMAC system returned to the state where it was not sending the second puff valve trigger or the bank triggers (besides the TF bank). The rest of the run day was spent testing the CAMAC system with several successful vacuum shots being taken after the swap including 97475, 97479, and 97482. On 97479 and 97482 gas was not puffed into the chamber which may indicate that the gas puff alone was not triggered in this case.

In terms of hardware changes to the machine, the section 8 feedback sensor cable (which was previously grounding that section before it was disconnected in August) was repaired, checked, and reinstalled.

Tuesday November 14 2017 11:42 pm Stewart/Levesque 97485-97520 Clean-up/SOL testing

Shots 97485-97520 went smoothly, without any of the CAMAC errors seen last week. Before running, the tree was remade and several crates only/gas puff tests were conducted with no errors from the CAMAC system observed for any of the ~40 tests. For shots 97486-97491, the bottom shell on section 2 was retracted while for shots 97492 to 97504, the shells were all in towards the stopper. The shot development motivation for this part of the run day was to create long pulsed plasmas that exhibit disruptions initiating on the outboard side of the machine (major radius ~92 cm). Most of the shots achieved this style, most notably shot 97501 that lasted ~ 5 ms.

For shots 97505 onward, the top shell on section 4 was retracted 4 cm from the stopper position. Shot 97505 was taken with the same settings as shot 97501 with the section 4 top shell retracted for comparison. Shot 97508 was the longest shot of the series and lasted past the 6.5 ms mark.

Jeff ran the last 12 shots of the day (97509-97520). From shot 97515 onward, the triggers for the Dalpha spectrometer and the USB spectrometer were fixed in traverser to correct for timing. Shots 97509 to 97515 were largely consistent shots with modifications to the VF settings and disruptive behavior initiating on low field side. For later shots (after 97514), the OHE was lowered, leading to a drop in the plasma current.

Thursday November 16 2017 11:36 am Stewart 97521-97579 Clean-up/BP testing

Shots 97521-97551 were designated as clean-up shots and shots 97552-97579 were delegated to shot development and bias probe testing. Shot 97533 was found to have a good shot style with long duration and high power input of ~180kW at the top of the plasma current ramp. The positioning on the inboard side would also allow for better capture of the outboard edge by the soft X-ray sensors.

Shot 97562 featured a large 2:1 mode with good SOL currents. Between shots 97569 and 97568 the bias voltage at the end of the ramp was changed between an input of +5 V and -5 V to the amplifiers, leading to changes in the plasma rotation from -35 kHz for the positive voltage and +20 kHz for the negative voltage. For shot 97575, the positive mode rotation reached ~30 kHz with a bias probe voltage ~ -230 V and a current of ~ -40 A. This is a much larger current than was recorded before for negative biasing with ion saturation currents of ~8-10 A. It was also found that the time base from the A14 taking the Bias probe data was changing variably with time. For the last couple of shots, the CPCI was also recording the Bias probe data( which did not have the same timebase issues). It can be seen from shots 97574 and 97575 that the soft X-rays at the mid plane increased during the biased shot compared with the unbiased shot.

Hardware changes: The spectrometer has been moved from the screen room to a shelf near the West rack. The bias probe data (BPS9) is now being recorded by the CPCI.

Friday November 17 2017 6:16 pm Brooks/Stewart 97597-97650 BP current control feedback

Good run day, today. I spent most of the morning finalizing hardware, wiring, and channel mapping. Once I started, it became clear that the current control is working. I sent out a few figures to various people throughout the run day.

The current control system is a PI (no D) classical control system. The current model is based on a linear approximation of the I-V profile, and the output from the GPU (voltage to drive a specified current) is dependent only on the measured current at BPS9.

I followed a prescribed PID gain tuning procedure and ended up scanning a range of proportion and integral gains. Fortunately, my simulink model did a good job at predicting starting gains to begin my tuning.

A few notes.

1. BPS9 is completely off of the A14 DAQ system and is now being recorded by both South:CPCI_10 as well as Caliban (GPU feedback computer).

2. I've also "T"'d the GPU output voltage (to the BPS9 amplifier) back into the South:CPCI_10:Input_93 so that I can confirm that I'm putting out the voltages that I believe I am. Having this information also allows me to calibrate the different time bases between the CPCI and GPU systems.

Ian finished the day using two amplifiers in series (QSC and AETech) to provide large negative BPS9 voltage swings.

Friday November 17 2017 11:35 pm Brooks/Stewart 97580-97594 Bias Probe Testing

Shots 97580-97584 were used to test the new data (current and voltage) recording setup for the bias probe at section 9. The current and voltage data is now being sent to inputs 94 and 95 for CPCI_10 in the South Rack. Shots 97585-97594 were used to test the effect of shell retraction on the shot style used in previous runs in the hopes that shell retraction would lead to less slowing of mode rotation by the walls of the shells and less heat loss from the plasma to the fully inserted shells. It was found from shots 97590-97594 that this configuration was undesirable in that the plasma became increasingly unstable with a lower q and earlier disruptions.

The rest of the run day was devoted to John's GPU testing and crates only testing.

Hardware changes: I was able to shield the Dalpha spectrometer (now next to the West rack) from the ambient light in the room and eliminated the spurious signal it was receiving.

Tuesday November 28 2017 10:00 am Stewart 97748-97759 Dalpha Spectrometer Testing

The run day on 11/28 was devoted to debugging the issues with the Dalpha spectrometer that was moved to the area near the West Rack from the screen room. The spectrometer was putting out a signal consistent with a standard time trace of measured light that lasted the entirety of the shot when connected to an oscilloscope (~5 ms of signal during the shot with peaks at the start up and disruption). However, when connected to the CPCI channel in the West Rack, no signal is recorded except "spikes" at the startup and disruption of the shot. The CPCI channel was tested using a function generator and is working (see shot 97753). 85k Ohm resistors were added to the inputs of the CPCI, but this did not lead to better results or a discernible signal from the spectrometer (see shot 97752).

The next step was to add a transimpedence amplifier to the spectrometer output into the CPCI. This is currently set up and will be tested 11/29.

Tuesday November 28 2017 2:20 pm Brooks 97654-97716 BP Control Studies

This run report covers 11/20/17 and 11/21/17.

This run attempted two experiments. The first was a continuation of my "current control" study from the previous run day. I requested 5 current-steps in total, and the previously calibrated control system did a good job meeting the requested current. I swapped from a linear to exponential model (derived from an exp. fit from the previous friday, same shot style) for the feedforward portion of the the FB code, and it performed worse. I did not spend time calibrating this new code so it's possible that its performance may have improved. I also attempted the code with the FF portion turned off, and it also had poorer performance. Again, I did not attempt to calibrate this new code so also may have improved performance.

The second portion was focused on developing a new run style for mode rotation control. Unfortunately, several things happened that hindered progress. First, a "good" shot style was never developed. Ideally, I wanted a large mode (over 2 Gauss) to last several milliseconds, but the shot style I settled on only had it for 1 millisecond before disrupting. The main idea behind this is that I need a sizable mode in order to get correct mode frequency measurements. I'm presently neglecting mode frequencies of anything below several gauss. Second, my post analysis mode analysis code does not agree with the existing GPU mode analysis code. I need to figure this out before I go much further. All this being said, it did appear that my code was functioning, and the feedforward portion of the mode rotation code certainly put the frequency about where it was supposed to be. The results from the FB portion is still being analyzed, but the previous mentioned issues may likely mean that a future rounds of tests will be required.

Wednesday November 29 2017 10:06 am Stewart 97760-97766 Dalpha Spectrometer Testing

This run day was, again, devoted to testing the Dalpha spectrometer. The spectrometer was putting out a waveform consistent with a time trace over the course of the shot duration ~5 ms with a "peak" region of around -400 mV and a "flat-top" region with an amplitude around -300 mV when connected to the oscilloscope. It is possible the unusual shape of the trace is due to smoothing or averaging in the oscilloscope. The same trace does not appear when the spectrometer is connected to the CPCI input, but rather a small positive deviation from the baseline voltage is observed with spiking noise (amplitude ~1 V) at the startup and disruption.

To test whether the spectrometer was correctly aligned to see the deuterium alpha line, a hydrogen lamp was used (from the teaching lab) and the spectrometer wavelength changed from the preset 656.1 nm. The signal dropped off at 649.5 nm and 662.3 nm and had a maximum peak (102 mV) around the 656.1 nm that the spectrometer was previously set to. The spectrometer was therefore left in the 656.1 nm configuration as it was before. For reference the hydrogen alpha line is at 656.285 nm.

Using a transimpendance amplifier between the spectrometer output and the CPCI input did not yield a better signal and there was a non-neglidgible amount of noise at the disruption on the time trace. Instead, the negative input into the CPCI was grounded for shot 97766 and a positive signal was gathered. Further testing is required to determine whether this is due to noise or due to the signal coming from the spectrometer.

Thursday November 30 2017 11:31 pm Stewart 97767-97778 Dalpha Spectrometer Testing

These shots were used to test modifications to the transimpedance amplifier setup between the spectrometer output and the West rack CPCI input (input 94). Using the transimpedance amplifier and a Megaohm resistor on the feedback circuit, a clear spectrometer signal with a peak of 4V and a flat region with an amplitude of ~1V was recorded on the CPCI with signal characteristics consistent with the Dalpha spectrometer data before the move. The success of this set up was contingent on modifications by Spence to the signal and power cables in order to minimize noise and the swap of the Megaohm resistor with the 50k Ohm resistor on the amplifier board.

As a side note, the Dalpha signal was measured previously using the oscilloscope instead of the CPCI. At first the scope showed a signal shape that included a flat region and a peak at the disruption only. This indicates that the oscilloscope may not have the time resolution necessary to record the full spectrometer output. Additionally, adding a 150k Ohm resistor across the input of the scope lowered the amplitude of the scope signal, but the peak that is usually seen at the breakdown of the discharge appeared in this case.

From successful shots 97787 and 97788, the new amplifier configuration produces a signal with approximately 25 times the amplitude of the previous A14 configuration (~1V at the flat-top region now vs 40mV previously) and does not encounter the same issues with "lost bits" of information that was apparent in the A14 case. This provides a workable setup for the next couple of runs but a permanent amplifier setup with the appropriate housing and power supply is still required.

Hardware Changes:

1.) The spectrometer output is now hooked up to a transimpedance amplifier with a +15V -15V power supply with a 1M Ohm feedback resistor (replacing the 50k Ohm resistor) and no input resistor. The output of the amplifier is connected to the West Rack CPCI input 94. Power cables and BNC wires used in the amplifier setup have been separated and arranged to reduce pickup and noise.

Friday December 1 2017 9:07 pm Stewart/Levesque 97789-97802 Bias Probe Testing

The first couple of shots from the run day (97789-97793) were tests of the Dalpha spectrometer noise level in relation to ambient light that could be making a contribution to the signal. It can be seen between shots 97790-97792, where the lights near the spectrometer were selectively turned off, that the noise level significantly decreased. For reference, shot 97792 was taken with all of the lights off as a baseline. The spectrometer housing was covered for shot 97793 and the Dalpha signal showed the same level of noise for the case where all lights were turned off, even with the lights near the West rack on.

Jeff's runs consisted of SOL tile pickup tests including a VF only vacuum shot 97798, an OH only vacuum shot 97799, and a TF only vacuum shot 97800. For the VF shot, the OH kryrtons were delayed by 50ms to avoid mutual current. Similarly, for the VF only shot, the OH krytrons were delayed by 50ms. The TF only shot had both the VF and OH krytrons delayed by 50ms.

The rest of the run day was devoted to testing the bias probe with the double amplifier setup. It was found that the Zener on the QSC amplifier circuit was nonfunctional after some additional testing. The other Zener and leg of the double amplifier circuit is working normally. In order to continue the high bias voltage testing, the Zener will need to be replaced.

Wednesday December 6 2017 11:13 am Stewart 97803-97825 BP Fast Rotation Study

Now that the D_alpha spectrometer is providing usable data, the goal is to look for drops in the D_alpha signal during fast mode rotation (with large positive biasing on the bias probe). Using the double amplifier configuration, voltages of +210V were applied to the probe with currents of around 90 amps. (Shots 97803 to 97809 were clean up shots). There appears to be a drop in the D_alpha signal in shot 97818 between 2-3ms (Full biasing begins around 2ms) however this could also be due to the drop in the mode amplitude during the fast rotation state (which is also from ~2ms to 3ms). The same type of drop can be seen in shot 97819 between 2ms and 3ms (also during the fast rotation state), however there is a strong 5 Gauss mode during this time. The soft X-rays (at the midplane) this time also increase significantly, but this could be due to the plasma moving outboard (the major radius changes from 92cm to 93cm).

The next few shots were the beginnings of a density scan. Shot 97825 saw a large 9 Gauss during the fast rotation state but a drop in the D_alpha at the corresponding time (the puff time was lowered from 1.5ms to 1ms). This puff time corresponded to a pressure of 61 microTorr. Below this puff time ~0.9ms, the plasma failed to breakdown consistently. It is now the goal to attempt the same experiments with a higher density and to perform a radial scan with the triple probe.

Thursday December 7 2017 11:32 am Stewart 97826-97840 Fast Mode Rotation Study

This run day was devoted to high positive bias/high density shots to determine the effects of density on the fast mode rotation state. For comparison shots 97831-97836 were conducted using a standard puff time of 1.5ms (corresponding to a pressure of 69 microTorr). Again between shot 97831 (no biasing) and shot 97832 (+200V of biasing), there is a clear drop in the D_alpha signal between 2ms (at the start of full biasing) and ~3ms. At 2ms the two SXR traces at the mid plane also diverge with the biased shot sharply rising higher than the unbiased shot (here the major radial positions of the plasmas are the same for the two shots). The largest mode frequency produced by this shot style is ~-30kHz.

Changing the time of full biasing to 3ms, there was an inconsistent appearance of the fast mode rotation state (and the drop in current/raise in voltage associated with the state). In this case, shot 97834 showed a fast mode rotation state at 3ms, where as shots 97835 and 97836 show no signs of the state at all, but rather a mode frequency that hovers around zero during biasing. This could indicate that the state is dependent on density and/or temperature, as would be indicated in previous work on H-mode thresholds being density dependent (Taylor 1989) and temperature dependent: CCT set the threshold as needing 10xTe of biasing to transition into H-mode. During the shot, both of these parameters should change between 2ms and 3ms (according to the plasma current changing from 13.5kA to 16.5kA the temperature should increase).

To test the effect of density and perhaps rule it out as the culprit in the "later biasing case" above, the puff time was increased to 2ms with a corresponding pressure of 100 microTorr. The timing of full bias was returned to 2ms. The result was a fast mode rotation state that lasted longer (~1.5ms-2ms) and saw a reversed mode frequency in excess of 40kHz. The soft X-rays in this case also increased in the biased cases after 3ms in comparison to the unbiased case (compare shots 97839/97840 to unbiased shot 97838). Only a small dip in the D_alpha can be seen for these shots at ~2ms with a corresponding dip in the mode amplitude. From this, it seems more likely that the lack of fast mode rotation during late biasing (after ~3ms) is temperature related rather than density related due to the presence of strong fast mode rotation when the puff time is increased by 45%.

During the high density shots (97840 and 97838) sensor #4 on section 4 of the SOL array saw current signals of ~250-270A. This is 100A larger than the maximum of the previous low density shots this week ~180-190A.

Tuesday December 12 2017 10:39 am Stewart 97841-97901 Triple Probe Scan

This series of shots was focused on a radial triple probe (now at section 5) scan during the fast mode rotation state from 103cm-108cm. There were, however, some issues with the triple probe including some arcing (most likely due to the up to air). Shots 97873 saw consistent arcing with ~2A current spikes with the triple probe in at 105cm. The subsequent shots saw less and less arcing with minimal (1-2 spikes per shot) arcing by shot 97887. Prior to these shots, it was found that the triple probe radial location needed to be recalibrate on the new stand. The radial positions for the triple probe recorded for shots 97843-97865 are therefore no longer correct.

As a quick check, all of the TP tips (A, B, and C) were set to float and compared against the bias probe (BPS9) floating potential at the same radial location. The two probes showed a floating potential around -55V (See shot 97894). It could clearly be seen that when the triple probe was biased, the measured floating potential on Tip C lowered to ~ -70V. The triple probe measured currents were also abnormally high at ~0.3A. The radial scan from 105cm to 109cm (97888-97892) did not show the expected drop off of floating potential at 107cm. This may indicate the same phenomenon that was exhibited by the bias probe when it was first used after the up to air with large ion saturation currents and a linear I-V curve. It therefore is prudent to continue the process of "cleaning" the triple probe before any accurate measurements can be taken.

Friday December 15 2017 10:55 am Stewart 97902-98016 Mode Rotation Study & TP Diagnosis

This run report is for days: 12/12 and 12/14. From the radial scan with the triple probe, it seems tips B and C agree on the floating potential of the plasma (at ~-80V), while tip A (currently the middle probe) sees a floating potential around -40V. Upon visual inspection through the quartz piece on section 5, the triple probe is both intact and all of the tips are in the correct positions. The discrepancy between floating potentials might be due to the differences in the voltage divider used for Tip A (A 1 Megaohm is used). During the radial scan from 103cm to 109cm with the triple probe tips floating, the floating potential did not change from ~103cm to 108cm (staying around ~-80V) but dropped off quickly to ~0V at between 108cm and 109cm. These measurements are not in line with previous scans with the triple probe where the potential should drop off at ~107cm.

While debugging the triple probe, a short fast mode rotation study was conducted. The triple probe was positioned at 107cm during shots 98008, 98009, 98013, 980015; all of which saw Dalpha line drops in the spectrometer signal. These shots also saw the floating potential of the triple probes at 107 cm change from ~ -40V to ~ +73V during the Dalpha line drop. Although the magnitude of the voltage cannot be trusted on these measurements, there is obviously a large change or transition at the edge of the plasma into the SOL since 107cm is the position of the limiters. The shots surrounding these 4 shots do not show such large transitions at the edge and also do not show appreciable drops in the Dalpha signal. This evidence further lends itself to being related to an L-H transition however, a working triple probe with reliable voltage measurements is required to confirm this with a radial scan of the density, temperature, and pressure.

Friday December 15 2017 11:48 pm Levesque 98018-98040 Biasing Chamber Section 4 w/r/t Section 3

Ran according to run plan JPL11. This involved biasing an otherwise-floating chamber section with respect to its grounded neighbor. Was able to successfully connect circuitry from one of the bias probes across the Section 3.5 midplane quartz break, with the whole chamber of Sections 4-5 as the "hot" lead and Sections 1-3 as the ground. Ran with a variety of connection configurations and voltage waveforms. External Rogowski Jumper's A and B were located as they have been for months, with A across the Section 9.5 midplane and B across the 3.5 midplane. Jumper B was only connected at the start of the day while verifying safe current and voltage levels, and connections were on the same studs as the biasing leads. Current along the west rack ground strap was measured all day, and the north rack ground strap measurements started at 98031.

Without the control amplifier in the circuit, first checked that there would be safe voltage and current demands that would be seen by the to-be-connected amplifier. Configurations include:

• 98019-98020: Jumper B connected and bias circuit set to "grounded," shorting Sections 3 and 4 through the bias current+voltage measurement box. Much more disruption current (i.e. ~50kHz oscillating current) went through the Jumper than through the bias box due to the Jumper's lower resistance and inductance. The Jumper had up to ~600A during disruptions, while the biasing circuit had up to ~9A.
• 98021-98022: Jumper B disconnected on right side, and bias set to "grounded." The Rogowski stayed close to its original position, so this may be used as an approximation to subtract magnetic pickup from the shots where Jumper B is connected, though the Rogowski is very close to the biasing leads. The biasing circuit now had up to 200A during disruptions with the Jumper disconnected (since the bias circuit is now a much better path).
• 98023-98024: Jumper B connected and bias set to "floating." Expected low current and voltage seen by the bias circuit. There were still significant voltage spikes, possibly due to noise.
• 98025-98026: Jumper B disconnected and bias set to "floating." Expected to measure chamber floating potential. During the main discharge, Chamber 4-5 floating potential is read as being around -5V to -10V. During the disruption, measurements are not clear with a quick glance, but voltages stay at reasonable levels. Jumper B stayed disconnected for the rest of the day.
• 98027-98028: Bias set to the control configuration, but leads were just shorted together near the control transformer (transformer and amplifier out of circuit). Current during disruptions dropped to ~90A due to higher resistance/inductance of the long leads.
• 98029-98030: AE Techron amplifier connected through the large 5:3 transformer. No command was sent to amplifier, so it should act similar to a short.

Shots 98031 onward included a variety of bias settings. Driven currents between Sections 3 and 4 were in the range of -80A to +60A. Shots 98038-98040 included biasing during the disruption. Biasing was able to change the direction of current between Sections 3 and 4 along the connected leads during the disruption, from 40A co-Ip to about 50A counter-Ip. Further analysis is needed before discussing implications of currents through the SOL.

Some effect of the biasing is seen on the equilibrium currents through SOL tiles in Section 4. Considering the various bias settings during the main discharge, around half of the bias current conducts along the ground bus between the north rack and west rack (i.e. Sections 3 and 8). This current path still requires at least one jump through the plasma. Disruption biasing gives no obvious effect on current through Jumper A (on the opposite side of the tokamak).

Notes regarding the choice of biasing Chamber 4: In January I tried to run a similar experiment by biasing Section 10 with respect to Section 9, but there were problems with apparent ground loops (between the chamber ground sourced from the west rack and the digitizer in the south rack) for that configuration. After the recent change for the Section 9 biasing diagnostics to use a CPCI instead of the A14 in the south rack, and including a Jensen isolation transformer in the bias voltage measurement circuit, there are still problems with the Section 9 bias probe voltage readings during the breakdown and disruption. The CPCI measurements clamp to +-0.6V during periods of large flux swings, presumably due to protection circuitry in the CPCI, suggesting there's still a large common mode problem with the South bias probe voltage measurement, despite including the isolation transformer in the measurement line. Rather than continue troubleshooting this problem, I decided to try the biasing hardware for Sections 4-5 instead, which doesn't have this potential ground loop problem (chamber ground and digitizing connections are to the north rack only). There were no problems with voltage spikes or the digitizer cutting out for this configuration.

Thursday December 21 2017 11:55 am Stewart 98041-98075 Triple Probe Radial Scan

This set of shots was devoted to a radial triple probe scan during the Dalpha drops for the positive biasing experiments that exhibit H-mode like behavior. The goal was to determine if the density and temperature at the edge of the plasma was changing during the Dalpha drop interval between ~2-3ms for this shot style. Shots that exhibited a drop in current and D_alpha line emission include:

• Shot 98064: TP at 102 cm with up to an approximately 50% drop in Dalpha
• Shot 98056: TP at 103 cm with up to an approximately 50% drop in Dalpha
• Shot 98073: TP at 104 cm with a harder to discern drop in Dalpha (~30% drop)
• Shot 98063: TP at 103 cm with up to an approximately 50% drop in Dalpha, transiently higher
These shots saw a drop in the Dalpha signal from ~0.8V to ~0.4V during the time span around 2ms and 3ms. This coincides with the time of the current drop on the bias probe from the full value of ~90A to a minimum value around 30A. During the Dalpha drop, the density and temperature do not change significantly for the shots outside of 103cm (The bias probe radial position). But there is an increase at the end of the drop for the inner position at 102cm at around 3ms. Each of these shots showed large changes in the floating potential during the Dalpha drop with the shots with the triple probe at 102, 103, and 104cm seeing a peaked floating potential of +130V and the shots with the triple probe at 106 and 107 cm seeing a peak in the floating potential of ~+64V and ~+10V respectively.

Additionally, shot 98063 with the triple probe at 106cm has a significant reduction of fluctuations in the density and temperature measured by the triple probe during the interval that the Dalpha drop occurs. Note: To verify that the triple probe was giving reliable data, a bias probe IV curve was carried out (shot 98051) and compared with the measured density and temperature of the TP.

Thursday December 21 2017 6:34 pm Brooks 98076-98100 BP Current Control

The intent of today's run day was to continue testing the GPU's control of BP current. Specifically, I wanted to more sophisticated request profiles to test the response of the feedback system. After a fair amount of testing and debugging, I requested I=15+15*sin(f*2*pi*t) where f was 1000, 3000, and 10000 Hz over the course of several shots. The system did a fair job in meeting these requests, and the total latency seemed to be frequency independent and on the order of 30 microseconds. The feedback system is thought to have a total delay of around 22-24 microseconds. I should double check these numbers.

Tomorrow, I will continue to investigate current control but also transition to studying mode rotation control.

Finally, take_shot_hbt.py gave me a number of %SS-W-NOMSG errors associated with Spitzer and various CPCI units around the machine room. Resetting Spitzer seemed to solve the issues.

Friday December 22 2017 6:48 pm Brooks 98078-98151 GPU latency test, current control and mode rotation control

There were three parts to today's run day.

1. I performed a latency test of my feedback system using two different power transformers (ACME and right-3-phase). Despite some initial confusion, both seemed to have on the order of 21-24 microsecond latency.

2. I repeated some of the current control studies I did yesterday using the right-3-phase transformer. Yesterday, I used the ACME. These tests worked, and upon quick inspection, appear to be similar in performance to the ACME results from yesterday.

3. Finally, I spent a few hours attempting mode rotation control, and I may have achieved some success, most notably in shot 98147. In this shot, I used a proportional only feedback system with no feedforward system. I request a step function in frequency from 6kHz to 3kHz, and the system appears to respond correctly. I'm not happy with the frequency mode algorithms on the GPU and in my personal post-processing library. I plan to overhaul these over the break, and reinvestigate later. Assuming I still think the system is working, I will spend some time trying to tune the feedback system, include voltage or current control alongside the frequency control, and include integral gain as well.

Thursday January 11 2018 10:52 pm Stewart 98152-98206 TP and Bias Probe Testing

This run report includes run days 01/09 and 01/11. The first day was devoted to cleaning off the triple probe at section 8 and comparing the density, temperature, and floating potential data with the triple probe already in use at section 5. Before testing could begin, it was found that an A14 (A14_18) in the North Rack would not prepare, despite the other adjacent A14's preparing in the same rack. Since this records control coil currents and they are not currently in use, we decided to continue without the A14.

Despite the initial arcing on the triple probe at section 8, little or no arcing was seen by shot 98166. Unfortunately, this triple probe was floating higher than TPS5 and the two probes disagreed on the position of the edge of the plasma by roughly 2cm (from shot 98178 for TPS8 @ 105.7cm and shot 98179 for TPS5 @ 107.7cm). The floating potential of TPS5, which was already in use, tends to agree with the bias probe floating potential (as was seen in the triple probe diagnosis before break). It is therefore prudent to double check the radial calibration of the probe at section 5 and once again compare I-V curves from the bias probe to the triple probe measurements for density and temperature to make sure that the measurements from this probe are accurate.

For the second day (shots 98188-98206), the shaping bank was tested as a power source for the bias probe at section 9. The bank itself was disconnected from the shaping coil and cables were run to connect the bank to the terminals of the transformer used for the QSC amplifier. The transformer was then connected to the bias probe in the usual fashion. From preliminary testing on a 2 Ohm resistor, it was found that the presence of the transformer prematurely truncated the exponential decay of the bank voltage (see shot 98192). When the transformer was removed, the voltage applied to the resistor from the bank decayed normally (exponentially) with no truncation (see shot 98193).

After testing the bank on a resistor, the bank was connected to the bias probe and used on 7 plasma shots (98198-98204), where the voltage on the bank was incrementally increased. Shot 98200 saw very fast 60-70kHz modes with an applied voltage of approx. 200V and a current of around 90A. For this shot there was a drop in current at ~2.2 ms but the drop only reached 42A and was not as clear as the drops seen using the amplifier power sources. The is also a subtle D_alpha drop during this time but not quite as strong as in the previous run day looking for H-mode behavior. Shot 98201 saw about 5 of these drops in current over the course of 1.5 ms and mode frequencies that fluctuated with the current (transitioning in and out of the fast rotation state). The floating potential as seen by the triple probe at 106 cm also saw these fluctuations with corresponding rises and drops in the floating potential. Despite reaching above 300V on shot 98204, the current did not raise appreciably from around 100A outside the drop.

Of note, however, the section 4 SOL tile #3 drops at the same time (2 ms in) as the bias probe begins full bias, especially for shot 98200 where is drops from a measured 50V to 35V rapidly. The same sort of drop is evident in the next 4 shots as well, albeit not as stark. For shot 98204 a popping sound was heard and was subsequently investigated. No damaged could be found on the banks or any of the diagnostics (the shaping bank was tested again and cleared and the diagnostics were checked on shot 98206 and cleared). There was also no apparent issues with the bias probe stand or the current sensor/voltage divider housing. A large current (~500A) was found to have flowed through the ground at the West Rack. Further investigation is required to diagnose the issue.

Friday January 12 2018 2:42 pm Levesque/Stewart 98204 Further info on shot 98204 arc

The popping/arcing sound during shot 98204 was found to be due to an arc across the MOV protection component within the bias probe measurement box. The applied voltage (~320-340V during plasma) likely exceeded the specifications of this MOV (P/N: 14D201K, spec'd max clamping voltage of 340V), since we applied higher voltage than we had originally intended to when the measurement circuit was built. The circuit effectively worked as designed, where the large current pulse went through the MOV at high voltage. This was originally meant to protect the control amplifiers against a surge produced by the plasma, but our applied voltage now exceeds this rating. The MOV is visibly damaged; for replacing the MOV, we'll consider using a higher voltage component, or will use an identical MOV and restrict ourselves to applying lower voltage in the short-term. The following information is more detail about what happened during the arcing event:

• The current pulse occurred about 4ms after the end of the plasma current quench.
• A 550A pulse to the West rack was measured as flowing from the rack to its grounding strap (direction of conventional current) that goes to the main grounding bus. The FWHM duration of the pulse was around 0.3ms.
• This was accompanied by a 1080A pulse of current returning to the shaping/bias bank, as measured along the grounded lead at the return to the bank (conventional current flowing to the bank ground bar from the biasing leads).
• There was no measurable current flowing to the bias probe at this time, as measured by the bias diagnostic box. This means that the current did not arc across the feedthrough or bias probe itself on the way to ground.
• The North rack ground strap had an accompanying 30A pulse flowing from the ground bus to the rack, with a longer pulse duration compared to West rack. The pulse shape matched for the North rack and shaping bank, and was different for the West rack ground -- this might be due to saturation of the West rack Pearson current monitor (not investigated).
• Given the current measurements and locations, the West rack ground current and the current returning via biasing leads should add at the shaping/bias bank ground bar, meaning that at least 1600A peak current flowed from the bank at the time of the surge.
• The pulse can also be seen in the loop voltage (about -2V response) and the regularly-broken SXR fan array channels.

Tuesday January 16 2018 7:00 pm brooks 98207-98245 BP voltage steps to study mode frequency response

Today's experiments focused on generating BP plots that distinctly showed the relationship between a BP voltage step and the resulting change in n=1 mode frequency. Ideally, this BP voltage step would go from approximately floating voltage to some other value (+100 and -100 V for example) that would show that we have the ability to spin up and down the plasma at will. These plots would be contrasted against a control where the probe was left floating without a voltage step.

By the end of the run day, I finally managed a reasonable shot style and BP feedforward configuration that was producing the type of plots (discussed above). This data still needs to be studied. More than likely, however, more data will be required.

A number of issues cropped up today. Several of these are discussed below.

1. Jeff noticed that the Control coils were plugged into the FB amplifiers at section 8-top. This was fixed.

2. The deuterium may be leaking from the line? In the past, The pressure increase caused by the puff valve never changed much shot to shot. Today, however, and after 30 shots, the pressure was observed to be noticeably low and caused a strong deviation in shot results (61 microTorr instead of 68). By repressurizing the line, the puff valve returned to 68, and the shots returned to their normal behavior. Ian and I speculate that the line may be leaking because we normally do not need to repressurize within the same run day, especially after only 30-35 shots.

3. Ian believes that the BP is better able to spin up or down the mode at earlier times (2 ms) as apposed to later times (4 ms). I've seen enough to believe that there may be something to this.

4. I ran with both the QSC and AETech7224 amplifiers powering BPS9. While this certainly increases the total power delivered to the probe, the QSC has the habit of destroying expensive fuses and zener diodes, particularly when we program the input voltage poorly. Likely, the new (second) AETech7224 will replace the QSC.

Monday January 22 2018 2:26 pm Stewart 98265-98291 Fast rotation state study

This run campaign was devoted to looking at more global changes to the plasma during the fast mode rotation state including the signals from the SXR fan array. A new shot style was used with a flat top profile for the plasma current instead of the current ramp that was used before in order to help ensure plasma conditions were not changing drastically during the time interval that the bias would be applied. Shots 98287-98291 are good examples of the unbiased case with relatively flat plasma current profiles that last ~8.5ms. With biasing, the plasmas only lasted ~5ms (see shots 98280-98286).

The plasma was biased with the double amplifier set up with a waveform that ramped between 1.5ms to 2ms and then ramped down again from 3.5ms -4ms. Fast modes were detected for shots 98280-98286 in excess of 50kHz in some cases (This is higher than in previous studies that had higher Ip). By comparing the unbiased case to the biased case, the core soft X-rays increased late in the biasing to 2 A.U. (shot 98285) from the unbiased (shot 98288) ~1.3 A.U. This corresponds to a roughly 50% increase across four SXR detectors that see the core. At the edge of the fan array, the SXR signal jumps from ~0.5 A.U. in the unbiased case to ~0.75 A.U. The rise in the SXR profile is global across the whole viewable plasma with all of the sensors detecting an increase between the two cases however, the increase is steeper at the core than at the edges of the array.

In addition to the increase in the SXR profile, in the biased shots there were groups of D_alpha spikes between 2.0-3.5 ms for shots 98280, 98281, and a single sharp ~2V spike on shot 98285 during fast mode rotation. These spikes did not show up on the other diagnostics such as the soft X-rays. They also have the characteristic sharp rise and slow decay of elms with the spikes lasting ~100 microseconds. These results are, however, clouded by the fact that there are noise spikes from the spectrometer itself. But, the amplitude (~1.5-2V on a background signal of 0.5-0.6V) and grouping of these spikes during the fast mode rotation state suggests these might be bursts of recycling at the edge.

Lastly, in terms of hardware changes, the USB spectrometer set up (including the lenses and mirrors) were pointed towards a tangential view from section 10 to the limiter between sections 2 and 3. The intent was to use the fiber optic and route the light at the limiter to one of the Thomson scattering polychromators, but either the light level was too low or not aligned correctly to discern a measurable signal. Special thanks to Jeff who was integral to this part of the run day.

Wednesday January 24 2018 11:33 am Stewart 98292-98309 SXR edge scan

Since there were global increases in the soft X-ray profile in the core and edges of the SXR fan array during the last run day, this series of shots was focused on viewing the edge of the plasma during this high confinement state. Using a lower VF setting, the plasma major radius was pushed out to 94.5-95 cm during high biasing to see if there was a change in the edge profile and to check for evidence of a pedestal.

Shot 98305 was used as the unbiased reference shot and had a consistent exponential decay shape at the edge, eventually reaching a value of 0.4 A.U. at the last SXR fan array sensor (for all time slices at 1.5 ms, 2.5 ms, and 3.5 ms). Shot 98308 was biased with a ramp from 1.5ms to 2ms and saw a period of fast mode rotation from around 2 ms to 2.9 ms and a marked increase in the edge SXRs during this same period. There was also a drop in the D_alpha signal from ~0.7 V to ~0.4 V during this time period. The edge profile increases from a value around 0.1 A.U. to around 0.4 A.U. (at the outer two sensors) and the shape is not longer the gentle decaying slope but a more linearly decreasing trend. Between time slices 2.85 ms and 2.93 ms, the edge profile transitions back to the exponentially decaying shape with the edge value back around 0.1 A.U. corresponding to a rise in the D_alpha signal and a transition out of the fast mode rotation state. Although a pedestal is not directly in view during this shot, the behavior is indicative of pedestal formation and collapse. This is further evidence of H-mode behavior during high biasing, however, further shots will seek to push the major radius out further to verify if a pedestal has indeed formed.

Sunday January 28 2018 11:13 pm Levesque 98385-98409 Biasing chamber Section 4 w/r/t Section 3, outboard plasmas

Ran according to plan in experiment ID JPL12, biasing chamber sections 4-5 with respect to 3, with far-outboard plasmas, and started moving shells in the biased sections. There were problems with banks not triggering near the beginning of the run day when VF timing settings were being changed. This has been a common problem lately, where triggers for multiple banks will stop working when the VF timing is changed. Bank triggering worked after several test shots, and the run commenced with VF delayed 220us later than typical recent settings. This gave plasmas that were much farther outboard than standard operation, including at the breakdown. Major radius during the target period was around 96cm. The intent was to make the LFS SOL tiles be the most preferable path for SOL currents during biasing, rather than HFS vessel contact. Modes were relatively large and rotated very steadily, producing clean LFS SOL signals; however, the possibility of direct magnetic pickup from modes on the SOL sensors has not yet been quantified.

Biasing leads were connected across the Section 3-4 quartz break. Using both the QSC amplifier and AE Techron amplifier together was not producing the desired waveforms (characteristics looked like there may have been a bad connection), so only the AE Techron was used for successful shots. A phase-flip-style waveform was used for the biasing, with a 1ms bias period that reversed sign half way through. Clear signals were seen on the Section 4 LFS SOL current tiles in response when the applied bias current was above 30A. Responses in other sections are not as clear, as expected, though Section 1 might have a correlated response. The run will continue on Monday.

Some noteworthy shots include:

• 98404 and 98406: These shots are similar, but have the opposite sign for biasing -- an obvious LFS SOL response can be seen when overlaying these two shots.
• 98405 was a partially-failed shot due to the OHST not firing. The discharge fizzled with a plasma current below 5kA for about 1.5ms before drooping to no plasma current. Despite having little or no measurable plasma current when the bias fired, bias current was still driven and is seen in the Section 4 SOL tiles. A fully-failed breakdown (98403, which might have failed due to low fill pressure) did not have any bias current at all in response to the same amplitude of bias voltage, as was expected.
• 98407-98409 involved retracting shells Sections 4 and 5. Less biasing current was driven with the shells retracted, and less current was measured in response to rotating modes.

Monday January 29 2018 10:57 am Stewart 98310-98384 SXR and TP Edge

This run report encompasses the edge scans from two days 01/24 (98310-98336) and 01/25 (98337-98384) using both the soft X-ray fan array and the triple probe at section 5. The first section of the first day was devoted to pushing the plasma outboard with the VF settings in order to capture the very edge of the plasma in the fan array view. The unbiased shot 98317 definitely had the very edge of the plasma in view by 3.5 ms with SXR values close to 0 at the two farthest edge sensors. Shot 98319 (a biased shot with a D_alpha increase) had the same shot style and a rise at the edge SXR, especially around 2.5 ms however, no pedestal could be discerned from this shot from the soft X-rays.

To elucidate why, a triple probe scan was conducted from R=108 cm to R=103 cm. The shots include 98325 (108 cm), 98329 (107.5 cm), 98326 (107 cm), 98331 (106.5 cm), 98327 (106 cm), 98332 (105.5 cm), 98336 (105 cm), 98335 (104 cm), and 98334 (103 cm). The shots around 106 cm (including shot 98327) saw depressions in the TP measured density during the current drops of the bias probe. Meanwhile at 103 cm, there is a huge increase in the density during the current drops/fast rotation states with a rise from a value of 1.2 to 3 (x10^18 m^-3) from 2.5ms to 3 ms. Looking at the floating potentials across these shots during the fast rotation state, the TP V_float increases from ~0V at 108 cm to ~224V at 103cm. The increase is more gradual than one would expect with other machines that show an E_r well and change in V_float in a width the size of a centimeter and the pedestal forming within a cm of this E_r well. For these shots, the E_r well extends out from 108cm to 103cm (over 5 cm) where, according to the Triple probe data, a density increase occurs. At radial positions within the well (around 106cm) the density shows depressions in the state.

This shallower behavior for the floating potential change (radially) for applied bias probe voltages between 100V and 200V was also seen on TEXTOR. By applying biases up to 500V, they were able to steepen the floating potential profile so that there was an increase from 0V to 500V within 1cm at the edge. This resulted in a density increase at the very edge of the sheared layer from about 1.2 to 2.1 (x10^18 m^-3). These results may explain why a pedestal is not present at the very edge of the plasma in these shots, since the density increase is seen at the edge of the sheared layer/E_r well at 103cm rather than closer to the edge around 107 cm. However, it is important to note that coupled with the SXR increase pointed out in the earlier runs with the fast mode rotation state, this increase in the density measured at 103 cm also points to a density increase in the core of the plasma during high biasing.

The next day of shots (98337-98384) was designed to be an unbiased analog to the TP scan with biasing. For the two cases, the density profiles follow each other closely from 108cm to 104cm until 103cm, where the biased shot shows a large density increase. The rest of the run day was devoted to increasing the plasma current in the hopes of changing the floating potential profile and hence changing the size and magnitude of the E_r well. The shots with a higher I_p showed the same V_float trend as was found before for lower I_p. Of note is shot 98374 that saw a collapse of the core SXR profile. The shot continued with this hollow profile for 3-4ms.

Monday January 29 2018 10:40 pm Levesque 98410-98438 Biasing chamber Section 4 w/r/t Section 3, outboard plasmas

Continued chamber biasing experiment JPL12, with far-outboard plasmas and Sections 4-5 biased with respect to Section 3. Shells in Section 5 were fully retracted for all shots, Section 4 shells were placed at four different positions, and all other shells were fully inserted. All probes were retracted for this and my previous run day on Friday, and external jumper B remained disconnected so that the biasing apparatus was the only non-plasma connection across the Section 3.5 quartz break. A vacuum shot was taken for each used shell position to help with LFS SOL pickup subtraction. Most shots had minor disruptions before or during the period of interest due to the shot style, so only one good target biasing shot was taken for each shell position. Good shots are as follows:

• Shells 4T/B fully inserted: 98437 biased, 98434 non-biased, 98438 vacuum.
• Shells 4T/B back 1.0 cm: 98431 biased, 98432 non-biased, 98433 vacuum.
• Shells 4T/B back 2.0 cm: 98426 biased, 98427 non-biased, 98428 vacuum.
• Shell 4T back 4.25cm (full retraction to stopper), 4B back 3.65cm (2mm from retraction stopper due to concern of SOL wing hitting quartz): 98420 biased, 98419 non-biased, 98410 vacuum shot but with different bank settings.

The amount of bias current driven by the amplifier decreased as the wall was moved farther from the plasma. In most cases the amount of current passing through Section 4 LFS SOL tiles in response to modes and biasing decreased as the shells were pulled back farther. However, at 1cm retraction, Section 4 tiles #1 and #4 (bottommost and topmost) measured more current in response to biasing compared to when they were fully inserted -- this should be investigated to see if it's typical.

Tile LFS04_S4 might presently have the wrong polarity in the tree; I've suspected this for these biasing experiments and in its characteristics for several other discharge conditions, and will try to confirm the polarity. If I find that it's wrong I'll correct the tree and back-correct all shots taken so far. Regardless, polarity should be checked at the next vent period when the poloidal EUV system is installed.

Tuesday January 30 2018 5:21 pm Brooks 98439-98472 BP feed-forward tests

The goal of today's run was to generate plots that distinctly show the relationship between BP voltage/current and n=1 mode rotation. In today's run, I used a voltage step at 4.5ms to slow and speed-up the rotation. The shot style was a fairly consistent 2/1 mode with roughly 2-3 gauss mode amplitudes. The runs went well, and the data looks to be good.

Note: All shells were in against their bumpers, and Rogowski jumper B was put back in place for the entirety of the run day today.

Saturday February 10 2018 12:23 pm Brooks 98521-98535 Transformer characterization

I ran a test on Friday attempting to characterize the right-most 3phase transformer. I ran the following shots:

1. Crates only. 2 ohm resistor simulating the BP. AETechron7224 at 100% gain. Right-most 3phase transformer (5:3 ratio).

2. Crates only. 2 ohm resistor simulating the BP. AETechron7224 at 100% gain bypassing the transformer.

3. Plasma shot. BPS9 is the load. AETechron7224 at 100% gain. Right-most 3phase transformer (5:3 ratio).

Monday February 19 2018 12:08 pm Levesque 97249-98575 Corrected polarity for an LFS SOL sensor

The run report on January 29 mentioned that the polarity of SOL tile LFS04_S4 might have been incorrect in the tree. I scoured through the raw calibration data for about half of the tiles, and found that this tile did most likely have the wrong polarity in the tree. There is conflicting data in the raw calibration files for this sensor, as if the polarity was wrong for the calibration current pulse, but comparing its polarity to its neighbor tested in series strongly implies the need for correction (unless there were other problems in that setup). I have corrected polarity for this sensor in the model shot, and in all shots taken since it was installed (97249-98575). A correction will also be made to the make_tree script for future tree rebuilds.

Wednesday February 21 2018 5:04 pm Stewart 98576-98615 Triple Amplifier/H-mode study

This set of shots was designed to utilize the new three amplifier bias probe set up, which was able to drive 100 amps over a 2 Ohm resistor with a voltage of 200V (shot 98575). This corresponds to 20kW of power from all three amplifiers, an increase from the 162 A and 85 V (14kW) over 2 Ohms from the double amplifier setup (shot 98504).

The results of biasing using higher current and higher voltage were two fold: the H-mode like state was sustained for longer (almost the entire biasing time saw fast modes and a decrease in current with fleeting ~100 microsecond transitions out of the state) and the soft X-ray profile at the core of the plasma increased another ~50% from the previous setup. High biasing beginning at 2ms was employed in shots 98600, 98601, 98602, and 98604. Here, mode rotation reached a maximum in excess of 70kHz.

Shot 98605 was a reference shot with no biasing that showed peak core soft X-ray values of ~1.2A.U. at 3.5ms. The values of the SXR at the core of this shot do not increase substantially throughout the entirety of the shot with the max core SXR at 6.5ms around 1.4 A.U. In comparison, shot 98603, taken with the same shot style but with biasing, showed core SXR values of ~2.5A.U. at 3.5ms. This represents a doubling in the core soft X-rays during the H-mode like state, most likely corresponding to large changes to density, temperature, and/or impurity confinement. The triple probe (positioned at 103cm) also shows a marked increase in density during this time, verifying some previous results with the two amplifier setup and reinforcing evidence form the increase in the soft X-rays. Comparing shots 98063 with 98064, at 4ms there is something that looks like a fast beta collapse with a sudden approximately ~50% reduction in SXR but no disruption or current spike.

A temporal scan with high biasing included shot: 98606 (full bias at 2.5ms), 98607 (full bias at 2.0ms), 98608 (full bias at 3.0ms, 98609 (full bias at 2.75), 98610 (full bias at 3.25ms). The shots with biasing before 3.0ms showed a current threshold for the H-mode like state of ~68A, while the shots with full biasing after 3.0ms showed a threshold around ~110A. On shot 98610 the state was sustained for the entire length of high biasing from ~3.5ms to 5.5ms, showing that this state can be achieved later in the shot. Of note, the shots with the current thresholds also had larger drops in current and larger voltage increases (~300V in comparison to the ~250 from the lower threshold shots). This increase in BP voltage corresponded to an increase in the V_float measured by the triple probe (again around +300V for these higher current threshold shots). This further supports the idea that the floating potential can be increased with higher bias probe voltage with a corresponding steepening of the E_r well at the edge.

Shot 98612 was fully biased at 3ms and was still in an H-mode like state when the bias probe switched signs from +200V to -150V. Here, an ion saturation current of ~30A was drawn and the mode frequency increased to +30kHz. The triple probe also saw the floating potential decrease to around the bias probe voltage, which is indicative of a negative E_r well forming. This is similar to the results of CCT and TEXTOR, which were able to induce negatively biased H-modes with sufficient ion current. The SOL tiles also saw an increase in current during the period of high biasing for this shot.

Hardware changes:

*Important* The outboard, top, and bottom limiters have been retracted by 2cm from their previous positions prior to the start of this run day. They have been moved from R=107cm and Z=+/-15cm to R=109cm and Z=+/-17cm. The LFS SOL tiles now form limiting surfaces with a minor-radial limiting edge centered at R=92cms and horizontal SOL tile surfaces at +/-15cm.

The MOV on the bias probe stand has been exchanged for one with a clamping voltage of 455V.

Thursday February 22 2018 3:41 pm Levesque 98617-98643 SOL biasing with most walls retracted, double puff

Ran with the Section 9 bias probe inserted in the scrape-off layer and most walls fully retracted, in an attempt to measure field-align currents directly between the probe and a small number of SOL tiles. Started with only shells in Sections 1, 4, and 8 inserted (the locations with the LFS SOL tiles), then had only Section 1 shells inserted for the second half of the run. Pulsed the bias probe in 0.7ms periods using the triple-amplifier setup from Ian (example biasing waveform in crates-only shot 98624). Up to ~110 amps was driven from the probe with it located between 106.5cm (front tip) and 107.5cm (back face of hemisphere) and biased positively. Tiles on fully-inserted shells nominally have their innermost edge at 107cm when mapped to the midplane along a surface of a full-size plasma. Many of the plasmas were far outboard at R₀ >=94cm as intended, with others drifting inward early in time and being around R₀ =92cm during the biasing.

I had expected to see a clear response on a small number of tiles along field lines connecting to the probe, and to be able to account for most of the biasing current since there were few potential contact surfaces. However, specific current paths still remained unclear for the most part, even with only Section 1 shells inserted and the plasma far outboard. When driving 80-100A from the probe, individual tiles only see up to a ~4A response. It looks like most of the current driven by the probe is still distributed over undiagnosed regions of the wall in this configuration. In many cases (but definitely not all cases), most of the current from the probe returned along the grounding bus between the North and West racks, meaning that the current was collected within Sections 1-5 (or just within Sections 1-3 in a subset of cases). Measurements are consistent with this current entering Section 1 along a surface other than the tiles. It's possible that ion saturation limits the current allowed to each tile, and the current is then necessarily distributed among other surfaces; however tile areas should be large enough to accommodate more current than was measured during the pulses. Some transient events like minor disruptions occurred during the biasing and produced a lot of ion current at a few tiles (e.g. tile LFS01_S5 in shot 98640), but this is not necessarily influenced by the biasing; tile currents exceed the biasing current in those cases.

The SOL current pulses were seen to induce fast reversed rotation in some shots, similar to our standard biasing cases with the probe much deeper in the plasma. As more shells were retracted, the rotation effect seemed to increase, as if there was less drag on the plasmas -- two possible causes could be reduced EM torque from nearby shells, or reduced neutral drag due to having less surface area involved in recycling on the time scale of the bias pulse. Faster rotation cases also corresponded to less current and higher voltage on the probe. Discharge parameters for these cases don92t overlay well though, so this might just be coincidental. This rotation effect is seen in shots 98628 (Sec 1, 4, 8 inserted), 98629 (Sec 1, 8 in), and 98630 (only Sec 1 in), during the 3-3.5ms window. A counter-example with the best overlaid equilibrium is shot 98641 (only Sec 1 in), which does not rotate strongly with the biasing.

Strong bias current pulses can be seen in the TA and PA magnetic sensors for some shots, where a perturbation appears while current is being driven. In example shot 98634 with a smoothly evolving equilibrium, the response opposes the field from the plasma current. The plasma current Rogowski also shows a reduction in measured plasma current during the pulse. The response in many of the SOL tiles is unclear however -- instead of measuring the square pulse, many tiles indicate a gradual drift of collected current during the bias period, and then remain at a different current level after the biasing turns off. This appears to be an instrumental error on the SOL tiles though -- it looks like the tile current offset that accumulates during this drift remains as an offset from reference shots long after the plasma ends. I can't find an explanation for this yet. This measurement drift effect can be seen when examining shot set [98632,98636,98637,98638,98639,98642], especially tile LFS04_S2.

The probe was also moved out 3 more cm over two steps to see if an obvious direct current path would appear. This reduced the amount of current driven by the probe, but there was still no clear direct response on a small set of tiles. Shots with the probe farther outboard include 98632-98640.

Five shots (98634-98637,98642) also included a second gas puff after the biasing period to study the effect of puffing during the disruptions. Here, the hypothesis is that the second puff will cool the SOL and edge plasma, and could change characteristics of the current that is inferred to jump across quartz pieces during the current quench. The second gas puff made the current quench slower in all cases. I haven't yet looked at dynamics of the modes or SOL currents. Note that in the past few years we observed that the second puff always results in a disruption within a millisecond of starting the puff -- this continued for shots in this run. Shots were highly variable in this run, so the double-puff conditions are not ideal for directly comparing shots with or without the second puff.

Surfaces were probably dirtier than normal for this run. One day before the run, Jim found the cryo gate valves closed in the morning and the chamber pressure in the 10^-4 torr range. There was a lot of water and air on the RGA. The cause of the gate valves being closed is unknown.

Thursday March 1 2018 11:21 am Stewart 98644-98669 H-mode Study with Bank

This run day was devoted to using the shaping bank to bias the electrode located at section 9 (R=103cm) in the hopes of attaining higher voltages than the triple amplifier configuration. The results can be seen in shots 98663 (biased to 73V, no change in Vfloat for the Triple probe [R=103cm]), 98664 (biased to 100V with a brief increase in Vfloat to +64V), shot 98665 (biased to 157V with a more sustained transition to the H-mode like state with Vfloat steady around +72V for ~1ms), shot 98666 (biased to 208 V, Vfloat steady at +91V for the entire duration of biasing until the disruption), shot 98667 ( biased to 303V, Vfloat steady at +127V for the entire duration of biasing until the disruption), and shot 98668 (biased to +367V, V_float steady at +140-+150V for the entire duration of biasing). Using the bank, the consistency of the H-mode state and the duration are improved from the case of the triple amplifier setup, where transient 100 microsecond transitions out of the state appear. These shots also show that the Vfloat of the triple probe is continuing to increase with increasing bias voltage as was seen on TEXTOR, most likely sharpening and raising the E_r field at the edge (a radial scan will be conducted to confirm this).

The core sxr reached a peak value of ~2.7 A.U. for shot 98667, a doubling from the unbiased case in shot 98661 of 1.3 A.U. The shape of the soft X-ray profile (both radially and in time) in shot 98667 is analogous to the shape of the sxr profile in shot 98603 biased to ~300V during the H-mode using the triple amplifier setup. Shot 98668 (biased to ~370V) had a core collapse of soft X-rays around 3ms and did not reach the same 2.7 A.U. at the core but the edge soft X-rays doubled in comparison to shot 98667. The triple probe, located again at 105cm, showed a ~50% increase in the density from the unbiased case and a temperature increase of ~15eV during the biasing. This behavior was not seen in the triple amplifier setup as far out as 105cm but rather at 103cm.

Hardware changes: At high biasing (shot 98669), the new MOV failed. A higher rated MOV was placed in the bias probe setup with a clamping voltage rating of 650V.

Monday March 5 2018 5:10 pm Stewart 98670-98704 H-mode Study Radial Scan

This set of runs used the shaping bank to power the bias probe once again, but a step up transformer was used to increase the applied voltage. The progression from shots 98678-98685 showed a range of biasing values from +100V and 50A to +500V and ~125A. The shots with +500V of biased had Triple Probe floating values (located at 103cm) in excess of +350V (see shot 98693).

Comparing the soft X-ray profiles of 98689 (biased to +500V), shot 98680 (biased to 300V), and shot 98704 (unbiased), the edge soft X-rays of the +500V biased shot were ~1.25A.U. whereas the core SXR at the same time slice for the unbiased shot peaked at ~1.1A.U. with an edge SXR value of 0.2A.U. This represents a 6X increase in the edge SXRs for the biased shot from the unbiased case. Shot 98680, which was biased to +300V, had edge SXR values of ~0.5 A.U. in agreement with previous shots with the triple amplifier setup charged to +300V (see shot 98603 for comparison).

Since it was hypothesized that there may be a radiative collapse during biasing leading to fast inward radial motion of the plasma, the OHE was raised from 225V to 245V in shot 98687 to inject a higher heating power. Here, the probe was biased to +500V again but the edge soft X-ray values ~1.25A.U. remained the same as for the lower OHE case (98689). However, the core X-rays increased to 3.5 A.U., which is 2.7 times the amount of an equivalent unbiased shot 98704 with a core X-ray value of ~1.25A.U. Given the much higher edge soft X-rays and consistency of the profile while injecting more power (while the core SXRs were increasing by a substantial amount) a radial scan with the triple probe was undertaken to determine if a pedestal was present at the edge.

Shots 98688-98696 were used as a biased (+500V) triple probe radial scan of the edge of plasma from 103cm to 108.5cm. During biasing, the density was consistent at ~1.5E+18 all the way out to 107cm. At 107.5cm, the density dropped slightly to 1.33E+18 and at 108cm, the density dropped significantly to 3.6E+17. This indicates the presence of a steep density pedestal out around ~107.5cm. For comparison, shots 98697-98704 were an analogous triple probe scan of unbiased shots which had a density ~5E+17 at 107cm. The sharp density change over the centimeter between 107-108cm in the biased shots seems to corroborate the evidence of a pedestal like structure hinted at in the 6X increase in the edge SXRs.

Monday March 12 2018 4:54 pm Stewart 98804-98846 H-mode Study with Higher OH

The goal of this set of shots was to determine the effect of raising the Ohmic heating power on the H-mode like state and to see if the radiative collapse-like behavior could be mitigated by increasing the input power. To this effect, the shot style from previous runs was altered to accommodate an OHE setting of 200 V and and OHST setting of 5,000 V. Comparing shots 98816 with an OHE value of 180 V and shot 98817 with an OHE value of 250 V (both of these shots were biased to +300V using BPS9), one can see that the lower powered shot crashes inward at ~4ms while the higher powered shot moved radial outward at this time and proceeds to crash inward at 5.5 ms. The soft X-rays measured by the fan array also show corresponding crashes in the soft X-rays at 4 and 5.5 ms for these respective shots. However, the SXR values for the low powered shot only reach ~2.4A.U., whereas for the higher powered shot these values exceed 4A.U.

With a new shot style, combining high OH power input and a flattop I_p profile, a triple probe scan was conducted (shots 98833-98837) from 103-107cm. During the H-mode like state, the temperature at 105cm abruptly increases from 35eV to ~70 eV as measured by the triple probe (shot 98834). The density rises more gradually, eventually doubling over the timespan of 1 ms. A similar behavior is also present at 106cm and 104cm. At 103cm the density measurement almost triples over 1ms. Looking at the fan array for shot 98836, the peak SXR value is ~1.2 A.U. all the way until 3ms, when the bias is turned on (this is the same value obtained for the shot style used in earlier run days). At 3.9ms (0.9ms after biasing to +450V), the core SXR value increased to 4.2A.U. This represents an increase of 3.5Xs over the millisecond of high biasing. The edge SXRs here are 4-5 times higher than in the unbiased period. An analogous unbiased shot for comparison to 98836 is shot 98822.

Wednesday April 11 2018 10:51 am Stewart 98890-98930 H-mode BP radial scan

This run report is for 04/05/18. The goal of this run day was to retract the bias probe from the usual position of R=103 cm to 104 cm and 105 cm to see if the transition into the H-mode like state could be achieved and to determine if the radial position of the probe effected the shape/position of the E_r well at the edge. As a secondary experiment, the two triple probes were placed at 105 cm and the TPS5 was moved in the range 102 cm - 107 cm to look for coherence between the two fluctuating TP (V_f) potentials during the biasing.

Shots 98890-98905 were clean-up shots with shot 98906 used as an unbiased reference shot. Shots 98907, 98908, 98917, and 98918 were biased (BP at 103 cm) with the triple amplifier set-up with a ramp at 1.5ms-2.0ms with the maximum possible positive voltage. Most of these shots had mode rotation of -40-50kHz during the H-mode like state with transient jumps to -65kHz. The floating potential during these shots at 105 cm was measured at ~+100V.

Shots 98914-98916 had the same biasing scheme but with the BP retracted to 104 cm. Here, the rotation was again around -40kHz with drops to -20kHz and transient periods at 60kHz. The floating potential at 105 cm was about ~110V for shot 98916 for a period from 2-3ms. Shot 98915 reached this 110V at the transition to the state but the V_f on the triple probe dropped over the same 1 ms window. During these shots, the bias probe saw more of the spike-like noise that is often seen when running with the bank. Retracting the BP to 105 cm (shot 98919), the issue became even worse. Mike pointed out that these spikes in voltage were associated with spike-like drops in BP current that look in time with the rotating mode. Given the radial dependence of the probe on this phenomena, it is likely that it is due to discontinuity as the "tongues" or troughs of the mode pass by the probe.

Of note in shot 98919, the triple probe floating potential at 105 cm only reached +25V and the mode rotation was only about -20kHz. It is therefore difficult to achieve this state when the BP probe is retracted to 105 cm, in agreement with the difficulty of effecting mode rotation in previous run days at locations closer to the limiter. In this shot, the current was still ~110A and the applied voltage +150V, as is the case in the other radial positions.

Shots 98925-98930 constituted a radial scan with the TPS5 from 102 cm to 107 cm. As an aside, shot 98921 was a dramatic biased shot with a minor disruption ~2.8ms (current spike, SXR drop, q below 2, D_alpha spike, and rotation reversal) however, the plasma recovered with the SXRs rebounding even higher, the D_alpha returning to pre-disruption levels and the q recovering and hovering above 2.

Friday April 13 2018 6:09 pm Stewart 98932-98948 Fast Camera Testing with BP

This series of shots taken on 04/09 were used to test the fast camera setup, timing, and data acquisition. Shots 98932-98936 were vacuum and clean-up shots testing to see if the timing/triggering/window of the Fast Camera was correct. Shot 98938 was a reference, unbiased shot that saw mode/turbulent? dynamics in the light and dark regions around the bias probe.

Shot 98941 had the same setup as 98938 but with a bias probe ramp from 1.5 ms to 2 ms into the H-mode like state. The beginning of the shot looks the same as the unbiased reference shot in terms of rotation of the mode and structures around the bias probe but when the H-mode begins, the flow is fast and the details are ether less discernible or no longer present. The frame rate of the camera was set at ~66,000 frames per second while the mode frequency maxed out around 40 kHz. It is possible that this frame rate coupled with the fast rotation essentially "blurred" out the features that were seen in the unbiased case. It remains a topic of a subsequent run day to determine 1.) If the 100,000 frame can still discern these features or if they are actually stabilized by the fast sheared flows 2.) If the very edge of the plasma becomes sharper and more circular due to this sheared rotation (This would confirm that the rotation is intact influencing the structures and not just rotating them faster than the fast camera can perceive).

The remainder of the shots were taken at different Fast camera viewing angles and exposure levels. Of note when, the camera was pointed at the probe and the BP was retracted to 104 cm (e.g shot 98943), there were no signs of arcing or sputtering even through the time traces of the BP voltage and current showed the spiking behavior that was seen in the previous run day.

Thursday April 26 2018 11:42 am Stewart 98956-98979 Fast Camera Testing/BP Study

This run report is for 04/18. The goal of this run day was two fold, to test a new mirror angle and orientation as well as to continue the Long distance correlation (LDC) study on the two triple probe floating potentials. Shots 98977-98979 produced good fast camera cine files of the bias probe induced mode rotation (BP ramp from 2.0ms-2.5ms). Shot 98975 had a disruption starting at 3.5ms and the fast camera file shows a flash of light at the probe tip of the BP. There is visible sputtering off the probe on the fast camera data for this shot after the flash and the BP current measurement shows a large -60A ion saturation current during this time. This provides a potential explanation for the unusually large BP ion saturation currents that have been seen in some of the other shots this year.

Wednesday May 23 2018 3:49 pm Brooks 99035-99060 BP double probe configuration

The goal of my runs yesterday and today was to use the BPS2 and BPS9 in a "double probe" configuration to study the perpendicular and sheath impedences. In these results, BPS9 was placed at 105cm, and BPS2 was incremented from 103, 104, 105, 106, and 107 cm. The goal was to look at the resistance between the probes as a function of radius and try to determine how much of the resistance, again, is due to perpendicular or sheath impedance. The probes were driven in the "double probe" configuration using the Rex transformer and the AETechron at 100% gain. The input to the AETechron was the waveform, 5*cos(2*pi*2500*t+0).

The data is still being studied, but a few interesting observations were made. The currents definitely dropped at BPS2 was moved radially outward (meaning the resistance went up). The voltages of the two probes seemed to be centered around their floating potential (or perhaps an averaged floating potential) as we'd expect from double probe theory.

Wednesday June 6 2018 5:49 pm Stewart 99099-99118 Double Probe/ITB study

The goal of this set of shots was to determine if an internal transport barrier (ITB) could be set up with the double probe configuration while BPS2 was still emissive. Shots 99099-99110 were clean-up shots and tests of the fast camera to find the correct exposure and number of frames needed. Shot 99111 was the first biased shot with the negative BPS2 positioned at R=103cm and the positive BPS9 positioned at R=105cm. The current running between the probes reached a maximum of 20A for the input voltage ramp of [-2V +2V] on the triple amplifier/transformer setup. The mode rotation increased from the usual ~10kHz to ~20-25kHz during the biasing. From the fast camera video, it could be seen that bright filaments were emanating from the probe tip in a localized circular region during biasing. This bright circular region changed location on the hemisphere of the probe throughout the shot, sometimes moving in circular motions around the probe surface.

The voltage was increased to the maximum +10V input to the amplifiers for shot 99113 and the current between the two probes increased to ~30-40A. The mode rotation increased to ~+30-+40kHz briefly near 3.5ms. Since the double probe configuration (which was confined to the region between 103-105cm) was effecting the MHD modes, a TP scan was conducted to see if the electric field at the edge was modified. Shots 99114-99116 were a TP scan from 103 to 106cm to see the effect on the floating potential. For shot 99114, which had the triple probe located between the two probes, the V_f dropped from -44V (no bias) to -200V (with bias). At 103cm (the interior edge of the biasing region) the V_f dropped from -55V (unbiased) to -190V (biased), see shot 99116. Surprisingly, for shot 99115 (outside the biasing region) the V_f also dropped from -35V (unbiased) to -87V (biased). The soft X-rays also show a marked increase for this shot, almost identical to the positively biased H-mode (BP to limiter) case for shot 99021. After this shot, the probe ceased being emissive and the double probe configuration rarely got above 6A of current.

This series of shots show that by confining the bias to a narrow region between 103-105cm (2cm from the limiters), fast rotation can still be achieved at the edge as evident in the MHD rotation frequency. This is corroborated by the large change in the electric field even beyond the biasing region, in the development of an E_r well similar to what might be expected by a negative biased H-mode with the probe to limiter configuration. This behavior may explain why optimum rotation and access to biased H-mode requires the bias probe to be inserted to a certain depth into the plasma: biasing this region alone effects the plasma all the way out to the edge.

Lastly, equal and opposite current is expected to be measured between both of the bias probes during this experiment. However, for shot 99117, only BPS9 saw an increase of ~7A at about 2.5ms, where as BPS2 did not show this increase in current. The two traces of I_bias and V_bias were not mirrored in this case.

Thursday June 7 2018 6:30 pm Brooks 99119-99173 Mode rotation control part 3

The goal of today's experiments was to continue studying the performance of the BP Feedback and Feedforward control system in controlling n=1 mode rotation. I was able to reproduce a shot style characterized by high mode activity and HFS limited, but only too late did I realize that having a HFS limited plasma was making it hard for my feedback system to do its job. I moved the plasma to be LFS limited, and the control system's performance immediately improved. Unfortunately, the strong mode activity vanished, and the control system had trouble measuring the mode frequency. I believe that there is some usable data gathered today, but more than likely another run day will be required.

Thursday June 21 2018 4:49 pm Brooks 99174-99180 Mode rotation control part 4

Today's experiments were plagued by issues. Caliban (GPU feedback system) refused to put out a signal for half of the day. On top of this, the vacuum interlock kept tripping every time I plugged in the AETechron power supplies. I blew a number of fuses on AETechron system today until I added a jensen transformer to the input of the AETechrons. After that fix, the circuit breakers associated associated with the Techrons kept tripping.

Earlier this week, I put the AETechron's into their new parallel configuration and adjusted their various gains to make sure they didn't generate a circulating power between them. Once setup, I performed a number of tests at full power with the AETechrons in parallel using a signal generator and a low-inductance 1 ohm load. I had no problems with the setup in this configuration. Now, in the setup discussed at the beginning, I'm having a lot of problems.

Tuesday July 3 2018 11:01 am Stewart 99187-99227 LDC with the Bank

This run report is for shots taken on 06/29. The goal of the run day was to conduct a long distance correlation study of V_f and I_sat measured by the two triple probes using the capacitor bank as a power supply for the bias probe in the hopes that this would create more reproducible shot to shot behavior. One of the main goals was to investigate the correlation length of the fluctuations associated with the higher (7-9kHz) frequency peak in the Fourier spectrum that roughly matches with the sound speed divided by the machine circumference.

The bank was charged to +300V for the radial scan using TPS5 (TPS8 was held fixed at R=105cm). Shots 99215, 99216, 99218, 99220, 99222, 99223, 99225,and 99226 served as the scan from 108-102cm. From the floating potentials, there was a mean E_r well located between 103-107cm (~4cm width) with a maximum field strength of about 130V/cm. The reduction of Vf,Te,and Ne fluctuations was greatest between 103-104cm and between 105-106cm corresponding to the two sheared layers of the Er well. It is important to note that from the scan, Vf fluctuations of about 10 percent or ~20V for some shots could lead to large changes in the local electric field on the order of ~50V/cm. A 20V change to a potential difference measured across 0.5cm (the increment of the fine portion of the radial scan) results in a 40V/cm change in the local electric field. Looking at Vf from shot 99219, the Vf oscillations swing above approximately 20V. The oscillatory behavior of the measured V_f therefore plays a non-neglidgible role in the total electric field well (mean+fluctuating E_r).

Unfortunately, the frequency spectra for each of these shots were not entirely consistent with "high frequency" peaks moving between 7kHz to 10kHz from shot to shot. The Isat measurements from TPS8 were also lost for the later part of the radial scan, most likely due to a battery issue. Shot 99215 of the scan scan also exhibited the low frequency, bursty mode behavior that was seen during the triple amplifier series of runs. This may indicate that a rake probe is necessary to perform a valid radial scan of a single shot, given the inconsistent nature of the fluctuations.

Thursday July 5 2018 10:04 am Stewart 99228-99247 Negative Biased H-mode Study

The goal of this series of shots was to use the bias probe-"limiter" setup with the newly re-introduced BPS2 to get enough negative current to initiate a negative biased H-mode again. Unfortunately, all of the commands to Caliban resulted in no signal output to the amplifiers. After some checking, John found that outputs 41 and 43 on the breakout board were not allowing signals through. The channels in the board have been changed to 34 and 35 respectively. Hopefully this will allow the Caliban generated signals to reach the amplifier setup.

** Hareware change: breakout board outputs 41 and 43 to the bias probe amplifier setup are now outputs 34 and 35 respectively.

Thursday July 5 2018 5:25 pm Brooks 99273-99324 Shot development

The goal of today's run was to develop a shot style for my mode rotation control work. The shot would ideally be outward limited and have a consistent mode activity > 2ms and >2 Gauss. I spent all afternoon following Ian's run attempting to develop this shot style but never succeeded.

BPS9, which was floating at 103cm, started giving strange voltage measurements starting around shot 99293. I will investigate.

Monday July 9 2018 9:36 am Brooks 99325-99400 Shot development for mode rotation control

Friday's goal was to develop a shot style with the following characteristics: outward limited, 2 gauss instabilities for at least 2 ms, and repeatable. Unfortunately, the repeatability trait proved difficult to find. Possibly due to the nature of the shot style I was trying to develop or because of the machine running at a higher pressure due to excess D2 or possibly because we put in a dirty BPS2 into the chamber earlier in the week. In addition, mode activities were weaker as well, but this may simply be due to the presence of the BP in the machine. However, I think I may be very close to developing a shot style. Shots 99392 to 99394 looked good but became unrepeatable. Scanning through adjacent parameters resulted in the shot style almost redeveloped.

Similar to the prev run day, BPS9 voltage measurements seem off again. Need to investigate.

Monday July 16 2018 10:15 am Brooks 99403-99433 Shot development

The goal of Thursday's (7/12/18) run was to develop a shot style suitable for my mode rotation control work. Specifically, the shot style is outward limited and has mode that is at least 2 Gauss and 2 milliseconds long. Shots 99432 and 99433 seem to meet this requirement and also appear to be mostly repeatable.

Of note, the lower leg of the voltage divider for the BPS9 voltage blew for the second time that day. The previous time it broke was the previous week. Mike suggested swapping to a higher wattage resistor as well as changing from a 1:100 divider to a 1:200 divider as voltage spikes during disruption or breakdown were likely responsible.

All shells were in except S9 (both), S10 (both), and S2 (top only).

Monday July 23 2018 10:18 am Brooks 99441-99468 Moe rotation control part 5

Continuation of my mode rotation control work. A shot style was developed and mostly reproducible today. The control is working and has shown some intermittent success in mode rotation control. More data is required, and it would be nice if I could pin down why the control system works sometimes and not on other shots.

Good news. The AETechrons can now be run in parallel allowing for additional current. The techrons are now powered from the south rack to remove 60 Hz and some HF noise. The jensen was removed between the south rack and techrons' input. With it installed, grounding issues (leading theory) causes large power draw from the techrons, tripping of TC based interlocks, and techron power breakers.

A new TA sensor failed on shot number 99446. Looking at older shots, it looked a little flakey, but now it looks dead.

Shells in against bumper except S9 and S10, top and bottom. Also S2 top.

Thursday July 26 2018 12:18 pm Saperstein 99469-99535 troubleshooting SOLC sensor pickup

I missed the run report for the previous three run days as well, so I will be covering all four here. The purpose of all four run days was to determine the source off excess pickup in the SOLC sensors. For all shots, the recording time for diagnostics was increased to cover -50 ms -> +150 ms, with the exception of some shots where the sampling frequency was lowered in order to recorded over even larger times. This was done to observe a more complete profile of low frequency pickup, and to verify whether or not digitizer measurements returned back to their original offset at the end of the shot. All, but one of the following shots were also vacuum shots.

Day 1:

(Shots 99469-99480)

This run day was mostly dedicated to initial looks at the long term pickup profiles for shots with either only the VF or OH banks firing. Unfortunately, the unknown artifacts being picked up in the signal didn't have the completely sinusoidal nature I thought they might have. I also noticed that the measurements from the SOLC sensors and VF current sensor didn't always return to their offset by the end of the shot. In order to investigate this, the section 8 bottom feedthroughs were disconnected from their amplifier box, and two jumper wires were set up to connect only sensor 1's feedthrough pins to the amplifier. The purpose of this was to test if the new offset introduced by the end of the shot would reverse the same way as the rest of the signal if the polarity of the feedthrough connections were reversed. Unfortunately, the jumpers we made to test this were made too short, which caused problems for reversing the polarity without the wires popping out of the feedthrough, as well as making a good electrical connection.

Day 2:

(shots 99481-99491)

The run day began by trying to fix the problems seen at the end of the previous run day. Longer jumper wires were made, and attached to the d-sub connector pins for for sensor 1 in section 8. This time, we were able to correctly reverse the polarity of the signal. We found that the offset didn't reverse completely, and while the offset did have the opposite polarity, it's magnitude was much closer to ~0 than the previous offset. Some shots were also taken with gain for section 8 bottom increased from ~100 to ~400. This helped to reduced the offset change in some of the shots. Shots 99490-91 were also taken with the feedthroughs from the same shell completely disconnected from the amplifier box, without any jumpers connecting any of the pins. In these shots we found that a small amount of pickup was still being measured, even with the feedthroughs disconnected.

Day 3:

(Shots 99492-99510)

The majority of the shots taken for this run day were testing to see if the strength of the offset change was dependent or not on whether the current shot has different banks firing than in the previous one. So I went back and forth between taking a couple VF only or OH only shots. We did find that only the first shot after switching from one bank to the other had a relatively large change in offset, and that the following shots with the same bank settings no longer had the offset change. Some digitizer drift was still noticeable towards the end of the shot, but it was negligible compared to the offset change introduced when changing banks. We believe this change in offset is due to the d-sub connectors between the SOLC feedthroughs and amplifier boxes being magnetized in different ways, depending on the banks settings. If the previous shot's bank settings were different than the current one's, then it makes sense for the magnetization at the end of the current shot to be different from the magnetization of the previous, changing the offset. We also tried waving a magnet around the d-subs to see if this would change the offset in the following shot, even if the bank settings weren't changed. We found that the offset did change, although not as drastically as it does when changing which banks are being fired. Shot 99510, the last shot of the day, was a plasma shot taken in order to make sure the spectrometer was still working correctly.

Day 4:

(Shots 99511-99535)

This run day started with taking shots with both the VF and OH firing, but with varying strengths of their fields. The purpose of this was to get an idea of what kind of effects would become present due to the coupling of the two fields. I also took some shots with all banks firing, and some with only the TF. For shots with only the TF firing, the recording time was extended to ~2 seconds for some of the shots, in order to verify that the current measurements were returning back to their original offsets by the end of the shot. All of these shots were taken in at least sets of two, in order to compensate for the offset changes introduced by changing the bank settings. We found that if only the magnitudes of the bank's fields were changed, and no new banks were introduced or removed, the offset introduced was typically negligible. In addition, it's also worth noting that the pickup from the TF field is significantly larger than that from the VF and OH fields, and that within the timescale of the plasma lifetime, this pickup doesn't appear to simply be some constant shift in the pickup. After these sets of shots, I took some shots with varying gain settings in the Section 8 bottom amplifier box, while the feedthroughs were disconnected from it. The purpose of this was to determine if the pickup we were seeing coming from the amplifier box was primarily a result of the ground loops introduced by the various dip switch options, used to set the gain, that changed the amplifier circuit. What I found seems to agree with this theory. The last two shots I took, were ones with the output of the amplifier box disconnected. The purpose of this was to see if the pickup would disappear if the amplifier was disconnected from the digitizer, and this is what we saw.

Monday July 30 2018 1:54 pm Brooks 99536-99594 Moe rotation control part 6

The goal of this day's [7/26/18] run was to perform more mode rotation cases to generate more plots for the BP RSI paper.

The shot style, a continuation from last time, took almost 20 shots to redevelop. It had been about a week since the machine ran last so maybe that was it. In addition, several of the shells were in a different location. Once the shot did develop, however, it was better than before. I experienced about 4ms of steady 3/1 mode behavior with a major radius between 91 and 92 cm. It was also a very consistent shot style.

The run day went very well. Unlike previous mode control tests, the control system clearly worked (or at least mostly worked) on every shot. I performed FB operation with the gain higher than previously, Kp=0.10. This caused the system to be underdamped (overshoot followed by oscillations), it worked. I performed several steps from between 5 and 8/9 kHz. I also recorded floating only shots with the probe in the same location to provide a comparison.

Setup: All shells in. AETechrons in parallel configuration. Rex transformer in 3:5 voltage step up. Limiters out. 'HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_86' and 'HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_87' record the transformer92s primary voltage and current, respectively. The primary voltage measurement isn't quite correct for some reason.

Monday July 30 2018 2:16 pm Brooks 99595-99638 BP radial scan for "controlability" study

The goal of this day's [7/27/18] run was 1) to study the mode rotation control's 'controlability' as a function of the BP's radius and 2) to take another look at the amount of current available to the probe in the SOL.

The shot style was a continuation from the previous day. I experienced about 4ms of steady 3/1 mode behavior with a major radius between 91 and 92 cm. It was also a very consistent shot style.

The run day, again, went very well. I moved BPS9 in 0.5 cm steps from 103.0 to 108.0. At each location, I ran with full positive power and again with full negative power and recorded the n=1 mode rotation frequency.

The main set of results show that the probe has the ability to control the plasma from about 4 kHz to 12 kHz in this particular shot style and with the probe between 103.0 and 104.5 cm. During this range, the power supply was limiting the amount of voltage and current delivered to the plasma. When the probe was moved to 105.0 cm, the amount of voltage and current was the same but the resulting frequency range was narrower. From 105.5 to 108 cm, the probe's voltage and current continued to drop, but the resulting frequency range was effectively zero. This tells me that the BP had no ability to control mode rotation when the probe was too close to the walls. The reason is not yet known.

Another observation was that there was a substantial amount of "arcing" on the BP voltage and current when the probe was located between 106.5 and 108.0. In addition, the arcing was observed on many of the other sensors including the SXR array and OH current. It is not known if this behavior is truly "arcing", but it only happens when the probe is in or near the SOL. Perhaps due to the mode activity and the proximity to the SOL, the probe is moving in and out of high and low density plasma regions.

Setup: All shells in except S9 top and bottom all the way out (same section as BPS9). AETechrons in parallel configuration. Rex transformer in 3:5 voltage step up. Limiters out. 'HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_86' and 'HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_87' record the transformer92s primary voltage and current, respectively. The primary voltage measurement isn't quite correct for some reason.

Friday September 21 2018 11:15 pm Levesque 99781-99816 Poloidal EUV system inaugural run

Made the first plasmas following vessel venting and installation of the new 64-chord poloidal EUV system with 100nm Al filters (matching the SXR fan array filter material and thickness). Other hardware changes included adding a gate valve and another puff valve, but they're not connected to anything yet. Base pressure was around 6e-8 torr without the e-gun on before the run began, dominated by water. The bake was operating for about 5 days, with a couple of several-hour off periods due to the controller or thermocouples having errors. No GDC was done due to concerns about causing EUV filter degradation.

Data collection from the new EUV system was successful! Plasmas were typical for the first operating day after an up-to-air period. Emission signal levels from the central chords are comparable to the SXR fan array, while edge chord emissions are very weak. Signal quality should improve with plasma quality, especially for the outer chords. A star shot is 99812, which was long enough to have coherent 4/1 mode activity on the magnetic sensors. The EUV system sees fluctuations from the mode, but detailed analysis has not been done. HFS-viewing EUV chords strongly increase as the plasma crushes radially inward following the disruption current spike for shots that were studied closely.

Fast camera data were collected for many shots, though the view (looking counter-clockwise from overhead, tangent to HFS limiting flange) is quite narrow since we don't have the coherent fiber bundle yet. Rotating mode activity is seen in some videos. Fast camera images had some speckling of bright spots during the start of some disruptions, especially those disruptions which generated extremely high D_alpha light. Each speckle covers <= 5 pixels and lasts only 1 frame. Speckles are mostly randomly distributed, and some appear outside of the area where the camera should see light. I speculate that the speckles are caused by hard x-rays depositing energy in the CMOS image sensor as fast electrons are lost. Shot 99815, frame 366 had the most speckles by far -- it had over 30, while most frames have only 1 or 2 when they are present.

The Flame USB spectrometer was setup again with a mostly radial view at Section 10. Spectrometer triggering worked today, though the timing might be wrong. Signal levels are very low such that only deuterium light was observed. We'll have to look into improving signal collection and/or transmission.

Tuesday September 25 2018 4:35 pm Stewart 99817-99870 Clean=up Shots

Run Report for 09/24. This was the second day of running with the new EUV system in place. It was found by shot 99827 that the EUV TIA board #7 had a bad +15V power plug. Jeff fixed this is issue and the remainder of the shots during the day had working EUV chords. The majority of the run series consisted of 50 clean-up shots with the pulse length increasing to approximately 4.5 ms by the end of the day (see shot 99863).

Sunday September 30 2018 2:51 pm Stewart 99925-99945 H-alpha Photodiode/USB Spectrometer Testing

The goal of this run day was to test the newly constructed photodiode/amplifier circuit designed to look at H-alpha light at different angles in the chamber using a fiber optic and the fast camera viewport. The main goal of this would be to get a better view of the plasma edge/limiter area during H-mode shots for better diagnosis of recycling changes at the edge. The second goal was to continue testing the USB spectrometer by attempting to make longer plasmas so that the second spectrum would show impurity lines.

The first couple of shots were mainly clean-up oriented with some D-alpha spectrometer testing involved. Shot 99926 had the lab lights off and no black bag covering the spectrometer, while shot 99927 had the lab lights on with the black bag off go the spectrometer. The final shot in the series (99928) had the black bag once again covering the spectrometer with the normal improvement in noise, comparable to the lights off case.

The rest of the day had several of the shots Jeff is interested in with very low SXR radiation. The five shots that exhibited this behavior were: 99935, 99937, 99940, 99941, and 99942. The shots were characterized by almost no SXR signal, lower plasma current , higher edge q, mode activity mostly below 5 Gauss, lack of a prominent current spike at the disruption, and very high spikes in the D-alpha spectrometer signal (In comparison to the shots interspersed between these shots). In particular, shot 99935 had semi-periodic D-alpha bursts that were larger than the H-alpha signal signals seen during normal disruptions and start-up (see shot 99936 for comparison). Lastly, shot 99942 had the fast camera recording data and showed the same "sparkling" behavior that Jeff had pervious mentioned in his run report (hard X-rays from run-aways?).

Friday October 5 2018 12:38 pm Stewart 99946-99961 Photodiode Testing/Clean-up

Run report for 10/01. This run day was focused on testing the new photodiode with outputs into the South Rack CPCI instead of the scope. It was also intended as continued clean-up to eliminate the shots that randomly have lower SXRs and plasma current. Shots with this behavior included 99951, 99953, 99954, 99955, 99956, 99957, and 99959, all of which had the new EUV system running. On shot 99961, when the photodiode system was connected to the South Rack, the major radius calculation and the edge q calculation began to have a "ringing" behavior. This was isolated to the ".sensors.rogowskis:cos_1:raw" node in the tree (this sensor is connected to the West Rack) that is used for the calculation. Attempts to fix this issue were saved for the next run day.

Wednesday October 10 2018 12:31 pm Stewart/Levesque 99962-100001 Clean-up/Double Puff

Run report for 10/05/18. Most of the day was devoted to continuing clean-up since the chamber pressure is still higher than normal and the pulse length of each shot is still shorter than normal. Shots 99962-99979 continued to exhibit the "ringing" on the q calculation and major radius due to the cos theta Rogowski in the West Rack. Jeff swapped the cos and Ip Rogowskis, as well as the AM502's in the West Rack, for shot 99980. The ringing was not present on either of these sensors for this shot. The hardware was then swapped back to the original configuration for shot 99981, and the ringing behavior was gone. The rest of the shots from the run day did not have this issue.

Shot 100,000 was also taken by Jeff after the series of clean-up shots. The shot only lasted about 5 ms and featured some 4/1 mode activity reaching around 10 Gauss. The last shot of the day was a double puff experiment with the second puff occurring at 3.5 ms and lasting 20 microseconds. The bottom EUV array (DET270) shows a ~50% drop in SXRs in some of the outboard chords and a ~50% increase in the more central chords about 0.5 ms after the puff. This may indicate cooling at the edge since the puff would be increasing the density there (the SXRs should go up if the density increases and T remains constant), while the core could be heating or getting denser. This may be related to the results from cold puff experiments done on C-mod where the core heated and the edge cooled, but more shots (along with TP data) would obviously be needed to make such a conjecture.

Thursday October 11 2018 4:43 pm Stewart 100002-100036 Clean-up/Photodiode Testing

Run Report for 10/10/18. The pre-existing goal for the run day was to investigate the double puff experiments further, with the tripe probe inserted to determine temperature and density changes that could corroborate the EUV SXR data from the prior run day. However, the plasma continued to have intermittent issues with breakdown. Several of the shots had no breakdown at all and some had low D-alpha signal during breakdown (see shot 100017). These shots also had low SXR readings on the SXR mid plane sensor and high spikes in the D_alpha spectrometer, clipping +10V on the CPCI input (again see shot 100017).

Shots 100024-100026 were tests of the new photodiode for the H-alpha sensor. 100024 had the diode and amplifier connected to the CPCI input for BPS9 current, both were off for this shot (the CPCI read zero volts for this shot). 100025 had the lights off and amp powered on. The CPCI read an offset ~0.07V for the shot. Shot 100026 had the lights around the photodiode turned on and the CPCI trace shows oscillatory traces they may be due to the lighting. Further testing will have to be performed when the fiber optic arrives next week.

Thursday October 11 2018 4:43 pm Saperstein 100037-100069 looking at MHD pickup in SOLC sensors

10/11/2018 The goal of today's run plan was take take several shots with shells hosting SOLC tile retracted ideally beyond the SOL. Types of shots include ones where all shells are fully inserted, others where one of the SOLC sections (1,4,8) has its shells fully retracted, while all other shells were fully inserted. The first 8 shots were used to look for a shot style that would produce strong, consistent MHD modes that could be analyzed. In the end, the shot style that was decided on was one that John Brooks typically uses. We did run into problems throughout the day where no breakdown (or poor breakdown) of the plasma was occurring consistently shot after shot. We tried moving the e-gun closer to the plasma (3mm closer), as well as trying to increase the puff time from 960 to 1000. This didn't have an apparent effect on the breakdown. Then Ian noticed that the charge on the e-gun battery was slightly low (~11 V), so the battery was recharged before continuing to take more shots. However, we still weren't getting breakdown. Lastly, we tried increasing the puff time a little more to 1020, and after doing this the plasma did start breaking down again. It's unclear as to whether or not the successful breakdowns were a direct result of increasing the puff time, or whether it was just a coincidence, but we were able to achieve good breakdown for most shots the rest of the day. However, occasionally we would still get either bad breakdown, or no breakdown at all. At the end of the run day, the sect 8 shells were still fully retracted, and the e-gun was left pushed in ~3mm.

Friday October 19 2018 12:45 pm Levesque 100070-100107 Radial decay of SOL current

Ran far-outboard shots while retracting shells in Section 4 to a few different distances. The goal of this campaign is to measure fine detail of the radial drop off of SOL current to Section 4 SOLC tiles. Not enough good shots were achieved for a satisfactory radial scan. The campaign may be continued on a later run day.

Before the run, I fixed amplifiers for 3 feedback sensors in Section 4: FB04_s3p, FB04_s3r, and FB04_s4r now appear to be working. All 3 amplifiers had bad 402ohm resistors at the amplifier input; this might be a common problem to look for among other broken sensors. Jumper A at Section 9.5 is now disconnected (was previously still connected with the ~8" extension), Jumper C is connected at Section 10.5 (previously disconnected), Jumper B is connected at Section 3.5 (previously disconnected while commissioning the new EUV system), and Jumper D is still not connected. Section 5 shells were fully retracted all day in order to study Jumper B currents with Section 5 being minimally impactful. Note that the new poloidal EUV system housing hardware is close to the fully-inserted top and bottom nominal shell faces, so the EUV hardware could significantly impact SOL currents for full-size plasmas. Limiters still remain at 2cm retraction from their standard R=107cm and Z=+-15cm positions.

Much of the day was spend trying to reach the target discharge evolution of shot 98434. This has an edge q slightly above 2 during the target period. Raising the e-gun filament current from 9.7A to 11A might have been slightly helpful. There were no breakdown failures today, but SXR signals were slower to rise than they have been in the past. Jim purged the puff line before shot 100086, and this appears to have helped the shot progression. After the puff purge, shots mostly followed the target evolution, but most terminated early due to 2/1 mode activity. Star shots and Section 4 shell positions were: 100088 (Shells fully in), 100090 & 100091 (Shells back 3.0cm), 100099 & 100101 (Shells back 0.5cm), and maybe 100107 (Shells back 1.5cm). There were several other shots that might be useable leading up to their disruptions around 3.0ms. Measured SOL currents in the tiles and Jumper B definitely dropped off with shell radius, but I haven't quantified it yet.

The ion gauge tripped off for every shot when in its normal state (viewing IG1 instead of IG2). An anomalous event in the pressure reading trace occurred very close to -90ms for all shots, though there are no particular triggers that happen around that time. Pressure evolution before the anomaly and plasma behavior and resulting pump-out suggest that there's no extra puffing beyond what was desired. Activating IG2 instead of IG1 caused the gauge to stay on the whole time, though it would automatically switch to IG1 each shot and the recorded trace still had problems. No ideas about what's causing this for now.

Full bank charge was dumped several times during the run. After the run, one of the TF dump resistors on the orange bank was found to have boiled over and splashed about 10 ounces of water around its base and a little on the capacitors. Jim cleaned this up and refilled the resistor -- normal operation can continue, but users should be aware of this suspect resistor during setup and operation. In addition to the multiple dumps, this may have been worsened by the affected resistor being a little shorter than the other dump resistors in its family, so it might have a higher energy deposition and temperature increase during each dump. This specific resistor has also been damaged in the past. The water level in this resistor was just at the top of the upper electrode at the start of the run. Security cameras were inoperative during the run so the banks could not be directly monitored to see the problem as it happened. One of the power cables for the cameras was found unplugged; plugging this in restored camera functions so the banks can now be monitored again during runs.

Friday November 2 2018 3:59 pm Stewart 100108-100122 Fast Camera Testing

This run day served as the first day of testing the new fast camera fiber optic system during plasma operation. The testing was successful and the fast camera view closely matched that of the previous fiber optic system used by Sarah Angelini. Unfortunately, half of the shots had "bad breakdowns" and the other half only lasted 2-3 ms with prompt current spikes about 1 ms into the pulse. This continued throughout all of the shots 100109-100122, even though Jim purged the puff line. One shot stood out and was of the bad breakdown/ low SXR type: 100117.

Shot 100117 lasted approximately 7 ms and had a comparatively low start up plasma current. The major radius calculation exhibited an erratic behavior, jumping ~1cm every ~0.5ms. The magnetics showed "spikes" of B signal between 2-4 ms, while the D_alpha spectrometer clipped 10V of signal at the current quench. The fast camera showed rotating filamentary structures in this shot, one lighting up much like the filaments from the emissive bias probe videos.

Wednesday November 14 2018 4:43 pm Brooks 100123-100183 Cleanup shots

The goal of today's run day was purely cleanup shots. Of note, Jim installed the new egun filament since we ran last. In addition, Jim opened the new proof-of-concept graphite bias probe to vacuum, but we haven't yet pushed it into the plasma. The plasma conditions still have not recovered from the up-to-air for the EUV fan array upgrade, and the insertion of the new filament and probe likely did not help.

The run today was characterized by two shot styles. When the fill pressure was low (<65 micro Torr) the plasma did not seem to break down correctly, and the various optical measurements (spectrometer, xray, etc) did not show much even though the plasma current was non-zero. The other occurred with higher fill pressure (>70 micro Torr). The plasma only lasted about 1 millisecond (very consistently), but it appeared to be much closer to the way HBT plasmas were before the upgrade. Despite many shots of this second type, no improvement on plasma conditions were observed.

Jim notes that the Deuterium regulator is leaking and is working on it.

There has been an issue with the Granville-Phillips Vacuum Gauge Controller in the control room. Its internal control logic gets activated on during every plasma shot, and therefore it needs to be reset after every shot. Jim is aware and is looking into it.

Ian is planning on doing another cleanup day tomorrow and possibly Friday as well.

Friday November 16 2018 10:38 am Stewart 100184-100213 Clean-up

This was a series of 28 clean-up shots, since John's run day did not show significant improvement in pulse length. The first couple of shots kept the same shot style as the previous run day, which only lasted approximately 2 ms. Lowering the OHS allowed for a lower plasma current and the pulse length increased modestly to 3 ms.

Jim checked the RGA and found that there was a large concentration of water in the chamber (which may be due to the introduction of the graphite probe). The graphite probe was retracted out of the chamber for the entirety of the run day since the plasma did not noticeably improve.

The day, however, produced some interesting slow moving modes. Since the plasma current was lower than usual, the edge q at the beginning of the shot was between 4-5. A good example was shot 100197 with a slow rotating ~5/1 mode.

Wednesday November 28 2018 4:32 pm Brooks/Saperstein 100215-100252 Cleanup shots

Today's run focused on cleanup shots. We baked HBT over the last past long-weekend (~6 days), and Jim reports that the water content is down by an order of magnitude. Over the course of the run today, the shots significantly improved from 1ms long plasmas to 5ms long plasmas. In addition, the loop voltage, spectrometer, and SXR plots are looking significantly improved. Ian plans on doing another run day tomorrow.

In take_shot, the north rack was throwing errors on one of the A14's all day. This has happened semi-frequency in the past as well.

The 307 granville-phillips vacuum gauge controller continues to reset itself after every shot. This is presently under investigation.

Jim finished baking and turned on the second cyro pump this afternoon which significantly reduced the base vacuum level.

Friday November 30 2018 6:58 pm Stewart 100253-100282 Graphite Probe Testing

Run report for 11/29/18. This run day was devoted to testing the new graphite electrode that Jim constructed with the fast camera viewing the probe to inspect for sputtering or damage. The first plasma shot of the day 100254 lasted ~5 ms, so it seems the prior day of clean-up shots was a success. It is however important to note that the shot style had a plasma current ramp instead of a flat top for these longer duration shots. With the flat top shot style (shots: 100257-100272) the pulse length was not as long and disrupted prematurely (compared to shots before the up to air), despite having an edge q above 3 for the duration of the shot. This did not change when the major radius position was changed from 90-92 cm. There was a quick growing magnetic signal before each of the disruptions (perhaps tearing modes are the culprit?).

On the fast camera videos, one can see that the plasma rotation was faster for the flat top shot style (e.g. shot 100257) ~8-10 kHz while the ramped Ip shot style (e.g. shot 100256) had very slow mode rotation. This difference in rotation characteristics is corroborated by the mode frequency analysis using the Mirnov probes. From shot 100255 onward, the fast camera video also showed the triple probe at section 8 (retracted behind the shells at first, but then placed at R = 103 cm for shot 100256 and higher) lighting up. The triple probe has not been inserted fully into the plasma since the up to air, so it is most likely due to impurities from the up to air.

Finally, for shot 100275, the graphite probe (now BPS9) was inserted to 105cm. The plasma only lasted a few milliseconds and the probe was giving off a lot of light on the fast camera video, so we checked the probe visually for damage (none was found), the pressure in the chamber (which was normal), and Jim used the RGA to see if any carbon was present. He found that the CO line was higher than usual but it was not a dramatic increase. So, we forged ahead with 3 more shots with the probe inserted (100276, 100279, 100280). The pulse length for the shots increased after each successive shot and were only about 0.5 ms shorter than the shots without the probe inserted by 100279. The light from the probe/around the probe also decreased after each shot, which might indicate that it was "cleaning-up". However, on shot 100280 sputtering like that on the emissive molybdenum electrode (after it is exposed to atmosphere) was visible from the graphite probe on the fast camera video. It was decided that the probe should be retracted and the results analyzed before further shots with the probe are taken (no damage was visible on the probe).

After the probe was retracted behind the shells, the shots went back to normal: Same pulse length, plasma current values, major radius evolution, etc. Hardware update: The shielded power cable for the EUV system was installed prior to running.

Wednesday December 12 2018 11:58 am Stewart Fast Camera and Bias Bank Testing 100283-100310

Run report for 12/06/18. The first half of the run day was devoted to shot development, since the shot styles with flat top plasma currents were terminating prematurely (most likely due to tearing modes since the edge q was above 3). The second part of the day was devoted to using the shaping bank to bias the molybdenum electrode at section 2 with the intent of creating the first biased H-modes with both the new EUV system and the new fast camera set-up. Unfortunately, although the shaping bank was charging normally and to the proper preset voltages, the probe showed no biasing. Instead, the probe collected ~6-8 A of current and behaved as if it was grounded. The run day was cut short in order to debug the biasing circuit to determine the issue in the biasing system.

Monday December 17 2018 2:30 pm Stewart 100311=100340 Negative Biasing H-mode w/ Fast Camera

Run report for 12/13/18. The goal of this run day was to increase the negative voltage applied to the bias probe (BPS2) using the shaping bank to (1) see if higher electric fields and rotation could be achieved and (2) use the fast camera and new EUV system during the shot to further assess changes to the edge. The start of the run day turned into a couple of hours devoted to determining why the shaping bank was not biasing the probe. It was found that the trigger for the shaping bank to discharge was not present in the current tree (it may have been removed due to an absence in the make_tree file). Even after this correction, it was difficult to get the timing right for the trigger to be at 3 ms. After a change in outputs in the basement rack and some trial and error with the take_shot interface, a timing of 2 ms for the trigger was finally settled on.

The two shots with positive biasing (+300 V) included shots 100335 and 100336, which had n=1 frequencies of -60 kHz to -70 kHz during the bias. During the H-mode window, the current was kept around +80A but during the disruption, the bias current increased to +160 A. The soft X-rays also tripled on the fan array during these shots (for a reference unbiased shot, see shot 100331).

Next up, the negative biasing cases were set to -300 V by swapping the outputs to the probe stand. Shots 100338 and 100339 were biased at 2 ms to -300 V and saw n=1 mode rotation of +60 kHz (a record for HBT-EP?). The SXRs also tripled for these shots however, the most striking behavior was found in the magnetics. Both 3/1 and 4/1 modes are present during the biasing, but the 4/1 mode is rotating at ~60 kHz, while the 3/1 mode rotates just beneath it at ~15 kHz. Using a model for the safety factor given in Wesson, the q =3 surface and the q=4 surface (if the edge q is 4) should only lie 2-3cm from each other. This supports the idea that there is a thinner E_r-well in the negative biasing case with very strong rotational shear across the last 2-3cm of the plasma. The fast camera also shows a dramatic change in the edge of the plasma that might indicate a sharper edge, however the resolution and amount of light going into the camera make discerning what is going on difficult (the light may even be increasing). The video does show an increase in rotation as one would expect from the MHD signals. During the biasing, the probe collected above -100 A of current (well above the I_sat limit of 4-6A). After the probe had been "cleaned" by these two negative shots, the very next shot (100340) went back to being I_sat limited and only drew -4A of current despite being biased to -300 V.

Thursday January 10 2019 10:08 am Stewart 100363-100390 Thomson Testing

This run day was devoted to testing the Thomson scattering system (aligned this week) on plasmas to see if scattered light could be detected and if the data from the oscilloscopes would be stored in the tree. Despite early setbacks, the run day was a success with about ten good (largely reproducible) shots with Thomson measurements on 2-3 of the polychromators.

First, it was found that the trigger to fire the Thomson laser was not being sent out (the trigger to turn off the flash lamps however, was working). Changing the firing trigger output to a different channel was unsuccessful (changing the tree in traverser did not change the outputs on the J222 in the North Rack). Jeff set up a delay generator to send out the fire pulse 112705 microseconds after the flash lamps off pulse. All of the A14s in the North Rack were also disconnected since one of the A14s (A14 #18) was failing during the prepare phase of take_shot. After this adjustment, the triggering system worked.

Shots 100380-100383 were the first couple of successful Thomson shots and measured temperatures on polychromator 4 of about 100 eV (R=96 cm) and on polychromator 5 Te ~ 40-60 eV (R=88cm). The density measurements for these shots were also fairly consistent shot to shot. Polychromator 6 was added on shot 100384 and started taking data on shot 100387. Shots 100388-100390 were repeats of the same shot style and had consistent results for density and temperature (100 eV for channel 4 and 40-60 eV for channel 5).

Monday January 14 2019 3:27 pm Stewart 100391-100436 Thomson Radial Scan

Combined run report for 01/10 and 01/11. These two run days were used to determine if Thomson measurements could be made outside of the core and to get Thomson data during biasing experiments. Some good Thomson shots during the first day, used as reference shots with the same shot style, include:100394-100396 and 100400-100405. Thereafter, the fiber optic connected to Polychromator #5 was moved from 88cm to 102cm for shot 100406. Only stray light (at very high levels with clipping on CH1 of the polychromator) was found for this position so the fiber was moved to 98cm. Here, Thomson data was again reasonable (shot 100407). The next radial scan shot, at R=100cm, 100409 saw some scattered light but this was masked by the amount of stray light being detected. The results indicate that reasonable Thomson measurements most likely can be made for radial positions: R= 86cm, 88cm, 90cm, 92cm, 94cm, 96cm, and 98cm (or 60% of the planned ten Thomson chord locations from 88cm-106cm). It is possible that the fiber termination housing is not in line with the beam for the last few radii (100,102,104 and 106cm). More work will have to be done during nitrogen calibration to determine the root cause.

Three good reference shots for Thomson for the second day were 100430-100432. Between both of the days, only three shots with H-mode biasing were taken: 100411, 100433, and 100434. Unfortunately, the thermal quench occurred at ~3.1ms for all of these shots- the same time as the Thomson scattering. Not surprisingly, the temperatures recorded for these shots were half that of the reference unbiased shots. These experiments will have to be repeated with careful timing of the Thomson to get accurate temperature and density changes during H-mode (and determine if Te actually does drop).

Wednesday January 23 2019 10:52 am Stewart 100462-100473 Thomson Fiber-holder Alignment

Yesterday's shots were devoted to changing the alignment of the Thomson scattering fiber-holder on the machine (the "helmet") in order to see if better signals could be acquired on the outer Thomson channels (R=100cm and above). Turning the holder counterclockwise (looking at the machine center) did not improve the signal to the outermost channel on polychromator #5. However, when the holder was rotated clockwise as far as the turning dials would allow, the stray light on channel A for all of the polychromators decreased. This decrease was larger than a factor of two for polychromators 4 and 6 (see shot 100473). The signal to channels B and C also increased for this shot in comparison to the original holder position in shot 100467. The signal level on polychromator #5 (at R=100cm) was comparable to the core channels for 100473, indicating that reliable temperature and densities should be able to be measured at this location. It remains the goal of the next run day to see if channels even farther out now have sufficient signal to make density and temperature measurements, as well as to see if the signal level can be optimized even further.

Friday January 25 2019 10:23 am Stewart 100474-100504 Thomson fiber-holder Alignment

Yesterday's run day was also conducted in order to achieve better alignment of the Thomson fiber holder. The helmet was able to be turned slightly more clockwise looking towards the machine centerline, however this led to a loss of light on the chord looking at R=102cm (Polychromator #5). See shot 100483 for reference. The holder was then turned back counterclockwise in a series of successive increments for shots 100484-100497. The optimum light collection configuration looks like it was found at the last position in shot 100497, where there was an 8-9V signal on channel B for poly 5 positioned at R=102cm. But there was no discernible signal on channel C for this shot. Moving the holder even further in the counter clockwise direction may have led to worse signal, however it is hard to discern given shot to shot variation (the pulse at R=102cm was stonger on shot 100501 for Poly 5: channel B). The "optimum" position for shot 100497 gave very good results for the chord R=100cm (shot 100498), with a 10V signal almost matching the core chords. This alignment therefore would most likely give good temperature and density measurements out to R=100cm but it remains to be seen if further radii can be diagnosed.

Tuesday January 29 2019 10:55 am Brooks 100506-100536 GPU testing and transformer testing

The original goal of this run day was to test the new graphite probe and also characterize the new stangense transformer. All of the shots were "crates only" and not plasma shots.

As it turns out, the graphite probe apparently detached from the copper rod back in November, and it has been sitting askew in the chamber sense then. All plasma shots between now and the tests in November have had this probe intruding on the plasma. The probe tip appears to be several centimeters below the centerline and in at least to 105cm if not further in. An up to air is required to remove it.

The GPU originally had issues working, but I eventually sorted through these issues. Now, the voltage measurement on the primary side of the Stangense transformer is recorded by the CPCI, but it lags behind the secondary voltage measurement on the CPCI. The o-scope does NOT show this delay, therefore I suspect the cabling is causing a delay in the signal. This still needs to be figured out.

Thursday January 31 2019 11:00 am Brooks 100537-100581 BP testing of Stangenes transformer and curret mapping

Two goals for yesterday's run: 1. Perform a radial scan of the BP while it drives high frequency currents into the plasma. I will next use a targetted FFT or similar analysis to attempt to track the currents through the plasma at the Mirnov and SOL current sensors. 2. Characterize the transfer function of the Stangenes transformer in the plasma for all 4 electrical configurations.

All shells but section 10 t&b and section 9 b were in. Section 2 t&b shells were pulled out briefly for a few shots mid-run. Also, section 5 shells were both found to be out at the beginning of the run day and were immediately inserted. It's unclear how long these shells were out.

The detached graphite probe at section 9 is still located deep inside the plasma, and all shots today were taken with this in place. An up to air is planned within the next few weeks and will remove it

The high speed camera took a few recordings of the graphite probe. The usb spectrometer was not operating.

Friday February 8 2019 9:56 am Saperstein 100583-100639 Radial scan of Sect 10 shells

The purpose of this run day was to perform a radial scan of the section 10 shells, and investigate how the currents running through jumper A, between sect 10 and sect 9, decay as the section 10 shells were retracted. Results from this experiment would give insight into the size of the SOL width. Shots 100589-100594 were dedicated to shot development, in order to find the ideal shot style that would be useful for analyzing the currents through jumper A. We ended up going with a shot style that resulted in strong mode activity (see shot 100594), so that the jumpers could have a strong clear pickup, who's strength could be normalized to the mode amplitude. This shot style was also found to be very consistent. Shots 100585-100588 were crates only shots that were taken to try and diagnose why the west rack wasn't initializing or recording data. Fortunately, it just turned out that the west rack accidentally got disabled, and this problem was an easy fix. Shots 100595-100639 were dedicated to the fine radial scan, taken in 1/2 cm increments. Measurements were taken for shell positions of 0., 0.5, 0.72, 1.0, ~ 3.5, and ~ 4.0 cm (not in that order) from the shells' fully inserted positions. The current appeared to significantly drop somewhere between 0.5 -> 1.0 cm, however, this statement is based on pre-normalized jumper pickup, and may slightly off. Starting with shot 100611, the same bank settings started giving slightly different shot profiles, with the shot lasting longer, and the strong mode activity occurring ~ 1 ms later than the original shot style. Several shots were taken to try and get back to the previous shot style, but this rarely happened. Fortunately, the mode activity still strongly resembled that seen in the previous shot style, it was just occurring later in the shot, which meant that this shot style could still be used to compare jumper pickup data with the previous style. As a result, this new shot style was used for the rest of the day.

Wednesday March 6 2019 9:37 am Brooks/Levesque 100654-100668 Cleanup shots after up-to-air

First run day following an up-to-air period. During the vent, the tokamak was at air pressure or ~1-100 Torr nitrogen for about 1 week, and was followed by a bake for 11 days. The following hardware changes were made since the previous run:

• Two broken bias probes were removed from Section 9. One was at the bottom of the chamber, while the other was previously stuck penetrating far into the plasma volume at the midplane.
• Rayleigh calibration was done for the Thomson scattering system. Calibrated density measurements for points at R₀={92cm,96cm,100cm} will be available for the next Thomson operation. A new capacitance manometer was temporarily installed on a gate valve for improved pressure measurements during the calibration -- its specifications state 0.5% absolute accuracy over most of the pressure range used for calibration.
• A replacement RGA was installed in place of the old RGA. The old RGA would no longer turn on after a long period of finicky operation.
• Transimpedance gains and compensation capacitors started being changed for some of the poloidal EUV system chords. Data from some chords in this system will be questionable until the changes are finished in the coming week or two.
• USB spectrometer triggering and timing was changed due to ongoing erratic trigger behavior. The spectrometer is now triggered via a single J221 pulse, fed into a pulse-train generator, fed into an Eagle Harbor fiber optic transmitter/receiver, fed into the spectrometer trigger input. Data are being taken now for the correct number of pulses, but the integration periods are questionable until further determination/improvement.
• Changes in midplane ports:
• All midplane probes have been removed for repair or replacement. Probe additions will be noted as they are installed.
• Section 4: a new gate valve was added for upcoming probe usage. The GDC electrode was removed, and remains on the bench.
• Section 7: a gate valve was moved here from its original Section 8.
• Section 8: the small pump stand was moved here from its original Section 7.

The shots yesterday were intended to start cleaning HBT with plasma shots after the up-to-air. Jeff describes shot number 100655 as "long lived erratic shot with low x-rays and possible run away electrons". The HS Camera was operating and shows interesting filamentary behavior. Cleanup shots will likely continue for at least the next few days.

Friday March 8 2019 9:48 am Stewart 100743-100785 Clean-up

These shots were devoted to the continuing clean-up after the up to air. The first plasma shot 100745 of the day lasted to about the 5 ms mark, which was surprising for only the second day of clean-up. By the end of the run day, plasmas were lasting beyond the 7 ms mark (see shot 100774). These shots were close in performance to the pre-up to air shots.

The last couple of shots were meant to test the newly aligned Thomson scattering setup however, the laser had a laser head interlock error displayed and the problem could not be fixed this run day.

Friday March 8 2019 11:45 am Brooks 100669-100742 Cleanup shots after up-to-air

Wednesdays (3/6/19) run day consisted entirely of cleanup shots. Plasma duration increased by throughout the day from 1-2 ms to maybe 5ms.

All shells in except 7T and 10T&B. HS camera was not on, and the 4-fan EUV array was not turned on until shot number 100696.

Tuesday March 12 2019 10:10 am Brooks 100798-100861 Double Probe (DP) initial testing

Yesterday's (3/11/19) run was intended to test the new DP's (double probe) ability to survive in HBT's plasma as well as drive current. The second objective was to validate the Shapal ceramic used by the probe to isolate its electrodes.

Preliminary results show that the probe was able to drive 40 amps (at positive voltages) in the plasma and 5 to 6 amps for ion saturation currents at 103cm. Unmagnified, visual inspection of shows the probe to be unchanged throughout the run day. Spectrometer data is still being analyzed, and the lab's telescope will be used to get a better view of the probe. Star shot for the day is 100845.

Notes: The probe was opened up the plasma in the morning and promptly dirtied the plasma. Conditions never fully recovered throughout the day. All shells were in except S7T and S10T&B. When the probe was inserted into section 2 the chamber, a minor alignment problem was detected. This required that the S2T&B shells to be retracted for the remainder of the run day.

Thursday March 14 2019 10:12 am Stewart 100862-100885 Thomson w/ Rayleigh

These shots were used to test the Thomson scattering diagnostics on the plasma after the Rayleigh calibration had been performed. Shots 100869-100873 had the same shot style, with the Thomson triggering at 3 ms. Shots 100874,100877-100883,100885 also kept the same shot style with a variation of ~1cm to 1.5cm in the major radius at the time of the new Thomson trigger (5ms). It was found that in cases with the plasma more inboard, as well as earlier on in the shot, that the signal to the chord at R=100 cm saw poor signal compared with the core channels at R=92cm and R=96cm. For shot 100885 (with the new Rayleigh calibration), the core channel at 92cm (poly #2), which was kept at the same voltage settings as when the system arrived (no relative calibration required), saw a T_e of 136 eV and an n_e of 3.7x10^19 m^-3. With a standard parabolic profile given in Wesson, the fit to this peak has a mean density that is 1.06 times the Greenwald density limit (n_G=2.3x10^19). The toroidal beta (beta_T) is around 1% using the fit profile, while the calculated poloidal beta (beta_p) was exceptionally high (around 230%). Using the simple fit model and the equations given for Bootstrap current in Wesson, the fit gives a total bootstrap current on the order of 1kA. In order to verify these numbers and further constrain the profile fit, the other channels must have a relative calibration to account for the low voltage on the polychromators used for the Rayleigh calibration.

Sunday March 17 2019 5:18 pm Brooks/Levesque 100909-100941 Double probe (DP) testing day 2

On 3/15/19, testing continued with the new double probe at section 2.

The first set of tests pushed strong current (around 45 amps) through the probe and into the plasma, and the HS camera was used to see if the probe deflected at all. There was no sign of deflection.

The second set of tests rotated the probe by 90 and 180 degrees. At 0 deg., about 40 to 45 amps were measured. At 90 degrees, 3-8 amps were measured. At 180 deg., about 40 amps were measured. This is as expected. I attempted to rotate the probe by 95 and 100 degrees so better align the probe parallel with the field lines (to further minimize current). There was no noticeable drop in current from the 90 degrees case.

The new probe's instrumentation box was tested, but signs are arcing were present. The box was immediately removed and was rebuilt that evening. It will need to be tested again later.

Finally, the probe was placed in double probe configuration. There was some odd pickup on the probe's sensors as well as many of the sensors in the south rack. Debugging in ongoing.

All shells in except S7T, S10T&B, and S2T&B.

Thursday March 28 2019 9:17 am Brooks/Levesque 100988-101037 Directional probe (DP) testing day 4

Summary for 3/22/19.

This was the fourth day of testing the new directional probe (DP) installed at section 2. The probe primarily operated in double probe configuration for the day. The following notable observations/tests were made: 1) The ion saturation current appears in one direction than the other. 2) "Arcing" occurs in double probe config. The HS camera shows that the "arcing" appears to be going through the plasma and not around the probe. The "arcing" also appears to have a toroidal preference. 3) Raising the edge q to 4 reveals a mode=4 structure in the sourced 14 kHz current measured by the PA mirnov sensors. This suggests that the sourced current remains spatially coherent at the edge and matches the edge q value. 4) Jeff has an interesting shot style where the edge q and minor radius "bounce". If the probe is discharging, this causes it to arc, and it the arc appears to sustain itself for at least 1 ms after the "bounce".

Jeff tried various configurations of powering and signal connections for the poloidal EUV system to troubleshoot noise problems during disruptions, but the dominant source of noise was not found. Further noise hunting in for the EUV poloidal system will continue in further runs.

All shells were in but S7T, S10T&B, and S2T&B.

Thursday April 4 2019 9:39 am Brooks 101046-101133 Directional probe (DP) testing experiment 5

This experiment was conducted on 4/2/19 to 4/3/19. The goal of these experiments were to continue testing the new Directional Probe (DP). On Friday (3/29), the old-style bias probe (BP) was opened to the vacuum and then inserted into the plasma on the morning of 4/2. The addition of this second probe opened up the range of experiments with the directional probe.

Specifically, the following experiments were conducted:

1. Both DPS2 and BPS7 were placed in at 103cm and driven with identical IV sweeps. DPS2 shows a lower floating potential suggesting that the probes might not be on the same flux surface even if they have the same major radius. DPS2 had larger electron currents and possibly larger (more negative) ion currents.

2. 14kHz currents were injected from BPS7 which was biased in reference to the local chamber ground. The goal is to see if the current map (using Mirnov and SOL sensors) looks different when sources from section 7 as compared to section 2. Analysis forthcoming.

3. Toroidal ion Mach number measurements as a function of mode rotation frequency. BPS7 drove several a range of flat current and voltage signals to induce different mode rotation frequencies. DPS2 was placed in a single probe setup. Both electrodes sourced ion sat. currents, and the current through each was measured. Preliminary results show a very noisy and inconsistent toroidal mach number measurements. The ion saturation current from the left (counter IP facing) electrode kept reading a ion sat. current very close to zero (and occasionally positive) and this made the mach number measurements difficult.

4. Poloidal ion Mach number measurements as a function of mode rotation frequency. The directional probe at S2 was rotated 90 degrees to measure the poloidal ion mach number. These measurements were much cleaner than the toroidal and shows a strong correlation to the influence by BPS7. Analysis forthcoming.

5. BPS7 was placed at 104.5 cm and left floating. The DPS2 electrodes were shorted together and also left floating. The radial location of DPS2 was scanned from 103.0 cm to 106.5 cm. The goal was to determine at what radial location of DPS2 showed the the same floating potential as BPS7 at 104.5. Preliminary results suggest that DPS2 placed at 106.0 had the best match. Do two different probe geometries experience the same floating potential while on the same flux surface. If so, does this mean that there is a 1.5 cm shift in major radius of the flux surface 180 tor. deg. around the machine?

6. BPS2 and DPS7 were placed in a 180 deg. toroidally separated double probe configuration and both at 103cm. The probes sourced 14 kHz currents. Both DPS7 electrodes were shorted together for one test, then only the right used, and then only the left used. Ion sat. current appears to be 2.5 amps, symmetric (unlike when the DP was in double probe config), and periodically arcs. The current path will attempt to be mapped, but they low current amplitude may make it difficult. Analysis forthcoming.

All shells in except S7T&B, S10T&B, and S2T&B. Probe current and voltage channels and gains are located in the shot comments for each shot.

Thursday April 11 2019 8:47 am Brooks 101134-101172 Shot development

The run day yesterday (4/10/19) was intended to start with shot development to reproduce shot number 99605 (characterized by 2 to 4 ms of a consistent mode instability), followed by mach probe measurements with the directional probe, and then end with some preliminary control work. Despite being able to very closely reproduce all of the plasma parameters (MR, edge q, Ip, etc), the shots repeatedly disrupted early during the mode's growth, never more than 1 ms from the start of growth. Reproducibility of the plasma has been difficult after the recent up-to-airs.

Starting at shot 101170, the plasma stopped breaking down correctly. After some investigation and inconsistent measurements, the e-gun filament resistance was measured to be about 50 kOhms (with no short to ground), suggesting that the egun filament is broken and needs replacing.

All shells were in except S7T&B (section 7, top and bottom), S10T&B, and S2T&B.

Tuesday April 23 2019 8:49 am Brooks 101228-101272 DP testing day 6

Thursday's and Friday's (4/18/19 to 4/19/19) run had several objectives. First, continue to develop a shot style. Second, conduct Mach probe measurements with the DP setup as a Mach probe at 104.5 cm and the BP driving mode rotation (variable but steady voltages) at 103.0cm for both poloidal and toroidal orientations of the DP. Third, start using both probes for feedback (FB) mode suppression.

The shot style from the previous run day quickly returned. Shots 101234 to 101237 are indicative of the intended shot style for the rest of the day. Shots like 101238 also cropped up occasionally but appeared to provide roughly the same mode activity despite the change in MR and qStar.

Poloidal and torodial mach probe measurements were successfully carried out and are documented here: http://sites.apam.columbia.edu/HBT-EP/wiki/lib/exe/fetch.php/2019_04_20_brooks_machnumbermeasurements.pdf. In summary, the toroidal ion Mach number showed a strong dependence on the BP voltage & current and the n=1 mode rotation frequency. In contrast, the poloidal ion Mach number lowed no discernible dependence. Were I to repeat these tests, I would prefer that they be conducted later in shot time as more mode activity is present after 3ms. In addition, I would want to rotate the probes 180deg from its toroidal and poloidal orientations as a sanity check.

Finally, I set up both DPS2 (as a single probe with both electrodes shorted) an BPS7 at 180 deg. apposed probes also drove 180deg apart signals. Both probes were ground referenced (and not to each other) as more current can be driven this way. They were also programmed so that they only drove positive current. This allowed the ion sat. current to be avoided and therefore allow more current to be sourced. It also caused the mode rotation to slow down. Only a few shots were conducted in this configuration but showed some promise. A phase offset of 0 deg. to both probes resulted in very early (at roughly 4 ms) and abrupt disruptions. A constant (positve) current with no phase-matching oscillations resulted in disruptions around 6 ms. The only one 180 deg. phase attempt resulted in a disruption near 7 ms, suggested a increased plasma lifetime. Unfortunately, this was not statistically significant, and the next run day will continue this work.

All shells were in except S7T&B, S10T&B, and S2T&B.

Sunday April 28 2019 3:00 pm Brooks 101273-101369 Mode suppression feedback DP and BP testing

This run report covers the two run days of 4/24/19 and 4/25/19. The main goal of these days were to continue optimizing the feedback system, and being my first set of feedback tests.

DPS2 (electrodes shorted together) and BPS7 were grounded to the chamber and driven with a phase locked signal (180 deg. apart to account for their toroidal separation). A common phase offset parameter was scanned at 0, 0.5, 1.0, and 1.5 pi, and a baseline case with no control was run. The 0pi case showed a marginally shorter time to disruption, and the 1pi case showed a marginally longer time to disruption. However, the error bars for the 5 cases overlapped substantially so no strong claim can be made about the results. A report summarizing my findings will be made available on the wiki. An improved feedback system should improve performance, and I also wish to find a more consistent shot style.

I spent the second day optimizing the feedback system. I was using the Crown CE4000 amplifier on the BP, but it suffers from an approximate 20 microsecond delay. I have now removed this amp, disconnected the parallel AETechron amps, and put one amp on each probe. Crates only testing shows the system is working, and feedback testing will resume later this coming week.

All shells were in except S7T&B, S10T&B, and S2T&B.

Sunday April 28 2019 11:29 pm Levesque/Brooks 101370-101385 CPCI and control coil testing

Tried to set up and troubleshoot a new CPCI for recording the control coil currents. Control coil current readings in the North rack A14s have had problems for several years, so we've finally routed measurements to a CPCI instead in preparation for upcoming control experiments. make_tree was run during the previous run day to regenerate the tree and add CPCI_40 to the North rack. mdsplus logs suggest that the CPCI triggers successfully, but so far it has not been storing data to the tree. The log shows other CPCI's connecting to the mdsplus server during the shot storage phase, but this CPCI does not connect despite a request being issued. Troubleshooting will continue at a later time as available.

The working north rack CPCI (usually for half of the high resolution sensors) was temporarily used to test the control coil amplifiers and new signal routing. 34 of the 40 coils are working, as expected from the last working configuration and present hardware set up. The inoperative coils are: FB01_C3 and FB06_C4 (bad coils), FB02_C1 and FB02_C2 (broken amplifier), and FB08_C3 and FB08_C4 (disconnected last year due to presence of a ground loop when connected -- this has not been investigated further).

Three good plasma shots were taken before we encountered CAMAC problems. All shells were fully inserted, and all probes retracted behind the shells. Plasma shots were 101380-101382. This is the start of shot development for the EUV control experiments.

For near-term feedback control, plans are as follows: 1) develop shots with all high resolution sensors connected, 2) when a target shot is ready, use the working north CPCI to record control coil currents instead of the high-resolution sensors during control experiments. PA1 and TA01 through TA05 won't be recorded during these experiments.

Monday April 29 2019 10:55 pm Levesque/Brooks 101387-101398 Preparing for EUV-based FB control

Continued preparation work for running the EUV-based mode control feedback system. Developed target shot style. 101393 and 101394 are star shots, along with several shots around them. The 4-5ms window in shot 101393 will serve as the baseline 3/1 mode for now for setting up a EUV basis vectors. BD analysis of EUV data fluctuations during this time window gives a mode pair with amplitude and phase that are well-matched to analysis using from the magnetic sensors.

GPU feedback control files have started being edited to accommodate the change to EUV control. The algorithm will use only EUV input data, with no magnetic sensors. EUV data are successfully being read by the GPU, though the channel mapping still needs to be confirmed. During upcoming feedback shots, a 3/1 mode shape will be imposed by the magnetic control coils, though this shape is not drawn from the EUV data for now.

Thursday May 2 2019 10:33 am Levesque/Brooks 101399-101436 EUV feedback control system commisioning

Report for 2 days of operation. Shot development and feedback system operation/troubleshooting for using the EUV control system. Discharge evolution and mode activity returned to target qualities by the end of the run.

Swapped control coil power amplifiers for Section 2B (previously broken amp) and Section 8T (problematic coils) in order to get another CC working. Coil FB02_C1 now works; now there are 35 working control coils. Coil FB02_C2 still has a problem somewhere in it's circuit from the GPU to the coil, though the coil resistance/inductance/isolation measurements are reasonable, so the input and output of this amplifier channel are disconnected for safety of the amplifier.

Completed an initial phase scan with a single shot per angle, plus a feedback off case. These are shots [101427,101428,101433,101431,101435,101430,101434,101432,101436] having respective feedback phase angles of: [off, 0, 45, 90, 135, 180, 225, 270, 315] degrees. Feedback was active during 3.5-5.5ms. Fourier analysis of the 4-5ms window in each discharge shows mode suppression and excitation dependent on feedback phase angle. The next run will continue the compass scan to take more points, including more near suppressive phase angles.

Thursday May 2 2019 9:49 pm Levesque/Brooks 101437-101467 EUV feedback control campaign

Continued EUV feedback campaign. Discharge evolutions and modes were not nearly as reproducible as yesterday. Although equilibrium parameters were in a fairly narrow range, the mode activity appears to be very sensitive to small differences in discharge evolution for today's shot style. "Good" shots fell into 2 different groups, with both being slightly different than yesterday. Feedback results were not very conclusive, though there may have been some evidence of mode suppression or amplification like yesterday. Only tried 2 feedback phase angles, where I expected to see amplification or suppression based on yesterday's results, in addition to feedback-off cases. Most of the shots had higher feedback gain than yesterday since the feedback effect wasn't as apparent as yesterday.

Very high gain for the amplifying feedback phase angle (shot 101467) did cause large mode growth and a disruption. Very high gain at the suppressive phase angle (shot 101466) did not yield a disruption, but may have been too strong to get optimal feedback performance.

4 shots were taken with the feedback coils modulated on and off in 250us periods, though mode was being tracked for the whole shot. These shots (101462-5) had inconclusive results.

One cryo pump was active for the run yesterday and today. Base pressure was 1 to 3e-7 torr yesterday, and about 4.0e-7 torr today, which is much higher than typical operating conditions. However, the RGA shows that it's mostly deuterium. The unused cryo was regen'ed but not opened to the vessel in the interest of maintaining similar vacuum and impurity conditions across the run.

Monday May 6 2019 12:45 pm Brooks 101492-101512 FB testing with DPS2 and BPS7

Friday's (5/3/19) run day was primarily composed of crates only and plasma feedback tests for debugging purposes. Of interest, Jeff's FB shotstyle from the previous week was recovered with several shells out and two probes in. This is important because his shot style is characterized by at least 3 ms of consistent mode activity before disruption. The major difference is that the shot style with the probe's in has a quiescent period (mode free) from 2 to 3 ms. Shot 101510 is a good example with the probes driving a flat probe current profile, and 101504 is a good example with the probes floating.

All shells in except S7T&B, and S2B was halfway. S10T&B were both in. S2T was in.

Friday May 10 2019 9:23 am Stewart 101550-101632 Shot Development and TP Conditioning

Combined run report for 05/08 and 05/09. These two days were used as preparation for the final, full triple probe and bias probe scan of the L-mode and H-mode temperature and density profiles. The first run day was devoted to full shot development with the optimum shot (101584) having a flat top I_p profile and an edge q above 3 for the time of interest (I wanted to avoid the H-mode MHD behavior when q_a drops below 3). The shot style is MHD quiescent for the time of interest (between 2-4ms), which should be ideal for looking at floating potential fluctuation changes between L and H-mode using the triple probe. The last shots of the day 101582-101599 were also used to clean the triple probe tips by charging them (+150V) and slowly moving them radially inward until arcing on the probe tips was no longer present.

Since the triple probe was giving a result of T_e ~ 20eV at every radial position tried on 05/08, the second run day was devoted to a full conditioning of the triple probe and debugging of the issue. It was determined by 101607 that swapping tips A and C on the TP gave results similar to those obtained with the prior triple probe (TPS5), with the temperature increasing radially from about 10eV at the edge to about 45eV at R=103cm. This configuration will therefore be used for the full scan. The rest of the day was used to continue conditioning the probe by moving it inward radially and switching which tips were biased negatively (collecting ions).

Note: shots 101617-101619 had prominent reverse sawteeth on the SXR fan array.

The last few shots 101628-101632 were a test of the shaping bank with the new 33 Ohm power resistor in parallel with the bias probe to prevent an open-loop scenario when the plasma detaches from the probe. The test shots showed at most a 0.5 degrees Celsius change on resistor for a bank bias above +300V. This however did not eliminate the noise on the diagnostics inherent to large positive biasing shots. With the new bias probe diagnostic box wire runs to the CPCI, the noise level not he bias probe voltage measurement was intolerable (clipping on the CPCI). These wires needed to be replaced in order to lower the noise on the bias probe voltage measurement.

Thursday May 16 2019 9:44 am Stewart 101685-101770 H-mode with Thomson and TP

Combined run report for 05/14 and 05/15. The first day was devoted to a coarse radial scan with the triple probe of density and temperature. The other main goal was to begin using the shaping bank to power the bias probe and develop a sufficient shot style with a flat major radius throughout (using the vertical field to correct for changes in pressure). A good target shotstyle was found in shot 101712, which was centered even during biasing.

The next day was a full radial scan using the triple probe in both L and H-mode with Thomson scattering in both timeframes. The bank was biased to +350V (SHST) for each of these shots with the probe turning on at 3ms (using a delay generator). The Thomson data was collected at 2.5ms (L-mode) and 3.8ms (H-mode). With a flat major radius and a flat top plasma current, abrupt changes in temperature and density should be due to the biasing. As was seen in earlier scans, the temperature at the edge increases abruptly in these shots and the density increases on a longer ~1ms timescale. The triple probe density measurements (time traces) also match with some of the EUV and fan array X-ray chords. Fluctuating quantities (V_f and I_sat) were also quenched as usual in the region of the electric field well or hill in this case. The major shots of interest were 101746, 101735, 101768, 101730, 101765, 101753, and 101759. Most notably, during H-mode the density at the core measured by Thomson scattering was between 4.3-4.8E19 m^-3 for the H-mode measurements and 3.1-3.3E19 m^-3 for the L-mode measurements. This confirms what the triple probes and SXR data indicated all along and serves as the final confirmation of a major change in confinement due to the biasing (i.e. biasing H-mode).

Unfortunately, the run day was not without its issues. For several shots during the run day, the magnetic data (FB,TA, and PA) data was not recorded along with the SOL data and the Bias probe measurements. This appeared to be an issue with almost every other shot taken, leading to a radial TP scan that took twice as long as necessary. Hopefully this issue can be resolved in the next run day.

Friday May 17 2019 8:53 am Brooks 101527-101549 FB debugging and some testing

This is for the run day on 5/7/19.

The run day today was intended to be a testing of the BPS7 and DPS2 (both in single probe setup) FB testing, but the day became more of a debugging day instead.

Two notable occurrences:

1. I developed a new shotstyle that might be useful for FB testing in the future. Shotnos 101539 and 101540. 2. Jeff and I believe that we've fixed the timing issue associated with the North Rack. Previously, the A14's in the north rack has a skewed time basis. At the start of the day, they were unplugged from the rack but their trigger and clock signals were still connected (in series) with each other and the CPCI. We noticed today that the CPCI was encountering the same skew in time. We removed the clock and triggering lines from the A14s and went straight to the CPCI. This appears to have fixed the timing issue. I speculate that having so many A14s and the CPCI on the same clock and trigger cable may have collectively had too large of an impedance and dragged down the voltage so that the signals became inconsistent.

All shells in except S7T&B (out) and S2B (half in).

Friday May 17 2019 9:04 am Brooks 101634-101684 FB testing with DPS2 and BPS7. Also 14kHz FF tests.

This run report is for 5/13/19.

The goal of this day's run was primarily feedback (FB) testing of mode suppression with the BPS7 and DPS2, both in single probe configurations. The target shot style was 101650, and the current delivered by both probes were matched and 180 deg out of phase. This made it such that no not current was delivered from their oscillating currents. A feedback scan of the offset phase was performed alongside several baseline (control-off) cases, and the results suggest that the system may be having some impact. The results are summarized here.

In addition, more feedforward cases were driven with the probes driving current at 14 kHz at 103cm. This was done with both probes simultaneously, with the DPS2 only, and twice more with each electrode of the DPS2. Should have done a BPS7 only case. In addition, having no offset (voltage and current) appears to help the coherence analysis perform better. Some of the results are summarized here.

All shells were in except S7B and S2B. This allows my FB4 Mirnov array to have better measurements, and because both probes sag a little, only the bottom shells were left out.

Friday May 17 2019 5:37 pm Brooks 101801-101890 Double probe FB for mode suppression with DPS2 and BPS7

The goal of today's run was to use the BPS7 and DPS2 in a double probe (toroidally separated by 180 deg) setup with the GPU FB system to suppress mode activity. Unfortunately, preliminary analysis suggests that the probes have no impact (positive or negative) on the mode or even disruption. Detailed analysis forthcoming.

A few other points of interest. The recently rebuilt triple probe operated all day today, and the results seem reasonable. Second, the base pressure continues to hover around 2e-7 to 3e-7 Torr range. Third, the system failed to trigger twice today, and the auto-dump system kicked in. The second time, I managed to press the "fire" button before the auto-dump, but it didn't fix the problem. Perhaps this suggests that the problem is with take_shot_hbt.py or oscar?

All shells were in today except S2B and S7B which were fully back.

Thursday May 23 2019 3:07 pm Brooks 101893-101997 DPS2 feedforward and feedback testing

This run covers 5/21/19 and 5/22/19. The goal of this run day was to test feedforward and feedback operation of DPS2 in a double probe configuration (with itself). The feedforward experiment drove 14 kHz voltage and current with DPS2 at 103, 104, 105, 106 and 107 cm. The feedback portion placed the probe at 104 cm (close to the q=2 rational surface for that particular shot style), and a phase scan was performed. Data analysis for all of the above is still ongoing.

All shells were in except S2B and S7B. Base pressure was again high (4e-7 Torr) at the start of both run days. Also, the shot style was not particularly repeatable today.

Friday May 24 2019 11:20 am Saperstein 101998-102042 tearing -> kinking shot style

The motivation for this run day was to try and see if the SOL current sensors could see any differences between kink and tearing modes. For this purpose, we were looking for shot styles that exhibited the following profiles: edge q drops from above 3 to below 3 (a shot style where a 3/1 tearing mode turns into a 3/1 kink), edge q raises from below 3 to above 3 (a shot style where a 3/1 kink mode turns into a 3/1 tearing mode), edge q hovers slightly above 3 (a shot style with a predominant 3/1 tearing mode, with little interference from a 4/1 kink), and an edge q slightly below 3 (a shot style with a predominant 3/1 kink mode). And ideally, we wanted all of these profiles to have a roughly outboard and constant major radius, so that the SOL current sensors see strong, consistent mode activity. Shot styles where a tearing mode turned into a kink mode, and vice versa, were useful because they allow for a very straightforward and accurate comparison of the SOL92s normalized (norm to FB mode amplitudes) mode amplitudes between a kink and a tearing mode. However, the mode amplitude profile for some of these types of shots can make the normalization method used qualitatively difficult to interpret. Therefore, the shots containing only one predominant mode can make normalization easier to interpret, at the cost of making it more difficult to compare results between shots. The run day schedule can be divided into 4 parts. The first set of shots (101998-102008) were dedicated to cleaning up the plasma. The second set of shots (102009-102020) were dedicated to finding and repeating shots that had the tearing->kinking shot style. Among this shot set, shots 102010-102012 were especially promising. The third set of shots (102021-102027) were devoted to trying to develop a shot style with a negative current ramp, in order to try and induce a shot style with the 3/1 kinking->3/1 tearing shot style. This was ultimately unsuccessful, since we could not achieve edge q profile we wanted, and the major radius changed wildly over the course of each shot. The final set of shots (102028-102042) were dedicated to getting a good tearing mode only shot style. This was difficult considering we had to keep the edge q high enough to get 3/1 tearing modes, but low enough the 4/1 kinks would dominate over them. We also weren92t able to get good shots of this type without 4/1 kinks dominating in the first 2-3 ms of the shot. However, this may be good, since these shot style may also be good for acting as 4/1 kink->3/1 tearing shot style. Of this shot set, shots 102033 and 102040 looked the most promising. Unfortunately, I didn92t have enough time to develop a 3/1 kink only shot

Wednesday May 29 2019 8:53 am Brooks/Saperstein 102054-102090 DPS2/BPS7 feedforward testing and vacuum field testing

Yesterday's run (5/28/19) was an eventful day. My feedforward testing of BPS7 and DPS2 in a toroidally-separated double probe configuration ran into complications. 102078 was an example where both probes were at 103cm, in their toroidally separated double probe configuration, and driving 14 kHz voltage and current oscillations.

First issue: CAMAC issues caused the OHS and OHE bank to trigger incorrectly. An example of this is 102064. After several reboots of the highway (and eventually rebooting Spitzer), the system began working correctly.

Second: The plasma stopped breaking down on shot 102080, and the egun was observed to be off despite being powered. Upon investigation, Jim and I believe that the filament has broken, and it was suggested (not confirmed) at lab meeting that the e-gun may have been placed too close to the plasma. Jim plans to replace the filament today.

Third. With the plasma not functioning, Alex took over the remaining run day to perform vacuum shots on specific banks for his Rogowski studies. On shot number 102089, the TF bank was having trouble charging up to its maximum voltage. In hindsight, the TF charging resistor electrode had worn down to a very small surface area and likely caused the issue. While troubleshooting, we repeated the charging on the next shot, 102090, and this time a bright flash (likely arcing) was obliquely observed on the security feed. Upon closer inspection, a small but observable amount of smoke was drifting up from the TF charging resistor, and its overfill line was swinging. We dumped the banks, hurried down with a fire extinguisher, shorted the TF banks, and inspected the damage. To be clear, there was no sign of fire or major damage. However, the overfill tubing for the TF charging resistor has melted and spilled less than a liter of conducting fluid on the floor. The fluid had puddled but not splashed onto the nearby capacitor banks. In addition, the TF charging resistor electrode (which had slowly worn down to a VERY thin disc) had disappeared. I SPECULATE that the sudden disappearance of the electrode may have caused the arcing, and excess heat buildup from the prolonged charging and sudden arcing may have caused the tubing to melt and spill some fluid. Or perhaps the arcing occurred when the fluid ran down the length of resistor's exterior and briefly shorted across the resistor or to ground. Alex and I took pictures of the scene, reported the event to Jeff, shut down HBT, mopped up, and went home for the night. Of note, the security camera has been showing a "hard drive full" error for the past week, and I'm uncertain if it recorded the incident.

All shells were in except S2B and S7B.

Monday June 3 2019 8:09 am Brooks 102091-102125 New egun testing and new DPS9 testing

The goal of Friday's (5/31/19) run day was to test the new egun (e-gun) and also test the new DPS9.

Jim installed the new e-gun earlier in the week and baked on it a few days. On Friday, we tested a few radial positions and egun heating currents until breakdowns were consistent. Presently it is inserted to the "usual radial position" (according to Jim) and is set to about 13 amps of heating current. Jim notes that the old egun did not break from plasma exposure (or whatever) but from having a loose connection internally in the vessel.

The DPS9 gate valve was opened up early in the run day, and many plasma shots were taken to clean up the plasma. At 101120, the probe was inserted into the plasma at 106 cm, but the SCSI cable in the south rack had accidentally been disconnected. At the final shot of the day, 101125, the SCSI cable was reconnected, and reasonable floating voltages were observed on the probe. Of note, this probe appears to sag worse than the other probes, and much care needs to be given to it as it is pushed radially in and out, even when the bottom shell is completely pulled out. Also of note, DPS9 has a small crack near where the copper rods enter the probe. Jim has patched it, but we need to keep an eye on it. The probe has been positioned so that this patched crack is visible from the nearby section 10 window.

The TF charging resistor incident resulted in a hairline crack forming in the pyrex glass fluid vessel. Jim has taped it with a high-tensile strength and high temperature tape for the time being, but we are looking to replace it. We have not yet found a supplier.

Jeff setup the HS camera to record the probe, but unfortunately, the camera seemed to receiving significantly less light than usual. Jeff changed f-stop values, sampling rates, exposure times, etc., but could not get the default HS camera settings to receive a bright enough image. In the end, Jeff reduced the sampling rate and increased the exposure rate to the point where the probe could be observed but only with about 2 to 3 frames per shot. The reason for this difficulty is unclear.

All shells in except S2B, S7B, S10B, and S9T&B.

Tuesday June 4 2019 10:45 am Brooks 102127-102165 DPS9 cleaning and DPS2/BPS7 FF 14kHz testing

Yesterday's run (6/3/19) had several goals. First, the DPS9 probe was left floating and moved from 106 to 105 to 104 cm in order for the plasma to clean the probes. Of note, the plasma lifetime significantly drops and the soft x-ray measurements show a much colder plasma. After some shots, the plasma starts getting better, but when the probe is pushed in again, the plasma quality gets poor again. The HS camera, which is working better with a narrower depth of focus, shows much light emanating from the new probe.

I also placed DPS2 (electrodes shorted) and BPS7 in a toroidally separated double probe configuration. They were connected to AETechron #1 and Stangense transformer #1 (in config #2) and driven at 14kHz. The probes were radially scanned from 103 cm to 107cm in 1cm steps with two shots at each location. Of interest, the mode activity was very quiet with the probes at 103cm and grew much stronger when the probes were further out. These tests occurred between 102148 and 102158.

Initially, all shells except S2B, S7B, S9B, and S10B were in. Near the end of the day, all shells except S2B and S7B were in. S9B was half in.

The bank triggering messed up twice yesterday. 102136 and 102165. Each time, the entire data highway, spitzer, and oscar were reset. This appears to fix the problem in the short term.

Tuesday June 4 2019 1:42 pm Brooks 102166-102191 BPS9 cleaning

Today (6/4/19) was spent cleaning the DPS9 probe, which was left floating at 104cm all day. The cleaning was very slow. The soft x-ray signal only provided reasonable results after many shots, and even then, it was intermittent. I operated at a high OHS attempting to achieve a hotter plasma in order to clean the probe faster.

Shot 102190 failed to trigger some of the banks. Same as the previous run day. I restarted spitzer, oscar, and data highway which seems to have fixed the problem (again) in the short term.

During shot 102191, the plasma failed to break down, and upon investigation, it appears that the egun (e-gun) has again broken (last week, the filament became loose inside the vessel). The reason for broken filament is presently unclear. Jim plans on removing the broken filament, investigating the cause, and replacing it with a new one today. Jim says he can insert the egun into the chamber tomorrow morning for a full run day.

All shells in except S2B and S7B. S9B was half in.

Friday June 7 2019 4:31 pm Brooks 102197-102248 new egun testing and DPS9 cleaning

Wednesday's (6/5/19) run day was mostly spent calibrating the new egun (e-gun) as well as cleaning the new DPS9 with plasma.

After Jim installed the new egun, we spent some number of shots calibrating its radial position and its current so that breakdowns were still consistent and also so that the egun would likely have a longer life. In the end, the egun was pulled out 2 cm from it's previous radial location, and the egun was also running about %60 to %70 of its max (and previous) current.

DPS9 was tested/cleaned in floating, grounded, and at 14 kHz frequency sweeps as a function of radius. Both electrodes were shorted together and referenced to ground (driving a radial current). The probe slowly cleaned over the day and experienced fewer "arcing"-like events. Arcing is characterized by large impedance (e.g. current) deviations from the standard I-V relationship. Voltage was also observed to rapidly "oscillate" during these events. Some of the extreme arcing events drove quasi "H" modes. The 14 kHz radial scan occurred from shots 102219 to 102222 with probe radial positions of 106cm, 105cm, 104cm and 103cm, respectively.

Finally, DPS2 was setup as a mach probe and measured toroidal, counter toroidal, poloidal, and counter poloidal mach numbers.

All shells in except S2B, S7B. S9B was half in.

Friday June 7 2019 4:43 pm Brooks 102249-102330 Amplifier testing and toroidal/poloidal mach measurements

Thursday (5/6/19) morning was spent setting up for both FB tests and Mach probe measurements. A third amplifier was neede(5/6/19) d, but I spent many hours tracking down an issue which turned out to be a bad TVS diode (Thanks, Ian).

Thursday afternoon, I setup DPS2 as a poloidal mach probe (by rotating it 90 degrees at the rotatable flange) and DPS9 as a toroidal mach probe. I fixed BPS2 at 104 cm, and the Mach probes were scanned together at the same radial position. I scanned both BPS2 voltage (-100V to +100V roughly) and radial position of both DPs. After more probe cleaning, I recorded most of my data from 106cm to 104cm. When DPs were pushed in further, DPS2 continued to arc, and very little data was recorded within the BPS7 surface. The good Mach probe data was between 102314 to 102328.

Please refer to "post shot remarks" for shell positions. Likely, all shells in except S2B and S7B. S9B was also likely half in.

Friday June 7 2019 5:21 pm Brooks 102331-102374 FB testing with DPS2

Today's run was intended to be FB testing with BPS7, DPS2, and DPS9 all set into a single probe (radial current) configurations. Each with their own amplifier and transformer. However after talking with Jeff, we realized that we could ground the transformer's centertap on the secondary side to the local chamber ground. The primary side was placed in a parallel configuration. The positive lead (secondary side) was connected to DPS2's right electrode, and the negative lead (secondary side) was connected to DPS2's left electrode. The resulting probe is one that can effectively bypass Isat (although each electrode is still Isat limited but on opposite phases) AND inject current in co-IP and counter-Ip directions at different phases. Mike suggested the possibility of adding a power supply to the centertap-to-ground connection so that any offset can be achieved for future work.

A phase scan was conducted using DPS2 setup this way. Shots 102353 to 102374. Unfortunately, no obvious change in mode activity was observed for any phase offset. When I return in late August, I plan to repeat this setup but with all four directional probes installed.

All shells were in except S2B and S7B.

The North Rack gave an error on shot 102337 when recording the data, and much of the data was lost for that shot.

Wednesday June 12 2019 11:30 am Saperstein/Levesque 102375-102426 Testing CC effect of SOLC signals

The purpose of this run day was to verify 2 things. First, how much magnetic pickup do the SOL current sensors see coming from the control coils (CCs). Shots 102375-102382 were dedicated to this question. Over these shots, we pulsed ~45-50 [A] of current through each of the coils (except sect 8 top, and section 2 bottom following shot 102378, and cc1-3), with pulses being 0.4 ms long and 0.1 ms between the end of one pulse and the start of another (the coils were not pulsed simultaneously). The model shot for these sets of shots would be 102379, where all coils (but the aforementioned) were pulsed with up to 50 [A] of current, and the SOL current tiles didn92t see any pickup even comparable to their background noise. It should also be noted that during initial tests of the CCs, while cc2-2 has its input removed, it was still amplifying a signal that could be seen by the jumper Rogowski92s, as well as SOL tile 4-3. The signal seen by these sensors has the appearance of raw magnetic pickup, which was odd since we were looking at the integrated data. My theory is that even though cc2-2 had no input attached, it was inductively picking up current within the amplifier from cc2-1, which would explain why the current amplified looks like raw magnetic pickup. The fact that this pickup appeared at the same time that cc2-1 was fired also supports this. We completely disconnected the output of the amplifier for cc2-1 and cc2-2, and this removed the pickup we were seeing in the jumpers and the one SOL current tile. The output is currently still disconnected, since we'll hopefully be continuing this campaign later this week.

Second, we wanted to see if pulsing a single CC at up to 50 [A] would have a local effect on the SOLC tiles. Shots 102383-102426 were dedicated to this. Most of these ended up being for cleaning/shot development, with the CCs turned on occasionally throughout the process. Following shot 102410, Jim switched one of the cryo pumps with one that had been regened, and this lowered the pressure from the e-7 range to the e-8 range, and plasma got a lot nicer, however we did have to basically restart the shot development process. We found our ideal shot style with shot 102416. It92s a long pulse (7->8 ms), with roughly constant major radius and mode growth rate throughout the shot. This shot had cc7-3 turned on with a positive pulse between 4->5 ms. Following this shot, we kept the same shot style, and toroidally varied which CCs were turned on (retaining the positive pulse style) between 4->5 ms, while preserving the poloidal position of the CC. We were able to get shots for a single CC in all sections except 1 and 8, where the CCs at that poloidal position were disconnected. After the shots consisting of only a single CC firing, I wanted to see what effects pulsing an entire section would have. I began with pulsing all coils in section 10 (shot 102423), and then moved on to sections 7 and 3 as well. The last shot of the day was one with the ideal shot style, except with all of the CCs left off.

For future runs of this campaign, I92d like to try negative CC pulses, pos->neg swing pulses, and a CC pulse style where ~3 toroidally adjacent (but same poloidal position) coils are pulsed at the same time.

Additional notes:

1. Some shots in the shot database aren't labeled with whether or not the a CC is on. Some shots have in their comments whether or not the previous shots had them on, but just in case, the physical logbook should have this information as well.

2. Right now the CC current measurements don't have a permanent CPCI that the data is sent to. Right now, the data is being sent to the CPCI in the North rack, to channels 1-28 and 33-42. Right now, the mappings can be found be checking Jeff's cc_currents_april_2019 jScope configuration.

Monday June 17 2019 12:35 pm Saperstein 102427-102472 continued testing CC effects on SOL

The purpose of this run day was to continue the previous run day's work in testing how the SOLC tiles respond to low frequency (compared to MHD), non-resonant pulses from the control coils (CCs). During the last run day, we tested only pulses with a positive polarity (basically disregarded isotropic effects), in either a single CC or a poloidal array of them (all CCs in the same section). On this run day, we continued the tests by introducing a polarity sweeping pulse (that swings from positive for the first half of the pulse, to negative during the latter half), which allows us to both get stronger responses from the SOLC tiles (pos->neg swing provides ~90 A change in current through the coils, as opposed to the ~45 A we were limited to with only a positive polarity) and tests isotropic effects. We also got the chance to test firing an array of toroidally adjacent CCs (with the same poloidal position, slightly above the midplane) to see if this could introduce an even stronger response in the SOLC tiles (as opposed to the poloidally adjacent coil tests we ran on the previous run day, which may have had some interference between coils, because of the similar length scales between the coil separation and MHD poloidal wavelength).

Shots 102427->441 were dedicated to cleaning up the plasma, and reproducing the same shot style we had for the previous run day (shot 102416). The model shot for this set of shots was 102441, which had all CCs off, with little MHD variation in the 4->5 ms window, and serves as a good reference shot to compare all the CC shots to. We did also occasionally pulse some of the CCs during this cleanup phase as well.

Shots 102442->462 were dedicated to single CC pulse shots, but now using a positive -> negative polarity swing in the coils. Similar to the previous run day, the pulses were fired between 4->5 ms, and the best shots were those that had little MHD variation during this period (MHD amplitude spikes can make it more difficult to differentiate between the SOLC's response to the CCs and its response to the MHD). Also similar to the previous run day, we rotated which coils were pulsed toroidally around the machine (retaining the same poloidal position), only skipping sections 1 and 8, because their CCs in these locations weren't working. Some good model shots to look at are 102442 (cc5-3 pulsed) and 102458 (cc2-3 pulsed).

Shots 102463->470 were dedicated to studying shots where 3 toroidally adjacent CCs were pulsed simultaneously. Because the section 1 and 8 top CCs weren't working, and we wanted to retain the same poloidal position we've been using throughout these runs, we were constrained to using arrays that didn't involve sections 1 and 8. This meant that we had 4 unique toroidal array settings to work with. In this run, we used 3 of them, the ones centered around sections 4, 5, and 6. Also, for some of these shots, the pulse start time was changed to 3 ms to try and put the pulse time in a region of less MHD variation. A good model shot to look at for this shot set is 102469, where sections 5, 6, and 7 had their sensor 3's fired.

Shots 102471->472 were just another set of reference shots with the CCs turned off.

Additional Notes:

1. When I say little MHD variation, I mean that the mode amplitude doesn't change vary much over the millisecond that the CCs are pulsed. Check out shot 102452 for a good example of MHD amplitude varying too much over the 4->5 ms period, when the CCs are pulsed.

2. A positive current on the CCs should imply they are generating an outward radial field, i.e. pulling the field lines outboard.

Thursday June 20 2019 11:10 am Saperstein 102481-102528 static RMP tests and SOLC responses

The purpose of this run day was to perform 2 tests, and maybe a 3rd if we had time. The first test was to fire only a single control coil in section 1 and then section 8. This was to make up for missing these sections during lasts weeks similar tests. The second test was to see how the SOLC sensors responded to static RMPs. The control coils were given an m/n = -3/1 helicity, and ran between 3-4 ms, the timeframe where q in our usual shot styles (102416) drops below 3. The control coils had a peak current of ~15-20 A, depending on the shot, and the sect 2 bottom and sect 8 top CC amplifiers were left off (sect 1 coil 3 just wasn92t working too). While the main focus of the RMP tests were to look at the SOLC responses, this entire part of the run day ended up just being shot development to find a shot where the RMPs could 93consistently94 couple well enough to the modes to lock them. We did find some shot styles that were successful at temporarily locking the modes, however, the locking usually wasn92t permanent nor reproduceable. The 3rd test we wanted to try to get to, but didn92t manage to, was to look at the SOLCs response to rotating RMPs. We92ll have to come back to these later after we find a reproduceable shot style where the RMPs couple to the modes.

Shots 102481-494 were dedicated to getting the remaining single CC shots that we didn92t get last week. The model shot for firing cc8-2 was 102488, and the model shot for firing cc1-2 was 102493.

Shots 102495-516 were dedicated to developing a shot style with consistent RMP coupling to the modes. I started from the usual (102416) shot style, and then tried to vary settings from there to get reproduceable coupling. It interestingly turned out that this usual shot style was actually able to get good coupling and mode locking, see shot 102500. However, the mode coupling wasn92t consistent in the shots taken after. Throughout the day, I tried varying settings to reach similar edge q ranges to those used in the past to get coupling on HBT, but unfortunately was not able to find a suitable shot style before running into bigger problem.

After shot 102516, and some analysis of some unintentional vacuum shots, we found that the SOLC sensors were now somehow seeing magnetic pickup from the RMPs. This pickup, varying for each sensor, could sometimes be as large as a couple amps when about 15-20 A was running through the CCs. This was first noticed in shot 102497, which was unintentionally a vacuum shot after the puff failed to fire. We noticed that the SOLC tiles were picking up RMP signals, in addition to the other applied fields. This was ultimately concerning because we had performed pickup tests for the SOLC sensors last week, and found that they saw no pickup even when the CCs were putting out their max field. We performed the same test again today, and found that they were now picking up some RMP fields. The pickup comparison can be made between shot 102379 (where we saw no pickup) and shot 102520 (where now we did see significant pickup), where these two shots should have the exact same run settings.

Shots 102517-528 were crates only shots dedicated to troubleshooting this problem. We switched back to the segmented pulsing of the CCs like that in shot 102379, and tried reverting any changes to the machine made between last Wednesday (2 run days ago) and today. The only big difference, was that the section 10 shells were retracted today so that fast camera data could be taken during these shots. But after inserting the section 10 shells all the way in, the problem continued to persist.

I tried narrowing down when the SOLC sensors might have started picking up CC fields, and I found that they were seeing pickup as early as the first shot on Friday (the previous run day). Shot 10242792s vacuum shot with the section 3 CCs on at full blast saw pickup in the SOLC sensors. This means that the change must have occurred sometime between Wednesday and Friday morning. Unfortunately, we didn92t take any vacuum shots later in the day on Wednesday, so it92s difficult to tell if something changed over the course of the day. As for Thursday, nobody was running that day, but I did have some undergrads run a plastic fiber optic around the lab, part of which wrapped around the base of the machine, which may have done something. As of right now, I don92t know what else might have happened on Thursday to cause this problem. It also appears that this pickup problem is local within the SOLC sensors. Based on shots like 102528, where only the coils in a single section were fired, it looks like only SOLC sensors around these coils (sec 4 sensors) saw pickup, while the others saw nothing. If the pickup problem was coming from something like a new ground loop, then I would expect that a local firing of the control coils would still induce a global pickup in the SOLC sensors. As of right now, we are still looking into why these sensors may have starting seeing magnetic pickup from the CCs. Fortunately, this pickup isn92t too problematic, since subtracting out this pickup can actually be very easy, and should even be able to be automatic, outside of jScope.

Additional notes:

1. The puff circuit 12 V battery was found to be at about 6 V today, after shot 102499. I believe that the voltage at the beginning of the day was around 11.4ish V, but this may have dropped around shot 102493, when we first started noticing less gas being puffed in during shots. This is also roughly correlated with when we power cycled the basement rack power after shot 102490, accidently discharging a puff into the machine and closing the pump valves. This battery was replaced with another temporary 12 V battery, which had been working just fine since. And the old battery was hooked up to a charger to charge overnight.

2. The fast camera was having trouble triggering correctly today. It would have the tendency to trigger when the banks began charging, and still ran into some other mis-triggering problems when we tried starting the wait for the trigger to come after the banks already began charging. We tried power cycling the basement rack power to see of this would help correct the problem, and temporarily it did. However, after some amount of time, the problem returned again.

Wednesday July 10 2019 3:03 pm Saperstein 102535-102638 RMP coupling shot development

The following is relevant to the Monday (7/8/19), Tuesday (7/9/19), and Wednesday (7/10/19) run days. Also worth noting for these run days, is that the base pressure was pretty high, thanks to problems with the pumps. As a result, the pressure was usually in the low to mid -7 range. Also, section 10 shells were fully retracted, and the rest were fully inserted.

Monday (7/8/19), shots 102535-102551:

The purpose of this run day was to both do clean up shots and begin shot development to look for a shot style where a 3/1 RMP could consistently couple to the plasma. Jim also installed a new external x-ray detector that92s attached to the outside of the machine at a little above the outboard midplane between sections 2 and 3, directly above the limiter port. Shots 102535 and 102536 were crates only shots that checked to make sure the x-ray detector was correctly wired the N rack CPCI channel 96, and it was. The rest of the day92s shots, 102537-102551 were just clean up shots, with a static RMP applied once every 3 shots. Shot 102541 is a good example of the new x-ray detector seeing lots of light during the plasma lifetime. Unfortunately, the detector also sees a lot of noise during the disruption, which makes it difficult to do analysis in that region. The gain was at it92s max, but it would be nice if we could make it larger.

Tuesday (7/9/19), shots 102552-102607:

The purpose of this run day was to continue RMP shot development. Tried to reproduce the shot style of 102500, and then continue development from there. Also tried varying strength of the RMPs. Using the RMP code, we varied the strength between 1.5 and 3.5. We found that at about 2.5 the plasma started disrupting following the end of the RMP, and by 3.0 and 3.5, the plasma disrupted more often than not. Shot 102588 is a good example of a str = 3.5 RMP disrupting the plasma. Some shots that we found that are good examples of great mode locking are 102600 and 102604. However, these shot styles weren92t very reproducible. Lastly, the batteries for the x-ray detector were left connected overnight, and at some point died, meaning x-ray data from this run day isn92t very reliable. However, shots 102557, 582, and 603 did see some significant x-rays.

Wednesday (7/10/19), shots 102608-102638:

The purpose of this run day was to finish RMP shot development. The previous days shots weren92t very consistent, so today we tried lower the OHS in order to make the shots more reliable. This worked pretty well, and managed to get some relatively consistent coupling shot styles. These include the shot styles for shots 102621 and 102631. These shots styles are relatively similar, with slightly varying OHS, VFS, and VFE values. 10262192s shot style managed to couple 4 shots in a row, while 10263192s shots style couple about 7 times in a row. The consistent shots did have varying degrees of coupling, but they should be consistent enough to use for future experiments. 102638 is also a good, no RMP reference shot. Lastly, the e-gun battery was pretty low on charge (~8.5 V), so I started charging it, and temporarily replaced it with a different one. And the the x-ray detector batteries were replaced, although we didn92t get any good x-ray shots today.

Wednesday July 17 2019 10:22 am Stewart 102639-102672 RBM/ITG Experiments

Run report for 7/11. The goal of this series of experiments was to categorize the turbulence type (main candidates: resistive ballooning mode and ion temperature gradient mode) in the edge region using a triple probe at the LCFS and gas puffing to cool the edge down. However, it was found that during the flat top shot style (see shot 102668), the SXR profile exhibited peaking over time with the edge Te measured by the triple probe decreasing over time. The floating potential and Isat fluctuations correspondingly decreased with decreasing temperature in an exponential fashion. This pointed towards ITG as the dominant mechanism for the turbulence drive even prior to the gas puff experiments.

The setup for puff experiments ended up being more difficult than expected since the second puff trigger from the CAMAC system would not puff at the right time. Instead, the puff trigger was sent to a function generator to output both the first and second trigger at the proper time. Shot 102663 was a good example of the puff cooling the edge of the plasma with the edge Vf and Isat fluctuations decreasing with decreasing temperature. This also seems to indicate ITG turbulence drive over RBM.

During the initial attempts to get the puff time correct, there was a shot with large hard X-rays (as measured by Jim's new X-ray detector) as well as a low fill pressure. The X-ray signal was actually high enough to clip on the CPCI channel it was connected to.

Wednesday July 17 2019 11:10 am Wei 102763-102715 Shot style development

Tuesday (07/16/2019) Shell 7 & 10 were fully retracted. Bias probe was retracted to 113 cm. The purpose of the run day was to develop a shot style with stable edge q and major radius while having strong MHD mode activity. This shot style was intended to be used to study correlations of the fast camera images with SOLC and magnetics. 102673-77 were vacuum and clean-up shots 102678-81 imitated the shot style of 77324 done by S. Angelini. This shot style was found to be undesirable as it tends to develop minor disruption immediately after breakdown. 102682-102715 was modeled on 102668. OHE & VFE were adjusted to increase mode amplitude (3/1 kink) by pushing edge q slightly below 3 during shot. Camera videos were taken for shots 102690-715, excluding 102704. Star shots: 102709, 102710, 102712; 102690, 102692, 102697, 102700, 102702, 102705

Monday July 22 2019 10:18 am Saperstein 102716-102764 RMP offset tests

The purpose of todays run day was to look for the ideal RMP phase offset to use for the SOLC radial scan. We already had a few sensors in mind that see clear locking that could be used for the radial scan, so we wanted to test what phase offset would introduce the strongest locking in these tiles.

CC current channels were reattached to their temporary position in the N rack CPCI. Only section 1092s shells were fully retracted for the entirety of the run day, while section 7 top was retracted for the first 3 shots. All other shells were fully inserted. All probes were removed, and the e-gun battery was replaced with its original, recharged battery.

Shots 102716-102728 were shots that started looking at the theta = 0 and pi offset values, but we were running into the problem with the puff pressure dropping quickly over the course of these shots. Eventually, we found out that this was due to the relatively low charge on the 9V battery within the puff circuit. We replaced it, and the shots that followed managed to maintain a good puff pressure. Section 7 top shell was also retracted between shots 102716-102718, but we fully inserted after that.

Shots 102729-102756 were dedicated to testing different phase offset between 0 and 2pi, in increments of pi/4, for a total of 8 different phase offsets. These offsets were not taken in continuous order, and were semi-randomly chosen to be tested in an arbitrary order. Each phase offset also had a crates only shot taken so that the SOLC measurements from those shots could be removed from their respective plasma shots. The only exception is phases of pi and pi/2, who92s crates only shots were taken at the end of the run day, shots 102763-102764.

Lastly, shots 102757-102762 were dedicated to once again getting reference shots with no RMPs.

Thursday July 25 2019 10:43 am Saperstein 102765-102907 RMP radial shell scan

This run report covers 2 runs days, Tuesday (7/23/19) and Wednesday (7/24/19). For these runs, all shells except those in section 10 were fully inserted at the beginning of both days (section 1092s shells were fully retracted) and section 4 top92s shell was retracted over the course of these 2 days (the positions the shells were retracted to were chosen to go back and forth). All probes were outside of the plasma as well. Section 4 top92s CCs were also left off when turning on the CC amplifiers.

The position of section 4 top92s shell was measured using a caliper that measured the distance between 2 arbitrary (but constant) points along the axis the shell retracts into. The positions of these shells were measured with an accuracy of about +- 0.3 mm. The position of the shell when fully inserted was about 14.1 mm (measured on the first day) to 14.4 mm (measured of the second day), and I believe the 14.4 mm measurement was more accurate. The shell was retracted up to its fully retracted position at ~56.7 mm, spanning a distance of about 42 mm. For each shell position, about (at least) 3 shots were taken with the static RMP on, then a shot with them off, and then a crates only shot with the RMP on to get a reference for the RMP pickup in the SOLC tiles for the shell at that position. A 3/1 static RMP with str = 2.25 (defined in prep_awg.py of the hbt/control/ library) and phase offset of pi/4 was applied to the plasma in order to induce locking. The pi/4 phase offset was chosen to maximize the strength of the locking measured by the SOLC sensor 4-3.

Tuesday (7/23/19)

The nominal shot style was fairly reproduceable for this day. 11 shell positions were tested. These included 14.4 mm (fully inserted, 102787), 56.7 mm (fully retracted, 102791), 34.4 mm (102799), 45.0 mm (102803), 25.0 mm (102806), 30.0 mm (102813), 19.7 mm (102818), 17.1 mm (102823), 15.0 mm (102829), 16.2 mm (102835), and 18.0 mm (102840). The shot numbers give in parenthesis are the model shots for each shell position, and the order in which the positions were given reflect the order they were measured/tested in. The model reference shot (no RMP) for this run day would be 102784.

Commenting on this run day, the following day I noticed that for Tuesday92s RMP shots that control coil cc5-4 was not working between at least shots 102716 and 102807. It was at least working up until 102542, but stopped working somewhere between then and 102716 (Stewart and Wei were the only two who had run days during this time, but neither of them were using the control coils). Somehow though, the cc5-4 coil just started working again for shot 102808 (nothing special happened between 807 and 808). After noticing this difference on Wednesday, I felt I had to redo the shell positions taken before 102808, these included 14.4 mm, 56.7 mm, 34.3 mm, 45.0 mm, and 25.0 mm.

Wednesday (7/24/19)

The nominal shot style was not as reproduceable as it was the day before. Settings frequently had to be tweaked to get the nominal shot style back, which resulted in testing significantly less shell positions than the previous day. Positions 14.4 mm and 25.1 mm were retested to make up for the temporary 93breaking94 of cc5-4 (the other positions have yet to be retested, because the signals at these positions were ~0 A, and I felt retesting them was a lower priority than getting newer, more relevant shell positions). The other positions that were tested include 19.1 mm (102880), 24.2 mm (102890), and 22.1 mm (102900). Interestingly, an inversion of the current polarity was found at the 24.2 mm shell position (and maybe the 22.1 position as well). There are still more positions to test between 20-30 mm, to get better resolution of the relevant extent of the SOL.

Friday July 26 2019 11:26 am Saperstein 102957 RMP radial shell scan continued

Before I could really start running today, we had a problem with the power supply for the TF/VF control panels. The fuses behind the panels had blown, and the fuses we replaced those with blew as well after turning the control panel power back on. The power supply for those panels appears to need to be replaced.

We shut down after this, and made sure the banks were dumped successfully.

Monday July 29 2019 10:59 am Wei 102908-102956 Shot style development

07/25/2019 n Section 10 shells fully retracted; section 4 top shell retracted by ~ 8 mm; reinserted after (include) shot 102934 n The purpose of the day was to continue the shot development runs on 07/17/2019. The goal was to have strong mode activity and current signals at the midplane SOLC tiles (in particular S1_04), by keeping the plasma outboard limited and making edge q dips and then rebounds around q=3. 102908-102913 were clean up shots. 102914-102956 were shot development runs. Camera was turned on at the end of the day, for shots 102954-102956. n Star shots: (w/o camera) 102922 102925 102935 102938 102942 102943 (w camera) 102954 102956

Wednesday August 7 2019 11:50 am Wei 102959-102997 Shot development

08/06/2019. Section 10 shells fully retracted. Today's runs continued the shot style developments of the previous two run days. This was the first run day since the control panel power supply was replaced on 08/02, and nothing went wrong during the entire day. 102959-102968 were vacuum & clean up shots. 102969-102977 used the shot style of 102954. Strong solc signals were observed on the top & bottom tiles but not on the midplane ones. Star shots: 102972-74. 102978-102997 used a different shot style based on 100088. The plasma MR was pushed beyond 92 cm in order to make it outboard limited. Edge q typically dived to 2 and then bounced back. Strong 2/1 mode was observed, which typically rotated at ~20 kHz, potentially due smaller minor radius. SOL current at about 20 A (not exactly sure about the unit) was observed on the midplane tiles between 2-5 ms. Hard x-ray signal was weak in all of these shots. Star shots: 102986-89 (identical settings; 102988 was vacuum reference shots), 102991-92, 102996-97. Camera videos are available for all shots of the run day, except 102966, 102969, 102970

Tuesday August 13 2019 9:58 pm Saha/Wei/Levesque 103001-103048 Double-puff experiments with fast camera and RMPs

Experiments exploring applied RMPs during a second gas puff with the fast camera active were performed over 2 days. Section 10 shells were fully retracted for the fast camera view. 103001-103002 and 103025-103026 were vacuum shots. The OH bank rang for shot 103001 since a connection had fallen off one of the OH ignitrons. All shots used a loose target of shot 102974. We tried to keep edge q just under 3 for the period of interest in each shot. The earlier shots were spent developing a discharge style where the second puff (at 4.0ms) would not induce a disruption soon after the puff. The regulator pressure for the puff line needed to be reduced from the recent 29psig baseline to around 9psig in order to prevent an immediate disruption after the second puff when using the shortest programmable 2nd puff duration of 1us. Star shots: 103028, 103040, 103044. 3/1 RMPs were applied for shot 103035 onward, with the applied fields generally being present as the gas was penetrating. In shots 103044-103048 we applied the RMP earlier (starting at 3.7 ms) and had it turn off while the gas was still penetrating the plasma. For shot 103041, we ramped up the strength of the RMP to disrupt the plasma quickly. For shot 103048, we applied a 1/1 RMP, which had no obvious effect. For shot 103043, we applied the RMP at a 180-degree phase difference relative to the other shots.

Wednesday August 14 2019 1:35 pm Wei/Levesque 103054-103096 Shot development: runaway electrons

2019/08/13

Section 10 shells fully retracted. Fast camera looking toward section 2

Today's run aimed to develop a shot style that emits strong hard x-rays and produces speckles on fast camera images. This was done by using the bank setup of a long stable shot while lowering field pressure to ~47 uT, which seems to be a "sweet spot" for generating hard x-rays. Plasma's position doesn't seem to affect whether or not hard x-rays appear, although successive stronger bursts tend to occur more often in inboard limited shots. The strongest bursts in a shot almost always occurred milliseconds before (rather than after) the current spike, and preceding these bursts was a "build up" period of consecutive weaker bursts lasting ~2 ms. In some shots spikes of hard x-ray seems to have a periodic nature (such as 103090), although the frequencies of these bursts -- ranging from 1.4 kHz (103090) to 5 kHz (103084) -- are lower than that of mode activities.

Speckles were found in shot 103061 65 66 69 85 90 91 93 94. These events tend to be rare and hard to reproduce, and we are uncertain whether it's caused by run-away radiation going into the lens or by it directly hitting the camera's image sensor. Jeff mentioned he found a speckle in 103063 at the location of the lower shell, although I wasn't able to find it after playing the video for multiple times.

2-star shots: 103061, 103084, 103090, 103092, 103095

1-star shots: 103063-69, 103091 103093, 103094

Camera videos are available for all shots except 103054-57, 103080-81

Monday August 19 2019 2:11 pm Wei 103097-103125 Shot development

08/15/2019, section 10 shells fully retracted. Today's shots continued the shot development runs of 08/07/19. I was able to reproduce the shot style of 102986 & etc which had a nearly flat Ip profile and saw strong signals at mid-plane SOL tiles due to plasma being heavily outboard limited. Nevertheless, with this shot style I wasn't able to further increase the pickup on these tiles as their responses were highly sensitive to plasma MR, and the edge q was very close to the q=2 limit so that I couldn't increase Ip while keeping the same MR profile. At the end of the day (103123 & 25) I found out that by pushing the plasma slightly inward (plasma was still heavily outboard limited) I was able to ramp up the Ip profile, and the tiles' pickup did increase with stronger Ip as expected. This can be the starting point for further shot development runs in the future.

Star shots: (old shot style) 103105-108, 103111, 103117-18, (new shot style) 103123, 103125

103120 was an abnormal shot which saw a huge spike on the current tiles at 3.4 ms due to a minor disruption on the outboard side.

Monday August 19 2019 4:42 pm Saperstein 103126-103

Use "

" to force a line break. Carriage returns are ignored.

Monday August 19 2019 4:51 pm Saperstein 103126-103174 RMP radial shell scan continued

The purpose of this run day was to continue the RMP radial scans I had started 2 weeks prior. There were still some shell positions I needed data points for to fill in the radial profile. The shot style used before was recovered quite easily.

Shots 103126-103141 were dedicated to shot development in order to get back my previous shot style. Shots 103142-103173 were then dedicated to getting shots for shell positions 21.0 mm (103144), 28.9 mm (103149), 22.9 mm (103154), 25.8 mm (103163), and 27.9 mm (103171). Model shots are provided for each position. Shot 103174 was then taken to check how the no RMP reference shot at 14.4 mm developed throughout the day.

Wednesday August 21 2019 10:07 am Saperstein 103175-103213 RMP radial shell scan continued

The purpose of this run day was to try and finish up the RMP radial scan run campaign. It92s worth noting that section 10 bottom92s CC amplifier was accidentally left off during this run day, however this shouldn92t affect the analysis significantly, especially if the current measurements are normalized to each shot92s mode amplitudes.

Shots 103175-103191 were dedicated to clean-up and shot development, with 103189 being the model shot that had the shot style I92d been using return. Shots 1033192-103213 were dedicated to testing the last couple of radial positions, 27.1 mm (103194), 18.1 mm (103201), and 19.1 mm (103209). Model shots for each position also provided.

Tuesday September 10 2019 5:30 pm Stewart 103254-103296 Rake Probe Testing

The rake probe was finally installed prior to the first shot of the day, with channels 1-17 of the probe on CPCI_40 in the North Rack (CPCI inputs 65-81). The first 14 shots were devoted to clean-up and the gate valve to the new rake probe was opened for shot 103268. The probe was subsequently inserted to R=107cm for shot 103281. Fast camera video was used to determine if there was any arcing/sputtering and the probe was visually inspected periodically to check for damage. The next ~15 shots were a slow insertion of the probe to 104cm, moving at 1 cm increments. While there was some lighting up on the fast camera for these first couple of shots with the probe in the plasma, no damage could be discerned on inspection through the port and the voltage measurements were in line with what the triple probe had measured in previous scans. A good example is shot 103293, where the rake probe took typical floating potential measurements (including turbulence) on the first 5 tips of the rake probe with the other tips beyond R = 107cm reading ~0V. It will be the goal of the next run day to fully insert the probe in order to determine if all of the channels are in working order. But the dynamics and radial resolution of the probe already look promising.

As a hardware update: (1) The rake probe is now installed (2) The hard X-ray detector input was moved before shot 103254 and placed in South Rack CPCI_12:INPUT_05 for all shots after 103280.

Wednesday September 11 2019 5:35 pm Stewart 103299-103353 Finishing Rake Probe Testing

This run day was a continuation of yesterday's commissioning of the rake probe with the goal of pushing the ceramic tip/cap of the probe to 102cm. The rake probe was first inserted to 104cm, which caused early disruptions, so the probe was brought back to 105cm (shot 103308). Since the shots were behaving normally with the rake probe at this location, the probe was again inserted to 104cm. On shot *103313* there was some pretty dramatic fast camera video with a luminescent plume emanating from the end cap of the rake probe. The shot also encountered some hard X-rays from the detector Jim installed and low SXR emission. The probe was inspected visually for damage (none was present) and it was decided to retract the probe to 105cm. The rest of the run day was free of any similar incidents and the data from the rake probe matched data recorded prior to shot 103313.

The remainder of the run day was allocated to a very slow insertion of the probe to R=102.8cm. The plasmas behaved normally the rest of the day with shots lasting ~6ms even with the rake probe inserted beyond R=103cm. For reference, shot 103338 had a well centered major radius with floating potential data on ~10 of the upper tips and ~3 of the lower tips. It was very easy to see the LCFS position from the probe data since the zero crossing (negative Vf to positive Vf) occurs at the tip corresponding to R=107cm for all of the shots in the radial scan. Power spectra also show broadband Vf fluctuations at all radii within the plasma, with V_rms decreasing with radius. It remains the work of tomorrow's run day to compare the rake probe with a floating triple probe (at section 5) and to use the bias probe to induce an H-mode in order to carefully analyze the turbulence suppression.

Hardware Update: Alex installed a shunt resistor on Jumper B to compare with the Rogowski sensor already in use.

Saturday September 14 2019 5:07 pm Stewart 103355-103447 Biasing H-mode with Rake Probe and TP

This run report covers the two run days: 09/12/19 and 09/13/19 in consecutive order. The first half of the first run day was designated for shot development with the rake probe (section 2), triple probe (section 5), and bias probe (section 7) inserted to their optimal positions in the plasma. The rake probe was pushed to it's final position at R=102.3cm for shot 103380, which meant the first floating potential measurement could be made at R=103cm (i.e. the same radial position as the tip of the bias probe).

The best shots from the first day were 103403, 103406, and 103409 which had the triple probe (unbiased) at 105cm, 106cm, and 107cm respectively. It was determined (using these shots and verified with some from the next run day) that the radial correlation length for the turbulence was Lr~2cm and the correlation time t_c~10-30 microsecond, with the correlation time increasing towards the core of the plasma.

The next run day was allocated for biasing experiments to verify the effect of biasing on turbulence suppression. A reference shot before biasing includes shot 103433 with TPS5 floating at R=107cm. For the next 4 shots, the bias probe was grounded with V_bias = 0V and I_bias ~12A (Reference: shots 103436 and 103437). In these shots, the MHD activity completely flatlined (most likely due to the biasing) and the radial V_f profile measured by the rake probe increased as expected. However, the autocorrelation function for the Vf measured across all radial position changed character with a large drop in t_c near the probe. Comparing the triple probe measurements at R=107.0 (the LCFS) between unbiased shot 103433 and +0V biased shot 103434 shows an enormous increase in high frequency floating potential fluctuations. This evidence indicates that even for small amounts of biasing, the turbulence can be altered.

Shots 103439-103447 were biased H-mode shots, however these were of the "dithering" variety with the electric field-well collapsing during the biasing. The full analysis will require a more quiescent shot with a more constant Er structure. Even so, preliminary analysis confirms a reduction in high frequency fluctuations across the channels of interest on the rake probe. It also shows a large jump in Lr and t_c for the biasing phase. The radial resolution of the rake probe also allows for the direct observation of the collapse of the Er-well, which could not be resolved before with a single triple probe.

Hardware update: Jeff has remounted the USB spectrometer fiber optic on section 10, which allowed spectra to be taken during the second run day.

Tuesday September 17 2019 10:32 am Stewart 103489-103516 Biasing H-mode with Rake Probe and TP II

Run report for 09/16/19. This run day was again devoted to trying to get a stable Er well and H-mode for a long enough duration to perform analysis on floating potential fluctuations. The goal was achieved for shots 103509 and 103512, however the rake probe indicates oscillations across all of the tips inserted beyond R=107cm during biasing that match a small ~1G MHD mode (which pollutes the data for turbulence measurements). This mode dies off in shot 103509 briefly yielding ~1-2V oscillations that are most likely turbulence where during L-mode the fluctuation amplitudes were ~20V. The true suppression amount is then on the order of a factor of ten.

For these shots, the SOL tiles in section 8 also saw the characteristic drop in both fluctuation amplitude and average signal. However due to the presence of the small MHD mode, another run day may be required to achieve a shot with similar parameters but with ~0 MHD for a period of 1ms. Brief Note: biasing was attempted at lower pressures with the puff time decreased to 150 microseconds, but fill pressures below 60 microTorr began giving bad breakdowns. Pressures around 61 microTorr produced shots which were almost identical to fill pressures at 69 microTorr (standard operation), so the idea of lowering the fill pressure to lower the density was abandoned.

Hardware Update: (1) The puff power supply was replaced by the TPS7 power supply box which seemed to give more consistent pressures shot to shot. (2) The rake probe wire shields were connected to together (also grounded) on the CPCI side which reduced noise by a factor of ~10 during biasing (compare shots 103498 with 103500). The shields were already connected to each other at the Rake probe box.

Wednesday September 18 2019 2:00 pm Saperstein 103517-103537 radial scan vacuum shots

The primary purpose of this run day was to get vacuum shots for different radial positions of section 4 top shell, to be used in analyzing low frequency SOLC radial profiles. All shots taken during this run day were dedicated to this. Shot 103537 was also used to help estimate the amount of Ip pickup the current set of SOLC tiles saw (which turns out to be a lot, with a signal to pickup ratio of about ~1:2).

I also meant to get some plasma shots in, in order to get a range of shots taken with the section 4 top shell fully retracted. However, I ended up not having enough time for this phase of the run plan, and had to cut the plan short. Fortunately, these shots were not absolutely necessary, but would92ve been nice to have.

No significant model shots for this run day.

Monday September 23 2019 10:24 am Stewart 103637-103672 ITG + Double Puff Experiments

Run report for 09/20/19. These shots were devoted to recreating the earlier ITG experiments performed with the triple probe, but now with the radial resolution of rake probe. The goal was to verify that the turbulence was long wavelength, electrostatic in nature, propagating in the ion diamagnetic drift direction, and destabilized by increasing T_e. The triple probe was scanned from R=103-107cm while floating to attempt time delay estimation velocimetry and to determine the poloidal wavenumber spectrum. The shots of interest are 103642, 103642, 103647, 103656, 103659, and 103660.

Unfortunately, the double puff experiments for shots 103661 and 103662 caused disruptions early on and could not be used to see the effect of cooling at the edge on the turbulence level. However, shot 103665 (single puff experiment) had a slowly decreasing T_e over time and a corresponding decrease in Vf_rms over several milliseconds. Another good shot with low mode activity was shot 103672.

Hardware update: The cooling water reservoir for the basement rack was empty during startup in the morning. We refilled the reservoir but the pump may need to be replaced.

Friday September 27 2019 11:02 pm Chandra 103538 103635

The theme of this two day run campaign was EUV based GPU mode control.

No major hardware changes were made.

The run started off following shot style 101431, and attempted to replicate Jeff's mode suppression results from April (shots 101399-101436). Initial results using the identical SVD basis were promising, although the natural mode dynamics were higher than desired, and so precluded unambiguous claims of suppression.

Further attempts were made using alternate basies, and it is now at least partially understood why results here were particularly unimpressive: there was an inconsistency in the EUV array numbering schemes between the GPU/CPICI and imposed basis.

Some software difficultly was had in getting the GPU to respond to updated manipulated constants in the compass scan, but it is believed that this difficult is overcome by running "make distclean" to remove and recompile .cubin files after each "make fb_shot".

Friday September 27 2019 11:42 pm Chandra 103967 103995

The theme of the shot was GPU feedback code output verification

No hardware changes were made

Based on the results of shots 103565-103653, this short campaign focused on verifying gain and phase. The output mode gains modified the control coil request as anticipated (although the effect on control coil currents is less clear. Flipping the phase offset 180 degrees appears to have had the intended effect, based on a comparison of two neighboring control coil - feedback sensor pairs.

Wednesday October 2 2019 11:36 am Saperstein 103734-103778, 103795-103860, 103880-103944 Rotating RMP Frequency Scan

This report covers the run the campaign taken over 3 days, 9-24-19, 9-25-19, and 9-26-19. The purpose of this run campaign was to investigate how the SOLC responds to varying applied RMP frequencies as compared to the response of the MHD. It92s relevant to note that for these shots, the basement rack cooling water was left off, since it wasn92t working at the time. In addition, all shells were fully inserted except for section 1092s, which were fully retracted, and all probes were retracted out of the plasma.

1st day, 9-24-19:

The purpose of this run day was to develop a shot style and corresponding RMP structure that couple well together consistently. The main RMP structure parameters to figure out were the frequency ramp range, the frequency ramp speed, and the current ramp through the coils (to counteract the suppression of the current the coils apply to high frequency signals).

The first 5 shots were dedicated to making sure the rotating RMP code was still working correctly, and to see how the control coils (CCs) would respond to high frequency inputs. Following that, we had a vacuum shot, and 2 more shots to finish cleaning the plasma (finishing with 103741). Shots 103742-762 were then dedicated to adjusting bank settings and RMP parameters, in order to develop an ideal shot style.

Starting at shot 103763, the CPCI92s started failing to store, and then eventually, even to prepare. This problem was fixed by power cycling Oscar and Spitzer as well as leaving the basement rack off for ~30 minutes. It92s likely the problem was caused by overheating of the basement rack, cutting off the highway. A small fan was placed facing into the rack to try and cool it. This problem did not surface again over the course of the rest of the campaign. Shots 103763-766 were lost during this time.

Shot development continued between 103767-778, however, and suitable shot style wasn92t found by the end of the day. No good model shots for this day.

2nd day, 9-25-19:

The purpose of this run day was to continue developing a shot style that would couple well to a ramping, rotating RMP. Fortunately, an adequate shot style was found by the end of the day.

Shots 103795-827 were dedicated to shot development, with the addition of modifying the max 93waveform94 value as another parameter for the RMP. A nice shot style was found by 103828 (model shot for frequency ramping down). Over the course of the shot dev, we did accidently blow several of the fuses for the CC amplifiers. This typically occurred when we made a mistake changing the CYCLE_TIME of the GPU system.

Shots 103828-860 were then dedicated to taking several shots in the frequency ramp down and ramp up configurations. The model frequency ramp down shot was 103828, and the model ramp up shot was 103835. Vacuum shots for these shot styles were also taken, and include shots 103854-857.

3rd day, 9-26-19:

The purpose of this run day was to get multiple shots using the shot style found the previous day. Shots 103880-907 were dedicated to replicating the shot style from the previous day, and getting repeated shots, and looking for ones where the RMP coupling to the MHD. Model shots include 103898. Next shots 103908-944 were dedicated to getting shots with constant RMP frequency, instead of a ramp. Model shot includes 103920. Vacuum shots for these shot styles were actually taken on a different day. These include shots 104152-168.

Thursday October 3 2019 11:16 am Brooks 104021-104136 DPS7 cleaning and 14kHz FF current mapping

The goal of these two run days (9/30 and 10/01) was 1) to condition the newly inserted (on 9/30) DPS7 (directional probe at section 7) and 2) to drive 14 kHz sine waves over several parameters to attempt to measure the current path to the wall.

DPS7 cleaned up very quickly, particularly at higher frequency driven current. Low frequency (or even DC offsets) resulted in more arcing. The little arcing that did occur noticeably diminished over the day.

Shot 104175 is indicative my desired shot style for my campaign. It was characterized by: 1) little to no mode activity between 2 and 4 ms, 2) flat q profile, 3) flat radius, and 4) outward limited. My coherence calculations, in conjunction with a q profile fitting algorithm, appear to have revealed definite current paths in the plasma that appear to change throughout the shot sometimes and more consistently when I change some variables. Results forthcoming.

Shells 10, 7, 4, and 5 T&B were moved in and out throughout the run day. See shot notes for specifics.

Thursday October 3 2019 11:28 am Brooks 104169-104212 DPS7 and DPS9 active feedback control and phase scan

The goal of this run day (10/2/19) was to use the four electrodes on DPS7 and DPS9 to suppress mode activity. The electrodes were controlled independently and locked to the n=1 mode phase with the following offsets: DPS7CoIp (108 deg + mode phase + fixed offset), DPS7CounterIp (108 deg -180 deg + mode phase + fixed offset), DPS9CoIp (180 deg + mode phase + fixed offset), DPS9CounterIp (180 deg -180 deg + mode phase + fixed offset). Assuming the current sourced by each electrode travels the same direction the electrode is facing (co-Ip or counter-Ip), this system allows probe at each section to increase and decrease the total amount of co-Ip current.

Setup: I used both of the new OSC5050a amplifiers for a total of 4 independent channels. Each channel was connected to a Stangenes transformer in Parallel:Parallel setup.

After an extended setup period and verifying that the system was working the way I expected it to, I performed a brief phase scan of the fixed phase offset value. Unfortunately, there was a lot of variability in the target shot style (in terms of mode amplitude and disruption time) that it makes it difficult to determine if my system was successful. Shots 104188 shows an example where the mode amplitude may have decreased after turning on the feedback system. More detailed analysis forthcoming.

Shell 7B was the only shell out. TPS5 was inserted and powered.

Friday October 4 2019 9:37 pm Stewart 104217-104262 Spectral Shift Experiments

Run report for 10/03/19. The goal of the run day was to further investigate a shift in the radial wavenumber spectrum for different amounts of bias, this time with the double probe at section 9 since the hemisphere probe at section 7 was removed. The shots of interest are: 104261 (near floating V_bias ~ -100V), 104250 (V_bias = 0V), 104253 (V_bias = +40V), and 104254 with V_bias = +60V. From a preliminary analysis, the 'measured' poloidal phase velocity of the turbulence goes from negative in the ~floating case to positive in the 0V bias case. The phase velocity increases with increasing bias with a steeper slope in the wavenumber frequency spectra for +40V and +60V.

Above this voltage range, the plasma began transitioning to H-mode with a clear dithering transition seen in shot ***104258***. The shot was rather dramatic in the periodicity of the large drops in current and the large spikes in electric field (This shot was obviously very close to the threshold of biasing required for the transition). The oscillation from low electric field to high electric field can easily be seen by the rake probe, and these oscillations in potential permeate to the far reaches of the SOL (The rake probe tip at R=108.5 can see the 'dithering' oscillations in the measured floating potentials). However, it remains the task of brute force analysis of the potential fluctuations to show a clear spectral shift.

Monday November 4 2019 8:13 am Brooks 104556-104609 Current mapping

The goal of Friday's (11/1/19) run was to continue my current mapping studies. Each electrode in DPS7 and DPS9 were interchangeably driven at 14 kHz at 103cm in order to study if the toroidal location and current sourcing direction play a role in current's path. Second, phase flips and frequency sweeps were conducted at DPS7 in the co and counter Ip directions. Shot 104957, PA2_S28P is a good example of the DPS7 Co-Ip response. Probe current is on channel 'HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_95' with a -10 times gain. Analysis on the data is still ongoing.

Plasma cleanup took longer than expected, likely due to the multi-week break we took over the various conferences.

Shell configurations were mixed throughout the run. Please refer to the log book for details.

Thursday March 12 2020 4:32 pm Brooks 104556-104637 14 kHz current mapping. Moving shells to map current exit.

These experiments took place on 11/01/19 to 11/04/19. Shells at several configurations. Fist day: DPS7 (co and counter Ip electrode) driving 14 kHz current for current mapping. Second day: different shell configurations to attempt to map the current exit. DPS9 co-Ip electrode driving 14 kHz.

Tuesday November 19 2019 2:20 pm Brooks 104638-104710 Rake probe testing with biased electrodes

The goal of this run day (11/12/19) was to place the rake probe in the plasma from 103 to 109 cm at section 2 and bias DPS9 with co and counter Ip with phase flips, freq. sweeps, range of single frequencies, dc biasing, and some radial scans with the biasing probe. In particular, we are trying to measure the plasma response (using the rake probe primarily as well as magnetic and SOL sensors) to the biasing.

Data analysis for this data is still ongoing, but there is some amount of low-pass filtering observed between the input bias voltage to the rake probe floating voltages.

For the majority of the shots today, only section 2 shells T&B were retracted. See the lab notebook for specifics.

CPCI_40 (north rack) had several issues throughout the day. Resetting all racks fixed the issue.

probe_name	electrode	measure	gain	address	notes
rake probe, S2	T1, top	103	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_65
rake probe, S2	T2, top	103.5	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_66
rake probe, S2	T3, top	104	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_67
rake probe, S2	T4, top	104.75	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_68	shorted
rake probe, S2	T5, top	104.75	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_69	shorted
rake probe, S2	T6, top	105.5	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_70
rake probe, S2	T7, top	106	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_71
rake probe, S2	T8, top	106.5	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_72
rake probe, S2	T9, top	107	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_73
rake probe, S2	T10, top	107.5	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_74
rake probe, S2	T11, top	108	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_75
rake probe, S2	T12, top	108.5	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_76
rake probe, S2	T13, top	109	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_77
rake probe, S2	B1, top	103	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_78
rake probe, S2	B2, top	105.5	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_79
rake probe, S2	B3, top	107.5	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_80
rake probe, S2	B4, top	109	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_81

Thursday March 12 2020 4:21 pm Brooks 104713-104745 DP probe conditioning

These experiments took place on 11/20/19. Conditioning of DPS2, DPS4, DPS7, and DPS9. Probes floating, grounded, and then driven. TPS5 at 103 cm. High speed camera operating for some shots.

probe_name	electrode	measure	gain	address	notes
DPS2		Preamp	1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_92
DPS2		Current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_04
DPS2	R	Voltage	213.3	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_03
DPS2	L	Voltage	213.9	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_01
DPS4		Preamp	1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_91
DPS4		Current	-10	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_86
DPS4	L	Voltage	213.0	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_85
DPS4	R	Voltage	212.7	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_84
DPS7		Preamp	1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_89
DPS7		Current	-10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_95
DPS7	R	Voltage	213.2	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_94
DPS7	L	Voltage	213.5	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_96
DPS9		Preamp	1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_90
DPS9		Current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_09
DPS9	R	Voltage	212.8	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_06
DPS9	L	Voltage	213.1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_08
DPS2		Amp_current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_85	confirm channel mapping
DPS4		Amp_current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_86	confirm channel mapping
DPS7		Amp_current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_87	confirm channel mapping
DPS9		Amp_current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_88	confirm channel mapping

Triple probes:

probe_name	electrode	measure	gain	address	notes
Triple probe, TPS5	V+	Pos. V	106.55	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_95
Triple probe, TPS5	Vf	Floating V	-106.82	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_81
Triple probe, TPS5	Isat	Sat. Current	-0.4762	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_82

Thursday March 12 2020 4:10 pm Brooks 104813-104979 Active feedback in the plasm and with the quadrature array

These experiments took place on 11/25/19 to 11/27/19. Active feedback with probes inside the prasma. DPS2, DPS4, DPS7, and DPS9 (co and counter Ip electrode) at 104 cm. TPS5 at 104 cm.

probe_name	electrode	measure	gain	address	notes
DPS2		Preamp	1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_92
DPS2		Current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_04
DPS2	R	Voltage	213.3	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_03
DPS2	L	Voltage	213.9	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_01
DPS4		Preamp	1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_91
DPS4		Current	-10	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_86
DPS4	L	Voltage	213.0	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_85
DPS4	R	Voltage	212.7	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_84
DPS7		Preamp	1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_89
DPS7		Current	-10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_95
DPS7	R	Voltage	213.2	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_94
DPS7	L	Voltage	213.5	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_96
DPS9		Preamp	1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_90
DPS9		Current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_09
DPS9	R	Voltage	212.8	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_06
DPS9	L	Voltage	213.1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_08
DPS2		Amp_current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_85	confirm channel mapping
DPS4		Amp_current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_86	confirm channel mapping
DPS7		Amp_current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_87	confirm channel mapping
DPS9		Amp_current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_88	confirm channel mapping

Triple probes:

probe_name	electrode	measure	gain	address	notes
Triple probe, TPS5	V+	Pos. V	106.55	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_95
Triple probe, TPS5	Vf	Floating V	-106.82	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_81
Triple probe, TPS5	Isat	Sat. Current	-0.4762	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_82

Thursday March 12 2020 3:52 pm Brooks 104980-105082 DP testing. 3 kHz testing from each probe. Then DPS7 radial scan.

104980-105082 These experiments took place on 12/3/19 and 12/4/19. First experiment: DPS2, DPS4, DPS7, and DPS9 (co and counter Ip electrode), one at a time, driving 3 kHz oscillations to provide a low frequency map of the sourced current. TPS5 at 104 cm. All DPs at 104 cm. Second experiment: Radial scan of DPS4 (co and counter Ip electrode) from 103 cm to 106 cm. TPS5 also radially scanned with probe.

probe_name	electrode	measure	gain	address	notes
DPS2		Preamp	1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_92
DPS2		Current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_04
DPS2	R	Voltage	213.3	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_03
DPS2	L	Voltage	213.9	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_01
DPS4		Preamp	1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_91
DPS4		Current	-10	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_86
DPS4	L	Voltage	213.0	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_85
DPS4	R	Voltage	212.7	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_84
DPS7		Preamp	1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_89
DPS7		Current	-10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_95
DPS7	R	Voltage	213.2	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_94
DPS7	L	Voltage	213.5	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_96
DPS9		Preamp	1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_90
DPS9		Current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_09
DPS9	R	Voltage	212.8	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_06
DPS9	L	Voltage	213.1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_08
DPS2		Amp_current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_85	confirm channel mapping
DPS4		Amp_current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_86	confirm channel mapping
DPS7		Amp_current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_87	confirm channel mapping
DPS9		Amp_current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_88	confirm channel mapping

Triple probes:

probe_name	electrode	measure	gain	address	notes
Triple probe, TPS5	V+	Pos. V	106.55	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_95
Triple probe, TPS5	Vf	Floating V	-106.82	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_81
Triple probe, TPS5	Isat	Sat. Current	-0.4762	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_82

Thursday March 12 2020 3:36 pm Brooks 105083-105125 Active feedback in the plasma with the quadrature array

These experiments took place on 12/6/19. These experiments were by active feedback experiments inside the plasma. These results were largely exploratory. All four quadrature probes at 104. All shells in. TPS5 at 104 cm.

Probe channels/gains:

probe_name	electrode	measure	gain	address	notes
DPS2		Preamp	1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_92
DPS2		Current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_04
DPS2	R	Voltage	213.3	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_03
DPS2	L	Voltage	213.9	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_01
DPS4		Preamp	1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_91
DPS4		Current	-10	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_86
DPS4	L	Voltage	213.0	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_85
DPS4	R	Voltage	212.7	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_84
DPS7		Preamp	1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_89
DPS7		Current	-10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_95
DPS7	R	Voltage	213.2	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_94
DPS7	L	Voltage	213.5	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_96
DPS9		Preamp	1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_90
DPS9		Current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_09
DPS9	R	Voltage	212.8	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_06
DPS9	L	Voltage	213.1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_08
DPS2		Amp_current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_85	confirm channel mapping
DPS4		Amp_current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_86	confirm channel mapping
DPS7		Amp_current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_87	confirm channel mapping
DPS9		Amp_current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_88	confirm channel mapping

Triple probes:

probe_name	electrode	measure	gain	address	notes
Triple probe, TPS5	V+	Pos. V	106.55	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_95
Triple probe, TPS5	Vf	Floating V	-106.82	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_81
Triple probe, TPS5	Isat	Sat. Current	-0.4762	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_82

Thursday March 12 2020 3:31 pm Brooks 105126-105187 Active feedback in the plasm with centertapped transformers and with the quadrature array

12/10/19 These experiments took place on 12/10/19. These experiments were by active feedback experiments inside the plasma. The transformers were set in the centertapped configuration. All four quadrature probes at 104. All shells in. TPS5 at 104 cm.

Probe channels/gains:

probe_name	electrode	measure	gain	address	notes
DPS2		Preamp	1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_92
DPS2		Current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_04
DPS2	R	Voltage	213.3	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_03
DPS2	L	Voltage	213.9	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_01
DPS2		SecondaryCurrent	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_02
DPS4		Preamp	1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_91
DPS4		Current	-10	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_86
DPS4	L	Voltage	213.0	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_85
DPS4	R	Voltage	212.7	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_84
DPS4		SecondaryCurrent	10	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_87	right channel?
DPS7		Preamp	1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_89
DPS7		Current	-10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_95
DPS7	R	Voltage	213.2	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_94
DPS7	L	Voltage	213.5	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_96
DPS7		SecondaryCurrent	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_93
DPS9		Preamp	1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_90
DPS9		Current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_09
DPS9	R	Voltage	212.8	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_06
DPS9	L	Voltage	213.1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_08
DPS9		SecondaryCurrent	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_07
DPS2		Amp_current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_85	confirm channel mapping
DPS4		Amp_current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_86	confirm channel mapping
DPS7		Amp_current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_87	confirm channel mapping
DPS9		Amp_current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_88	confirm channel mapping

Triple probes:

probe_name	electrode	measure	gain	address	notes
Triple probe, TPS5	V+	Pos. V	106.55	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_95
Triple probe, TPS5	Vf	Floating V	-106.82	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_81
Triple probe, TPS5	Isat	Sat. Current	-0.4762	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_82

Friday December 20 2019 11:02 pm Levesque/Wei/Boyes 105201-105231 Hard x-ray (HXR) localization studies

Ran plasmas similar to our runaway/slide-away electron experiments from August 13. Made plasmas that had strong hard x-ray (HXR) emission by operating at the lowest fill pressures we could while still achieving breakdown, which drifted a bit during the day. Successful fill pressures for this experiment were in the range of 50-53 uT. Fast camera data were taken, simultaneously looking in both toroidal directions (co-Ip and counter-Ip) from Section 10 by using 2 mirrors and splitting the view. Speckles were seen in some images. No obvious signs of synchrotron radiation were apparent, though the radiation spectrum should be far longer wavelength than what the camera could observe based on quick spectral estimates, if the radiation is even present at all. The camera could see standard MHD oscillations in both directions; this viewing configuration may be appropriate for n=1 mode studies.

Attempted to poloidally localize the source of HXR emission by using our two HXR detectors: one was in a fixed position at Section 2.5 attached to the outboard midplane limiter, without additional shielding; the other was placed directly above the first detector, and had 8 layers of thin lead tape shielding over its diode/scintillator box and collimator. The shielding over the 2nd box and collimator was not sufficient to prevent HXRs from penetrating for these plasmas 96 x-rays still apparently passed through the shielding, and did not strictly come through the collimator opening. This was determined by pointing the detector above the tokamak, away from possible source locations. Pointing the detector in 3 poloidal directions (above/away from tokamak, straight inward above midplane, and slightly downward toward midplane) gave qualitatively similar results. Thicker shielding is required to perform this type of poloidal/radial localization experiment.

The possibility of significant electronic noise pickup or coupling between the detectors has not fully been excluded, but it appears that noise being the source of the main signals is very unlikely given the tests performed so far. Moving the shielded detector about 5 feet farther away from the machine produced small signal levels that were consistent with a ~1/R^2 drop-off in intensity (e.g. shot 105222). The power and signal cables for the lead-shielded detector needed to move for this test, so we cannot yet fully rule out electronic coupling between the shielded/unshielded detector as producing the measurement in the shielded detector; more rigorous tests should be done.

Since the poloidal localization was not feasible with the prepared setup, we removed the lead shielding and changed to toroidal asymmetry measurements. Here, both detectors had the same configuration, with one at Section 2.5 limiters (unmoved) and the second at Section 8.5 limiters. Toroidal asymmetry was observed during HXR bursts and rotating MHD activity, but this has not yet been quantified.

This shot campaign did not have clear HXR avalanches that appeared at the end of some shots in the previous HXR experiment (such as 103061 and 103084). Star shots were: 105211, 105214, 105215, 105227, 105228.

Friday December 20 2019 1:15 pm Brooks 105232-105292 DPS2,4,7,9 FB testing at 106 cm

Overview: In these tests, the DPS2,4,7,9 array was placed at 106 cm, each of their two electrodes were shorted, and the probes were ground referenced. The GPU FB system drove an n=1 phase locked signal to match the mode's (n=1 analysis on the FB_S4 array) frequency. Phase offsets were applied to scan the various phase angles. Auto-FB was done prior to the experiments to determine the optimal phase angles.

Setup: All shells were fully inserted, S9B was 1 cm retracted, S4B half in, and S7B full retracted. TPS5 was at 106 cm. AETechrons and QSCs were used in conjunction with the Stangenes trasformers in parallel:parallel configuration to provide power to the probes. GPU provided the FB signals for the amplifiers.

Results: Are still tentative, but some mode suppression was observed.

Thursday March 12 2020 2:48 pm Brooks 105232-105292 Active feedback in the plasma edge with the quadrature array

These experiments took place on 12/12/19. These experiments were by active feedback experiments in the plasma edge with all four quadrature probes. All shells in except S4B (half) and S7B (full out). DPS2, DPS4, DPS7, DPS9 were placed at 106cm. TPS5 at 106 cm.

Probe channels/gains:

probe_name	electrode	measure	gain	address	notes
DPS2		Preamp	1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_92
DPS2		Current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_04
DPS2	R	Voltage	213.3	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_03
DPS2	L	Voltage	213.9	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_01
DPS2		SecondaryCurrent	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_02
DPS4		Preamp	1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_91
DPS4		Current	-10	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_86
DPS4	L	Voltage	213.0	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_85
DPS4	R	Voltage	212.7	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_84
DPS4		SecondaryCurrent	10	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_87	right channel?
DPS7		Preamp	1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_89
DPS7		Current	-10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_95
DPS7	R	Voltage	213.2	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_94
DPS7	L	Voltage	213.5	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_96
DPS7		SecondaryCurrent	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_93
DPS9		Preamp	1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_90
DPS9		Current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_09
DPS9	R	Voltage	212.8	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_06
DPS9	L	Voltage	213.1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_08
DPS9		SecondaryCurrent	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_07
DPS2		Amp_current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_85	confirm channel mapping
DPS4		Amp_current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_86	confirm channel mapping
DPS7		Amp_current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_87	confirm channel mapping
DPS9		Amp_current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_88	confirm channel mapping

Amp_current 10 HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_88 confirm channel mapping

Triple probes:

probe_name	electrode	measure	gain	address	notes
Triple probe, TPS5	V+	Pos. V	106.55	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_95
Triple probe, TPS5	Vf	Floating V	-106.82	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_81
Triple probe, TPS5	Isat	Sat. Current	-0.4762	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_82

Friday December 20 2019 12:22 am Chandra 105309-105333 GPU Feedback

The theme of the run was evaluating position control on the GPU. The major radius was perturbed via a preprogrammed waveform to the control coils from the GPU, to enhance or cancel out the vertical field (by running the top and bottom coilsets in opposition). The goal was to compare the plasma position as observed by the magnetic sensors, and reconstructed by the EUV.

The plasma equilibrium was observed via a simple, 6-basis tomography algorithm, ( +/- 2cm and 0cm in the x direction, +/- 2cm in the y direction). A plot comparing the reconstructions will be appended. Rough agreement in the spatial and temporal dynamics between the EUV and magnetics is observed. However, offsets in the data may complicate the use of this technique as a calibration tool for the EUV or R_Maj calculation.

Friday December 20 2019 1:08 pm Brooks 105337-105393 Auto FB testing with DPS2,4,7,9 and TPS5

Overview: In these tests, we explored the ability of the DPS2,4,7,9 probe array to do "Auto FB" (automatically drive or suppress MHD modes in HBT-EP). The idea would be that oscillating currents, due to the modes, would be sourced and sinked with n=1 phasing in the four probes. By driving a DC offset in voltage (~50V), these current oscillations (on top of the DC current) would be on the order of 5 to 10 amps. This may be enough to provide some passive stabilization of the mode.

Setup: All shells were fully inserted, S9B was 1 cm retracted. TPS5 was at 103 cm. AETechrons and QSCs were used in conjunction with the Stangenes trasformers in parallel:parallel configuration to provide power to the probes. GPU provided the AWG for the amplifiers.

Results: 1. Of the 30 something shots I performed, easily half of them disrupted instantly when the first current pulse stepped up. 2. Those that did not disrupt tended to be shifted toward the HFS instead of the LFS. 3. Of those that did not disrupt, the current pulse appears to have drastically slowed the mode's rotation to the point where it briefly locks (and presumably often disrupts). However, this behavior isn't 100% consistent. 4. It's unclear if the mode amplitude ``does actually'' change during the periods of high probe current. It might go up a little, suggesting that the lower rotation frequency is less stable. I believe I've adequately compensated for my Mirnov sensor subtraction of low-frequencies and this is not an issue in this data set.

Thursday March 12 2020 2:34 pm Brooks 105393-105490 Frequency dep. characterization of probe's influence on the plasma. With Mach probes

These experiments took place on 1/8/2020 and 1/9/2020. DPS9 driven electrode at 104cm. DPS4 is poloidal Mach probe and is radially scanned. DPS7 is toroidal Mach probe and is radially scanned. TPS5 is radially scanned. Rake probe at 102.3 cm (103 cm to 109 cm). All shells in except S2 (because of rake probe).

Performed both square/step pulses and single frequency sine waves with each probe (at a range of frequencies). Both step and sine waves are used to characterize the plasma and MHD's response. Step is good for equilibrium and transience. Single frequency is good for generating Bode plots.

Probe channels/gains:

probe_name	electrode	measure	gain	address	notes
DPS2		Preamp	1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_90
DPS2		Current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_04
DPS2	R	Voltage	213.3	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_03
DPS2	L	Voltage	213.9	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_01
DPS2		SecondaryCurrent	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_02
DPS4		Preamp	1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_91
DPS4		Current	-10	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_86
DPS4	L	Voltage	213.0	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_87
DPS4	R	Voltage	212.7	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_84
DPS4		SecondaryCurrent	10	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_85	right channel?
DPS7		Preamp	1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_92
DPS7		Current	-10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_95
DPS7	R	Voltage	213.2	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_94
DPS7	L	Voltage	213.5	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_96
DPS7		SecondaryCurrent	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_93
DPS9		Preamp	1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_89
DPS9		Current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_09
DPS9	R	Voltage	212.8	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_06
DPS9	L	Voltage	213.1	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_08
DPS9		SecondaryCurrent	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_12:INPUT_07
DPS2		Amp_current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_85	confirm channel mapping
DPS4		Amp_current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_86	confirm channel mapping
DPS7		Amp_current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_87	confirm channel mapping
DPS9		Amp_current	10	HBTEP2::TOP.DEVICES.SOUTH_RACK:CPCI_10:INPUT_88	confirm channel mapping

Rake and triple probes (these should be valid for my previous experiments for the past several months as well):

probe_name	electrode	measure	gain	address	notes
Triple probe, TPS5	V+	Pos. V	106.55	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_95
Triple probe, TPS5	Vf	Floating V	-106.82	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_81
Triple probe, TPS5	Isat	Sat. Current	-0.4762	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI:INPUT_82
rake probe, S2	T1, top	103	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_65
rake probe, S2	T2, top	103.5	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_66
rake probe, S2	T3, top	104	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_67
rake probe, S2	T4, top	104.75	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_68	shorted
rake probe, S2	T5, top	104.75	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_69	shorted
rake probe, S2	T6, top	105.5	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_70
rake probe, S2	T7, top	106	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_71
rake probe, S2	T8, top	106.5	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_72
rake probe, S2	T9, top	107	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_73
rake probe, S2	T10, top	107.5	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_74
rake probe, S2	T11, top	108	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_75
rake probe, S2	T12, top	108.5	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_76
rake probe, S2	T13, top	109	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_77
rake probe, S2	B1, top	103	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_78
rake probe, S2	B2, top	105.5	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_79
rake probe, S2	B3, top	107.5	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_80
rake probe, S2	B4, top	109	101	HBTEP2::TOP.DEVICES.NORTH_RACK:CPCI_40:INPUT_81

Tuesday January 21 2020 2:06 pm Saperstein 105491-105667 e-gun current collection scan

This report covers the run campaign made over 3 days last week. The purpose was to determine if the e-gun was collecting a significant amount of current from the plasma by retracting it out and seeing if the current collected by the jumper between sections 10-1 dropped.

Day 1, 1/15/20 [105491-105551]:

This run day was dedicated mostly to shot development with all shells in and all probes retracted. The goal was to get an outboard shot with decent mode activity. Unfortunately, the plasma was very inconsistent throughout the day, and a relatively consistent shot that met these criteria wasn92t found until around the end of the day (see 105540). Some shots with the section 10 shells retracted were made at the end of the day (see 105546).

Day 2, 1/16/20 [105552-105623]:

This run day was actually dedicated to performing a scan of the e-gun position. Shots 105552-105569 were dedicated to clean up and shot development (model shot 105565). The section 10 shells were then retracted, and shots 105570-105599 were dedicated to getting shots with varying e-gun positions (model shots 105572, 105574, 105577, and 105584). The positions used varied from 0 -> ~15 mm retracted out. We stopped getting breakdown somewhere between ~10.3 to 15 mm out. Shots 105600-105622 were also dedicated to varying the e-gun position (model shots 105602, 105606, and 105617), but this time, bias probe 7 was inserted to 103 cm and biased sinusoidally about ~0 V. This was done because previously the jumper wasn92t seeing much high-frequency current, and the probe was an easy to track current that could be investigated with e-gun position.

Day 3, 1/17/20 [105624-105660]:

The purpose of this run day was similar to the previous day92s, but this time all shells except sections 1, 4, and 8 were retracted, in order to best emulate the shots conditions Brooks was using when he noticed a significant amount of current in jumper A. This run day was made difficult by the fact that we weren92t getting perfect breakdown with so many shells retracted. We couldn92t reproduce the same shot style from the previous day, and the shot settings we did end up using had a bifurcation that led to one of two differing shot styles. In addition to getting some shots with the e-gun retracted, we also got some with it slightly pushed in.

Rian Chandra also made 3 shots during the run day (105636-105638) in testing the GPU system.

Friday January 31 2020 11:22 am Chandra 105674-105805 GPU: Position Control

The goal of this campaign was evaluation of various feedback control laws for major radius control.

The first phase of the run was marred by storage capacity issues on Spitzer, which have been resolved. The second phase tested further open-loop operations, and gathered fast-camera data on shots 105727-29, visually verifying that the plasma is being pushed inward and outward, in line with the request. The third phase tested various gains for proportional and integral control. Results were inconclusive, and further work here is needed.

One final result of note: shots 105742-50 were crates only, in which a triangle wave signal was sent into both the CPCI and GPU, and it is noticed that when the GPU "misses" samples, there is a corresponding skip in the time-base, which manifests itself as a progressive phase-slip.

Friday January 31 2020 1:19 pm Saperstein 105809-105895 HFS limiter current source

This run report covers the 2 run days dedicated to determining the current source potential of the HFS limiters. Obviously the HFS limiters will collect current when the plasma is inboard limited, but it92s important to know at what major radius the HFS limiters stop collecting a significant amount of current. All probes were out for this campaign. And all shells were in, except for the cases where section 10 was retracted.

Day 1, 1/29/2020 [105809-105873]:

This run day covers shot development for two different types of shot styles. The first [105822-105832] was a style that was inboard limited, and the second [105833-105864] was a style that had a large sweep over the major radius. Shots 105809-105821 were cleanup shots that had section 10 shells in. Following these shots, the section 10 shells were retracted out for the rest of the run day. Shots that had something close to the first shot style we wanted include 105827-105832, with 105829 being the model shot. Shots that had something close to the second shot style we wanted include 105856-105864, with 105856 being the model shot. It92s worth noting that the shot development for the R sweep shot style required lowering Ip considerably. This is because sweeping R above ~92 cm drastically drops q as the plasma shrinks. In order to compensate for this, we had to drastically raise q throughout the shot. This led to a rather interesting shot style that had a flattop q of ~6, which had clear 6/1 modes with an amplitude of ~1 G. This shot style also lasts relatively long, and sometimes doesn92t even have what could be identified as a disruption. The major radius also jumped out as far as ~104 cm in some cases [105856,105857,105859].

The next part of the run day was dedicated to trying to correct for a shortcoming in the Rogowski current measurements for jumper A [105865-105867]. The large magnetic pickup the Rogowski was seeing interfered with the measurements, likely introducing a linear drift in the data. In order to correct for this, I tried to change some of the digitizer timing settings, so that the Rogowski would start recording data prior to the TF firing. Unfortunately, I was having trouble getting the timing right, so I just returned all the settings back to normal.

The last part of the run day was just dedicated to seeing how long I could get this R sweep shot style to last for [105867-105873]. The shot style by default already had the plasma lasting until ~10 ms, and I wanted to see how long we could get it to last for, so I we tailored R to stay within the boundaries that don92t change q. The longest I could get the plasma to last for though, was only ~12 ms [105869].

Day 2, 1/30/2020 [105874-105895]:

The purpose of this run day was to retake some R sweep shots, except this time, with a shunt also measuring the jumper A current in series (this was to correct for the linear-ish drift the Rogowski was seeing). I moved the temporary shunt I had previously set up for jumper B (sec 3-4) to jumper A (sec 10-1). This required replacing the jumper B cable with a new cable. This new cable juts out from the machine ~7.59292. I then replaced the old jumper A cable with the 93new94 one that had the 20 mOhm shunt in series. The shunt leads were run to a Jensen, to separate the grounds, that then fed to the S rack CPCI 12 input 11. It92s also worth noting that the old jumper A cable jutted out from the machine ~7.59292.

The first shot of the day was a vacuum shot [105874], but apart from that, all other shots pretty consistently still had the R sweep style from the previous run day, without any cleanup or shot development necessary. The first several shots (including the vacuum shot) had the section 10 shells still retracted [105874-105883], then the shells were inserted again between shots [105884-105889], but after that the shells were retracted again for the rest of the day [105890-105895]. Nothing settings wise was changed between any of these shots, with the exception of the sec 10 shell positions, I just took many shots for statistical purposes. The model shot for the day was likely 105891.

Tuesday March 10 2020 10:58 am Stewart 105896-105918 Clean-up Shots

These were the first shots after the up to nitrogen for the Raman and Rayleigh calibration of the new Thomson scattering collection optics. The first plasma shot 105898 had a low fill pressure and a large amount of toroidally asymmetric hard X-rays. The next ~20 shots were all devoted to clean-up, with the pulses lasting beyond the 6ms mark by the end of the run day (see shot 105917). Therefore, Thomson testing on plasma shots will begin during the next run day.

Thursday March 12 2020 11:15 am Stewart 105919-105975 Thomson Testing/Probe Clean-up

Combined run report for 03/10/20 and 03/11/20. The first day was devoted to testing the Thomson scattering system on plasmas with shots 105930-105949 having Thomson measurements on multiple scopes (all L-mode). Small note: channel C on polychromators 1,3,5,7,8,9,10 were off up to shot 105936. Laser light was seen on several of the radial positions including: 91cm, 95cm, 97cm, 99cm, 101cm, and perhaps 103cm. It was also found that the high voltage to polychromator #8 had been wired incorrectly and has been rewired for future runs (channels, A, C, and D were swapped on this poly). Throughout the day, several of the shots varied between being centered and migrating inboard and outboard with Thomson signal going up or down on the "edge" Thomson positions. However, in order to get the full picture, the spectral calibration will have to be performed and a high to low voltage conversion used. Even so, the results (Thomson signal on 6-7 of the channels) are promising.

The second day included full insertion of the bias probe, triple probe, and rake probe. One of the voltage dividers on the rake probe needed to be repaired, so shots up to 105961 will show abnormally high voltages on the third probe tip. After shot development and proper placement of the probes, we are ready for the full Thomson experiments in L and H-modes (along with an edge triple probe scan).

Friday March 20 2020 3:52 pm Brooks_/_Chandra 106047-106073 EUV/GPU: Position Control

The theme of this run campaign was verifying R_maj control with the EUV-GPU feedback system.

The operator was Brooks, the Physics leader was, over a remote connection, Chandra. The shotstyle was 101430.

No major hardware changes, all shells in, all probes out.

Based on a transfer function model incorporating a simple R-L circuit model for shell-coil mutual inductance, proportional gains were found and tested here.

Results were disappointing on two fronts: Early in time the EUV saw a persistent and highly repeatable +1cm offset from the Rogowski. Once the demand function was turned on, R_maj slowly evolved to meet it, but with much worse time constant than modeled.

Comparison with the SXR fan array in terms of filtered emissivity profiles and impact parameter weighted averages seem to suggest that the early-time offset may be physical, however, gain miscalibration can also not be ruled out, especially in light of how asymmetric the EUV profiles appear to be, at low amplitudes. Hopefully the upcomming system hardware calibration will fix this. This effect was not visible in offline tests, as the Rogowski major radius was also pushed up early in time, in the shot analyzed.

Later in time, we find that model is failing to capture some sort of a damping effect, in which the plasma is responding much slower, and seemingly only to lager amplitude currents. This effect was similarly obscured in offline analysis, as the shot in question was a ±40A Square wave.

Plots will be attached.

Thursday July 2 2020 12:03 am Wei/Levesque 106074-106080 First run after pandemic shutdown

Today we tried to start up the tokamak following the COVID-19 shutdown. The chamber had been left under vacuum with pumps off for about 3.5 months. Pressure had risen to around 1.4 Torr before pumping was resumed last week. We baked for about 5 days after turning on the cryos. Chamber base pressure is now strongly dominated by deuterium gas, which seems to be leaking in from the puff line. Base pressure is still enough for plasma operation since the background is our working gas anyway.

While setting up, we found the OHST voltage divider was low on insulating oil. This may have been at the same level during previous runs -- we can't recall. Refilling the oil disturbed the electrical connections, so we ran OH-only first shots to verify that the voltage divider was unchanged.

After several lower power TF-only shots (3.1kV, 5.1kV, 6.1kV) that had no significant noticeable issues, we starting charging for full vacuum 106080 shot with the TF set to 6.4kV (below our imposed 6.5kV limit). At or near the full charge on the TF, there was an arcing event on the TF bank. Based on our observations of basement conditions and analyzing camera footage, the following is our hypothesis about what happened:

While setting up to run, one of the rope dump shorting rods appears to have been stuck down, with the conducting tip near the hot rail but probably not actually touching it. Camera footage shows that the horizontal bar was far from level as soon as it was secured before starting the run, with the lower end being a site of the later arcing during shot 106080.

Before the arc, TF-only shots 106077-106079 showed no indication of problems on the bank current traces. When the TF was nearly fully-charged for 106080, an arc initiated on the end of the TF bank closest to the exit door. This may have been caused by the TF bank voltage reaching a threshold needed to arc over the presumed small gap between the hot bar and the rope dump rod. Water/CuSO₄ solution in the short water resistor immediately erupted upward, leaving the glass tube empty. The solution splashed down over much of that section of the bank about one second later, causing additional arcing/discharge of the bank. Prior to the arc, there was no steam coming from the subject water resistor, and the bank appeared to charge at its usual rate, so we think there was no direct contact of the rope dump prior to the arc flash.

Some minor damage needs to be repaired on the TF bank, followed by rigorous cleaning and inspection of the bank. This would take a couple of days with full staffing of the lab, but it might take a little longer since we can't have as many hands helping at once. Jim is well underway on the repair/cleanup. Capacitors with blown fuses that we've found so far are temporarily shorted for safety. We plan to start operating again early next week if possible.

Monday July 6 2020 11:09 pm Levesque/Wei 106081-106093 Recommissioning TF

The TF bank has been restored to operation. 11 fuses were replaced along the rows closest to the entrance, along with 2 sets of polyethylene+mylar sheets.

We started charging the TF to only 500V for initial tests. Strange behavior was seen in the field measurement traces, which made us cautious about proceeding. We later found the behavior is caused by a problem with the TF's A14 digitizer -- at least one of its recording/storage bits is faulty. The problem is very clear when viewing the raw digitized data, but the source is not obvious when viewing the integrated result. This problem has been present for quite some time, but its effect is much more significant at low TF voltages. Roughly correcting the raw data gives a result that's reasonably close to the expectation, so we continued raising the TF voltage in ~500V increments. We stopped at 3100V for today, and will continue to full voltage tomorrow as long as no further problems arise.

We also quickly attempted to measure the toroidal field using a Rogowski coil that was already partly set up on the machine, but this was unsuccessful. We did not spend time trouble-shooting this, nor did we try to estimate the signal levels we would receive. A J221 trigger and North CPCI timing were changed temporarily to accommodate this test, but the J221 didn't seem to respond to the programmed change.

Tuesday July 7 2020 5:56 pm Wei 106094-106119 TF bank testing & clean up shots

Following yesterday’s tests we gradually increased TF voltage from 3100V to 6100V in 500V increment, during which no problem was found. We then tested the OH & VF banks, and after making sure all banks were working correctly we turned on e-gun and were able to make plasma. 106106 was the first plasma shot taken since the covid-19 lockdown. 106106-106119 were clean up shots. We were able to gradually extend shot duration to around 3ms. The USB spectrometer was working and we noticed more lines than we usually see during normal plasma operation. D-alpha was also higher than usual, which might be because of higher background deuterium level or carbon impurity since C-II line (656.976nm) is very close to D-alpha (655.949nm). Base pressure was high throughout the day at around 6.6e-6 torr. More clean up shots are needed before running experiments.

Wednesday July 8 2020 5:33 pm Levesque/Wei 106120-106152 Clean up shots & testing control coils

We continued doing clean up shots throughout the day. Background pressure was still high at around 6E-6 torr. HXR level was high for most of the shots. It looks like plasma condition has been gradually improving, although more clean up shots are still needed. We have also tested using CCs to change plasma motion and we were able to move it inward or outward for a few cm during the pulse.

The TF rope dump got stuck in the plastic guiding cylinder again when we setup the machine this morning, but fortunately we noticed it before the first shot was fired. People should be extra cautious and make sure the shorting bars are properly lifted before charging the banks.

Thursday July 9 2020 5:28 pm Wei 106153-106198 Clean up & testing CCs

We continued doing clean up shots and testing control coil pulses throughout the day. Plasma condition has been gradually improving. Base pressure decreased from 5.8E-6 to 2.0E-6 by the end of the day, and D-alpha level is still higher than normal. A few shots during the day (notably 106177) show multimode activity, but unfortunately EUV and North Rack CPCI were not turned on between 106153 and 106176.

106181--106188 (ref: 106184 & 106185) are a list of shots for testing CC VF pulses at different time using various pulse strengths, although their shot styles are different from the one shown in the run proposal.

Jim turned on the RGA after the last shot to check the impurity level after 4 days of clean-ups.

Monday July 13 2020 11:03 am Wei 106200-106234 Clean up shots & shot development

Base pressure was still high at around 2E-6. RGA result from previous day shows overwhelmingly A=4 component so the plasma should be clean if there’s no He leakage. We were having trouble maintaining a stable puff pressure in the morning. Jim tried fastening the values on the other two puff lines and the puff line was working fine afterward.

Interesting shots: 106212, 106227, 106228

Wednesday July 15 2020 11:11 pm Levesque/Wei/Chandra 106289-106314 TF current measurement and TF crowbar testing

The programmed trigger for the TF crowbar ignitron hasn't been working reliably for quite a few years (over 20000 shots) due to CAMAC problems, however there has consistently been a sharp inflection point in the raw toroidal field measurement at around +135 to +150ms, indicating some non-smooth behavior in the TF current circuit. Today we directly measured the TF coil current and the current through the crowbar ignitron to determine if the crowbar is passively triggering to produce this change in TF current evolution. No current at all was found to flow through the crowbar ignitron for the hardware configuration we've been running for years. The inflection in the field measurement is due to conduction through the main (TF start) ignitron stopping when the current reaches near zero after a half-period of LRC oscillation.

We were able to use an Avionics delay generator to trigger the TF crowbar with a delayed signal from the TF ST trigger. The TF crowbar is now working as intended, triggering around +30ms, bypassing the broken software trigger. The delay generator was taken from the shaping/biasing bank trigger circuit, thus the shaping/biasing bank cannot be used with consistent programmable timing until a replacement trigger is installed.

Crates-only tests were also done throughout the day for testing signal propagation through the GPU feedback system.

Thursday July 16 2020 7:33 pm Chandra 106235-106357 GPU-EUV Verification

The theme of the run was verifying various realtime scripts toward EUV tomographic control, and vacuum tank cleanup. Namely: Verifying that changing the cycle time on the GPU produces expected results, verifying realtime dark channel subtraction, verifying real time tomographic tracking of the equilibrium, and real time mode discrimination. Cleanup was largely successful, partial pressures continue to fall and the run ended with a base pressure of 1.2e-6. Shots appear largely repeatable, and reproducing pre-shutdown results, see attached. Realtime dark channel subtraction appears to be successful. During crates only shots, a signal generator was used probe the difference between the GPU and CPCI timebase. Something like a drift may be observed in some cases Tomography testing was largely unsuccessful due to issues in correctly transferring the matrix of basis functions onto the GPU threads. It is however worth noting that the algorithm was able to run in real time, during the shot.

Friday July 24 2020 12:29 pm Levesque 103235-106558 Fixed control coil current nodes in tree

Retroactively corrected named control coil current nodes in the hbtep2 tree from August 30, 2019 through present. This affects nodes with the format ".SENSORS.CC_CURRENTS:FB##_C#". Retrieving data from these nodes now returns measured CC currents in amps (instead of raw voltage to the digitizer), and Sections 8-10 are now pointing to the correct hardware channels on CPCI_40.

Friday July 24 2020 5:55 pm Saperstein 106559-106625 kinking v tearing campaign - day 1

The purpose of this run day was to develop a good kinking shot style with a flat R at ~92 cm and an edge q between 2.5-2.75. Some time towards the end was also spent developing a good tearing shot style. A good enough kinking shot style was found (star shots 106589, 106590, 106591), and they were surprisingly reproduceable, with me being able to reproduce this style ~30 shots later with 106619. Shots 106621-106625 were dedicated to doing some quick shot dev to get a good tearing shot. I was also able to maintain a constant major radius quite well as I raised the edge q.

Of note, two shots during this run day ran into problems with the diagnostics, shots 106559 and 106605. The diagnostics digitized in the same CPCI as that belonging to the OH and VF current diagnostics appeared to measure only small amounts of pickup and no sign of a plasma or the correct equilibrium fields. The reason for this is likely that there was an issue with triggering this CPCI, and that these diagnostics started recording at the wrong times. Despite these possible trigger time issues, the other shots in today’s run didn’t seem to have any timing issues. I compared the signals of TAR 5 and 6 over the course of most of the shots. TAR 5 is digitized in the same CPCI as the OH/VF current, while TAR 6 is not, so lags between the two measurements would have signaled a trigger timing issue. However, all shots apart from 106559 and 106605 had no issues.

Also of note, the puff battery was found still left on before startup, and the flux loops were also grounded.

Monday July 27 2020 5:17 pm Saperstein 106626-106695 kinking v tearing campaign - day 2

The purpose of this run day was to develop a consistent tearing shot style. Unfortunately, I had a very hard time maintaining a relatively flat major radius profile, and any shots that did manage to have decent R profiles (106674) weren’t reproduceable. All in all, this run day was relatively unsuccessful. Tomorrow I will try flattening the edge q profile more, and raising the puff pressure if the modes are too weak.

We also ran into problems with the OH/VF current digitizer again. Once again, two shots today had this issue. One of which was the first [vacuum] shot of the day (106626). Fortunately, the rest of the shots showed no issues.

Tuesday July 28 2020 1:21 pm Wei 106358-106558 NeuralNet-informed position control using preprogrammed 0/1 pulse

This run report summaries the 4-day run campaign between 7/20/2020 and 7/23/2020. The goal of the run was to develop a shot style with inward scooping, marginally stable major radius. Then use the VAE neural network to identify a warning time for the incoming disruptive event, and use the CCs to apply a pre-programmed VF pulse to see if the plasma survived the potential disruption. 59 shots were taken using CC VF pulse applied using different pulse strength, start time, and pulse duration.

Please refer to the corresponding run proposal and summary under the Machine Learning Studies page on the wiki.

Wednesday July 29 2020 5:22 pm Saperstein 106696-106760 kinking v tearing campaign - day 3

The purpose of this run day was to develop a consistent tearing shot style, with this day differing from the previous by focusing on flatter Ip and q profiles, and a raised puff pressure. Flattening Ip and q was relatively easy, but I did have some trouble raising the puff pressure. The puff time at the start of the day was 1050 us, which gave a puff pressure of ~69 uT. However, increasing the puff time to 1400 us only had the effect of raising the pressure to ~70 uT (106740). We tried refilling the gas line, and replacing the puff battery in order to resolve this issue, but this didn’t help. Instead, when I increased the puff time to 1800 us, I found that the pressure shot up to ~101 uT (106742). Lowering the puff time down to 1600 us, the pressure came down to ~80 uT (106743), which I continued to use for the rest of the run day.

I tried to develop the shot such that the nominal analysis period was aligned with that of the kinking shot style (i.e. 5-7 ms period), however this wasn’t really doable. I found that the mode amplitude consistently dropped off at ~4 ms, regardless of the edge q and major radius profiles, as well as the puff pressure. So instead I focused on the 2-4 ms region where there was significant mode activity. The shot style was then developed to get the ideal profiles during this region. The shot styles weren’t as consistent as the kinking shot styles, but I did manage to get some pretty good shots that should be useable (106744, 106746, *106754). An emphasis should be placed in 106754, which is probably the star shot of the day.

Further analysis of these shots is still required to verify that they are tearing shots.

Lastly, we once again had the same problem we’ve had for the last two run days where there was an issue with one of the digitizers for two of today's shots. Once again, the first shot had this issue (106696), and then another shot towards the end of the day also had this problem (106756).

Tuesday August 4 2020 2:14 pm Saperstein 106761-106787 kinking v tearing campaign - day 4

The purpose of this run day was recreate the “kinking” (106619) and “tearing” (106754) shot styles from last week, except with the rake probe (RP) inserted so that it can be used to look for the q=3 inversion radii that the EUV system is not well equipped to see [at the moment].

The beginning of the run day was dedicated to cleaning the RP, and trying to first reproduce the “tearing” shot style (106754). The probe started off inserted with its tip at 105.1 cm and took about 13 shots to clean (finished cleaning at about 106774), and the cleaning process went relatively smoothly. While it took longer to get the shot styles to match 106754 for later times, by 106774 the shots at least matched the tearing style up to about 4 ms, which covers our primary region of interest. When clean, we found that the mode amplitude during the 2-4 ms period of 106774 (RP at 105.1 cm) roughly matched that of 106754.

The rest of the shots of the day saw the RP probe position moved further into the plasma. Shots 106775-106776 had the RP tip at 102.8 cm. However, with the probe tip inserted so far, the mode amplitude during the 2-4 ms period dropped to almost nothing, resembling the mode activity seen in some of John and Ian’s “heavy probe density” shots (e.g. 103352). I then started retracting out the probe tip, looking for the boundary where the RP tip location starts significantly effecting the mode amplitude. Shots 106777-106780 saw the probe tip moved to 103.8 cm and did some very small amount of mode activity, but nothing compared to what was seen for the probe at 105.1 cm (106774). Moving the tip to 104.75 cm for shots 106781-106782 saw the return of mode amplitudes similar to the ones with the RP at 105.1 cm. Moving the tip to 104.3 cm, shots 106783-106874 continued to see good modes, but with the probe tip moved to 104.0 cm for shots 106785-106787, the mode amplitudes returned to something similar to the small modes seen with the probe at 103.8 cm. From this, we can conclude that a significant drop in mode amplitude occurs when the probe tip is located somewhere between 103.8-104.3 cm.

Today’s run day did end relatively early, so didn’t get to try and reproduce the “kinking” shot style (106619). This will instead be continued with the next run day.

Lastly, it’s worth noting that we did have some shots today that ran into the digitizer issues we’ve been seeing for the past several run days. However, the first [vacuum] shot of the day did not have these problems, which differs from what we have seen before. Three shots today had the digitizer issue. These include 106768, 106770, and 106784.

Sunday August 9 2020 1:58 pm Saperstein 106788-106854 kinking v tearing campaign - day 5

The purpose of this run day was two-fold. First, I wanted to continue getting shots with the rake probe (RP) inserted, this time reproducing the “kinking” shot style (106619), instead of the “tearing”. And second, I wanted to take the RP back out, reproduce the “tearing” shot style (106754), and then lower the edge q (while keeping the profile flat) so as to get a scan of the edge q for the “tearing” shot style. The section 2 shells were retracted out for ease of moving the RP.

Shots 106788-106814 were dedicated to reproducing the “kinking” shot style with the RP in at various positions. I started with the RP at 108.0 cm (106788-106800) so that the plasma could clean-up without the RP perturbing the plasma too much. Shots 106801-106808 then had the RP moved to 105.0 cm, so the that the probe tip could start be cleaned, and so that slow insertion of the probe into the plasma didn’t change the plasma evolution too much (the star shot from this set was 106805). We found that having the probe at this position didn’t seem to change the plasma evolution all that much (similar to the RP in the “tearing” case). Finally, the probe was moved in all the way to 103.0 cm for shots 106809-106814, close the furthest we could move it in without potentially damaging the probe. Surprisingly, this didn’t appear to change the plasma evolution as much as we though. The equilibrium changed slightly, but the mode activity remained active, something we didn’t see with the RP at 103.0 cm in the “tearing” case. The two star shots from this set were 106811 and 106814.

With the “kinking” shot style successfully reproduced in the presence of the RP, the focus of the run day then switched to reproducing and modifying the “tearing” shot style, starting with shot 106815. The RP was pulled out all the way to 115.0 cm. Reproducing something similar to the original “tearing” shot style (106754) was actually very quick, with the first shot (106815) being a relatively faithful reproduction. Shots 106815-106845 then tried to modify this shot style by changing the edge q to be lower than that in 106754, but still well enough above 3. This was done by raising Ip, with appropriate changes in the VFS and VFE settings to compensate for the effect on the major radius profile. Unfortunately, the shot development didn’t go very well. Making significant [consistent] changes to the edge q proved difficult, and even worse, we had a very hard time maintaining a flat major radius profile (which is absolutely necessary for a reliable analysis of the SOLC sensors), even during the 2-4 ms period of interest. The only star shot from this set was 106843, but not because it achieved a lower q while maintaining our other desired parameters (its q was actually almost exactly the same), instead it’s because it was actually a pretty faithful reproduction of 106754 that may be useful to have for verifying the consistency of the “tearing” shot style.

The last shots of the day (106846-106854), were dedicated to restarting the shot development of 106754, except this time, we attempted to lower the edge q resiliently by simply lowering the toroidal field, instead of raising Ip. This method seemed to work well enough, with 106853 being much closer to what I was looking for than any of the shots from the Ip varying shot dev. However, we didn’t start this shot dev process until late in the day, so only so much shot development was done. The star shot for this set would be 106853.

Lastly, I’d like to talk about the issues with one of the digitizers we’ve been having for the entirety of this run campaign. We didn’t actually have any digitizer issues today, which was pretty surprising. I’m not sure what might have changed to stop it from happening, but it’s possible it was just a coincidence it did not happen today.

Monday August 10 2020 4:59 pm Stewart 106855-106905 Upgraded Rake Probe Testing

This run campaign was a test of the new rake probe located at section 4 (RPS4). The initial dozen clean-up shots were required due to a loss of power on Friday. By shot 106875, the new rake probe was inserted to R=107cm. These first shots with all of the probe tips outside the LCFS and in the SOL showed an incredibly strong correlation between the poloidally separated tips (see shot 106879). The main feature of the turbulence here was large intermittent positive spikes, indicative of the blob turbulence already observed. The strong correlation between these tips is promising for Reynolds stress measurements and phase velocity estimates.

The new rake probe was fully inserted to R=102.9cm without any arcing incidents by shot 106903. The radial profile of the floating potentials (for the top tips) matches what is expected given the profiles found with the last rake probe. The poloidal and radially separated probes also have good correlation (within a correlation length) from the preliminary analysis. One issue that came up was that the mean floating potentials of the tips on the side of the probe body were not the same as the tips on the top of the probe(see shot 106902). This may be due to shadowing effects. Even so, the fluctuation amplitudes were comparable and look well correlated. Further in depth analysis will indicate whether the correlation is sufficient for the planned Reynolds stress analysis, although the results so far are encouraging.

Tuesday August 11 2020 5:49 pm Stewart 106909-106959 RP-Upgrade and TP Scan

This run day was devoted to a detailed scan of the edge with the triple probe (TPS5), while the new rake probe was present. The scan included a high resolution (2mm increment) scan with the triple probe from R=108cm to R=105.8cm. These shots are contained within the range 106926-106959. One striking observation was that the blob-filaments were found in the Isat measurements even at R=112cm (a full 5cm from the limiter)[see shot 106943]. This corroborates what was seen on the rake probe, with some "strong" blob perturbations reaching 6cm from the limiter. The spikes in the Isat data were highly correlated (above 60% correlation in some cases) with spikes in the floating potential. Tomorrow's run day will look further into this.

Wednesday August 12 2020 6:52 pm Stewart Detailed SOL Scan 106961-106998

This series of shots was meant to provide a detailed picture of blob-filaments in the SOL. The first half of the run day was compromised by CAMAC issues, where the OHE was firing too late. The ultimate remedy was to restart all of the crates and restart Spitzer. The rest of the day (after shot 106967) went along smoothly. Shots 106979-83 were a 2mm increment scan with the new rake probe, with no MHD modes present from 2-4ms. The rake probe was then moved to the SOL for shots above 106984. Shot *106987* was the ideal/target shot for this entire run campaign with the optimum flat q(a) over time, flat major radius, , flat Ip, and very little mode activity .

By switching the positions of the triple probe tips (shots 106988-106994) located at R=109cm, it was found that the strong positive correlation between Isat and Vf occurs when the tip measuring Vf is below the tip measuring Isat. When the Vf tip was above the tip measuring Isat (shot 106994), the correlation was negative. In this case, when a positive spike in Isat was seen by the TP, a negative perturbation was seen in Vf. This indicates the dipole structure of the blobs and is in the proper direction (positive potentials below and negative potentials above) for the ExB direction to move the blobs radially outward.

Lastly, shots 106996-106998 consisted of a rough scan (1 cm increments) of the blobs with the SOL tiles on the upper shell of section 1. The behavior seen by the tiles matches that of the Isat measurements. Closer to the LCFS, the blobs are more numerous and the turbulence more broadband. Moving out 2-3 cm from the LCFS, only stronger, intermittent spikes are seen at less frequent intervals.

Saturday August 15 2020 8:43 pm Stewart 107000-107113 Biasing and Reynolds Stress

This is a joint run report for 08/13 and 08/14. The first run day was partially compromised by mislabeled GPU outputs to the amplifiers used for the bias probe. By process of elimination, the correct output was finally found and the rest of the run day went smoothly. The goal of the run day was to measure the Reynolds stress at different radial locations under biasing conditions. The shot range with useful biasing data includes 107041-107053. Preliminary analysis of 107052 shows that the fluctuation amplitudes decrease at the very edge during biasing, leading to a reduction in the time average Re= as expected. The sensors 0.5mm inside the plasma show an enhancement of the Reynolds stress component (Re) when biasing is applied (perhaps due to the eddy tilt but limited suppression in this region). This results in a large radial gradient in Re just at the edge, which could be driving a significant flow.

The next run day was a continuation of these experiments with a full 2mm increment scan with the Rake probe for shots 107068-107079. Jeff performed EUV noise testing after this was completed. He removed a ground loop at section 4 and removed Jumper B, which caused the noise in the EUV channels to decrease dramatically. This testing is contained within shots 107084-107091. The rest of the run day was somewhat hampered by the loss of the cryopump at section 1. The cryopump was becoming too warm and causing the gate valve to close. Without this pump, the shots had higher loop voltages, higher SXRs, lower Ip, and tended to crash inward. The impurity level was then intolerable to perform the required experiments until the cryopump cooled down again. The final shots (107103-107113) included a scan of applied bias voltage (both negative and positive values) completed after the cryopump had cooled down and the gate valve opened at section 1.

Wednesday August 19 2020 4:56 pm Saperstein 107194-107198 SOLC mock shunt testing – Day 1

The purpose of this run day was to simply test a mock SOLC tile shunt. The shunt was placed in series with jumper A, connecting sections 9 and 10, and was being digitized in the south rack CPCI 12 - channel 12. However, a ~20 mOhm commercial shunt was also placed in parallel with it. The drop across this commercial shunt wasn’t recorded for this run day, but it was placed there so that we had the option of taking measurements from it later. Also, I forgot to add that the shunt was present in the shot database, so this is the only online documentation of the shunt being installed for this run day (it’s also documented in paper in the lab notebook), but I’ll also try to add it to the change log on the wiki.

Ian’s probes from the previous run days were left in, so I emulated a “kinking” shot (106619), so that there would still be MHD activity for the shunt to see. Shot 107194 was a vacuum shot (no digitizer issues) and shots 107195-107198 were pretty generic cleanup plasma shots. Out of the 4 of them, I wouldn’t say any of them were exceptional, but they are all good enough for investigating accuracy of the SOLC mock shunt.

We found that while the mock shunt had great accuracy pre-disruption, its accuracy was hindered during disruption, when we saw that the signal was frequently being “zero-ed out” in this region. By “zero-ed out”, we meant that the signal would suddenly drop to a value close to 0 (relative to the points around it) for some amount of time. Jeff thinks this is being cause by magnitudes of the voltages on each of the shunt leads being well above what the CPCI can handle, causing the signal to drop to a value close to zero, as opposed to clipping at 10 V. This isn’t the first time we’ve seen this happen. It happened when I did some mock shunt tests months ago (back in January/march), and Jeff has also recently started seeing it in the EUV system. We figured that this might be due to current running from the CPCI to ground, so we did some “path-to-ground” tests at the south rack.

The tests showed that while the resistances between the positive and negative leads of a channel were about 8 kOhms (as they should be), the resistance between each lead and ground was +- 1 MOhms (+- depending on the orientation of the resistance measurement). We also found a 100 mV DC drop between each lead and ground, implying that there is some DC current running from the CPCI to ground, even when the machine isn’t running. We also checked the channels on each of the south rack CPCI’s, and all of them had this problem. We tried removing as many connections from CPCI 12 to ground as we could (with the exception of channels 1-32 being connected to a breakout board via a SCSI cable) to see if this made a difference, but it didn’t. Since the EUV system also started seeing this issue after it’s grounding scheme was changed at shot 107091, we tried disconnecting it and changing its grounding scheme again, but this had no effect on the presence of the 100 mV DC drop to ground. Further work will need to be done in the future to track down the source of these issues.

All in all, it’s likely the “zero-ed out” issue that the mock shunt is having has something to do with the rack, and isn’t inherent to the shunt itself. Tomorrow, we’ll try digitizing it at the north rack, and seeing if that makes a difference.

Thursday August 20 2020 4:29 pm Saperstein 107199-107214 SOLC mock shunt testing – Day 2

The purpose of this run day was to move the digitization of the V5 mock SOLC shunt from the south rack to the north rack. We found yesterday that the shunt’s signal was “zeroing out” frequently during disruptions, so we decided to switch the rack its being digitized in so as to determine if the source of the “zeroing out” had something to do with the south rack (since the EUV system also now sees it). For convenience, as the digitization of the shunt was moved to the north rack, the shunt was also moved to now be in series with jumper B (secs 3-4), and there is no longer an extra shunt in parallel with the mock shunt. Since the shunt was moved to jumper B, and jumper B was reconnected, we also disconnected the alligator clip grounding sections 3 and 4 to the TIA box, implying that the EUV system will likely be very noisy for these sets of shots. Lastly, the mock shunt was now being digitized in CPCI 13 channel 89 of the north rack.

The first several shots of the day (107199-107203) had a Jensen in the digitizer circuit, and found that the mock shunt was still seeing zeroing out, even when being digitized in the north rack. Shots 107204-107206 had this Jensen (the one also used yesterday) replaced with a brand-new Jensen, to see if maybe this artifact was a defect in that specific Jensen. However, the zeroing out was still present. Shots 107207-107208 had the Jensen removed entirely from the digitizer circuit (this was fine considering the sections 1-5 were being grounded to the north rack anyways), but still saw zeroing out, implying that the Jensen isn’t the source of this artifact. Shots 107209-107210 still had the Jensen removed, but now had a simple RC filter was put together at the digitizer (R ~ 300 Ohms, C ~ 100 nF, f_cutoff ~ 33 kHz). The idea was to determine if the zeroing out could possibly be the result of real current spiking and reaching beyond the threshold, after which the CPCI might introduce this zeroing out effect. The RC filter should suppress any high frequency spikes, hopefully preventing large spikes from occurring. Unfortunately, these shots still saw zeroing out occurring. Shots 107211-107212 had the capacitance raised to ~1 uF, lowering the cutoff frequency down to about 3.3 kHz, hopefully heavily suppressing any large spikes. Even still, these shots still saw zeroing out. Shots 107213-107214 had the Jensen reintroduced to the circuit, but this still had no effect of the presence of the zeroing out.

Based on the raw data collected from the heavily RC filtered shots, it’s seems very unlikely that the zeroing out is occurring because the real current is getting too large. Instead, it seems to imply that something else is causing the voltages at each lead to rise significantly. Since this is only occurring during disruption, it could be that current running to ground during the disruption is somehow causing this zeroing out effect.

Friday August 21 2020 4:53 pm Saperstein 107215-107237 SOLC mock shunt testing – Day 3

The purpose of this run day was to try to track down the source of the “zeroing out” effect we’ve been seeing in the mock SOLC tile shunt, as well as the poloidal EUV system when sections 4 and 5 are grounded though the TIA box. Once again, the shunt is located in series with jumper B and is digitized in the north rack (CPCI 13 channel 89).

The firs several shots of the run day (107216-107220) were devoted to looking for a relationship between zeroing out events and current running to ground in the north rack. The north rack ground current diagnostic (a Pierson) had been disconnected for some time, so I reconnected it, digitizing its signal in the north rack at CPCI 13 channel 88. For these sets of shots, sections 4 and 5 were grounded through jumper B instead of through the TIA box. Comparing the current to ground to the RC filtered shunt’s zeroing out showed that zeroing out tended to occur at [local] peaks and troughs in the current going to ground. Analysis of the derivative of the current to ground with respect to time also more clearly showed zeroing out in the shunt occurring around its zero-crossing regions, implying that the zeroing out may have more to do with the current to ground changing direction, as opposed to the magnitude of the current going to ground. No shots in this group are worthy if singling out.

The next set of shots (107221-107222) were dedicated to determining how sensitive the zeroing out was to specific CPCIs. To test this, half of the poloidal EUV arrays (arrays 025 and 090) were moved to have their signals digitized in CPCI 12 (channels 33-64), as opposed to CPCI 38 (channels 65-96). For these shots, jumper B was disconnected and sections 4 and 5 were grounded through the TIA box, so as to get the zeroing out effect to occur in the poloidal EUV signals. With this configuration, we could then compare the zeroing out of the signals in CPCI 38 vs CPCI 12. We found that zeroing occurred at the same time in all chords, regardless of CPCI, implying that the CPCI was not responsible for shunt (when digitized in the south rack) zeroing out at different times than the EUV signals.

Lastly, we decided to see if the shunt still saw zeroing out if its signal was recorded by a scope instead of a CPCI. The shunt was disconnected from CPCI 13, and instead fed to a scope (still through a Jensen). The shunt signal was being recorded through a single scope channel (hot to hot, negative to scope ground), as opposed to the differential setup used by the CPCI. The RC filter was also removed for this. Jumper B was reconnected, and sections 4 and 5 were no longer grounded through the TIA box. It took a while to figure out the settings necessary for the scope to trigger correctly. The best we got was catching the trigger at the end of the disruption (with a threshold of 30 mV for 100 us), and then recording about 10 ms prior to that. Shots 107223-107232 were dedicated to developing the correct trigger settings, while shots 107234-107237 managed to have the scope capture [at least] the correct timing. Special emphasis should be put on shots 107235-107237 for capturing the entire plasma lifetime on the scope. Unfortunately, however, the scope measurements turned out to be incredibly noisy. The scope measurements don’t appear to see any zeroing out effects, however, the noise makes the reliability of this statement uncertain.

Monday August 24 2020 9:59 pm Stewart 107117-107192 Reynolds Stress + RMPs

Run report for 08/18/20. This run day was focused on finishing the scan in applied bias probe voltage from 08/17 and measuring the Reynolds stress when both biasing and magnetic perturbations were applied. Shots 107132-107146 rounded out the voltage scan from the previous day with dithering L-H transitions present in shots 107140-107146 (the best shot of that series was 107145). It was found that with higher applied bias, the frequency of the dithering increased. The next five shots consisted of a 2mm increment scan with the Rake probe (RPS4) for shots with Vbias=0V.

The next phase of the day focused on determining if a lower biasing current was required to achieve an L-H transition earlier in the shot. A ramp in bias probe voltage was successively moved in 0.5 ms increments from 1.5ms to 5.0ms in each of the shots (107153-107162). No discernible pattern could be gleaned from this time scan, with threshold Ibias values ~50-60A for almost all the shots. Interestingly, shot 107154 had the highest threshold current ~100A and had the earliest biasing ramp at 1.5ms. This may have been due to MHD present before 2ms.

The last phase of the run day encompassed a detailed study of both resonant and non-resonant n=3 magnetic perturbations on the plasma, while biasing. For non-resonant perturbations, a (m,n)=(0,3) configuration was used and for RMPs, (m,n)=(9,3) was used. It was found (shots 107163-107190) that these higher n perturbations did not seem to change the L-H biasing threshold significantly, although there may have been some suppression on shot 107186 in the RMP case. It was, however, found that (m,n)=(3,1) perturbations did have an affect on both the L-H threshold (shots 107191-107192) and the Reynolds stress. In these two RMP shots, the L-H transition was suppressed (i.e. there was sufficient biasing for a transition but no transition) and the Reynolds stress was also suppressed across all of the Rake probe sensor pair positions. It is not completely clear why the n=1 perturbations performed better at suppressing the transition than n=3, however my next run day should include a similar/thorough set of experiments with (3,1) RMPs.

Tuesday August 25 2020 9:13 pm Stewart 107238-107308 Biasing and 3,1 RMPs

This run day was a continuation of last week's RMP studies while biasing. Shots 107238-107250 were clean-up shots that also served as tests for Alex of the new SOL tile design. An excerpt from Alex will be included at the end of this report including further details of the tests. The next part of the run (shots 107251-10760) focused on the application of 3,1 RMPs with various levels of bias probe voltage. The suppression of the L-H transition was clearly evident (compare shots 107260 [RMP] with 107262 [no RMP]) once again. Last week it was found that 3,1 RMPs completely suppressed the measured Reynolds stress when ~40A of bias probe current was applied. In shot 107260, the Reynolds stress reappeared with an increased bias probe current of ~50A. This makes sense if the Reynolds stress is to play a role in the L-H transition at higher levels of biasing (it must "overcome" the effects of the applied RMP and re-manifest).

The next several shots consisted of m,n=0,2 magnetic perturbations with different levels of biasing at Jeff's suggestion. The effects were not as pronounced on the threshold for the transition as the 3,1 RMPs (similar to the n=3 configurations from last week). Once this series was completed, the survey with the 3,1 RMPs continued. Shots 107266-107271 included a brief scan in RMP strength. It was found that weaker RMPs were incapable of suppressing the L-H transition (see shot 107267).

The last part of the run day consisted of a full scan of temperature and density with the triple probe while biasing at Vbias=+60V (as close to the transition as possible without any dithering). The scan showed that the applied bias did not alter the temperature profile but did change the density beyond the LCFS. This may be due to the suppression of the blob-filament turbulent transport in this region, which was also measured during the scan.

The following are excerpts from Alex and Jeff's tinkering throughout the run day: "Update to SOLC shunt ex-vessel tests: Replaced current unshielded Jensen model (DIN-2LFO) with old shielded transformer model (North HIlls wideband transformer), to determine if using a shielded transformer would remove the zeroing out effect we've been seeing. We found that all configurations using the shielded transformer had no zeroing out. Shots 107238-107242 had only the CT grounded, shots 107243-107244 had only the shield grounded, and shots 107245-207246 had nothing grounded (left floating). None of these schemes had any zeroing out. We also tried putting the unshielded Jensen back in and grounding its s-output, to see if this has a similar effect to grounding the CT. This was done for shots 107247-107249. When compared to a shot where the s-output wasn't grounded (107250), we found that while the frequency of zeroing out occurring was suppressed, zeroing out was still occurring, even if infrequently."-Alex

Jeff continued troubleshooting EUV noise during the run. "The EUV signals recorded by the CPCI have been zeroing out during periods within the disruption and startup ever since resolving the major EUV noise issue, which is suggestive of a large common mode voltage during the fast plasma current ramp up/down. Grounding configurations tried today are still consistent with CPCI self-protection when common mode voltage is too high relative to the CPCI ground in the South rack. External Rogowski D was connected to measure current to ground coming through the EUV amplifier enclosure box from the vessel; this measured up to 40 amps during the disruption, which is similar to disruption currents previously measured along ground between the North and West racks. The zeroing-out problem has not yet been resolved. The zeros tend to occur when the ground current from the local vessel starts rapidly changing, which may indicate that the ground path has significant inductance or capacitance relative to the amplifier signal path."-Jeff

Saturday August 29 2020 3:51 pm Wei 107312-107460 NeuralNet-informed position control using preprogrammed 0/1 pulse, 2nd run campaign

This run report consists of 3 continuous run days between 8/26/2020 and 8/28/2020. Runs were done using the same procedure as the previous run campaign, by first taking a marginally stable, inward crashing reference shot and then apply CC VF pulse at the warning time suggested by VAE-evaluator neural network algorithm using various supply voltages (2-4V) for 3ms. These shots explored 3 new operating regimes: 1) low initial Ip and maintain the same ramp rate (initial Ip @ 7.5kA, peak Ip @ 12kA), 2) high initial Ip and maintain ramp rate (initial @ 11kA, peak @ 17.5kA), and 3) high initial Ip and reduced ramp rate (initial @ 11.5kA, peak @ 14kA). In all cases the VAE-evaluator algorithm was able to predict a safe warning time, and we noticed at higher Ip the effect of CC pulse on plasma position was less significant, thus stronger pulse strength was more helpful at maintaining the plasma at a safe operating state. The relation between plasma current level and minimal pulse strength is important for deploying this algorithm for real time operation as stand-alone position control or integrated feedback system.

All probes were retracted and all walls were inserted during the run. Also note that the metadata for day 2 (107361--107414) was not correctly stored in the online shot browser but are accessible using list_shots_hbt.py and shot_browser_hbt on Spitzer.

Monday August 31 2020 1:47 pm Saperstein 107461-107469 Testing shielded Jensen

The purpose of this short run day was to test the new shielded Jensen transformer (DIN-2LI) to see if using it would remove the “zeroing out” effect we were seeing with the previous Jensen model (DIN-2LOF). This Jensen is being used to separate the common ground between the voltage drop at the mock SOLC sensor and the measurements made at the CPCI.

Sections 4 and 5 were grounded via jumper B to section 3 (north rack), instead of being grounded via the TIA box to the south rack. The V5 mock SOLC shunt was still installed in series with jumper B and was being digitized at the north rack CPCI 13 channel 89.

Shots 107462-107463 had the new Jensen installed in the SOLC shunt circuit, with the shielding left floating. From these shots, we fortunately found that there was no zeroing out occurring during the disruption (or anywhere else). Shots 107464-107467 should have had the internal and/or external shielding grounded through one of the S ports (input 3 for 107464-107465 and input 6 for 107466-107467). Based on the circuit labeled on the Jensen, grounding either port should ground both the internal and external shielding, but I’m unsure as to why there is a degeneracy. For all of these shots, there was no zeroing out. The last couple shots of the day (107468-107469) had the new Jensen replaced with the old one again, except the shunt leads were fed into the AC inputs instead of the DC. I was curious to see if this would have any interesting effect on the presence of the zeroing out. However, it appears that zeroing out is still present, even in this configuration.

The last thing I want to mention is that were some significant differences in the magnitudes of the currents measured by Rogowski B and the shunt that couldn’t be explained by vacuum field pickup.

All in all, this short run day showed that this new Jensen model (DIN-2LI) successfully removed the zeroing out effect we were seeing in out measurements with the old Jensen model (DIN-2LOF), but there does appear to be something else causing differences between Rogowski B and the V5 mock shunt circuit.

Tuesday September 1 2020 10:30 am Saperstein 107470-107497 Testing Rog B calibration against Pierson

These shots make up a set of crates only shots that were taken to check the calibration of Rogowski B against a x10 Pierson ammeter. We didn’t want to remove the Rogowski from its amplifier/digitizer circuit for the re-calibration, so the calibration circuit was set up just outside the machine between the north rack and where jumper B is typically located. The tests were digitized through the north rack CPCI and were ultimately triggered off the Thomson trigger. The x10 Pierson was digitized at CPCI 13 channel 89 in the north rack, but it was sometimes replaced with the V5 shunt that we’ve been testing for the past couple of weeks.

Shots 107470-107496 were used to troubleshoot triggering and amplifier issues (the first amplifier we tried using my be broken). The only crates only shot of these that’s of note would be 107496, which was the first to trigger correctly with the amplifier putting out the correct amount of current. This shot also had the V5 shunt (~26.5 mOhms) take the place of the x10 Pierson. The most import shot of this test is 107497, which has the x10 Pierson digitized instead of the shunt, and is a useful shot for calibrating the correct gain of Rogowski B.

Based on 107497, we found that Rogowski B’s gain calibration was off by a factor of about 0.76 (i.e. RogB_corrected = RogB * CalibrationFactor), which is roughly consistent with the error between the V5 shunt and Rog B that was seen in the previous run report (about 0.63), and very consistent with the error between the V5 shunt and Rog B seen in 107496 (about 0.73).

Based on these findings, it’s possible that all the external jumper Rogowski’s need an updated calibration.

Thursday September 10 2020 8:16 pm Stewart 107500-107635 Triple Probe Vph Scan

Run report for 09/09-09/10. These two runs days consisted of radial scans with the triple probe (TPS5) [left floating] in order to get a detailed picture of sheared flow at the edge with various amounts of applied bias. The first batch (shots 107517-107543) included a scan from R=103.5cm to R= 110.5cm with an applied bias voltage of -75V (near floating). The radial profile of the phase velocity of the turbulence matched what has been found before with the direction of poloidal rotation flipping near the LCFS (this also seen on many other machines). The next survey (107545-107561) was conducted at a biasing voltage of V_bias=0V. This enhanced velocity "hill" that was observed near the LCFS in the ~floating case and greatly increased the shear flow in the region between R=106.5-103cm. Similarly, at V_bias =+60V (shots 107576-107959), this "hill region" underwent even greater enhancement, with a larger peak phase velocity than the 0V biasing case (for comparison the peak value for the floating case was +3.5km/s, the peak for the 0V case was +4.5km/s, and for the +60V case, the peak value was estimated to be ~7km/s).

The next five shots (107596-107600) focused on a scan of the phase velocity located at R=106cm, with bias probe voltages that were successively changed in 20V increments. This will "fill-in" some of the gaps from the last run campaign with the older rake probe. Lastly, the triple probe was scanned once again while leaving the bias probe floating (the new rake probe also inserted to R=102.9cm). This scan is contained within shots 107602-107635 and yielded results that matched the V_bias=-75V case (as expected). This will also allow for a comparison between the triple probe measured phase velocity and the rake probe measured phase velocity.

Friday September 11 2020 4:59 pm Saperstein 107636-107698 kinking v tearing campaign - day 6

The purpose of this run day was to try and develop a shot style that would produce 2/1 tearing modes by keeping the edge q just above 2. All probes were out and all shells were fully inserted (with the exception of section 4 bottom, which was slightly retracted since we couldn’t get the motor to work). Sections 4 and 5 were grounded to the north rack via jumper B. Also of note, I found that the alligator clip grounding sections 4 and 5 had fallen off. It’s possible that the last several run days were running with sections 4 and 5 left floating.

Shots 107636-107685 were dedicated to trying to achieve the desired shot style by raising Ip relative to shot 106754, a “tearing” shot style with a Ip flattop and a flat edge q above 3. Ultimately, this shot development method didn’t work very well, since Ip at the start ended up saturating very quickly (for raising OHS settings) and we could no longer keep an Ip flat top while lower the later edge q via the OHE. No star shots for this shot set. The racks also failed after 107664, and we had to power cycle the racks and restart spitzer in order to fix the problem.

Shots 107686-107689 were dedicated to trying to achieve the desired shot style by lowering the toroidal field and therefore the edge q. This did not go very well, since the plasma became relatively unstable and inconsistent below the 5.9 kV setting for the TF. No star shots.

The last set of shots, 107690-107698, were dedicated to raising R in order to lower the edge q while keeping Ip flat. These shots were taken at the end of the day though, and not enough shot development was done to find a suitable shot style. Should try to continue this shot development later.

Monday September 14 2020 5:09 pm Saperstein 107699-107762 kinking v tearing campaign - day 7

The purpose of this run day was to continue shot development from the previous run day to find a shot style with the edge q just above 2, so as to try and produce 2/1 tearing modes. Sections 4 and 5 are grounded the north rack via jumper B, all probes are retracted, and all shells are inserted (with the exception of section 4 bottom, who’s shell is slightly retracted due to the motor breaking prior to the previous run day).

The first several shots of the run day (107699-107711) were dedicated to cleanup and trying to lower the edge q by raising the major radius. This did not go well, so we went back to trying to lower the edge q by raising Ip.

The rest of the shots for the day (107712-107762) were dedicated to raising Ip, but this time we drastically raised the OH settings in order to get a significant response. This proved to work quite well, and we managed to get relatively flat R and edge q profiles, with the edge q getting as low as 2.1-2.2. However, the shot styles themselves weren’t always consistent between shots, and the mode structures were found to vary over the evolution of most of these shots. Even with edge q’s this low, we still consistently saw 3/1 modes or possibly some combination of a 3/1 and a 2/1. We also sometimes saw shots with what appeared to be dominant 4/2 modes (107718, … etc.). As for 2/1 modes, something resembling these were occasionally found during the early periods of shots like 107751, 107754, … etc.

Overall, we were able to find shot styles similar enough to what we were looking for, with plenty shots that potentially have 2/1 (or 4/2) tearing modes to investigate. Probably the best shot of the day was 107754.

Friday September 18 2020 8:40 pm Chandra 107783-107835 GPU-EUV Position Control

The theme of the first part of this three part run campaign was investigating the linearity of the response of the plasma major radius to vertical field perturbations as imposed by the control coils.

A square wave pulse of roughly 2.5ms was applied to the up/down symmetric coil sets, varying the amplitude in steps of 10A to ± 40A. The initial derivative of the pulse was also varied, from a ramp to a step.

Results, appended, show that pulling the plasma inboard had a roughly linear response, which will be investigated quantitatively at a later date. Because the plasmas initiated around R₀≈92cm, they became likely outboard limited and therefore the response to outward directed perturbations was roughly a constant.

No significant impact was found in changing the initial derivative of the pulse.

The EUV's tomographic determination of the major radius tracked that from the large Rogowski better than in previous experiments, but the reason for this is as yet unknown.

The dead SXR channel was handled more correctly, resulting in the emissivity centroid as measured by that diagnostic much better matching the magnetic R₀.

These results conform to expectations, and therefore unfortunately do not explain the poor outcomes of the last EUV-GPU feedback experiments. The next phase of this campaign will incorporate these results into feedback experiments with proportional and integral gains.

Monday September 21 2020 10:28 pm Chandra 107836-107888 GPU-EUV Position Control

The theme of the second part of this three-part run campaign was PID gain scans for plasma R₀ control.

Results with proportional and integral gains were very promising (see attached baseline and feedback shots). After a rough scan of gains, the plasma was seen to track the demand waveform quite well (as determined by the magnetic diagnostics, SXR emissivity centroid, and EUV-GPU Tomography system), and resisted a net 3.5cm outboard natural growth excursion, to remain stable at a pre-programmed 91cm, as requested, and once control was removed rejoined the baseline evolution path.

The next phase of this campaign will look for cleaner examples of these results, and then investigate poloidal mode tracking.

Tuesday September 22 2020 5:30 pm Saperstein 107901-107937 kinking v tearing campaign - day 8

The purpose of today’s run day was to reproduce the 3 kinking/tearing shot styles of interest for this campaign (106619, 106754, 107754), except with the triple probe (TP) inserted into the SOL, so as to measure I_sat as a pseudo-measure of density. The goal was to analyze the intermittency of I_sat in the context of blobs, to determine if there might be a relationship between the MHD-like SOLCs were measuring and blobs.

Shots 107901-107923 were dedicated to shot development in order to reproduce the “kinking” shot style (106619) with the section 5 TP at 107.2 cm. 107923-107928 were then dedicated to getting repeated shots of this style to analyze. The start shot of this set was 107927. Other shot(s) of interest include 107923, 107926, and 107928

Shots 107929-107937 were then dedicated to getting shots of the same “kinking” style, except with the TP pulled back to 109 cm. 107929-107930 didn’t have the TP pulled back though, and instead they were used for testing how retracting the section 5 shells (20.2 mm back for the bottom shell and 20.5 mm back for the top shell) affected the plasma evolution. Fortunately, there was no significant effect. The rest of the shots that followed did have the TP at 109 cm. The star shots for this set were 107934 and 107937. Other shot(s) of interest include 107929.

We did not manage to get to the other two shot styles in time for this run day. Instead this will be continued on another day.

Wednesday September 23 2020 8:48 pm Chandra 107938-107985 GPU-EUV Position Control

The goal of part III/III of this run campaign was fourfold:

1^st: To show that, against a relatively unrefined plasma natural evolution (up to ≈ 93cm), the control system could keep the major radius centered around 91cm. This is demonstrated with some oscillation up to ± 0.5cm, with integral and proportional control.

2^nd: Maintaining a flat R_m and I_p concurrently. This was somewhat successful: With the same flat current profile, the control coils were able to hold the plasma relatively stable, and extend the shot lifetime, over the baseline case, but they ended up either railing or very close thereto. Similarly flat major radii could be achieved naturally, but at the expense of a more curved current profile.

3^rd: Test a higher resolution basis set, with geometric parameters (Gaussian and Lorentzian widths, etc) optimized in parallel against a test shot. This was semi successful: The higher resolution (1mm major radius step size) was able to clearly see the oscillations of the plasma due to n=1 rotation (at 9kHz), but the optimized parameters ended up fitting the data in the present shot style less well than the un-optimized ones, leading to a persistent offset in R_m

4^th: Using a basis pair calculated from an SVD, discriminate between m=3 and m=2, as q^* evolves. Success unclear: Results will be appended at a later date.

Thursday September 24 2020 9:46 pm Stewart 107993-108067 Ensemble Velocimetry

Today's run day accomplished two goals: (1) Get additional shots with turbulence phase velocities from the triple probe (and ExB velocities from the Rake probe) under different amounts of biasing to perform ensemble averages. (2) Scan the old rake probe (RPS2) in 2mm increments to get a better v_ExB radial profile, since previous results have shown rapid change in the electric field at the edge (R=107cm).

Shots 108004-108015 were taken with the triple probe at 107cm, measuring the phase velocity of the turbulence with the bias probe set to -75 V (near floating), 0V, and +60 V (close to the L-H threshold). The next section of the run day (108017-108049) had the triple probe located at 106cm and the bias probe increasing bias in 20V increments after two successful shots in each iteration. The voltage span went from V_bias=-120V to V_bias=+60V. From these measurements, it is once again clear that the turbulence phase velocity is ~1km/s higher than the measured ExB velocity (as expected for ITG turbulence). The trends were also the same, with an exponential rise in both turbulent phase velocity and v_EXB with increasing bias (matching what is expected from an I-V curve).

The last several shots of the day 108050-108067 consisted of a 2mm scan of the rake probe at section 2 with a -75V bias and +60V bias from the bias probe. The results corroborate what has been found before, but now with much higher resolution.

Hardware update: It was found this morning that the main floor rack cooling water pump had fallen through some of the wood supporting it. Jim was able to fix this temporarily and the pump operated normally throughout the day.

Friday September 25 2020 5:01 pm Saperstein 108068-108138 kinking v tearing campaign - day 9

The purpose of today’s run day was to reproduce the two “tearing-like” shot style of this campaign (106754 and 107754) with the triple probe (TP) inserted at both 107.2 cm and 109 cm. Shots were also taken with both section 4-5 grounding schemes (ground to N rack via jumper B and ground to S rack via TIA box). All shells were inserted except for section 5’s, who were retracted 2 cm, and section 4 bottom, which was slightly retracted. All probes were out, except for the section 5 TP.

Shots 108068-108088 and 108096-108097 were dedicated to cleaning up and developing the first tearing shot style (106754) with the TP at 107.2 cm and using the jumper B grounding scheme. Shots 108087 and 108097 were the star shots that had the best reproduced shot styles. Shots 108089-108091 then had the TP moved to 109 cm. The star shot of these was 108090. Shots 108098-108105 had the grounding scheme switched to the TIA box scheme, with shots 108098-108103 having the TP at 107.2 cm and shots 108104-408105 having the TP at 109 cm. The star shot with the TP at 107.2 cm includes 108103, while the star shots with the TP at 109 cm were 108104 and 108105.

Shots 108092-108095 and 108106-108133 were dedicated to reproducing the second tearing shot style (107754), with shots 108092-108095 and 108106-108128 having the jumper B grounding scheme while shots 108129-108133 had the TIA box grounding scheme. The star shots with TIA box grounding include 108114, 108122, and 108127 while the star shot with jumper B grounding includes 108129. Lastly, shots 108134-108138 continued with the jumper B grounding scheme and had the TP moved to 109 cm. Unfortunately, none of these shots were very good, due to the incredible inconsistency between type 2 (107754) tearing shot styles.

The last thing to note is that the type 2 tearing shot style was incredibly inconsistent. This made shot development very difficult, and many of the “useable” shots are only useable between the 2-3 ms time period, because after that they quickly disrupt.

Monday December 14 2020 5:20 pm Saperstein 108862-108894 SOL internal calibration and armature Rogowski calibration

Today’s shots are divided into crates only internal new solc tile calibration tests (108862-108884) and initial testing of the armature Rogowski magnetic pickup calibrations (108885-108894). The sol calibration tests were only done to verify that the tiles in section 6 were working correctly at least for low frequencies (1 kHz), before welding their SS sleeves to their respective shells. I won’t go into much detail about it here. See the installation log for more details.

Breakdown of armature Rogowski initial calibration tests:

Ramped up VF voltages over shots 108886-108889. 108885 and 108888 were accidently skipped. 108886-108889 had the bottom Rogowski digitized using sol 1-4’s amplifier circuit (how good the connections were for these shots is possibly questionable). 108890 had the bottom Rogowski disconnected at the feedthrough to see how much pickup was coming from the amplifier alone. 108891 added a jumper to the circuit that introduced a large pickup area down by the amplifier board (which was sitting on the base-pad) and verified that the Rogowski was correctly being digitized at sol 1-4’s CPCI channel. 108892 had this jumper removed, but swapped the polarity of the Rogowski leads relative to that used for 108886-108889. 108893 and 108894 digitized the top armature Rogowski at sol 1-4’s CPCI, with differing polarities for each shot. The magnitude of the amplifier board pickup can be removed (leaving behind the Rogowski pickup) by taking the difference between signals with opposite wiring polarities at the feedthroughs, and dividing by 2.

Wednesday January 13 2021 5:17 pm Saperstein 108972-109036 1st plasma and cleanup following SOLC/tangential EUV upgrade

This report covers the 1/12/21 and 1/13/21 run days immediately following the new SOLC tile/tangential EUV system upgrade.

1/12/21:

This run-day started with OH pulsing looking for tools left around the machine after the up-to-air. The OH pulsing includes shots 108972-108974, with a slow ramp-up in OH power. Fortunately, no tools were heard/found during these shots.

Next, we did a ramp-up of TF-only pulsing (shots 108975-108980). The TF had no issue at any of the power levels. However, at lower power settings, integration of bit noise made the pulse appear to die down slower than it actually did. Following this, a full-banks shot was taken [108981], and everything seemed to work well.

The last couple of shots we took were the first plasma shots (108982-108983). Both of which were very short, but non-the-less fine. It’s worth noting that the power supplies for most diagnostics were accidently left off for the first shot [108982], but turned on for the second [108983].

Lastly, it’s worth mentioning that most of the shells were left retracted for these shots. Notable shells (ones with SOLC tiles on them) include 1T/B, 2B, 4T, and 10T/B.

1/13/21:

This run-day continued cleanup of the plasma. All shells were fully inserted so as to clean faster. The wiring for the SXR was reattached, the filter wheel for the new tangential EUV system was turned to the closed position, and the GPU’s power strip was plugged back into the south rack. The flame spectrometer also appears to be working more consistently today. Of note, the first several shots of the day (108984-108992), as well as all of the shots yesterday, didn’t have the 6T’s SOLC d-sub secured well to the amplifier box, and as a result SOLC sensors 6-7 and 6-8 didn’t see any real current, and saw a significant amount of magnetic pickup (~5 A when converted to false current).

Plasma cleaned up pretty well over the course of the day, with 109036 being the last shot, lasting up to about the 4 ms mark.

We did run into a couple of CAMAC triggering issues, however. With the OH and VF banks occasionally failing to fire. At one point though, we did have the triggers failing two shots in a row (109031-109032), and decided to turn off the racks for ~20 min to give them time to cool. Following this, we no longer had issues with triggers failing, although there weren’t many more shots in the day anyways. Other shots with CAMAC triggering issues include 108994 and 109006.

Thursday January 14 2021 4:56 pm Saperstein 109037-109090 cleanup day 2

The purpose of today’s run day was to continue cleanup following the shunt SOLC/tangential EUV system upgrade. Cleanup started relatively slow, with us running into issues with recurring minor disruptions (see shots 109038-109055). We worked past this issue by emulating an old shot style (74780), which extended the plasma lifetime significantly (lasting up to ~5.5 ms mark, see 109056). The rest of the run development consisted of modifying this shot style to optimize the lifetime of the plasma (up to ~6.5 ms mark, see 109087). Star shots for the day include 109058, 109062, 109078, 109083, 109085, 109085. The plasma seems mostly cleaned up by this point.

Boting also installed the tangential EUV power supply this afternoon. However, it was left off and unplugged for all but two shots, since we found that it was overheating and that there was a issue with the circuit self-oscillating.

Also of note, had no CAMAC triggering issues today.

Tuesday January 19 2021 4:45 pm Saperstein/Li 109096-109126 Tangential EUV troubleshooting & RP cleaning

The purpose of todays run day was to both troubleshoot the source of the self-oscillating noise seen by the tangential EUV system, as well as to start cleaning the rake probe.

Shots 109096-109103 were crates only shots dedicated to tan-EUV troubleshooting. A more detailed description of the troubleshooting process can be found in the shot-database. In the end, we found that the low amplitude noise wouldn’t be much of a concern, but that it would be possible for the large amplitude noise to occasionally come up and ruin the data for that shot.

The next set of shots (109105-109115) were dedicated to swapping settings on the tan-EUV filter wheel, to see what the signal would be for each filter case. The first shots with each filter setting include 109105 (closed), 109108 (open & open), 109109 (Al & Al), 109110 (Al & Ti), and 109111 (Ti & Al).

The rest of the shots for the day (109116-109126) were dedicated to inserting and cleaning the section 4 rake probe (RP). The RP was first inserted to 106.9 cm for shot 109117. We saw that the RP was seeing a lot of noise though, and through some troubleshooting (109118-109119), we found that it was due to the sec 4-5 grounding scheme we were using. We originally had secs 4-5 grounded via the pol-EUV box (south rack), and the noise disappeared after grounding them instead the north rack (via the sec 3.5 jumper).

IMPORTANT NOTE: WHEN USING THE RAKE PROBE (RP), MAKE SURE TO GROUND SECTIONS 4-5 THROUGH THE SECTION 3.5 JUMPER (NORTH RACK) AS OPPOSED TO THE POL-EUV BOX (SOUTH RACK). THE LATTER CASE INTRODUCES SIGNIFICANT NOISE TO THE SIGNALS.

The shots following these (109120-109126) continued to push the RP further into the plasma, up to 105.0 cm. A good shot to look at with the RP in would be 109124.

Wednesday January 20 2021 6:07 pm Saperstein/Li 109127-109191 Cleanup/shot dev

The purpose of today’s run day was to clean up the plasma and do a little bit of shot development to get the cleanup shot style to have a flatter major radius profile, so that the data may be more useful for some preliminary data analysis. Sections 4-5 were grounded via the north rack, the rake probe (RP) was at 105.0 cm, and the section 4 top/bottom shells were retracted out roughly 1 cm (to allow room for the RP).

Shots 109127-109138 were dedicated to cleanup and shot dev, with use achieving a relatively good R-profile with shot 109139. Following this, we tried to see if we could reproduce an old shot style, 106805, but were unable to reproduce it well (109141-109144). Implying that there’s still plenty of cleanup to do. Shots 109145 and on then went back to optimizing and repeating the cleanup shot style. Shots of note include 109161 and on. After this point, the plasmas were relatively reproduceable, with only occasional deviations that were fixed with small changes in the bank settings.

Thursday January 21 2021 4:24 pm Saperstein 109192-109224 more cleaning

Today was just another cleanup day. Nothing of note really happened today. Some of the nicer shots include 109201, 109207, and 109215.

Friday January 22 2021 6:20 pm Chandra 109225-109293 Cleanup

The purpose of today's run was continued cleanup of the vacuum chamber. The goal was to increase ohmic heating as much as possible, to kinetically pump out the walls.

With comparison to the final shot of the previous run day (109224), the shot length and temperature (inferred from the magnitude of the SXR midplane cord) increased by roughly a factor of two using the same shotstyle, by shot 109247. Attempts to increase the ohmic heating power further increased the SXR radiation up to a factor of three, at the expense of shot length and stability (star shot 109283).

Caliban was taken offline following a detected leak of GPU cooling water. The hose connecting the reservoir to the GPU, which is put under tension during coolant refilling, had at some point slipped slightly off it's clamp. For safety, the unit has been unplugged from power till this is repaired.

Tuesday January 26 2021 5:26 pm Stewart/Levesque 109295 Tree rebuild

Rebuilt the MDSplus tree to contain nodes for the new tangential EUV system (located at Section 7; nodes within HBTEP2::TOP.SENSORS.EUV.TANG) and the newer LFS SOL tiles (located at Sections 2, 6, and 10; nodes within HBTEP2::TOP.SENSORS.SOL). Corrected polarity for sensor FB06_S4P, which was swapped during the recent up-to-air for diagnostic installation -- I'll also back-correct the polarity for shots where the polarity was wrong.

Friday January 29 2021 5:23 pm Li 109296-109319 clean up

Clean up shots. Tried troubleshooting DET1 channel 12 on tangential EUV system by swapping the connectors, did not affect the negative signal.

All of the shells were retracted except for the sections with the new SOL tiles for shots after 109300. The shells with the old tiles were fully retracted out to the stoppers, while the shells without tiles were retracted 1cm. Hardware update: the poloidal EUV grounding scheme was returned to the low noise configurtion, with the white alligator clip jumper cable removed and the section 3-4 jumper connected.

Tuesday February 2 2021 7:09 pm Chandra 109320 109362

The theme of this run day was continued cleanup, and verification of the EUV-GPU control system post up-to-air.

The EUV signals as processed and stored by the GPU were verified to be the same as those stored directly from the CPCI into the tree.

The GPU was verified to output to the control coils correctly.

The GPU process was verified by examining the tomographic R_m tracking algorithm, which produces measurements in line with the SXR and cosine rogowski, although with less accuracy than previously, due to either the quality of the plasma or the issue detailed below:

Unfortunately, the 90℃ (downward vertical) EUV array appears to have sustained minor damage to its filter. This is observed in the following: the signal across all channels on this array is roughly 2x higher than the other arrays, the array measures signal during breakdown when the others do not (ie: it measures H_a, which should be blocked by the Al filter), the array measures a much higher signal on channels which previously were dark (ie: pointing directly towards the wall).

Finally, the pickup between the control coils and the Cosine Rogowski (the standard R_m measurement) was measured with a squarewave feedforward pulse to be 3-10% of the signal during a standard shot.

Target cleanup shots were 109215 and 109283.

The grounding configuration is maintained as low noise for the SOL tiles.

The tEUV system routing to the CPCI was disconected due to SCIS cable limitations.

Thursday February 4 2021 5:55 pm Chandra 109368-109414 Cleanup

The theme of the run was post up-to-air cleanup

45 Plasma shots were taken, with a target shot of 109215.

Hardware changes:

- For shots 109379-408, channels 1-32 of the pEUV system were disconnected from the GPU (although still routed into the tree), and the SCSI cable instead used to connect the tEUV breakout board to CPCI 12 and the EagleHarbor BNC breakout board (for HF noise reduction). From 109409 onward, the SCSI cable was returned to it's prior position.

- The grounding scheme is through the EUV housing (EUV noise reduction scheme), the Sec 3-4 jumper is disconnected.

Note: On the final shot of the day, a small spark was observed in the middle of the OH bank in the basement, but upon examination no obvious scorching or debris could be detected.

Monday February 8 2021 5:34 pm Saperstein 109415-109436 Current path testing & Pierson-TF coupling testing

The purpose of todays run day was two-fold. First was to test to see if the Piersons saw a significant amount of coupling to the TF (as well as to see if they moved significantly in its presence), and second was to continue testing to determine the current paths in the vessel.

Notable machine settings: Sections 1, 4, and 8 were fully retracted, while the rest were out 1 cm out (this is the new fully inserted position for the section 2, 6, and 10, which have the new SOL tiles). Sections 4&5 were grounded via jumper 3.5 (B) to the north rack. The section 5 rake probe was at 105.0 cm. The north rack ground current diagnostic was digitized at north rack cpci (13) ch 89, while the west rack equivalent was digitized at west rack cpci ch 96. A x20 (adding 50 Ohms at the termination doubled it from x10 to x20) Pierson was attached to the section 9.5 top turnbuckle with jumper 9.5 (A) fed through it, oriented such that the symmetry axis of the Pierson was facing toroidally, and was digitized through the BNC panel at south rack cpci 38 ch 01 (the old sol tile channels were temporarily removed for the BNC panel SCSI). Another Pierson of the same orientation was also attached to the section 3.5 top turnbuckle, but had no jumper fed through it, and was only present so that we could view from the control room how much the TF was moving the Piersons. Lastly, the tangential-EUV system was left off, since we were using the BNC panel that it usually needs to suppress dangerous amounts of noise.

The run day started with a ramp up of the TF banks (109415-109418), to make sure that the large forces from the TF wouldn’t shake the Piersons too much. We found that even at max settings (109418), the TF didn’t move the Piersons at all. Vacuum (all banks, 109419) and plasma (109420) shots also saw no movement in the Piersons. Additionally, we found that while the Pierson had some pickup from the VF (~< O(1 [V]), see shot 109419), the fields had little-to-no influence on the Piersons measurements (comparing Pierson measurements to Rog A measurements, 109420), which bodes well for the external HFS SOLC tile measurement plan.

The rest of the shots (109421-109436) were dedicated to tracking down current paths in the vessel, which included a little bit of cleaning and shot dev (109421-109427). Sections 4 and 5 were floated for shots 109428-109431 to see if this had a significant influence on the net SOLCs, while the rest (109432-109436) were grounded via jumper 3.5 again. Section 3’s shells were fully retracted for shots 109434-109436, to see if this had a significant influence on the currents measured by jumper 3.5 (B). We transitioned to an outboard shot style (at 109430) to push the plasma further away from the HFS potential current sinks. And lastly, two vacuum shots were taken at the end of the day, that had the shot settings that were used for the majority of the day (109435-109436).

Tuesday February 9 2021 4:56 pm Saperstein 109437-109452 Current path tracking – day 2

The purpose of this run day was to continue investigating the current paths in the vessels by retracting out undiagnosed shells, to see what effect this had on the global currents.

The machine setup at the start of the day was the same as that for the start of the previous day, with the exception that the section 9.5 Pierson was still disconnected. Several shots were taken at the beginning of the day to set a baseline shot style and global current profile (109437-109443). We then retracted out the section 5 shells (109444-109445), the section 3 shells (109446-109447), the section 7 and 9’s shells (109448-109450), and section 10’s shells (109451-109452). The shell retractions were cumulative, meaning we kept the previously retracted shells still retracted when retracting the new shells. Implying that by 109452, only sections 2 and 6 shells still inserted, while all of the rest were fully retracted. We found that retracting the section 5 shells had a significant effect on the global current structure, retracting section 3 had little-to-no effect, and the combined the retraction of sections 7, 9, and 10 also had a huge impact (bring the net SOLCs closer to 0).

Important note: we found that we started having breakdown issues after retracting the section 7 and 9 shells, that were resolved by also retracting the section 10 shells. We currently think this may have to do with issues concerning the ability of the e-gun to sufficiently ionize the plasma when certain shells are retracted (specifically the section 9 shells).

Wednesday February 10 2021 4:08 pm Saperstein 109453-109471 Breakdown shell scan

The purpose of the run day was to reproduce and narrow down the conditions under which the plasma no longer breaks down well as a result of varying shell positions. This comes after yesterday, where after retracting all but sections 2, 6, and 10, we were no longer able to get good breakdowns. The day started with sections 2, 6, 9, and 10 at the 1 cm out position (this is fully inserted for 2, 4, and 6) and the rest fully retracted (109453-109455). We then varied the shell positions for sections 7 and 9 to see when breakdown would deteriorate significantly. Shots 109456-109457 had section 7 pushed back into the 1 cm retracted position, where it stayed for shots 109458-109465 as we retracted out section 9’s shells in 1 cm increments, until it was full retracted. These shots gave varying qualities of breakdown (based on size of inverted Ip during OHB period and number of soft x-rays), but still managed to have decent breakdown to where the shots could last about 4 ms or so. With the section 9 shells still out, we then started retracting out the section 7 shells in 1 cm increments (109466-109471). The varying qualities of breakdown seen when retracting the section 9 shells still apply here, but the most surprising finding was that we continued to get decent breakdown, all the up to, and even including, when the section 7 shells were fully retracted, which was the same configuration we had yesterday that failed to breakdown correctly. As a result, we were never able to reproduce the lack of breakdown seen yesterday.

Another point to mention is that we also started digitizing the south rack ground current at south rack cpci 38 ch 01. A x20 (x10 but with a 50 Ohm termination) Pearson was added around the ‘nominal’ south rack ground. I emphasize ‘nominal’ here, since we found earlier today that disconnecting this ground seemed to have little effect on the grounding of the south rack.

Tuesday February 16 2021 5:35 pm Saperstein 109472-109535 Varying limiter geometry and ground loop tracking

This report covers two run days of the same campaign, 2/15/21 – 2/16/21.

Machine Setup: RP4 @ 105.0 cm. Sections 4&5 grounded via jumper 3.5 (B). Same N, W, & S rack ground digitization as the previous run days. Added a x10 Pearson measuring current from the section 3 HD amplifier box to the north rack, digitized at north rack cpci (13) ch 88. Tan-EUV was left off. Shell positions varied with limiter configuration.

-- 6-limiter config: sections 1, 2, 6, 8, 9 & 4T were in. | sections 3, 5, 7, & 4B were out.

-- 4-limiter config (1): sections 3, 4, 6, 8, 9, & 10 were in. | sections 1, 2, 5, & 7 were out.

-- 4-limiter config (2): same as (1), except section 4T is out and section 2 is in.

-- 4-limiter config (3): same as (2), except section 7 is in.

-- 2-limiter config: sections 3, 5, 6, 7, 9, & 10 were in. | sections 1, 2, 4, & 8 were out.

Note on shell positions: “in” for the new SOLC tile sections means their fully inserted positions (which has the shell 2 cm from the LCFS). “in” for the old SOLC tile sections means they are in their fully inserted positions (shells are 1 cm from the LCFS). The only exception is the section 4 shells, where “in” implies 1 cm retracted from their fully inserted positions (their fully inserted position has the shells 1 cm from the LCFS). This was done to keep the section 4 shells from accidently hitting the RP in section 4. “in” for the undiagnosed shells, means they are retracted 1 cm from their fully inserted positions (which is 1 cm from the LCFS), implying their shells are also 2 cm from the LCFS. “out” always means fully retracted.

2/15/21 (109472-109513):

Shots 109472-109499 were dedicated to cleanup and shot development with the shells in the 6-limiter config, with shots 109500-109502 being the best shots in the 6-limiter config. The start shot for this config was 109502. Shots 109503-109510 then had the shells put into the 2-limiter config. The start shot for these was 109508. Shots 109511-109513 then had the shells in the 4-limiter config (1). Unfortunately, we weren’t able to get good breakdown with this shell config.

Other notes from this run day: Measurements from the added x10 Pearson showed there was a discrepancy between the currents running from the vessel to the north rack, and the north rack to ‘nominal’ ground. The currents measured leaving the vessel agree far better with the west racks currents than those leaving the north rack. Additionally, between shots 109478 and 109479, we added insulation between the pump hoses in section 8 and the section 8 amplifier box. This eliminated the large currents the west rack ground was seeing, as a result of this ground loop.

2/16/21 (109514-109535):

This run day changed the 4-limiter config around until we got good enough breakdown, and then did some shot development to reproduce the shots styles used for the other 2 configurations. Shots 109514-109517 had been in the 4-limiter config (2), but this still didn’t get good breakdown. We then transitioned to the 4-limiter config (3) for shots 109518-109535, which was able to get good breakdown. Start shots from the latter configuration include 109522, 109529, and 109534.

Wednesday February 17 2021 4:31 pm Saperstein 109536-109554 Varying limiter geometry – day 3

Today’s run day was a continuation of the previous two. The purpose was to get shots in the 3-limiter config (1), which has sections 2, 3, 6, 7, 9, & 10 in, and sections 1, 4, 5, & 8 out (see yesterday’s run report for details on what “in” means). The star shots for this configuration were 109545 & 109546, with their vacuum shot being 109547.

We also got did some shot development to raise/flatten the major radius, in this configuration (109548-109554). The star shot being 109550. We didn’t yet get a vacuum shot for the bank settings used though.

Tuesday February 23 2021 10:16 am Saperstein 109555-109603 Varying limiter geometry – day 4

The purpose of this run day was to get shots with the last several limiter configuration cases initially needed. These include the 3-limiter (2), 2-limiter (2), and 1-limiter (1) cases described below.

3-limiter (2): Sections 2, 6, & 10 are in, the rest are out.

2-limiter (2): Sections 6 & 10 are in, the rest are out.

1-limiter (1): Section 6 is in, the rest are out.

The definitions of “in” and “out” for each section are consistent with the ones defined in the “Tuesday February 16 2021 5:35 pm Saperstein 109472-109535 Varying limiter geometry and ground loop tracking” entry of the archives.

The day started in the 3-limiter (1) config, getting clean by about 109561. We then moved to varying the positions of the sections 9 & 10 shells specifically. The purpose of this was to investigate the magnitude of the effect of moving these shells might have on the Ip and major radius magnetic measurements, considering that these diagnostics are located between sections 9 and 10, and the positions of the shells may very well significantly influence the eddies in the section 9 and 10 shells, thereby influencing the magnetic measurements. Shots 109562-109568 had only section 9 retracted (starting from the 3-limiter (1) config), and shots 109569-109572 had both sections 9 and 10 retracted. The star shots of the section 9 retraction were 109566 and 109568, both of which provide support against the shell position playing a significant role in the diagnostic measurements (error introduced at least < 0.2 cm, where error is estimated by comparing the magnetic measurements of R to optical ones, i.e. the SXR fan-array). The start shots of the sections 9 and 10 retraction were 109569 and 109571, which imply that having both sections out may introduce an error on the order of < ~O(2-4 mm). These measurement errors should be double checked in the future though, considering that the error in the SXR measurement of the absolute R is ~O(1 cm), and its error in their deviations seems to be on the order of ~O(2-4 mm), which is what I’m currently using to set the upper limit of the error in the section 9 & 10 retraction case.

Following these tests, we proceeded with varying the limiter geometries for the cases provided above. We started with the 3-limiter (2) config for shots 109573-109581 (start shots include 109577, 109578, 109579 & 109581). The first two shots had very poor breakdown, but we were ablet to resolve the issue by raising the puff time from 1200us -> 1400us. The 2-limiter (2) config covers shots 109581-109591 (star shots include 109581, 109585, 109586). Half way through these shots we started getting poor breakdowns, which we resolved by increasing the puff time again up to 1600us. In hindsight, we should have just refilled the puff line, since we started running into breakdown issues again by the last two shots. And finally, the 1-limiter (1) config covers shots 109592-109603 (star shots include 109598, 109601, 109603). The first several shots of these once again had breakdown issues (109592-109596). Increasing the puff time was no longer working, and even though we thought to refill the puff line (which likely was part of the issue), we didn’t resolve the problem until we realized that the e-gun battery voltage had dropped down to ~6.5 V (from ~11.5 V at the start of the day). We don’t know what caused the battery to deplete so quickly over the course of the day, but we found that replacing the battery (and dropping the puff time back down to 1400us) allowed us to get decent breakdown again by shot 109597. At the end of the day, the puff time was still at 1400us (instead of the initial 1200us).

Lastly, I want to point out shot 109600, because the major radius for this shot went quite a bit outboard, which is of note since I found that the sum of all the new SOLC tile currents was much closer to 0 A during the outboard region than it usually is, likely implying that there exists some surface on the HFS that is acting as a significant current sink, unless the plasma is moved outboard enough. Based on this, more tests should be done with more outboard plasmas, to see if we can consistently account for a larger percent of the current for these shot styles.

Wednesday February 24 2021 12:08 pm Saperstein 109604-109652 Varying limiter geometry – day 5

The purpose of this run day to develop a shot style in the 1-limiter (1) config that was more outboard and with lower m# modes, in hopes of further limiting the potential current paths between the plasma and the vessel.

We started with cleanup for shots 109604-109610, with 109610 being a great relatively centered shot in this configuration. Shots 109611-109634 attempted to emulate and develop the 107754 shot style, which has strong, long lasting 2/1 modes. Unfortunately, we weren’t really able to reproduce the shot style at this time (plasma is still too dirty), let alone develop the style to be more outboard (in the 94-95 cm range). An interesting shot from among these to look at though is 109630. For shots 109635-109652, we went back to starting with the 109610 style, and tried to develop it too be more outboard (without regard for its effect on q at the moment). By the end of the day, we still didn’t really have the shot style we were looking for, and the best shot we got was 109652.

Wednesday February 24 2021 4:31 pm Saperstein 109653-109663 Varying limiter geometry – day 6

Today was a relatively short run day. The plan was originally to cleanup, then start inserting section 2 tiles to 2 cm retracted position, to get an idea of how large the currents out there might get, since this is the rough position of the section 6 shells when their tiles are limiting the plasma. We never ended up getting to that point though. Instead, shots 109653-109658 were cleanup, and 109659-109663 tried flattening the major radius of the 109610 shot style to stay closer to 92 cm. Star shots of which include 109662 and 109663.

The run day then ended a little early, so as to get ready for the tests that would be done tomorrow. To prepare, RP4 was retracted to about 110.4 cm, and the section 4 tiles were fully inserted (both top and bottom).

Friday February 26 2021 2:50 pm Saperstein 109664-109701 Varying limiter geometry – day 7

This is the run report for 2/25/21.

The purpose of today's run day was to start testing the effects of grounding the only limiting section to the south rack, so that high frequency currents would have to run through the high inductance ground before being redistributed somewhere else in the machine. We did this test for the section 4 tiles and started it for the section 10 tiles.

RP4 was at 110.4 cm for the section 4 tests (that had only section 4 inserted) and back to 105.0 cm for the section 10 tests. Hard x-ray detectors were disconnected, and the BNC panel was moved to south rack cpci 12 channels 33-64 instead. Nothing was digitized at the BNC panel for this run day. The old SOLC tile sensors’ SCSI was put back to its original spot, at south rack cpci 38 channels 1-32. The south rack ground measurements were disconnected.

2-limiter (3) config: Section 4 and 6 shells in, all others out.

1-limiter (2) config: Section 4 shells in, all others out.

1-limiter (3) config: Section 10 shells in, all others out.

The run day started off in the 2-lim (3) config, with sections 4&5 grounded to the north rack via jumper 3.5 (shots 109664-109673). Star shots include 109669 and 109673. Then we transition into the 1-lim (2) config for shots 109674-109680. Star shots include 109679 and 109680. Sections 4 and 5 were then grounded to the south rack via jumper 3.5 for shots 109681-109686. The star shot was 109681. We then moved to the 1-lim (3) config, re-grounding sections 4 and 5 to the north rack, and keeping section 10 grounded to the west rack via jumper 3.5, for shots 109687-109701. For this configuration, we ran into several issues with getting adequate breakdown. The e-gun battery needed to be replaced, and the puff time ended up needing to be raised to at least 1900us to get any breakdown (109700), with good breakdown not really occurring until the puff was set to 2000us (109701). And that’s as far as we made it with the section 10 tests today.

Friday February 26 2021 4:49 pm Saperstein 109702-109751

The purpose of today’s run day was to develop the 1-lim (3) config shot style with section 10 grounded to the north rack, and then ground section 10 to the south rack instead, and investigate the effect on the currents collected by the section 10 tiles. Shots 109702-109741 were dedicated to shot development, with shot 109717 being the star shot. Other notable shots include 109728 and 109734. Section 10 was then grounded to the south rack via the copper box by section 10 for shots 109742-109751. A x20* Pearson measured the positive current running from the vessel to the rack and digitized the data at south rack cpci 12 channel 33. The star shot was 109743, with 109742 and 109745 also being notable. The bank settings for these shots were the same as it was for 109717, but the shot style ended up being a little different (in part due to the large variance we’ve been seeing for the 1-lim configs), however, they are still comparable. A little bit of shot development was done at the end of the day to try and push the major radius out, to see if this could move the plasma further from the HFS limiters, hopefully without inflating the floating potential profile. Not much progress was made though, since the breakdown quality started declining for the last couple of shots, and we decided to end the run day.

Monday March 8 2021 10:08 am Saperstein 109752-109782 Triple probe cleaning

This is the run report for 3/4/21. The purpose of the run day was to start cleaning the triple probe in section 9 (TP9). TP9 wasn’t biased for these tests (or even digitized), this was just an initial cleaning of the probe. I also want to emphasize that the digitization location for the tan-EUV (and half the pol-EUV) was changed for different parts of the run (see Diagnostics).

Machine setup

Shell positions: All sections with a probe (in or out) were fully retracted, including 2, 4, 5, & 9. All non-probe limiter sections were fully inserted, including 1, 6, 8, & 10 (1 & 8 have shells 1 cm from 107 cm, 6 & 10 have them 2 cm away). All other sections were 1 cm out, including 3 & 7.

Diagnostics: Tan-EUV system was powered on for this run, and was digitized in two separate places over the course of the run day. The chords started off being digitized at south rack cpci 38 chs 65-96, where it replaced half of the pol-EUV chords, that were originally located there (half of the 64 pol-EUV chords weren’t being recorded at all in this scheme). Starting with shot 109762, the tan-EUV chords were then digitized in their original location, which is south rack cpci 12 chs 33-64. The pol-EUV chords that were disconnected were then reconnected (all pol-EUV chords were then being digitized). The last diagnostic(s) to note are those on the same breakout board as the hard x-ray diagnostics. While the tan-EUV chords were plugged into cpci 38, the cpci 12 chs were empty, so I reconnected the hard x-ray breakout board, which mostly includes probe data that we hadn’t been using for a while. RP4 at 105.0 cm.

Grounding Scheme: Sections 4&5 grounded via jumper 3.5 to north rack. Section 10 grounded via Co box to south rack.

The run day started with TP9 still separated from the vacuum. We decided to clean the plasma a bit first, since na issue with one of the pumps a couple days prior caused some oxygen to flow into the chamber. Cleanup includes shots 109752-109761. The plasma turned out to still be relatively clean, and even got better shots than we typically had over the past couple of weeks, with some shots having about 1 ms of ~7-9 G saturated modes prior to disruption (109759, 109760, 109761). Concerning the tan-EUV system, the noise level wasn’t too bad, on the order of O(40 mV), but the plasma didn’t appear hot enough to get good signal in some of the tan-EUV chords. So more tests with the tan-EUV system need to be done in the future with higher power plasmas.

Following cleanup (109762), the tan-EUV digitization was switched from cpci 38 to cpci 12. Once noticing that there was little effect on the tan-EUV noise after doing this, we then opened TP9 to the vessel, starting with shot 109763. This rose the base pressure, and had it settle somewhere in the mid-high 10^-7 range, but this dropped over the course of the day. Cleaning went pretty well with the plasma looking pretty normal by 109771. We then just inserted the probe further and further in, to get it clean. Shots 109772-109774 had TP9 at 107.0 cm. Shots 109775-109777 had it at 106.0 cm. And shots 109778-1098782 had it at 105.0 cm. Shot 109775 was a shot of note, considering it also had strong saturated modes that lasted for about 2 ms.

Monday March 8 2021 4:53 pm Saperstein 109783-109822 TP9 cleaning – day 1

The purpose of this run day was to start cleaning the TP9 while biasing it (to 150 V). Machine setup is the same as the previous run day’s except TP9 was moved back to 108.0 cm at the start of the day, and the tan-euv system was left off.

TP9 digitizer info:

The TP9 diagnostics are currently being digitized at south rack cpci 10 chs 93-96. Isat is at ch 96, V_ion is at ch 95, V_float is at ch 94, and V_electron is at ch 93. The gains and such for these measurements were taken from the TP5 nodes in the tree, considering that we are currently using the same box that TP5 used to use.

Shots 109783-109788 were cleanup, for which TP9 was left off. TP9 was biased starting with shots 109789, and we immediately found that there were tons of arcing events. Way more than usual, according to Ian. The number of arcing events per shot did somewhat drop as we continued taking shots, but never really disappeared between shots 109789-109800. We then retracted TP9 out a bit to 109.0 cm for shots 109801-109814, but found that there were still a significant number of arcing events per shot. Retracting TP9 out to 112 cm removed the arcing (109815), but pushing it back in to 110 cm brough them back, although at a much smaller frequency than we were seeing at 108-109 cm. The rest of the day’s shots were just taken cleaning the probe at 110 cm (shots 109816-109822).

I would also like to note that the shot style found here was quite interesting, and had very strong, saturated, long-lasting modes prior to disruption. The most notable shot would be 109794, which had a saturated 5 G mode last for about 3 ms before disrupting.

Tuesday March 9 2021 5:08 pm Wei 109823--109871 TP9 cleaning - day 2

The purpose of the run day was to continue cleaning up TP9. Machine setup was exactly the same as the last run day.

On shot 109871, a popping noise was heard in both the control room and the office area during the time when the TF bank was charging up, following which the shot was dumped. One water resistor on the TF bank (second one from the right on the orange bank) blew up and fell over at 15:17:40 according to security camera 3. Jim noticed the dump relay was mechanically stuck on closed position, so the water resistor could be heating up during TF charging. We shut down the machine afterward.

Thursday March 25 2021 12:09 pm Saperstein 109872-109880 Testing new TF relay

Today we tested to make sure the new TF relay that we got to replace the one that broke last time was working correctly. We found that all the relays worked correctly when hitting the reset button (with doors closed, keys in, and high voltage on).

We did a ramp up of the TF as well. The ramp up went fine, but we did notice for the first half of it that the TF crowbar didn’t appear to be firing. Upon investigation, we found that one of the banana clips for the crowbar fell out. Plugging it back in resolved the issue. We also ran into an issue with the first full vacuum shot (all banks powered), where the shorting stick for the OHE was accidentally left on, so the OHE wouldn’t charge. Fortunately, we noticed this quickly when the banks started charging, and resolved the issue. Aside from those tests, we took two plasma shots to make sure things look ok, and then we ended the run day.

Monday March 29 2021 5:10 pm Saperstein 109881-109909 TP cleaning, blade limiter test, and CC tests

The purpose of today’s run day was to clean TP9, while also doing some other tests. These include 1) re-inserting the blade limiters and retracting out all SOLC tiles to verify that the current they collect drops significantly and 2) driving the control coils (CCs) to see how the new SOLC tiles respond.

The blade limiters were inserted to their “standard” positions (R=107cm, z=+-15cm) prior to the start of the run day, which is 2 cm further in than they previously were. We did find though that one of the blade limiters [south(2->3)-top] was only about ~1.8 cm out instead of 2 cm, which Jim says was because that blade limiter seems to start hitting something, preventing it from coming out any further. The stoppers left behind from the last time the blade limiters were moved were used to set the standard position.

Machine setup

Probes: RP4 @ 105.0 cm, TP9 @ 110.0 cm (charged to +150 V), RP5 was open to the machine but out, DP2 was also open to the machine but out.

Shells: 2, 4, 5, 9 fully out. 1, 6, 8, 10 fully in (1 cm out for new tile sections). 3,7 1 cm out.

Grounding schemes: 4 & 5 grounded to south rack, 10 grounded to south rack.

Diagnostics: tan-EUV off. Sec 10 to south rack ground current not digitized.

Blade limiters inserted to standard positions.

The first several shots (109881-109886) were for cleanup and setting a baseline for the typical currents seen by the section 6 and 10 SOLC tiles. Shots 109881-109886 didn’t have the SXR-midplane chord working, but after fiddling around with the cabling, it started working again (109886). All shells except for sections 3 & 7’s were fully retracted for the blade limiter tests (109887-109888). We saw the currents in sections 6 and 10 dropped about an order of magnitude, with some of section 10’s tiles still retaining some finite equilibrium current. We then went back to a similar shell configuration that we had at the start of the day, except the section 2 shells were also fully inserted. We then left it in this shell config for the rest of the day (109889-109909). Starting with 109890, TP9 was retracted out to 111.0 cm to try and reduce the amount of arcing that was occurring. Starting with 109901, we started seeing deterioration in plasma breakdown, trying to fix this by refilling the puff line. At the end of the day, we realized this was because the e-gun battery had pretty much died (dropped to ~6.5 V).

Lastly, we started testing the CCs again, in the 3/1 helical phase flip config (109908-109909). The CCs fired correctly, except we noticed that cc6-1 was seeing large spikes of current during breakdown, even though the cc had been considered broken for some time. The MS connector and amplifier banana plugs were still connected. I disconnected the bananas and this removed the large currents spikes. The bananas are still currently disconnected, and labeled to show which amplifier they go to (top amplifier, ch 1). We unfortunately didn’t get any good SOLC response to the CCs though, since we still hadn’t tracked down the e-gun battery issue by the time we started testing them. So we don’t have any good shots with a solid plasma and the CCs on.

Tuesday March 30 2021 5:19 pm Saperstein 109910-109951 SOLC response to CCs tests – day 1

The purpose of today’s run day was to continue the controlled experiments where we very the control coil (CC) profiles and look at the responses in the new SOLC tiles. At the same time, TP9 was being cleaned.

Machine setup is the same as yesterdays.

We started with 3/1 helical RMPs that flip phase half-way through the pulse, with the full pulse lasting 1 ms (1/2 ms per phase) and at a max amplitude of about 20 A (109914-109921). The star shot was 109920. The natural mode typically locked late in phase 1, and then stayed locked for most of phase 2. We then moved on to a 2/1 helicity with the same other parameters (109923-109927). The goal was to no longer resonate with the natural 3/1 mode, however, we found that the 3/1 often still locked regardless. Didn’t really have any start shot for this one. We then took a couple shots with a 1/1 helicity (109929), which didn’t lock as well.

We then increased the length of the pulse from 1->2 ms for the remaining shots (109930-109951), and took some more 1/1 (109930-109931) and 3/1 shots (109932-109951). Star shots include 109931 and 109943. We also tried varying the amplitude of the RMP for the 3/1 case, using half-strength for 109945-109948 and 1/3 strength for 109949-109951. Star shots for these include 109946, 109948, and 109949. Some of the star shots (mostly the low amplitude RMP ones) are quite interesting since they seem to show significant response in the sec 6 SOLC tiles, while the magnetics remain relatively unchanged.

Over the course of the day, TP9 was also moved around to test how much it was still arcing. We tried moving it in to 107 cm (109926), but found that there still an incredible amount of arcing. So much so that it effected the plasma equilibrium and prematurely disrupted the plasma. For now, we have the probe at 108 cm, where it still arcs ( ~O(10 arcs/shot ) ).

Friday April 2 2021 4:48 pm Saperstein 109952-110026 SOLC response to CCs tests – days 2 & 3

This run report covers both the 4/1/21 and 4/2/21 run days. The purpose of both was to try different control coil (CC) pulsing configurations and see how the SOLCs respond. At the same time, we were continuing to clean the TP9 probe.

Machine setup for both run days was similar to that for SOLC response to CCs tests – day 1. With the exception that the TP9 probe was moved around a bit, and the shell configuration changed half way through the first run day covered.

4/1/21 [109952-109990]

The first portion of the run day (109952-109965) was dedicated to varying the CC pulse shapes (and magnitudes) with the same, MHD heavy, shot style from the previous run day. Instead of doing helical RMPs like for the previous run day, we pulsed entire sections with the same profile. Star shots include 109961, 109963, 109964, and 109965. 109961 and 109964 are of special interest, since they’re characteristics are incredibly similar, despite only having the CCs fired in the former shot.

The rest of the run day (109966-109990) was devoted to developing a quiescent shot style, in the hope that the SOLC response could be more directly tied to the CCs as opposed to the MHD mode. We attempted to emulate shot 106926, and with some tweaking of the puff time [1200us->1400us] and reordering of the shell positions (moved all shells, except 4&9, to be 2 cm from LCFS. Fully inserted for new SOLC sections, 1 cm retracted for all other sections.), we were able to get some kind of decent reproduction, if a little shorter (see 109976). We continued with developing the shot style, and at the same time, often pulsed the CCs to investigate the effects on the SOLCs in a more quiescent plasma. Star shots include 109985, 109987, and 109988. We also realized that moving more of the shells in and to the same positions relative to the LCFS was probably a good idea, since it makes the effects of each CC more uniform across the machine. Of note, in shots 109985 and 109988 we were able to see a relative clear response in some of the SOLC tiles, however, they were not in the tiles we excepted (those closest to the fired CCs).

Prior to 109976, we also disconnected the tan-EUV output to cpci 12 chs 33->64, and instead replaced it with the breakout board that the hard x-ray detectors are attached to. The hard x-ray detectors were then digitized at south rack cpci 12 chs 37 and 42. Based on the shots from this day, the signals had lots of pickup.

4/2/21 [109991-110026]

Today’s run day was dedicated to continued development of a quiescent shot style, with some CC tests thrown in. It was difficult to make progress due to the persistence of a large fast-growing mode that typically onset somewhere in the 3.5-4.5 ms range and either disrupted the plasma or kicked it into an unusable state. Shot development revolved around trying to avoid this mode, but this was made difficult by the fact that we didn’t really know what was causing it. The combination of a flat-top Ip and the presence of impurities would make the plasma more prone to tearing modes than usual, but the mode grew very quickly, and none of the optical diagnostics (pol-EUV sensors) saw evidence of an internal mode. The fast growth-rate seemed to potentially imply a kink/RWM mode, but the mode still continued to grow quickly, despite q_a being held close to 3.5 for the duration of the shot. More shot development still needs to be made to get a long enough quiescent period. Some shots of note with the CCs pulsed to a 3/1 RMP include 110015, 110016, 110020, and 110021. These shots saw significant responses in the SOLC tiles.

Tuesday April 6 2021 9:47 am Saperstein 110027-110071 SOLCs response to CCs tests – day 4

The goal for this run day was to continue developing a quiescent shot style, that can be used as an equilibrium for RMP tests. Machine setup was the same as the previous run days. Unfortunately, not much progress was made. I tried dropping the edge q to try and avoid fast 4/1 kink growth rates and even considered the error in our typical edge q measurements, however, I never managed to get a quiescent period to last more than about 1/2 ms. I realized towards the end of the day that the shot I was attempting to emulate (106926) also had many probes pushed far into the plasma, which was likely contributing to the quiescent period seen in that shot style. Trying to emulate this, I tried further inserting RP4 in to 103 cm, and continued shot development. This helped to create a short consistent quiescent period at the beginning of the shot (110071), but didn’t manage to get 2 ms long quiescent periods we were seeing before. This may require inserting more probes (like DP2) deep into the plasma.

Thursday April 8 2021 4:51 pm Saperstein 110072-110122 Quiescent shot dev and SOLCs response to CCs tests – day 5

The goal of today’s run day was to continue shot development for a quiescent shot style, with the help of RP4 being inserted to 103 cm to help suppress mode growth. I also tried varying the density via the puff time, in order to further stabilize the plasma. I varied puff times from 1000->1400us, and found that the best puff time was somewhere between 1100-1200us. I was able to get a handful of shots that had decent quiescent periods, ~O(1.5 ms) (110101,110105,110106,110110), but the issue was that they had significant major sweeps during these periods, which is inappropriate for SOLC analyses. Most of the run day was dedicated to trying to retain this significant quiescent period, while also flattening the major radius to a reasonable enough degree. Unfortunately, by the end of the day, I wasn’t able to achieve that.

I also started trying to insert other probes, to improve the mode suppression. I initially tried to insert the DP2 probe, but accidentally ended up stripping the insulating tube from the copper rods during insertion. We ended up taking the probe back out again, since we didn’t want to risk sputtering of the Copper. Instead, we started inserting RP5 to 109 cm (110119-), leaving section 2’s shells now fully retracted so that we could freely move the probe.

Machine setup was the same as the previous run days.

Monday April 12 2021 10:59 am Saperstein 110123-110176 Quiescent shot dev & CCs/SOLCs tests - day 6

The goal of today’s run day was to once again try to develop a shot style with a long enough quiescent period, so as to pulse RMPs and see how the SOLCs respond in the absence of strong background MHD. We continued the shot dev using the probe insertion method we started on previous run days (starting with on RP4 in the plasma and RP5/TP9 sitting just outside of it), and tried to lengthen the quiescent period by both inserting more probes deep into the plasma (RP5/TP9) and varying the puff times/plasma density.

The slow insertion of the probes was pretty standard, but optimizing the puff time required a bit more investigation. We were already limited on the lower end of the puff time due to breakdown issues that surfaced at puff times below 1100us, and increasing it beyond the standard puff time I’d been working with for this shot dev campaign (1400us) would likely only make the plasma more unstable to any density limits. Ideally, we’d like to lower the plasma density to the optimal point where we can avoid density limits while still retaining adequate breakdown, but this optimal puff time was something we would have to deduce empirically. Another concern was the high puff times we were currently using (>1000us), since optimal puff times in previous years had been closer to the ~800us range. We currently do not know if the increased puff time we need for adequate breakdown comes from the need for a higher density (for whatever reason), is because something has changed with the puff circuit (that requires it to have a higher set puff time in order to achieve the same plasma densities we’ve been getting in previous years), or some combination of the two. Regardless of the cause, we varied the puff times and found that a puff time above 1200us was required for an adequate breakdown, with the best breakdown still occurring in the 1400us range (breakdown quality was qualitatively assessed by comparing the inverse Ip dip just before startup of the shot to that of an older shot, 106926. The deeper the dip, the better the breakdown).

We also tried improving the e-gun heating by slightly increasing its heating current. Jim and Jeff mentioned that they noticed that the e-gun had been dimmer in the last year or so, and we thought that that may have been contributing the larger required puff times that we were seeing. We also found that the heating current was currently only set to ~4 A, which was about half the recommended current found in the past to optimize the e-gun heating (~8-9 A). However, we don’t actually know when this optimization was made, the 8-9 A just comes from a note left on the e-gun circuit that lacked a time-stamp. Regardless, we increased the heating current (marking the old position with tape and a marker) until Jim found that the e-gun was emitting light at a more typical intensity. We found this to occur at ~5 A (bias voltage went from ~33 -> ~31 V), and is the heating current setting that we are currently using. This change started with shot 110150. Unfortunately, we did not notice any differences in breakdown quality or puff time optimization.

Inserting additional probes did not appear to have a significant effect on the quiescent period duration. Jeff says that in the past, John’s double probes (DPs) have been the only probes found to create significant quiescent periods when inserted deep (~103 cm) into the plasma. We may need to wait for DP2 to be fixed to achieve the ~2 ms quiescent periods I’m looking for. At the moment, we can only achieve 0.5-1 ms quiescent periods.

Although we couldn’t achieve longer, more consistent, quiescent periods, we did still try a simple compass scan of 3/1 static RMP pulses. We set the duration to be 0.5 ms, and to start at 2 ms in the lab time-frame. At this time, when the quiescent period occurs is still inconsistent, so the RMP pulse did not always align well with a quiescent period, but we did repeat shots until we got good ones where the firing of the RMP aligned well with a quiescent period (the timing of which was consistent for the best shots). At the end of the day, we managed to finish half a compass scan of RMP phase, and found noticeable responses in many of the new SOLC sensors that were phase dependent, that may not have been as noticeable in plasmas with strong background MHD. The CCs off case was shot 110176*, and notable RMP shots include 110165, 110166*, 110168, 110170*, and 110172*. Star shots* are noted.

Shot dev notes: The shot style we’re currently developing bifurcates at ~1.5 ms into a centered or inboard shot, with it more often bifurcating into the centered branch, which is the one we are currently developing. The inboard shot does appear to have longer quiescent periods, however, it’s less consistent, and centering it has been found to be difficult.

Post run day notes: In trying to retract out and remove DP2, the probe’s rod ended up getting stuck in the gate valve, one of the heads come off, and the rods became detached from the probe stand. A lot of water and oxygen also got into the chamber as Jim tried to open the probe, but it’s been pumping out pretty well, and should hopefully clean up pretty quickly. As of right now, the probe’s rod is still stuck in the gate valve, and we have no way of moving it, since it became detached from the probe stand. The missing head is also lying either somewhere in the gate valve or at the bottom of the vessel (it doesn’t appear to be sitting on the bottom shells, based on the view from the section 10 viewport). Unfortunately, it doesn’t look like this issue can be resolved until the next time we go up to air.

Thursday April 15 2021 11:19 am Saperstein 110177-110212 Vessel potential fluctuation tests and CCs/SOLCs response – day 7

The goal of today’s run was two-fold. First, to finish the RMP compass scan during a short quiescent period to see how the SOLCs respond, and second, to measure the potential fluctuations of the vessel and see how much of it makes it through the Jensens as common mode noise. I managed to accomplish both today, although the compass scan could use a few more consistency shots for the phases I got on the previous run day, and I still need to replace one of the SOLC Jensens with the DIN-PB model during a reproduceable shot, to see how it effects the measurements.

machine setup

Setup vessel potential fluctuation (VPF) diagnostics. Measured the fluctuations of the section 10 vessel, which was currently being grounded to the south rack via a jumper to the SE Co box. Diagnostics include a direct voltage measurement of the vessel (who’s negative was the south rack ground), the output of the DIN-2LOF model Jensen that has both of its inputs shorted to the same point on the vessel where the direct voltage measurement was being taken, and the output of the DIN-PB mode Jensen (which has its Faraday shield grounded to the south rack), which was making the same measurement as the DIN-2LOF model. These diagnostics were digitized at the south rack CPCI 12 chs 33-35 (ch 33: direct voltage measurement, ch 34: DIN-2LOF measurement, ch 35: DIN-PB measurement). The diagnostics were not connected at the beginning at the run day, seeing as they replace the digitizers used for the external hard x-ray detectors, which we wanted to have connected for the first several shot, in case the water and oxygen that got into the vessel previously causes an excess of hard x-rays/runaways.

Reconnected two of the pol-EUV amplifier boards, which had been disconnected for some time. All pol-EUV chords are now being digitized again.

Otherwise, same machine setup as before.

The beginning of the run day was devoted to cleaning, expecting the air that got let in the other day to introduce a lot of impurities. Fortunately (or unfortunately, if you were looking for runaways), the plasmas weren’t very dirty at all, and cleanup effectively resembled the cleanup we usually have to do for the first several shots of the week. With cleanup finished early, we continued with the RMP scan (110187-110212), finding that the plasmas were significantly more reproduceable than they had been for previous run days, which helped significantly speed up the compass scan process. Most shots were useful enough to use in the scan analysis. One shot of note is 110198, which was a shot without RMPs, that reproduced 110176 very well.

The VPF diagnostics were connected and digitized starting with 110192. We found that direct voltage measurements were clipping during disruption, so starting with 110194, we also attached a [factor of 10.5] voltage divider, which kept the measurement from clipping. Starting with shot 110197, a 500 Ohm resistance was added in parallel to the output of both the Jensens to see if it would improve their common mode rejection. The addition appeared to have little effect.

From these VPF results, we found that the vessel potential can fluctuate as much as 70 V peak-to-peak during disruptions, and can actually take on a substantial positive potential (~20 V). However, the fluctuations were less than ~400 mV during flat-top. We also found that the DIN-2LOF Jensen let through a significant amount of common mode noise during the disruption (~2-3 V), while the DIN-PB model only let through ~O(50 mV).

Friday April 23 2021 10:37 am Saperstein 110213-110255 Rotating RMP, vertical control, and testing new SOLC Jensen

The goal of today's run day was fourfold. First, we wanted to double check the magnitude of the vessel potential fluctuations (VPFs) with the voltage divider now setup correctly (the negative of the output actually went straight to a ground). Second, we wanted to then try replacing the old [DIN-2LOF] Jensen for one of the shunt SOLC sensors with the new one [DIN-PB], that should have significantly better common mode rejection at high (disruptive) frequencies. Third, to complete the RMP compass scan campaign, we wanted to fire a rotating RMP during the quiescent period, so as to compare to the static case. And lastly, we wanted to test out the ability of the control coils to do feed-forward vertical control, to try and center the plasma, and see how this effects the SOLCs.

Machine setup is the same as the previous run day, with the exception that the VPF diagnostic circuit was corrected. We did find that the major radius and q measurements were a bit fuzzy at the start of the run day, but this was fixed (starting with shot 110224) by power cycling the Ip and cos Rogowski amplifiers and removing a small short between the copper and aluminum boxes by section 7. Also, we found before setup that the basement/machine room security cameras were no longer working. After messing around with the DVR box and monitor, we came to the conclusion that the DVR box was broken, and that we need a new one.

The run day started with cleanup and some preliminary vertical control tests (110213-110223). The only really notable shot of these is 110220, which was a decent shot with what appears to be successful vertical control, but is really only of note because it’s the best vertical control shot using the flat-top (110176) shot style.

Next, we did rotating 3/1 RMPs in the flat-top shot style (110224-110235). One thing we found while running was that SOLC response to these perturbations varied between shots, and that the SOLC response was somewhat suppressed relative to that of static RMPs with the same coil current. As an example, you could compare 110230 (rotating RMP) to 110172 (static RMP) and 110176 (baseline), to see that the amplitude of the rotating response is smaller than that of the static response in many sensors, even though the applied coil currents are similar, if not slightly larger in the rotating RMP case. Another star shot for this range of shots would be 110234.

With the rotating RMP tests done, we moved on to vertical control tests (110236-110255). For this set of shots, we wanted to emulate the Ip ramp (109920) shot style, since we thought it would provide the shot consistency we wanted for these tests. In doing this, we found that this shot style, in addition to be relatively consistent, also had a relatively long quiescent period during the first ~1.5 ms of the plasma lifetime. Star shot was 110242. Noticing that, we took some rotating RMP shots with this shot style, to see how the SOLCs responded (110245 and 110246). Shot 110246 was the better of the two.

With that done, and having found a consistent shot style, we replaced the old Jensen for SOLC sensor 2-4 with the new one (110248-) and found that the new Jensen removed any clipping during the disruption, and managed to retain a similar profile during the plasma lifetime. This meant that replacing the old Jensens with the new ones should allow us to study the currents during disruptions.

Finally, shots 110250-110255 focused on vertical control for this consistent shot style. We tried two different CC settings, both of which had the all coils pulsed at the same polarity, but with the midplane coils being twice as strong as the others. The first (1x) was a setting had midplane coil currents of about 25 A (110250-110252) and appears to have pushed the plasma down from about +0.7 cm to be roughly centered (centering style). The second (1.5x) had midplane coil currents of about 38 A (110253-110254) and appears to have pushed the plasma down from about +0.7 cm to about -0.5 cm (reversing style). All of these shots were really good, and the SOLC sensors had especially interesting and varied responses, both in their equilibrium (many of which didn’t budge) and in their response to MHD.

Friday April 30 2021 10:15 am Saperstein 110273-110328 Shell scan compliment to vertical position scan

The main goals of today’s run day were twofold. First, to perform a scan of the positions of the sec 6 and 10 top shells, so as to compare the effects the scan has on the SOLCs to those vertical control has on them. Second, to test the new amplifier board schemes to potentially use for the tangential EUV system. Of these goals, a section 6 top shell scan was performed out to 1 cm (with 2 mm increments) and all the amplifier board settings Boting wanted to test were tested.

Machine setup:

Tan-EUV system is being digitized again, at it’s typical CPCI location. The channels that were being recorded and the boards being used did vary over the course of the run day though. The Jensen test diagnostics, along with the BNC panel, were also removed, to allow for the Tan-EUV system to be digitized again.

The grounding scheme for both sections 4&5 and sec 10 were also changed. Secs 4&5 were grounded via the 3.5 jumper to the N rack, and sec 10 was grounded via the 9.5 jumper to the W rack. Otherwise, the machine setup was the same as it had been for previous run days.

Shot summary:

The run day started with a quick cleanup (star shots: 110281 and 110284), followed by some new vertical control tests to compare the shell scan to (star shots: 110286, 110287, and 110288). The vertical control tests were re-done because the grounding scheme had changed, which may have affected the SOLCs and their response to vertical control. This turned out to be a good idea, since it appears that some of the tiles saw a more significant response to the vertical control than in the previous tests. One of the more interesting vertical control shots was 110288, since it had the control coils powered on from 1.5->5 ms, which is almost the entire duration of the shot, and made it easier to see the responses in some tiles that weren’t necessarily clear in the vertical control cases that fired later.

Following the vertical control tests, I started the position scan of the section 10 top shell (star shot 110295). I started with a 1 cm retraction, but found that the shell being that far out may have been interfering too much with the plasma evolution, since I temporarily lost the consistent shot style I’d been using. Retraction of the sec 9&10 shells have been known in the past to interfere with breakdown and evolution, so this conclusion wasn’t unreasonable. As a result, I decided to do the sec 6 scan first instead, and to come back to the sec 10 scan if I had time. The sec 6 top shell scan went fairly well, with star shots for the following shell retractions:

1 cm – 110300, 5 mm – 110304, 2 mm – 110307, 8 mm – 110308, 4 mm – 110315, 6 mm – 110318

Reference shots with sec 10 top fully inserted were taken before (110299) and after (110325) this scan. The order of the scan was performed in the order provided above. The run day ended trying to get one more vertical control shot, with the settings set so that the magnetics would see the vertical position roughly centered (110328).

Boting’s amplifier tests went on in the background. I’ll list here which shots had which settings. Al 50K & 12 (110278-110281), Al 200K & 12 (110284-110287), Al 200K & 5 (110288-110290), Al 400K & 5 (110291-110293), Ti 400K & 12 (110294-110296), Ti 4.7M & 5 (110297-110301), Ti 4.7M & 5 edge channel (110302-110304), Ti 4.7M & 12 edge channel (110305-110307), all blocked and all channels connected (110308-110310), both boards are now initial components (110311-).

Interesting note: Occasionally, the major radius would be too low (by ~1/2-1 cm), and a 4/1 mode would grow at the beginning of the shot. This is likely because the probes were no longer far enough in the plasma to completely suppress its growth. This then highlights the necessity to keep the plasma relatively centered (or outboard) in order to maintain the quiescent period seen in this shot style.

Thursday May 27 2021 4:51 pm Saperstein 110333-110363 Replacement of sec 6 SOL tile Jensens & vertical position fine scan – day 1

The goal for this run day was two-fold. First, to check out the disruption data from the section 6 SOLC tiles now that their Jensens have been replaced (and their cages are now grounded), and second, to start the fine scan of the vertical position (vp), to see how SOLC measurements vary with vertical position. We managed to get good disruption data from the tiles, but we were only able to finish the fine scan of the vp.

Machine setup: Same machine setup as the previous run day except for the following. Section 6’s SOLC tiles had their Jensen’s replaced with a newer model (DIN-PB) that can adequately reject common mode noise and allow for disruption measurements. Tan-EUV and pol-EUV systems were left off. Section 5’s shells were inserted to the 1 cm retracted position, so that now all shells are at the 1cm* out (fully inserted for new SOL sections) position.

Caliban wasn’t working for the beginning of the run day, so we started with some quick cleanup and investigation of the sec 6 SOLC tiles’ disruption data. It turned out that the current Jensen setup we were using had a significant amount of high frequency noise (150->200kHz), which we didn’t really pickup during the tests because we tested a low-res tile, which had a higher signal-to-noise ratio, and since it was such high frequency. Shots 110343 (old Jensen) and 110344 (new Jensen) can be compared for SOLC 6-8 to see the differences in noise level between the two signals. Only SOLC 6-8 was swapped back to the old Jensen for this test. Fortunately, we found that adding a 500 Ohm resistor in parallel with the output of the new Jensen suppressed the noise quite well. Compare 110344 (no resistor) and 110348 (with resistor) to see the effect (on all sec 6 sensors).

Following the Jensen signal troubleshooting, we moved onto starting the vp fine scan. We used the 2/3x max strength (max ~ 40 A) setting for the CCs as the largest vertical control setting, and managed to scan 0x (110348), 1x (110354), 1/2x (110360), and 1/5x (no star shot) that max vertical control setting, in that order.

Run notes: SOLC tile MHD response magnitude was found to be heavily dependent on the major radius. Which required some shot dev and many repeated shots to maintain a consistent enough major radius.

Friday May 28 2021 1:11 pm Saperstein 110364-110390 vertical position fine scan – day 2

The goal of today’s run day was to finish the fine scan of the vertical position control started on the previous run day, and we did.

Machine setup: Same as at the end of the previous run day

We scanned from no vertical control to the max vertical control settings (~25 A) in increments of 1/5x the max vp setting (vp max), in a non-ascending order. While the plasma shot style wasn’t always consistent (although sometimes it was) the shot style itself was fairly reproduceable. Star shots for each vertical control setting include 110388 (0x max), 110374 (1/5x max), 110386 (2/5x max), 110373 (3/5x max), 110390 (4/5x max), and 110383 (1x max). Other good shots that may be of use include 110371 (4/5x max) and 110375 (2/5x max).

Thursday June 17 2021 5:02 pm Saperstein 110412-110443 Jumper resistor tests, shell armature rog tests, and troubleshooting SOLC disruption noise

There were several goals for this run day. First was to determine if digitizing one of the section 6 SOLC tiles in the north rack suppressed the high-frequency noise that that section’s tiles have been seeing during disruptions with the new Jensen. Second was to get data with the section 2 shell armature Rogowskis installed in the last up-to-air. And last was to scan jumper resistances with varying resistors (and brands) to make sure that the resistors could handle the disruption loads and to get a first look at a resistance scan. Fortunately, we managed to perform all of these tests today.

Machine setup:

Machine setup was the same as that for the previous run day, with the following changes.

Shell armature Rogowskis are hijacking SOL 1-8’s amplifier board. As a result, no section 1&2 top SOL sensors are being digitized. The armature Rogowski is digitized in SOL 1-8’s typical location, and which armature Rogowski is being digitized (along with which shell of SOL sensors in section 2 aren’t being digitized) can vary shot to shot. The run day starts with the top shell armature Rogowski being digitized.

The jumper resistance scan experiment was setup, with a series of resistors being swapped around and introduced in series with jumper 9.5 (A). The run day started with using a 1 Ohm PulsEater resistor in series with the jumper.

Run day:

The run day started with cleaning, getting more-or-less cleaned up by shot 110419 (star shot for 1 Ohm resistor – PulsEater). We did have an issue with the OH start firing late for shot 110415, but this didn’t happen again for the rest of the run day. Following cleanup, we began varying the digitization and faraday cage grounding schemes for SOLC 6-1. Up until now (and the default scheme) had the sensor digitized (and the cage grounded) in the south rack. Shots 110421-(110422 start shot) had the sensor’s digitization and cage grounding switched to the N rack (N rack CPCI ch 87). Shots 110423-(110424 star shot) had the cage grounded to the south rack, while the sensor was digitized in the north rack. Shots 110425-110427 had the cage grounded to the north rack, while the sensor was digitized in the south rack. We found from this test that digitizing the sensor in the north rack removed most of the high-frequency disruption noise, while changing the grounding of cage had little effect. For shots 110428 and on, the sensor and its cage were put back to their default schemes.

Following those tests, we moved onto to varying the resistors used in the jumper resistance scan. We started with PulsEater resistors for shots 110412-110435. Shots 110412-110429 used a 1 Ohm resistor (star shot 110429). Shots 110430-(110431 star shot) used a 10 Ohm resistor. And shots 110432-(110435 star shot) used a 100 Ohm resistor. We then moved on to testing some of the Riedon resistors. Shots 110436-(110440 star shot) use the 100 mOhm resistor. Shots 110441-110443 use the 2 Ohm resistor. We found that current in jumper 9.5 (A) dropped significantly for resistances above 1 Ohm.

As for armature Rogowskis. The top shell’s was being digitized for shots 110412-110440, while the bottom one’s was digitized for shots 110441-110443. When accounting for vacuum pickup, the measurements looked pretty good for the top Rogowski. For the bottom however, we only got one plasma shot with it, and it looks super funky and unreliable. More shots should be taken to test the bottom Rogowski. A good vacuum shot for the top rog is 110412 and a good one for the bottom one is 110443.

The tan-euv system was also momentarily on for this campaign, from shots 110429-110433. However, for shots 110429-110431, the power supplies’ fuse was blown, and the amplifiers weren’t being powered. So there was only good data for shots 110432-110433. Although, these were still apparently very noisy.

Wednesday July 28 2021 10:26 am Saperstein 110445-110475 Bt scan and Ip digitization

The goal of today’s run day was threefold. First, to get reproduceable shots with the jumper 9.5 (A) Rogowski connected and disconnected (to compare the Ip measurements before and after), second, to digitize the Ip Rogowski to a scope in the screen room (to avoid possible ground current noise during disruptions), and lastly, to do a scan lowering the toroidal field (to achieve a larger range of disruption rotation frequency scale-factors). By the end of the day, only the first goal had been accomplished. We started working on the second one, but didn’t manage to finish in time. The latter two goals will have to be continued on another run day.

Machine setup: Same as previous day, with the following exception(s). Jumper 9.5 (A) was replaced with a shorter jumper (about 10 cm), compared to the longer jumper (about 220 cm) used for the jumper resistance scan experiment. The TS trigger was sent to the screen room to trigger the scope for Ip digitization. The Ip Rogowski signal was disconnected at its amplifier in the west rack, and sent to an old SOL amplifier board in the screen room instead (gain: ~100, rc_time: ~50 us), to be digitized through a scope. This last change wasn’t made until shot 110462. Jumper A was left disconnected at the end of the run day.

The run day first began with a ramp up of the banks (110445-110452). The light bulbs in the bank room had been replaced the previous week, so we wanted to make sure that nothing had happened to the banks. The ramp up went fine, with the exception of the last several shots having triggering issues with the north rack CPCI. This issue was eventually resolved. We think the issue was with the grounding of FO transmitter power supply (used to transmit the TS trigger in the north rack to the screen room) being plugged in outside of the north rack. However, the solution was still never entirely clear, since grounding the mentioned power supply the north rack didn’t immediately resolve the issue.

Next, we got some shots with jumper A on (110453-110457) and off (110458-), with the star shot being 110459. Jumper A was left off for the rest of the run day, and left off at the end of the day as well.

The rest of the run day (110462-110475) was dedicated to trying to digitize the Ip Rogowski signal in the screen room. Unfortunately, we ran into what appears to be triggering issues, that weren’t resolved by the end of the day.

Friday July 30 2021 10:33 am Saperstein 110487-110532 Bt scan and Ip digitization – day 2

The goal of today’s run day was to get the Ip Rogowski signal to digitize to a scope in screen room and to start a scan of the toroidal field to extend the disruption rotation scale factor range, in continuation of my previous run day. Unfortunately, we were still unable to get Ip to digitize to the scope in the screen room correctly, but we did start the Bt scan and manage to get decent plasmas with TF bank values down to 4000 V.

Machine setup: same as previous run day, with the exception that jumper 9.5 (A) was connected for the entire run day.

The run day started with troubleshooting an issue with the N rack CPCI 13 (I_OH, I_VF, etc.) mis-triggering. This occasionally happened on the previous run day as well, and we were never able to narrow down the source of the problem. We unfortunately didn’t have much better luck today. We found that the issue went away after disconnecting one of the Thomson BNC’s, unplugging the FO transmitters power supply, and closing the N rack doors, however, the issue never came back after reverting these changes. This issue can likely just be chalked up to HBT being HBT. Troubleshooting this issue covered shots 110487-110499.

Shots 110500-110515 were devoted to trying to digitize Ip to a scope in the screen room. We did manage to get the scope to trigger correctly by shot 110504, but the Ip Rogowski raw signal never came out looking reasonable.

The end of the day (shots 110516-110532) was devoted to starting the Bt scan. We managed to get the TF bank setting down to 4000 V before we stopped getting breakdown at 3500 V. It may still be possible to get breakdown at <= 3500 V, but it will require more shot development. Star shots for each TF setting include 110518 (6000 V), 110521 (5500 V), 110522 (5000 V), 110529 (4500 V), and 110530 (4000 V). The 3500 V attempt was 110532.

Tuesday August 3 2021 10:46 pm Levesque/Wei 110535-110561 Data collection with new dual S710 fast cameras

Collected data using the two new Phantom S710 cameras viewing the plasma simultaneously through the viewport at Section 10. One camera (S/N 26731) remains on a tripod and looks toward Section 3 with a narrow view, while the other camera (S/N 26730) is connected to a fiber optic relay and has a wider view of Section 8. Frame rates of 100kfps and 400kfps were tested. Light collection was not as high as anticipated for the fiber-bundle-coupled camera; this will be problematic for taking videos with sufficient dynamic range if we don’t improve the light collection. Exposure times of 8us for 100kfps imaging were good enough during the main plasma for the tripod camera, while this gave very little (but still measurable) signal in the fiber-optic-coupled camera during most of the main discharge. Rotating bands of light due to mode activity are apparent on both cameras, but the camera with the wider view (fiber-coupled) may not have enough light for accurate quantitative analysis. So far, the observed brightness does not seem to have the expected linear dependence on programmed exposure for some of the test cases. A small set of videos with 8us exposure appear brighter than the 9us exposure for 100kfps imaging despite plasma parameters being comparable; actual exposure times should be tested before converging on final diagnostic settings. The 400kfps imaging (1.5us programmed exposures) saw a great amount of light during the disruption for the tripod camera; the fiber camera did not see nearly as much disruption light, but the videos are good enough for correlation analysis for the brightest dynamic structures. Star shots are 110555 (camera session #600) for the 10us frames, and 110558 (camera session #608) for the 2.5us frames.

We also took two low-fill-pressure shot attempts (1 successful plasma) to look for hard x-rays after reconnecting the two existing HXR detectors to CPCI 12, channels 37 and 42. Shot 110560 had features of our typical energetic-electron plasmas, and a couple of HXR bursts were observed during plasma events. There was no obvious HXR response (i.e. speckling) on either of the cameras upon brief inspection. A plastic scintillator was placed on top of one of the FB sensor amplifier boxes for this shot. No obvious flash of light was easily visible from the scintillator by eye from the control room, but this could be expected since the scintillator shouldn’t give off very much light anyway. Some of the brightest HXR shots we’ve measured in the past (e.g. 106273, 106274) have had at least 2 orders of magnitude higher fluence than for this specific shot.

Thursday August 5 2021 10:29 am Saperstein 110562-110633 Bt scan – day 3

The goal of today’s run day was to see if we could get breakdown at half our typical TF (~3200 V), and if we could, to get enough shots for statistical significance to be used in a scan. We were able to do both by the end of the run day.

Machine setup: Same as previous run day, with the following exception(s). Section 10 is completely retracted out. Pol-EUV system is on, but in the noisy configuration (grounded to N rack).

The run day began with reproducing typical (110566 – 6500 V) and low TF (110570 – 4000 V) shots. This was shots 110562-110570. The first 3 shots of the day once again had N rack triggering issues, which disappeared after arbitrarily moving around some wiring in the N rack. It seems likely that some cabling in the N rack is causing this issue, and that moving around some wires gets the cabling in the configuration where it no longer causes the issue. This is just a guess so far, but I recommend moving some cables around in the N rack around, whenever it runs into this triggering issue. Shots 110571-110590 were then dedicated to getting some shots at TF 3500 V. This did require increasing the puff time above 1300us. Increasing to 1500us resolved the breakdown issues, but I found that we got better breakdown at 1900us.

We did manage to still get breakdown at 3200 V, thanks to the increased puff time, but we did not attempt going to lower TF settings. The rest of the run day (110591-110633) was just devoted to getting shots at this TF setting, with some shot development snuck in to see how clean we could get the plasma evolution to be. The shot styles were incredibly inconsistent, and shot development did not make much headway. Fortunately, however, we were only interested in the disruptions, so pre-disruption evolution was not much of an issue. Some star shots include 110599, 110610, 110623, 110631, and 110632.

Wednesday August 25 2021 9:06 pm Levesque/Wei 110634-110651 Testing new HXR detectors in one toroidal location

Ran hard x-ray-producing plasmas in order to commission the newest 8 HXR detectors (scintillator + photodiode + amplifier units). We now have a working set of 10 HXR detectors, though only 7 can be powered at the same time until the power cabling is finished (tomorrow). Plasmas were made at the lowest fill pressures we could use while still achieving breakdown, following the recipe we've used a couple of times before. Plasmas were not reproducible, but most of the successful breakdowns did produce significant HXR signals on all detectors. The new detectors were arranged in a 4x2 grid to get them as close as possible to each other. They were located on top of the Section 1 FB sensor amplifier enclosure, facing the Section 1.5 convolution spool piece. Note that this is directly facing the Section 2 LFS SOL tiles, viewing through the thin metal of the convolution piece (thinner than the rest of the vessel, though the metal thickness will vary dramatically along the possible lines of sight). Light-blocking tape is placed in front of the scintillators in order to remove/reduce possible stray visible light at the detectors. The associated C7319 amplifiers were set to 10⁷ gain (the maximum available of 3 possible settings) and the highest bandwidth setting. The 2 older HXR detectors remained above the midplane at the limiter spool pieces. Note that the shields on the new signal cables and power cables are floating for now. New HXR detector signals are temporarily recorded at North rack CPCI 40, inputs 65-72. The power supply is also powered via the rack. No isolation transformers are used since the power source is from the diagnostic rack and the power cables run alongside signal cables. Older HXR detectors are connected to South rack CPCI 12 inputs 37 and 42, through separate power supplies and Jensen isolation transformers.

Many of the HXR bursts briefly saturated the detectors. All co-located powered HXR detectors saw the same signal characteristics and similar signal levels within a factor of ~3, which is a good sign. There was no evidence of magnetic pickup on these new detectors, while one of the older detectors still has severe pickup (probably due to cabling). Unpowered detectors had some response during the HXR bursts, down by a factor of ~100-300 from the powered channels. Measuring small but finite signals in the unpowered channels should be investigated sometime to find the cause.

Shells in Sections 4, 5, 9, and 10 were fully retracted all day. The remaining shells were in Alex's standard 1cm retracted configuration for benefit of the newer SOL tiles. The fast cameras were operating, with one camera looking at the plasmas and the other looking at a large plastic scintillator. The one looking at the plastic scintillator didn't show signs of measurable visible light emission from the scintillator, but this could be expected since it's not coupled to a high gain amplifier.

After shutting down we arranged 5 of the new HXR detectors to be in a toroidal array, evenly spaced, facing downward directly over the spool pieces, slightly below the top of the TF magnets. Tomorrow we'll run the same type of plasmas while measuring with this toroidal array. The remaining power cables should be finished by the end of the run day in order to test using all 10 detectors at once.

Thursday August 26 2021 7:44 pm Levesque/Wei 110652-110678 HXR shots with a toroidal array of HXR detectors

Ran HXR plasmas with 5 of the new HXR detectors (#1-5 from yesterday) arranged in a toroidal array just above the outboard midplane. Detectors were placed on top of the FB sensor amplifier boxes, facing the local spool piece toward the right of the detector. The five toroidal locations of the detectors are: Sections 1.5 (convolution, newer LFS SOL tiles), 3.5 (quartz, older LFS SOL tiles on retracted shells), 5.5 (quartz, newer LFS SOL tiles), 7.5 (convolution, older LFS SOL tiles), 9.5 (quartz, newer LFS SOL tiles on retracted shells). Wall and limiter configurations are the same as yesterday, with blade limiters inserted for a=15cm. Note that HXR detector 3 was connected to the wrong diagnostic channel until shot 110660. The mounting configuration that we had set up after yesterday's run failed the high-potting test due to conduction/tracking along ~8 inches of the black plastic mounting pieces, which is why we changed to the midplane configuration. The older HXR detectors remained at the outboard blade limiter locations, Sections 2.5 and 8.5. The fast cameras were running, with one camera looking away from expected light sources to check for speckling in its videos.

Successful HXR shots showed periodic behavior of HXR bursts with a toroidal phase dependence around the machine. The phase dependence presumably correlates with phase of the n=1 mode activity, but this has not been verified. Star shots for this are 110661 and 110662, with 110662 being an excellent shot with steady 3/1 mode activity to analyze. Detectors that view regions where local plasma facing components are closer to the LFS LCFS (i.e. limiters and limiting tiles) produce stronger HXR signals than the other locations, suggesting that the emission source is mainly from the low field side for these cases.

At 110671 we set up the remaining 3 new detectors (HXR #6-8) as a rough radial array viewing the spool piece at Sec 1.5 (unobstructed as much as reasonably possible) since power cords were now available. HXR detector #1 could also be considered part of this array, though its view is different in case there is sensitivity to that. Distances from the chamber end flanges to each detector are roughly #2=14", #6=26", #7=54", #8=77". Examining some peaks just in detectors #6-8, a clear ~1/r² dependence of the signal amplitude is seen. Star shots including this radial arrangement are 110673 and 110677.

Wednesday September 15 2021 9:28 pm Chandra 110680-110710 Poloidal EUV filter damage survey

The theme of this run day was predicting the extent of damage to the poloidal EUV (pEUV) 90^o (downwards-facing) array, to discern if the pinholes in the new Aluminum filters will let through an unacceptable amount of light.

The goal was to measure the signal ratio between the fully open and (assumed unbroken) aluminum filter settings of the tangential EUV (tEUV) system for the same line of sight, just after breakdown when the light signal is still dominated by D-alpha. This can be compared to the same ratio between the damaged and undamaged arrays of the poloidal system. The machine started up well today, and the "unmodified" tEUV TIA boards were used (the noise level was roughly 200mV in an offline test, but this seemed to decrease once the machine was started up, as can be seen in the attached plots).

Results indicate that the visible light to bremsstrahlung ratio for the tEUV early in time is between 20-30, and the equilibrium ratio is between 6-8 (for the two groups of shots tested). The appended plots show that the same ratio for the pEUV system between the damaged and undamaged arrays is roughly 8 early in time and 2 later in time.

This suggests that the damaged pEUV array is between 20-30% open, much much higher than even the most perforated of the newly procured aluminum filters.

Tuesday October 12 2021 5:17 pm Li 110712-110728 TangEUV Board Testing

Tested the new TIA boards for the tangEUV system. The noise became larger when the Al tube was taken off. The shots of Ti filter have an offset of around 3.5V. Got a good estimate of how to adjust the Al gain but not the Ti. Pressure gauge bypassed for gate valve closure.

Tuesday October 19 2021 6:56 pm Li 110857-110874 TangEUV Board Testing

TangEUV board tests. Plugged in channels one by one. Bottom Ch 3, 11, 12 and Top Ch 6, 7, 8, 14, 15, 16 are bad and have an offset. Still have some noise source before TIAs.

Wednesday October 20 2021 5:20 pm Li 110875-110902 TangEUV Board Testing

By adding channels one by one, different noises showed up. 110891 and 110892 had all channels connected, BOT ch11 had a large offset, BOT ch3 did not work normally. Starting from 110893, BOT ch11 is unplugged again. BOT ch3 back to normal during 110895, 110896, 110897, 110899 and 110901.

Friday October 22 2021 4:13 pm Li 110903-110934 TangEUV Board Testing

The lowest noise configuration is TOP Ch 1-15 and BOT 1-10+16. Tried shot development (starting from shot 110913). Shots to look at: 110913, 110918, 110925, 110929. 110931 and 110933 show clear sawtooth.

Saturday October 23 2021 2:01 pm Li 110935-110957 TangEUV Board Testing

TOP Ch15 and 16 are replaced with 10M boards. BOT Ch11 is replaced but still does not work. BOT Ch3 works unstable, might need to double-check. The power cable for BOT Ch 13 is plugged wrong in 110949, and unplugged afterward. All other channels seem to be affected but got better after cooling down for a while. Need to check on Monday to see if they still work fine. Sawtooth shots: 110938, 110940, 110943,110945. Minor disruption shots: 110940,110946, 100948. 110954, 110956, and 110957 show obvious sawtooth in the later time range, but have large oscillation at the beginning, should try to smooth the noise.

Monday October 25 2021 8:39 pm Chandra 110958-110995 GPU-EUV Tomography Control

The shots were a continuation of a multi-day campaign to observe if the tomographic tracking code, when used to actuate the control coils with a fixed m=2 or m=3 output flux structure, could impact the mode amplitudes.

Results were highly un-satisfying. A consistent shot style was found (ex: 110995, a baseline shot), and a compass scan was performed in increments of 45^o. No visually detectable change in the mode amplitude was observed.

The applied waveform included Chen's "control coil pole cancelation" transfer function.

Issues were encountered in the GPU with the output for control coils in sections 9 and 10. The phase appears be reasonably correct, but the coils are firing earlier and with a unstable waveform. This will be investigated further at a later date, but should not have produced the observed null result.

In comparison with previous successful feedback tests (e.g. Chen feedback on 86056) the magnitude of the control current was comparable (both between 20A-30A), as was the major radius (91.3 for this test, 92.3 for Chen's experiments).

Comparison plot between Chen's feedback off, on, and on for a short duration tests vs my baseline and a suppression shot attached, along with measured control coil vs Mirnov currents on the same shell and phase tracking, and a tomogram to verify m=2 emissive structure.

Friday January 14 2022 5:16 pm Li 111551-111670 clean up & tangEUV troubleshooting

Clean up shots. Got some good long shots on 01/14/2022.The poloidal EUV system appears to be seeing reasonable signal in the three arrays with intact filters. The broken upper-midplane array sees a signal close in magnitude to the previously broken downward facing array.

For tangEUV system, Ch3 and 4 are broken. Currently on a stage where adding more channels introduces various noises. Adding on the enclosure lid helped a lot.

Next step: Add more channels with the lid on. Try changing the ground loop.

Tuesday January 18 2022 5:22 pm Wei/Li 111671-111709 clean up

111629, 111634, 111640, 111643, 111644, 111651, 111653 are some nice long shots during the cleanup.

By unplugging the connector near the CPCI for a group of four disconnected channels (Ch 21~24), it is clear that there is a type of high freq noise is caused by the parts between the breakout board and the CPCI (look at Ch24, the noise is still there even it's unplugged). Another type of noise disappeared on all channels after unplugging that connector, which means it is possibly caused by coupling of the cables within the conduit or other setups before the breakout board.

Wednesday January 19 2022 4:48 pm Li 111710-111753 clean up & tangEUV troubleshooting

The probe is inserted today, the plasma seems to be ok though. The gas puff leaks quite fast and many shots are short today.

111737~111741 show a low freq noise after plugging in one more channel, need to figure out where it comes from.

111746~111749 have certain connectors unplugged from the breakout board, but still have the same frequency noise in all channels, even the totally unplugged ones (not connected to the input, output, power, or the breakout board). This means that this noise comes from somewhere between the breakout board and the CPCI. However, the phase of the high freq noise in different channels is not in phase, there is a gradual shift.

Thursday January 20 2022 5:03 pm Li 111766-111781 tangEUV troubleshooting

Most shots were taken with all the ribbon cables unplugged and only one channel's output connected to the floating scope.

By comparing the no-input noise among different setups: bench test, the input connected to the biasing board, and the input connected to the biasing board but having the GND of the D-sub connector directly connected to the GND of the power supply (bypassing the biasing), it is clear that the biasing board increases the noise.

Tuesday January 25 2022 5:13 pm Wei 111782-111811 Fast cameras black reference troubleshoot

Today’s goal was to troubleshoot the black reference code in the camera control software with Marco from SPICA Tech. The code was fixed at shot 111799. In the subsequent shots we tested the dual camera setup using different resolutions, frame rates, and exposures. In particular, in 111804-808 the tripod camera (26731) was rotated sideway to get a poloidal view of the inboard limiter, and during these shots filamentary features resembling turbulence were observed prior to and during disruptions.

Star shots: 111799, 111803-808. Detailed camera setups were documented in the logbook and shot comments

Wednesday January 26 2022 4:31 pm Li 111812-111847 clean up & tangEUV troubleshooting

Long nice shots: 111837, 111842, 111843, 111846.

All channels of the tangEUV system can work at the same time without having high noise. Adding inductors in the power cable does not seem to affect the noise. The boards with reduced RC show much lower noise, so I plan to reduce the RC on a few other high gain boards.

Thursday January 27 2022 5:23 pm Chandra 111848-111900 Shot Development for Mode Feedback

The goal of this run day was to produce a consistent shotstyle which would be kink-dominated, outboard limited, similar to Qian's baseline shot 86061.

Development was largely unsuccessful, shots were largely inconsistent. Closest modern shots to the target are 110993, and 106619 (Alex's database).

Implementation of GPU based tomographic phase measurements without the upper-midplane array were inconclusive, although the offline tomograms seems of acceptable quality.

Best shots were 111875 (low mode amplitudes, but likely kinking and correct R_m), or 111900 (probably tearing, but good mode amplitudes, bad R_m)

Friday January 28 2022 4:45 pm Li 111901-111931 TangEUV Board Testing

Got all 32 channels to work at the same time with low noise (starting from 111915 to 111931). The only change is switched from ribbon cable #1, #2, #3 and #4 to #3, #4, #5 and #6. My guess is that the insulation on #1 or #2 got cut in the conduit and the wires are touching the conduit surface, which might cause grounding issues. This needs to be further confirmed.

Further plan on the TangEUV system:

(1) Reduce the gain on Ch 12 (saturates quite often).(2) Deal with the offsets on Ch 15 and Ch 31 (~1V). (3) RC on Ch 15, 16, 27 and 28 might be too low. Test by reducing the RC on Ch 29 and 30 in only one stage and check the outcome.

To further reduce the noise, we can try to add RC filters on the BOB, but this would require a new BOB.

Shots are quite repeatable from 111927 to 111931 (long shots with a flat major radius).

Thursday February 3 2022 11:20 am Li 111932-111968 TangEUV Board Testing

The RC on all boards is back to the original value from shot 111943. Ch 29 and 30 now have gains of 60MOhm. Star shots: 111951, 111952, 111953, 111954, 111957, 111965, 111967, 111968. These are very long shots with a flat major radius. Many of them show signals on Ch 29 and 30, which is good. In 111955, 111957, 111959, Ch 28 has spikes that are not present on other channels. 111949, 111950, 111963, 111964, 111966, 111968 show weird bumps on Ch 28. In some of these shots, clear correlations can be seen on Ch 12 and Ch 28. The board for Ch 28 has been changed from Board 34 to Board 29 starting from 111967, but the bump is still there. Although the boards have been adjusted and verified to have zero offsets on the bench, Ch 15 and Ch 31 still have offsets of about 1V. Noise on Ch 13, 14, 15 and 16 seem to be slightly different from the other channels on the same diode, this is possibly caused by the broken connector on the breakout board. Plan to test with a new breakout board.

Monday February 7 2022 6:08 pm Chandra 111969-112022 Shot development for pEUV based control

The theme of this run day was further development of a repeatable shot style which is more likely to be kink-dominated.

The primary target was shot 86061, Qian's baseline shot. We were relatively successful in matching the profiles from this shot (see attached presentation). Delaying the VF start time helped. The shot style is characterized by 2-4kHz bursts of m=2-3 mode activity, corresponding to drops in loop voltage. It is hypothesized that this is due to a periodic sawtooth-like event (as seen by the tEUV system) relating to a drop in temperature and broadening of the current profile.

Evidence in support of this hypothesis is found in the SVD of the pEUV offline tomographic reconstruction (see attached). A periodic broadening of the emissivity profile (<~1% of the total magnitude) is seen in the 1^st, 2^nd, and 3^rd modes, while the 4^th and 5^th describe an internal m=2 structure, matching DCON, rotating at the magnetic mode frequency.

The next step in this campaign will be applying the discovered shot style to mode suppressing feedback.

Machine Changes: VF start timing changed fromm 700us to 800us.

Tuesday February 8 2022 4:38 pm Li 112023-1112051 TangEUV Board Testing

TangEUV system ran stably. Ch 28 did not show the bumpy signal in today's shots. 112050 is a long shot. 112051 has exponential mode growth.

Noise on Ch 9-12 is higher than normal, plan to test by swapping the connectors on breakout board.

Thursday February 10 2022 7:44 pm Chandra 112052-112106 pEUV GPU Mode Suppression

This run day attempted to apply the shotstyle developed from 111969-112022 to mode suppressing feedback, as well as verifying that the control coils were having an effect on the plasma.

Results were inconclusive. Overall, the GPU was much less successful in tracking the mode than previously. Some effect on the plasma was found by applying a static vertical field perturbation with the control coils. The applied suppressing flux shape was determined empirically based on the control coil pickup on a baseline shot. There remains an as yet unexplained timebase shift observed in the section nine and ten control coil sets. The problem appears to be digital. Suppression phases of 0, 180, 90 degrees were applied, to no discernable difference in effect.

Future work should focus on returning the plasma to a pEUV-GPU trackable state before resuming feedback work. Example plots will be appended.

Machine changes: All shells are inserted in. The Vf start time was changed from +700 to +800, and has been reverted. The puff time was changed from +1500 to +1700, and has been reverted.

Monday February 14 2022 4:54 pm Li 112108-112130 TangEUV System Ribbon Cable Testing

Ribbon cable testing. The combination of #3, #4, #5, #6 and of #2, #3, #5, #6 both give the lowest noise level. The combination of #2, #4, #5, #6 shows a slightly larger noise. Any combination including #1 shows very large noise. When #1 is used, Ch 12 shows a lower noise than other channels, while it's not even on cable #1.

Thursday February 17 2022 12:06 pm Wei 112131-112181 Dual fast cameras MHD analysis

The goal of today’s run is to first develop a shot style with stable centered MR, edge q right below 3, and growing or saturated n=1 mode. The 2 fast cameras were set up to look at Sec 9’s cross section (CAM-26731 with fiber bundle & demagnification setup) and Sec 1’s HFS-limiter (CAM-26730 on tripod) in order to investigate both the poloidal-toroidal and local poloidal-radial structures of MHD activities prior to disruption.

After much work in trying to reproduce the shot style of 102709 (112134-112160), it was found out that the stable MR=91 cm shot style did not produce sufficient light at the HFS edge, probably due to low recycling rate or wall contact with the inboard wall. We adjusted the shot style so that the plasma MR started at around 91 cm and falls inboard, staying at 89-90 cm for around 2 ms (112161-112177). During these shots CAM-26730 was set up to either a zoomed-in view focusing on the HFS-limiter (112161-112171) or a slightly zoomed-out view (112172-112177). Using this shot style, blob filament structures (single and double layers) similar to what previously found during current quench were observed on CAM-26730 during MHD activities prior to disruptions.

Both cameras run at 256x96 resolution, 100 kHz, 9 us exposure. We did not try to run CAM-26730 at 128x64, 250 kHz due to insufficient light collected. We also did not try to set CAM-26730 to look at LFS edge through the mirror since the viewing angle was too awkward. For future runs we would need to set both cameras up using the fiber bundles, or to try gas puffing to increase emission level at the edge.

Afterward Rian Chandra tested the triggering of the new USB spectrometer (112178-112181).

Star shots: (2 stars) 112167, 112170, 112171, 112173; (1 star) 112166, 112168, 112172, 112177.

Friday February 18 2022 7:41 pm Chandra 112184-112200 GPU Timebase Diagnosis

The purpose of this run day was to diagnose the offset in input and output of the GPU, for channels 32 and above.

The source of this offset is not clear. Directly applying an RMP from the CPCI itself, without the GPU, does not result in the observed offset. Further, the CCoil amplifiers observe pickup from plasma breakdown at the same time. This implies that the timebase shift is not in the current digitization CPCI or amplifier DAC CPCI, and is therefore a real shift being imposed by the GPU.

At the end of the run day, time was spent trying to find a linear shift function which would align the input and outputs. This was unsuccessful. Future work will solve this problem, before feedback shots resume.

Hardware changes:

It was observed that a scuzzy cable was missing in the connection between the pEUV breakout board and the GPU. It is hypothesized that this was the source of the very poor GPU tracking observed last runday. A cable was swapped with the tEUV, both systems observe data correctly.

The tEUV breakout board was placed in a plastic bag, so that it can be stored inside the South Rack, so that it can't be stepped on by accident.

All Shells are in.

Wednesday February 23 2022 8:20 pm Chandra 112201-112297 GPU Timebase Verification and Feedback

This two-day run campaign was devoted to developing a repeatable shotstyle with flat outboard R_m, flat q just below 3, and magnetics indicating 2<m<=3. The campaign also spent time correcting a linear timebase shift in GPU channels n>=32 (corresponding to the second set of analog input channels), and verifying that the output matrix amplitude and shape is correct.

The GPU timebase was corrected with a linear shift. The source of this shift is not clear. It appears some time between shots 109362 and 11962. The appended plot will show the GPU being directly driven by a signal generator for two channels in the correct and shifted set, and two measured control coil output current waveforms, with and without the correction.

The output flux surface appears to match the first two modes of an SVD of the induced pickup on the control coils. This is taken to be the optimal output flux shape. There is an as-yet unsolved question as to what the correct normalization of this vector should be.

None of the shot styles tested showed obvious effect of the control coils (although vertical field pulses show that there is some coupling from the coils to the plasma), or were otherwise too un-repeatable to be useful. One comparison example is appended, showing how the application of feedback (from 2-5ms) does not make an obvious change in the mode amplitude evolution, but does have a strong effect on the midplane SXR, which may imply that the control coils are having some perturbative effect on some part of the plasma edge. Note that this shotstyle has the "m=2.5"-like structure in the contour plots.

Future work will continue shot development.

A few example shots (including one very long one) are highlighted: 112262, 112256, 112238

Hardware changes: All shells are in. BP2 has been withdrawn to 115cm, it can be reinserted to continue cleaning.

Thursday February 24 2022 5:40 pm Li 112299-112358 Tearing shot development

112300-112306: Target shot 107751. Recovered the target shot pretty successfully.

112337-112358: Target shot 106843. Try to recover the target shot, did not match perfectly but got a few good shots.

Shot 112334 is long and has a minor disruption in the middle of the shot. 112327 and 112329 also have minor disruption during the shot.

Star shots to be further analyzed: 112300, 112302, 112303, 112304, 112305, 112306, 112350, 112355, 112356, 112357, 112358.

Friday February 25 2022 3:01 pm Li 112359-112401 Tearing shot development

Used shot 107751 and shot 106843 as target shots to develop two types of tearing mode shots.

The shot development targeting 107751 went pretty well and the star shots are 112360, 112368, 112372, 112379, *112383, *112385, *112388.

The shot development targeting 106843 was ok, the shots are not as long but the signals are good enough to analyze. The mode amplitude in these shots is pretty low though. The star shots are 112391, 112392, 112394, 112396.

Hardware related, TangEUV Ch3 becomes unstable and has an offset of about 9V during a few shots (ex. 112387).

Wednesday March 2 2022 5:32 pm Saperstein 112402-112431 Testing DP2 biasing

The purpose of today’s run day was to test the biasing of the double probe in section 2 (DP2). Unfortunately we ran into issue either grounding the probe and/or digitizing the voltage/current measurements, and we were unable bias the probe correctly.

All shells were in except section 2 and 4T’s shells, which were completely retracted and 1 cm out respectively. Both sections 10 and 4&5 were grounded via jumpers 9.5 and 3.5 respectively. All probes were out at the start of the day. The fuses for the DP2 amplifier were also not present at the start of the run day. Two different biasing boxes for DP2 were used over the course of the run day, one of John’s boxes and one of Ian’s. Digitizer channels and gains for John’s box were all in CPCI 12: voltage (left head) – ch 33 (*212.7), voltage (right head) – ch 35 (*212.7), current – ch 34 (*10). Digitizer channels and gains for Ian’s box were all in CPCI 12: voltage – ch 36 (/0.00535), current – ch 34 (/0.003).

The run day started off with some issues with “take_shot” freezing and issues with CPCI’s initializing, but the issues were resolved by shot 112407 by restarting Oscar and spitzer respectively. Cleanup was quick (112408-112413), and we started pushing in the probe with shot 112414. First pushed probe (floating) into 105 cm from 112414-112416 then pulled it back to 107 cm for 112417-112423. I “attempted” to start biasing with shot 112421, but the biasing setup turned out to be incorrect at the time, since one of the fuses on the output of the amplifier was still missing. Biasing attempts continued until 112426, with the value for the applied voltage specifies for the GPU gradually increasing. I stopped trying to bias the probe after that shot, and only “grounded” the probe instead. Starting with 112428, the second fuse for the amplifier was finally attached, but the amplifier still was not biased for the rest of the run day. This is when we realized the signals we were seeing for voltages and currents were questionable, since the probe being grounded didn’t set the voltages to near zero. So we tried a couple troubleshooting methods, including swapping digitizer channels and bias probe boxes (John’s -> Ian’s). Neither of these methods worked, and the run day ended with using Ian’s bias box and its nominal digitizer channels. We will likely need to do some crates only troubleshooting to resolve this issue later.

Apart from the probe biasing, the attempt to reproduce shot 110429 was somewhat successful (star shot 112429). The shot style does have its differences, but they are likely due to the lack of probes that were inserted to the plasma in 110429.

Thursday March 3 2022 3:16 pm Saperstein 112435-112456 Testing DP2 biasing/grounding – day 2

The purpose of today’s run day was to continue the testing of DP2, with the appropriate grounding of the DP using one of John’s DP boxes. We were able to successfully bias the probe and drive up to 40 A of current, but the current had a noisy response to the present MHD, and there was also some DC offset in the current that existed outside of the biasing period, but only during the plasma lifetime. This MHD response in the probe current decreased as the probe was pushed in, but approached some minimum limit in the effect. Switching to a more quiescent shot style also reduced this MHD response significantly. Hopefully this MHD response is an artifact of some DP circuit issue, that can be fixed in the future.

Machine setup: same as previous run day, except DP2 digitization was replaced with John’s box (see digitization info in previous report) and the grounding scheme was corrected.

The run day started with checking that the corrected grounding scheme was working (112436-112441). The current was zero and the probe voltages were fluctuating about zero with an amplitude around 10 V, which is reasonable. Biasing started with shot 112442 and continued on and off for the rest of the day. Both the biasing voltage and the probe position was slowly changed over the course of the day. The probe didn’t appear to be arcing significantly, so we managed to push the probe into 103 cm by shot 112450. Starting with 112445, a 2 Ohm load was added to the output of the transformer to try and stem the MHD response (assuming it was related to the total resistance of the circuit fluctuating with the sheath impedance at the probe). This helped mitigate the response, but only a little (about a factor of 2). By the end of the day, we were able to bias the probe to about +80 V and +40 A,

Starting with shot 112453, we began trying to reproduce shot 112396, which was a more quiescent shot style. Reproducing this shot style was not difficult (star shot 112455), and did see the MHD response of the current drive drop as expected.

Next steps to trying to reduce this noise include digitizing probe info in the North rack (this is where John digitized his data) and seeing If the noise is as significant when doing wall biasing.

Friday March 4 2022 4:41 pm Li 112458-112500 Tearing shot development

The purpose of today's run is to develop a shot style with low q (slightly higher than 2) and inward major radius (90~92 cm). Shot 107751 is used as a reference but a smaller major radius is aimed for. If the plasma is moved inward, the temperature flat region would show up in an inner or outer region, where the temperature reconstruction results are more reliable.

The star shots today are 112473, 112477, 112485, 112495, 112496, 112497, 112498, 112500. The safety factor could be kept at a stable value, but the major radius tends to flow outward. In 112500, VFS=7600 which is much higher than VFS=6800 in shot 107751, but no obvious difference is seen on the major radius. This needs to be confirmed further.

Shot 112496 can be used as a reference in future runs. Try to keep the major radius stable based on this shot. Some shots that have a small major radius but are not very long are also worth analyzing: 112463, 112467, 112481.

Monday March 7 2022 5:35 pm Saperstein 112501-112520 DP2 troubleshooting – day 3

The goal of today’s run day was to continue to troubleshooting the grounding scheme of the DP2 probe. We had found since the previous run day that the grounding between the DP2 box and the probe stand was unreliable, and wanted to test how better grounding the box would improve the signal measurements. It did improve the grounding of the probe voltage, although we also found that Jensen we were using was introducing some pretty bad noise. Removing the Jensen and directly connecting the voltage divider to the rack reduced the noise to a reasonable range.

Machine setup: Same as previous run day except… probe was at 103 cm the entire day, DP2 started off digitized in the south rack but was moved to the north rack during the run day, the probe head voltages may have been swapped the whole day, and we provided a better ground to the DP2 box.

The run day started with just some test shots to look at the noise, with the signals being digitized in the south rack (112501-112505). While the voltage measurements were about zero, they were still very noisy. From 112506 and on, we swapped to digitizing in the north rack, to see if this influenced the noise, but it did not. We then tried unplugging the probe voltage outputs from the box, and found the noise in one of the CPCI channels was still bad (112507-112508), and that swapping the Jensen outputs just prior to rack swapped the bad channel (112509), implying that the issue likely lies with the Jensen. We then reattached the voltage measurements, but this time without the Jensen (112510 and on, DP2 is already grounded to the north rack, so using a Jensen in this case wasn’t essential). This reduced the noise quite a bit, to a level we could tolerate (about 4V amplitude). The current still had the strong MHD response, but this was expected. We then moved on to comparing the probe signals for different grounding configurations. Probe shorted to stand ground (112510-112511), floating probe (112512), and probe grounded via transformer (112518).

We also tried to do one biasing shot, but for some reason no bias was applied (112520).

Tuesday March 8 2022 5:01 pm Saperstein 112521-112550 DP2 troubleshooting & wall biasing – day 4/1

The goal of today’s run day was to finish troubleshooting Jensen issues with DP2 and to start testing the wall biasing scheme. It seems like the DIN-2LOF Jensen that I was using before was faulty in some way, since the replacement DIN-2LOF Jensen seemed to work just fine. The wall biasing scheme also worked quite well, putting out very flat currents that were also seen on jumper 9.5 (A). Unfortunately, the Ip-Rogowski was not able to see a noticeable difference in current when jumper 9.5 was disconnected, forcing the current to travel through the plasma.

Machine setup: Same as previous day, with DP2 signals digitized in north rack and voltage measurements bypassing Jensens at the beginning of the run day.

The first set of shots troubleshot the Jensens, replacing the DIN-2LOF Jensen we were originally using with another of the same model. The new one had no issues. We also tried comparing the same model to the newer DIN-PB model that we’ve been using for the new SOLC tiles. There was no significant difference, even when the Faraday cages of the DIN-PB were grounded. The star shot to see for this was 112530, which had both voltages passing through the new DIN-2LOF with the probe in the shorted to ground config, and should be compared to 112522, which was the same except the Jensens were being bypassed. We ended this portion of the run day with some biasing shots of progressively stronger biases (112533-112535). 112534 and 112535 are of note because they managed to transiently output about 100 A, with 112534 reversing the direction of the mode rotation and speeding it up to almost 30 kHz.

We then swapped over to the wall biasing scheme (112540 and on), and were no longer digitizing DP2 data, even though DP2 was still inserted. An explanation of the scheme is as follows…

* Used one of Ian’s BP boxes. The positive “to probe” lead was attached to section 10 at the same location jumper 9.5 (A) is attached, and the negative “to probe” lead was attached to the same location in section 9.

* Before running to section 10, the positive “to probe” lead was sent through a 2 Ohm load.

* The voltage and current signals were digitized in the south rack CPCI 12. The voltage being in channel 33 with a gain of x100, and the current being in channel 34 with a gain of x20.

For the first several shots (112536-112539) we ran into issues with Oscar freezing and rack failing to prepare, but this was eventually resolved with a reboot of spitzer. We first tested the wall biasing circuit without any bias (star shot 112542) and then went on to bias with progressively larger biases (star shots 112543-112547). We found that at about 5V input to amplifier the voltage applied started saturating. Jeff seems to think this is likely a limitation of the amplifier. We then ended the day taking some biasing shots with jumper 9.5 (A) disconnected, to force the bias current to run through the plasma. The purpose of this was to force the current to run through the Ip-Rogowski and see if the Ip-Rogowski could resolve the change in current over the natural Ip (-ramp). The star shot for this was 112550. Unfortunately, we have yet to be able to resolve the change in the Rogowski signal. This was made difficult by the fact that the Ip ramp-rate was relatively fast compared to the timescale of the current pulse. It’s also possible the current ran co-Ip from section 9 to 10, bypassing the Ip-Rogowski. Future tests may want to try biasing current from section 9 to 10 instead, although the fact that section 10 is grounded through section 9 may introduce some grounding complications.

Wednesday March 9 2022 4:45 pm Li 112551-112605 Tearing shot development

Today's run target is to develop tearing shots with an inward major radius. Star shots are 112560, 112561, 112563, 112564, 112579, 112580, 112586, 112587, 112589, 112590, 112596, 112598, 112599, 112601.

112589 and 112590 are very long, but unfortunately tangEUV Ch3 was bad during 112590.

112596 has a major radius around 89 cm, and 112601 has an even more inward major radius.

Overall, the desired shot style is achieved. Whether it gives better temperature reconstruction results needs to be checked after analyzing the data.

Monday March 14 2022 5:59 pm Chandra 112606-112712 GPU-Feedback

The first part of this run campaign (112606-112654) was focused on verifying the output timebase and mode suppressing flux shape.

An error in the GPU code was identified as the cause of the linear timebase shift previously reported on. This has been corrected. The suppressing flux shape is now based on linearly shifting four sine/cosine pairs to match the induced current shape on the four toroidal control coil sets, instead of the previous method in which the correct flux suppressing shape was predicted analytically given the control coil poloidal positions.

The second half of this campaign (112655-112712) was spend applying these changes to a mode suppression compass scan. The target shotstyle was repeatable, outboard limited (R_m), and q=2.5-3. This gave a roughly 3ms period for feedback to be analyzed in. The compass scan had 45^o angular resolution, with three shots per phase and a baseline shot interspersed every four shots.

It is unclear if the mode amplitude plot from hbtplot shows discernable amplification or suppression, but an analysis of the power spectrum of an FFT of the Mirnov probes will be appended to this report once it is completed. Hopefully this will show a clearer effect.

Tuesday March 15 2022 5:14 pm Wei 112714-112785 Fast camera-based mode tracking using machine learning method, preliminary run campaign

The goal of today’s run is to develop appropriate feedback-type shot style and repeatedly take camera videos. The 2 fast cameras were set up to look at Sec 9 cross section (CAM-26731 with fiber bundle & demagnification setup) and Sec 1 inboard cross section (CAM-26730 on tripod). Both cameras operated at 128x64 resolution, 250 kHz, 3 us exposure.

2 shot styles were developed: 1) outboard limited plasma, MR between 92-93 cm, downward sweeping edge-q, rising mode amplitude (112717, 112723-112752, 31 shots, target shot 112723) , 2) centered plasma, MR between 90-92 cm, saturated mode amplitude (112763, 112765-112785, 22 shots, target 112763 & 112775).

During the shot development phase we also tried adjusting the VF start time. Shot 112758 had a bad breakdown which was recorded on the cameras.

Wednesday March 16 2022 5:16 pm Saperstein 112786-112822 Wall Biasing Shot Dev

The goal of today’s run day was to do some shot development to reproduce shot 110429, and more specifically, the small MHD response jumper 9.5 (A) saw during that shot. Other goals included beginning the wall biasing experiment once a suitable shot style was found. By the end of the day, we were able to more-or-less reproduce 110429, including the small MHD response in the jumper, and performed some biasing tests throughout the day. We also added resistors to the jumper A cable to increase the wall impedance, and encourage bias current to run through the plasma. This effect was noticeable with a 100 mOhm resistor, and dominant with a 1 Ohm resistor. Unfortunately, we have still yet to see a significant AC response in the jumper to DC biasing.

Machine setup (at start of run day): All shells were at their 1cm* positions except the sections with probes (secs 2, 4, 5, & 9), which were fully retracted to allow easy movement of the probes. For reference, the 1cm* position has the shells out 1 cm from their original fully inserted positions. For the new shunt SOL tile diagnosed sections, this is now their fully inserted position. RP4, RP5, and TP9 started off at 107 cm, while DP2 was at 104 cm. Sections 4/5 and 10 were grounded via their local jumper cables. The tEUV system was disconnected from its CPCI channels so as to digitized both the shunt SOL tiles and wall biasing diagnostics simultaneously. The wall biasing circuit was setup, with the same current/voltage digitization scheme in the south rack used previously (3/8/2022). When we switch to the long jumper A scheme, the jumper current is no longer being measure by the typical Rogowski. Instead, it’s being recorded by a x40 Pearson and digitized in the south rack CPCI 12 ch 35.

The run day started off with trying to reproduce 110429 with the shells and probes positioned as detailed above. We were actually able to reproduce it rather quickly, starting with 112790, and it appears to be relatively consistent (112791). Unfortunately, while the shot style was the same, jumper A was still seeing a lot of DC and AC (MHD response) current. To resolve this, we changed the machine setup until it matched that for 110429 well enough to reproduce the small AC response. First, we replaced the short jumper we’ve been using with the longer one we used during the 110429 campaign (changed jumper digitization here), and while this did reduce the DC response of the jumper, it had little effect on the AC response (see 112792). We then inserted RP4&5 into 103 cm over the course of several shots (112794-112797), but this alone did not reduce the AC jumper response. We also pulled TP9 for these and the following shots. We then finally pushed in the rest of the shells to their 1cm* locations (with the exception of 4B), and found that this change did in fact remove the AC response (see 112798). Unfortunately, the mode behavior changed a bit with this change, which made it a bit more inconsistent between shots. We tried retracting the RPs back out to 107 cm (112802), but this actually reintroduced the significant AC jumper response. Implying that both the probes needed to be in and the shells needed to be in to reproduce the small AC response. We left the RPs back in at 103 cm for the rest of the day. From then on, we tried a little bit of shot development to try and make the mode behavior more reproduceable and similar to that seen in 110429 and 112790. There was some mild success in this, with the final shot, 112822, having decent mode behavior, but it’s major radius could be kept flatter.

Biasing tests were also performed sporadically throughout the day. Starting with 112806, which demonstrated a clear jumper response to the applied bias, with seemingly all of the bias current passing through the jumper. We tried both 1 ms pulse biasing from 2-3 ms (1128036), as well as longer 3 ms biasing from 2-5 ms (112819 & 112822). Resistors were also added to the jumper. A 100 mOhm resistor was added for 112808-112819, and a 1 Ohm resistor was added for 112820-112822. The 100 mOhm resistor biasing saw about 75% of the current pass through the jumper, while the 1 Ohm resistor biasing only saw about 25% of the current pass through the jumper.

Unfortunately, even though the bias was more-or-less DC for the 1 Ohm resistor case, we have yet to see a significant AC response in the jumper that may suggest current path rotation effects (compare 112821 and 112822). The next step should be to retract out one of the section 10 shells and reperform the test. This should make the collection area in section 10 less prone to averaging over the mode rotation and smearing out the rotating current path effects.

Thursday March 17 2022 5:22 pm Saperstein 112823-112867 Wall Biasing Exp – Day 2

The goal of today’s run day was to try to see the AC jumper response to the DC bias, either with or without any natural MHD jumper response, with the ideal case being without any natural MHD jumper response. Unfortunately, we were not able to see this in either case, at least not anything that was clear and substantial. It may be possible that the unknown effects that suppress the natural MHD response of the jumper also suppress the effects of rotating current paths, implying that a natural jumper MHD response may be necessary to have in the background. This makes direct interpretation of the experiment a lot more difficult, but not impossible. A series of relatively similar shots were taken with the wall bias on and off, and their average characteristics may be used to identify the current path effect. We did try retracting out section 10B to reduce contact area and increase the dependence on the current path, but no significant effect was noticed.

Machine setup (at start of day): Same as end of previous run day. All shells at 1cm* positions, except 4B. RPs at 103 cm and DP2 at 104 cm. 1 Ohm resistor added to long jumper 9.5 (A), with jumper current still digitized through Pearson in south rack.

The run day began with trying to reproduce shot 112791, although the best we could do in this machine config was 112827. This shot style was not very reproduceable, and while it led to similar-ish features in mode amplitude profiles, they were different enough to make comparisons between shots difficult. We then retracted out section 10B (where is stayed for the rest of the run day) starting with 112828, to see if this might amplify the rotating current path effect by reducing the possible contact areas in section 10. It’s difficult to say if this really had any effect, though. Shots 1112828 (reference) and 112830 (biased) can be compared to see little-to-no effect.

We then retracted out the sec 2 shells between 112831-112838, to get back the natural MHD response in jumper A, and to see if biasing in this case made the current path effect more noticeable. Unfortunately, it was difficult to notice any effect. Compare bias cases (112833 and 112835) to reference case (112832).

We then swapped the 1 Ohm resistor on jumper A with the 100 mOhm resistor, to see if the effect was more noticeable. We still didn’t see any significant effect, though. Bias shots: 112843 and 112854. Reference shots: 112841 and 112851. Comparing 112841 and 112843, it does seem possible that there is some effect present, but it’s difficult to tell because the mode amp profiles between the 2 shots were too different. We also tried biasing negatively (112844), but there was no significant difference. Shots 112839-112855 were mostly dedicated to trying to develop a consistent and useful shot style in the small natural MHD jumper response scheme, but no such shot style was found.

The run day ended with a series of shots taken with the rake probes back at 107 cm, in the significant natural MHD response scheme (112856-112867). This shot style was at least relatively reproduceable, so a series of bias and reference shots were taken, so that the MHD response in both cases could be analyzed on average and compared, so as to maybe identify some current path effect on top of the natural MHD response. Reference shots: 112856, 112857, 112858, 112862, and 112863. Bias shots: 112859, 112861, 112864, and 112866. A vacuum shot for this shot style was also taken (112867).

Tuesday March 22 2022 8:45 pm Chandra 112868-112966 GPU-Tomographic Mode Suppression

The goal of this run campaign was to diagnose the incorrectly saved suppressing flux matrix, correct it, and then redo the kink mode suppression compass scan of shots 112655-112712.

It was determined that the matrix was not being mapped correctly to the GPU output threads. This has been corrected, and plots will be appended to show a good comparison between the output current and target shot pickup current at two half-period timepoints.

With the corrected matrix, it was found that activating the control coils at high gain immediately locks the plasma, leading to a disruption. By running the coils at roughly 1/2 their maximum current (+/-20A), we were able to get between 1-2ms of data collection period.

Preliminary analysis does not clearly show signs of suppression or amplification of the kink mode, however, there are signs that the mode may be being accelerated and decelerated in the suppression/amplification phases, respectively.

Plots to follow.

Repeatable baseline shotstyle with a 5-10G saturated mode, reasonably flat q=2.8-2.6, reasonably flat R_m90.5-91.5 can be seen in shots 112964-66. Stripy plot analysis indicates that the plasma is 3/1 dominated from 4-5ms, and mixed 3/1-2/1 from 5ms onwards.

Wednesday March 23 2022 6:01 pm Saperstein 112967-113004 Wall Biasing – conductivity scan – day 1

The goal of today’s run day was to bias the section 10 section/tiles while also retracting it out in order to measure the I_tile(r_shell) profiles, so as to deduce a conductivity profile. This would then inform how significantly the plasma displacement in the presence of a mode would effect the bias current emitted by the tiles/shells. By the end of the day we had retracted out the sec 10T shell to 2 cm from it’s 1cm* position, and still saw significant bias current in the SOL10-3 tile and with little change to the bias current passing through the jumper. The shell still needs to be retracted out further until the bias current leaving the sec 10T tiles drops to near zero.

Machine setup: Same as setup on 3/17/2022, except jumper 9.5 (A) has a 1 Ohm resistor and both sec 10 shells were inserted at the start of the run day.

The beginning of the run day was just clean and shot development (112967-112980), with us achieving something close to the desired shot style with shots 112978 and 112980. Following that, we began the bias shell scan (112981-113004). We started biasing with 10T fully inserted (0 cm), with the star reference shots being 112978 and 112980, and the star bias shots being 112981 and 112982. We then pulled 10T back 1cm, and got star reference shots 112984 and 112985, and star bias shots 112986 and 112987. We then pushed the shells back to 0.5 cm retracted, and got star reference shots 112990 and 112991, and star bias shots 112992 and 112997. The shells were then pushed back to 1.5 cm out, with star reference shot 113000 and star bias shots 113001 and 113003. Finally, we pushed the shells back to 2 cm out, with star reference shot 113005 and star bias shot 113007.

Looking at the data, we found that bias current in SOL10-4 dropped ~75% by 1.5 cm, the bias current in SOL10-3 dropped only ~50%, and the bias current in jumper 9.5 (A) only rose ~15%.

Thursday March 24 2022 5:19 pm Saperstein 113008-113052 Wall Biasing – conductivity scan – day 2

The goal of today’s run day was to finish the conductivity scan in the low natural MHD response (NMR) shot style, and then start the scan in the high NMR shot style. Both of these goals were reached by the end of the day.

Machine setup: same as end of previous day, with the 10T shells still 2 cm out.

The day began with cleanup until we reproduced well enough 113005 from the previous day (113013). We then pulled 10T back to its final position at 2.5 cm, with star reference shot 113015 and star bias shot 113017. We could not pull the shell back further than this, because it hits the retraction stopper just after 2.5 cm.

We then pulled both RP’s back out to 107 cm, to perform the same scan with the high NMR shot style (113014-113052). We started with 10T 2.5 cm out, but we had a hard time with shot dev, so we fully inserted it again starting with 113026. We got some okay shots at this shell position (113029,113035,113036), but they weren’t the best, and we came back to this position later to get better shots. We then moved 10T to 1 cm out (star reference: 113037, star bias: 113038), 2 cm out (star reference: 113039, star bias: 113040), and 2.5 cm out (star reference: 113043, star bias: 113047), before going back to the fully inserted position again (star reference: 113049, star bias: 113052).

It's difficult to say as of yet if there is a significant difference in the scale length of the low and high NMR shot style. More analysis is required, as well finishing of the high NMR scan with the 0.5 cm and 1.5 cm positions. We did notice, however, that the conductivity sometimes peaks about 1 cm outside of the LCFS, and that the rotating current path effect seem to be clear enough in when biasing 10T 1 cm out.

Friday April 1 2022 11:03 am Saperstein 113053-113063 Wall Biasing – blade limiters retracted 2 cm

This was a short run day, with the goal of reperforming the fully inserted wall biasing experiment with the blade limiters retracted out 2 cm instead of being fully inserted. This was to eliminate (or suppress) the blade limiters as possible bias current collectors, so that the bias current paths could be better diagnosed by the SOL tiles. The experiment was performed, but we still found that we were unable to track down the paths of the bias currents (no new significant currents showed up in SOL tile measurements).

Machine setup: Same as previous run day, except the blade limiters were retracted out 2 cm from their fully inserted positions (this will be their new semi-permanent positions for now) and the rake probes were inserted back to 103 cm, so as to work in the low Natural MHD Response (NMR) scheme.

Most of the run day was just cleanup and shot development to get back to the 112981 shot style. The star reference shot in this configuration was 113061 and the star bias shot was 113062. Unfortunately, comparing the two did not show significant (> 1 A) change to the non-section 10 tiles.

Wednesday April 13 2022 5:57 pm Li 113072-113121 Sawtooth study

The goal of today's run is to see if we can find the condition for the sawteeth-free shot.

The initial shots do not seem to show sawteeth features on the tangEUV signals and temperature reconstruction. This might be because the overall temperature is too low.

The probes are suspected to cool the plasma down. So from shot 113107, probes at sections 2, 4, 5 are retracted to 115 cm.

From shot 113110, the OHS is lowered from 6000 to 4500 and the current ramp is steep, but still have sawteeth. 113115 to 113121 are all good shots to look at. 113119, 113120 and 113121 show major radius drop at around 3.5 ms. Try lower OHS and stronger current ramp tomorrow.

Thursday April 14 2022 6:17 pm Li 113122-113171 Sawtooth study

Today's run continues to develop sawteeth-free shots. The plasma seems to have something going on at the first ms in all shots. Wiggles show up in the current and major radius profiles. Might need to replace the gas bottle, or the banks may have some issues.

113128, 113130, 113131, 113160, 113163, 113168: These shots seem to be sawteeth-free or have very weak sawteeth.

113133, 113136, 113137, 113138, 113146, 113149, 113150, 113151, 113152, 113153, 113154, 113166, 113170: These shots have clear sawteeth, even under strong current ramp condition. Need to have a closer inspection on the relation between the occurrence of sawteeth and the q profile. In some shots, eg. 113130, a rotating mode can be seen on the temperature reconstruction profile.

Sunday April 24 2022 10:43 pm Chandra 113172-113228 Magnetic Mode Suppression

The goal of this run campaign was to evaluate how sensitive mode suppression is to plasma shot style, by testing if the previously working magnetic feedback code produces results on the current shot-style.

We find that better acceleration and deceleration is observed than with the EUV mode phase observer, and that hints of mode suppression and amplification can be seen, but results are not as clear as in Chen's thesis.

The reason for this may be a slightly improved phase tracking. Results suggest tailoring the shot style closer to Chen's, if this is possible. Prior attempts to do so were very challenging.

Further analysis will be appended.

Note from Alex: " Swapped back to normal digitization for jumper 9.5A starting shot 113172 "

Tuesday April 26 2022 5:29 pm Wei 113237-113288 Fast camera-based multimode tracking using neural networks, second run campaign

The goal of today’s run was to reproduce the shot style of 112723 (shot-style-I from the 3/15/22 run campaign) and to take videos using both cameras now looking at the plasma cross-section toward opposite toroidal directions. Prior to the run day, CAM-26730 was set up using the 12.5mm F/1.4 fujinon prime lens with the (older) fiber bundle & (newer) demagnification lenses. This camera now looks toward Sec 1 cross-section directly through the port window. CAM-26731 still uses the previous setup and looks toward Sec 9 cross-section through the mirror. Both cameras operated at 128x64 resolution, 250 kHz, 3 us exposure. All shells except the Sec 10 ones were fully inserted; Sec 2 & 6 shells were out by 1 cm from the previous “fully-inserted” position.

31 shots (113241-113271) were taken. Multimode (n=1 & 2) activities were identified on FB2 array in these shots. Afterward, we adjusted CAM-26731’s object lens so that the camera’s view moved slightly (113272-113279) and significantly (113281-113288) toward the left and then took additional shots. These shots will be used to test the trained ML model’s robustness to shifted camera views.

Tuesday May 3 2022 5:45 pm Chandra 113289-113417 Reproducing Chen's feedback shots

The first day of this run campaign focused on reproducing target shot 85470, in which applied feedback was found to elongate the shot. This shot style was found difficult to reliably repeat, with no relevant star shots to report.

Phase two of this project was to spend more time trying to recreate Chen's feedback baseline shot 86061, and verifying that the GPU using his code was outputting correctly. We verify that all available control coils are firing as expected. The shot style was reasonably repeatable, but still with drift in q, Rm and the puff pressure for some reason. A plot showing this change is attached. Preliminary results suggest that the plasma is responding to the control coils, and the next phase of this work will be a compass scan with the magnetic code, followed by the same scan with the EUV-GPU code.

Star shot is 113347.

Hardware changes: Section 10 shells inserted fully, and the GPU Analog-Out to control coil amplifier breakout board scuzzy cable was re-attached after being seemingly unintentionally unplugged.

Friday May 6 2022 7:42 pm Chandra

Use "

" to force a line break. Carriage returns are ignored.

Friday May 6 2022 8:04 pm Chandra 113421-133572 Reproduction of Magnetic Feedback Results

[A previous blank summary may have been submitted by mistake]

This is the second phase of the magnetic feedback reproduction campaign. Shots 113431-113480 were a 30^o compass scan using Qian's feedback code on a reproduction of his shot style (original: 86061, reproduction 113347, featuring a longer saturated mode period). Phases 90^o±30^o had to be discarded, as the plasma locked and immediately disrupted. Overall, results from the scan were mixed. With two shots per phase, the plasma seemed to drift too much for useful results to be extracted. We note that no mode suppression is observed.

The second part of this phase implemented PID gains instead of Qian's static proportional gain with a changing phase. A combination of the PID gains and a better controlled plasma shot style gave very promising results. Appended will be plots showing clear, repeatable, unequivocal amplification of the mode during a natural low period, up to and slightly above the saturated level, once the control coils activate. There appear to be some small ability of the coils to target specific amplitudes, but only within an envelope of roughly ±3G. Note that it was not possible to amplify the mode above it's saturated value during the saturation phase for any stable amount of time.

Mode suppression is still not observed, in fact if anything the mode appears amplified in the suppression phase. However, strong deceleration of the mode is observed. To verify these results, the PID controller is used but the phase is perturbed ±20^o and ±45^o. Results from the ±20^o tests appear roughly identical to the purely amplifying/suppressing cases, if slightly less reproducible.

Finally, and the original goal of this project, the GPU input was switched back to the pEUV system. The system was verified to still track the mode, although the phase envelope is relatively poor (±19^o). Insufficient time was left over to attempt reproduction of the magnetic feedback results, two run days are requested next week to continue this study.

The goal of this campaign is to show that the EUV input system produces sufficiently similar results to the magnetic input system. Solely amplification results, while disappointing, may be sufficient.

Thursday May 12 2022 1:22 pm Chandra 113573-113652 EUV-GPU Mode Suppression

The first phase of this run campaign (113573-113596)was dedicated to implementing PID gains for the EUV-input controller, and verifying that the output from the control coils looked reasonable. The target shot style remained the same (86061/113347). It is noticed that the tomography system has difficulty returning clear profiles/tracking the mode phase when the mode amplitude is low, early in time.

In the second phase of the campaign, we attempted to solve this issue. It is observed that the system frequently looses the phase tracking, and accidentally applies a static flux surface, locking the mode and leading to a disruption. A few methods to detect and halt this process were tested, but to no success.

Finally, we tested applying a rotating RMP with half the coils, and running the suppression algorithm with the other half, with the goal of raising the baseline mode amplitude during the low-amplitude, controllable phase. The coil output appears correct, but the mode was not clearly amplified or suppressed.

Next steps will attempt to further investigate the artificially-amplified RMP mode.

There are no significant hardware changes to report, although the V_f start time has been increased to +800 from +700

Thursday May 12 2022 5:48 pm Saperstein 113653-113722 Bt scan 2nd campaign – CQ time control – day 1

The goal of today’s run day was to get low Bt (Bt ~ 0.17 T ~ ½ Bt_normal) shots that have short CQ times, such that the plasma remains large for the majority of the CQ. This may have been achieved for a select few shots, but most of today’s shots were very inconsistent and plagued with beta-collapses that had the major radius crashing too far well before the current spike (if any) was triggered. A shot style that keeps Ip (relatively) high with 2 < q* < 3 has been found to be the most promising so far, but further shot development is still required.

Machine setup: All shells fully inserted, except for 4B which was out. All probes out. Sec 10 grounded via jumper A (9.5) and secs 4&5 grounded via jumper B (3.5). A 350-450 Ohm resistance to ground was found from secs 4&5 even when in the “floating” configuration. This shouldn’t be an issue when we’re operating in the grounded sec 4&5 configuration like we are here, but may be an issue if left operating in the nominally “floating” configuration.

Started off trying to keep q* high (> 3), but these shot styles were very inconsistent and prone to single beta-collapse that results in the plasma crashing far inboard well before the disruption actually begins. The best shots I was able to get under these conditions were 113669 and 113678. They still began crashing far in well before the disruption, but they were stable enough from the beginning of the shot to potentially just need some shot dev. Unfortunately, this shot style incredibly difficult to reproduce, so we stopped trying to reach it.

Instead, we tried changing the operating space to have q* live nominally between 2 < q* < 3, with the hope that the additional current we could get in this regime would help to better heat the plasma and reproduce normal operating conditions with fewer beta-collapses. This worked moderately well. The first couple of shots (113681,113682,113683*,113685) were free of major beta-collapses and decently reproduceable at the time. They crashed in a bit before the disruption, but this could possibly be fixed with some shot dev. Unfortunately, after refilling the puff line, we had a hard time reproducing this shot style again. Instead, we later tried dropping the major radius so that we could pump up Ip some more and still live between 2 < q* < 3. This never really ended up working very well though, and we had issues where Ip always continued to have a negative ramp that was plagued with multiple beta-collapses. Messing with the puff time wasn’t able to resolve this either.

In the end, we just tried reproducing 113683 again, and we were fortunately able to get something close enough by the end of the run day (113722). Tomorrow we’ll continue shot development starting with this shot style, and try to prevent the major radius from crashing in at the end.

In terms of analyzing the scaling law for the decent shots we have today, only 2 managed to keep the major radius decently high enough over the course of the CQ, 113685 and 113722. However, R still managed to fall well under the critical threshold around half way though t80->t20. Additionally, the scaling still did a poor job prior to R crossing Rcrit, where it overpredicted the frequency quite a bit. It’s possible that these plasmas are so cold that the post-disruption temperature is significantly smaller than it is in normal Bt disruptions. Next time, we should also try raising the temperature of the plasma to the best of our ability. This is difficult for low Bt shots though, since our plasma is only heated Ohmically (and these low Bt shots are very resistive) and because the reduced field will decrease our energy confinement time. It’s possible we could attempt to raise the temperature by lowering the puff time/pressure, but this may not be enough. We may have to just try operating at higher Bt (maybe ¾ Bt_normal).

Friday May 13 2022 6:01 pm Saperstein 113723-113760 Low Bt 2nd campaign – day 2

The goal of today’s run day was to continue the low Bt campaign and further shot development by keeping the plasma further out, starting with the 113722-shot style, and varying the puff time if necessary. Unfortunately, we were not able to find a suitable shot style. Varying the coil settings away from those used for 113722 resulted in most shots being plagued with beta-collapses, crashing inboard before the disruption even began, and varying the puff time didn’t appear to have any beneficial effects. We were able to decently reproduce the 113722-shot style whenever we returned to its settings though, so at least that was a relatively reproduceable shot style (star shots 113732, 113734, 113736, 113745, 113748, 113754). There were also some shots that weren’t perfect, but still managed to disrupt at least relatively far outboard (113738* and 113741*). After today though, it doesn’t look like we’ll easily be able to improve the shot style to have the disruption major radii I’m looking for. Next time I should just try operating at higher Bt.

Another concern that came up over the run day was that these shot styles might have significantly lower post-TQ temperatures than our normal Bt disruptions. The concern for this stems from the fact that the scaling law still does a poor job of predicting the HC rotation frequency even before Rcrit is reached (see 113738 and 113741). There are other potential signs of concern as well, but those lines still need to be developed. This again supports the push to try operating at higher Bt.

Tuesday May 17 2022 3:54 pm Li 113762-113808 Snake & sawteeth shots

Today's goal is to get better data for shots with rotating snake structures and see if we can switch between shots with and without sawteeth by changing bank settings.

Snake shots are: 113765, 113770, 113779, 113784, 113785, 113786, 113791, 113793, 113797, 113800, 113804, 113806, 113807.

Shots with a drop in the middle (possibly caused by outboard major radius and contact with the outer walls): 113767, 113792, 113794, 113795, 113796, 113801, 113806, 113808.

Shots with sawteeth: 113791, 113802, 113803.

From shots 113800~113808, changing VFE seems to affect the existence of sawteeth, but the relevance needs to be further checked.

Wednesday May 18 2022 6:53 pm Li 113809-113886 Sawtooth study

Today is a continuing run day from 05/17/2022, to study if we can manipulate the occurrence of sawtooth.

When the major radius is more outboard and the plasma tends to hit the wall and cools down, shots like this (I call them 'hit wall shots') tend to have sawtooth. When the major radius is more inboard, they tend to be sawtooth-free shots. The actual cause for this might be the difference in q. 'Hit wall’ shots have larger Rm, leading to lower q, then leading to sawtooth. Example: 113812&113823, almost the same Rm, lower edge q comes with sawtooth.

Also tried to switch to a different current profile: higher flatter Ip since 113840. Need to look further to see if different Ip with a similar major radius has a correlation with the occurrence of sawtooth. But the above primary conclusion about the major radius impacting the occurrence of sawtooth is still valid.

113851, 113852, 113853: same bank settings, higher q shots are sawtooth-free.

Compare 113878&113880, 113881&113882: different Rm leads to different shot styles.

Thursday May 19 2022 4:54 pm Saperstein 113887-113951 Low Bt – cleaning triple probe

The goal of today’s run day was to use the triple probe (TP) in section 9 (TP9) to measure the density in the low Bt (1/2 normal) plasmas, and then compare them to the densities in normal Bt plasmas. However, because TP9 was never finished being properly cleaned since the last time (a little under a year ago), most of the run day was devoted to just cleaning and further inserting the probe. We were able to reduce the arcing frequency to reasonable amounts (a couple short arcs per shot) by the end of the run day. We also got some normal Bt shots with TP9 in, that can then be compared to the low field shots.

Machine setup: Same as previous low Bt run day, with the exception that TP9 has been in the plasma all day at various positions from 108->104cm. The fast cameras were also turned on starting with shot 113932, but the shells are in so that views aren’t great. TP9 digitized in south rack CPCI 10 chs 93-96. Isat – 96 (gain ~ -0.48x), V+ - 95 (gain ~ 106.55x), V- - 93 (gain ~ 106.55x), vfloat – 94 (gain ~ -106.82x). The gains are suspect though, so don’t assume they are correct.

Most of the run day was just slowly inserting TP9. It was at 107cm from 113890->113902, at 106cm from 113903->113917, at 105cm for 113918, and at 104cm from 113919 on for the rest of the run day. We also tried some shot dev involving lowering the puff pressure/time and raising VFS. We found that the shots became too funky for puff times below 1300us (~90uT), and that they were usually best between 1500->1650 us (>100uT). Star shots with TP9 at 104cm include 113943 and 113945.

Towards the end of the day, we also started trying to get shots at normal Bt, trying to emulate 110428, except with the shells still in and no other probes in the plasma. The reliability of the gains we have for TP9 are suspect at the moment, so directly comparing a low Bt and normal Bt shot is better than just looking at the absolute measurements. We were able to pretty successfully reproduce the shot style, with the star shot being 113950.

Tuesday May 24 2022 6:29 pm Chandra 113952-114016 EUV-GPU Amplification

The goal of this run day was to combine a rotating RMP with the EUV-tracked GPU output.

The first part of the run day was dedicated to testing a new CCoil pole correction algorithm, to avoid the HF signal amplification and potential locking in Qian's algorithm. This was largely unsuccessful.

In the second part of the campaign, half the coils were used to apply an RMP, to assist the mode tracking, or to provide an artificial amplitude to suppress. Sometimes, the plasma responded to the RMP. Other times, it didn't. Reasons for this are not clear. Plots appended.

Wednesday May 25 2022 7:16 pm Chandra 114018-114074 EUV-RMP Shot Development

The goal of the day was finding a shot style style which would reliably respond to the control coils.

The "Qian Reproduction" Shot style was reproduced reasonably well, and although the RMP did modify f_{mhd Comparison of successful feedback shot suggested using a style with lower q, and potentially higher Rm. A semi-repeatable and noticeably different style was found, but neither the coils nor the EUV feedback system were found to have an effect at all on the amplitude, or frequency.}

It appears that the plasma is only narrowly sensitive to the control coils, and that natural variation in shots strongly complicates observing a clear effect in mode amplitude, although the effect in mode frequency is, as previously reported, very clear.

It may be worth spending a few more days tayloring the shotstyle, but after that it is unclear how to proceed. It is further unclear why reproducing the previous magnetic mode suppression results, on very similar shots, is seemingly so challenging. The only major seemingly relevant change is the SOL tiles, and the ferritic wall.

Plots (including star shots) appended.

Thursday May 26 2022 5:10 pm Saperstein 114075-114130 Ip and density scan – effect on post-TQ Te – day 1

The purpose of this run campaign is to determine how the post-TQ temperature and HC rotation frequency scaling depend on the pre-disruption density and plasma current. The goal of today’s run day was to first vary the density via the puff time/pressure. We tried both lowering and raising the puff pressure relative to the usual puff pressure (~67uT), but found that while the puff pressure could be lowered down to ~54uT, the plasmas were relatively inconsistent and moved a lot (which makes making TP comparisons difficult). Additionally, the amount the density was being lowered by at these pressures didn’t seem all that significant anyways, so we mostly focused on raising the puff pressure as much as we could. By the end of the day, we were able to effectively double the density (as measured by the TP) by roughly doubling the puff pressure. However, we did not find that either the HC rotation scaling or the post-TQ Te (approximated dynamically using the loop voltage and Ip) changed significantly with the doubling of the density. The lack of an effect on the HC rotation scaling was somewhat expected, but the lack of an effect on the post-TQ Te does come as somewhat of a surprise. Although, it could be possible that either our post-TQ Te approximation is invalid or that the post-TQ density is not strongly related to its pre-TQ counterpart.

Machine setup: Same as my [Saperstein] previous run day, with TP9 @ 104cm and same digitizer nodes/gains. The only change worth noting is that the deuterium bottle was replaced before the run day began.

The baseline shot style was established at the beginning of the day by shots 114084-114086, with 114086 being the star shot because its CQ major radius stays above the critical (reliable) major radius for the longest portion of the CQ. Some more baseline shots were also taken at the end of the day (114129&114130). This shot style was chosen as the baseline since it is the one that was used to establish the HBT HC rotation frequency scaling in my paper. We then moved on to lowering the puff time/pressure down to 600us/54uT. We did not get very good breakdown at these pressures, so we raised them up to 800us/57uT instead, where we managed to get better breakdown and decent shots, but TP9 starting arcing quite frequently at the beginning of the shot. In the end, the densities that TP9 seemed to see didn’t vary significantly from the baseline, so we decided to back off trying to lower the density any further (or really at all). The best shot from this part of the campaign was 114093.

We then moved on to raising the puff time/pressure above the baseline (1000us/67uT). We were able to raise the puff time/pressure up to about 2400us/143uT without issue, before we found that raising the puff time no longer raised the puff pressure any further (going from 2400us->2800us). This resulted in more-or-less a doubling of the density relative to the baseline, with a corresponding drop in Te that left SXR signal varying weakly with changes in density. Star shots and rough density increases for each puff time/pressure include…

(1600us/99uT) star shot(s): 114099*&114102, ne increase: ~1.5-2x. (2000us/128uT) star shot(s): 114107,114109,&114112*, ne increase: ~2x. (2400us/143uT) star shot(s): 114124&114126*, ne increase: ~ 2x.

Friday May 27 2022 4:45 pm Saperstein 114131-114158 Ip and ne scan – day 2

The goal of today’s run day was to drop the pre-disruption plasma current relative to that of the baseline (114086) to see what effect it would have on either the post-TQ temperature and/or the HC rotation frequency scaling. Only some progress was made though, as we ran into fundamental issues with running at low current as well as many issues concerning beta-collapses. The fundamental issue was that we wanted to run at significantly lower current density (about a factor of 2 smaller), but this meant we had to operate at much higher q* (q* ~ 5), which isn’t always conducive to disruptive conditions, meaning the plasma may first crash inboard due to loss of horizontal position control before disrupting at the nominal centered position. The other issue was that when operating at too small a plasma current the plasma would often have some kind of beta-collapse and recovery prior to actually disrupting. When this happened, the plasma would then crash inboard very fast before the TQ even began, leading to very small/inboard pre-disruptive plasmas, which in-turn lead to both higher pre-disruption current densities than anticipated as well as major radii that were below the critical (reliable) threshold for most of the duration of the disruption. As a result of these complications, we had to start operating at higher current (about 2/3x baseline instead of 1/2x baseline). The results of operating at these lower currents are still unclear, and require further analysis, primarily concerning when the post-TQ temperatures should be compared.

Machine setup: exactly the same as the previous run day. Although it’s worth noting that we found that the TP9 battery was dead towards the end of the run day, and it’s unclear at what point it died. So just in case, I wouldn’t trust any of the TP9 measurements between 114131-114157. Leaving the final shot of the day (114158) as the only shot with reliable TP9 data.

The baseline shot style was reproduced at the beginning of the run day with 114132&114135. The only star shots worth mentioning are 114147, which was a low (~1/2x baseline) current shot that demonstrates how inconvenient (and misleading) the beta-collapses just before the actual disruption can be, and 114157&114158*, which are good mid (~2/3x baseline) current shots that do not suffer from beta-collapse complications. These are the shots that should be investigated before continuing any further with this campaign.

The last thing to mention were a set of static RMP shots we took towards the end of the day (114151-114153), which tried to avoid the fundamental high q* issue by using the control coils to initiate a disruption at high q*. Unfortunately these shots still often just suffered a beta-collapse prior to disrupting, making any analysis of them very difficult.

Thursday June 2 2022 4:52 pm Li 114185-114227 Sawtooth study

Since peaked plasma current profile is more likely to have sawteeth, and q approaching 2 seems to lead to tearing mode which leads to disruptions, today’s run goal is to create plasmas with flat-top Ip curve (peaked current profile) and q away from 2 and check if there will be sawteeth and if sawteeth will help to keep the shots long and stable.

Clear high-frequency modes are detected in many of the shots. Start shots are: 114198, 114201, 114202, 114203, 114206, 114207, 114208, 114212, 114214, 114215, 114218, 114220, 114221, 114225, 114226, 114227.

Closer examination needs to be done to see which shots have sawteeth.

Further shot development to be done: see if the major radius can be kept flat, rather than going inboard at the beginning of the shot; see if the shot can be maintained longer.

Friday June 3 2022 9:13 am Levesque 114159-114181 Testing for noise generated by 8520 power supply

For most of the shots I connected one of the two new AE Techron 8520 power supply units to Flux Loop D (first flux loop above the outboard midplane) to check for possible noise induced on diagnostics during the power supply operation into a load coil on the tokamak. Check digital shot comments for variation in the connections and driving waveforms between shots. All diagnostic systems and bank power supplies were powered on to mimic the setup for a normal shot, but capacitor banks were not charged for each pulse -- only the 8520 ran power to its load.

Noise appears in diagnostics when the 8520's output voltage to a load on or near the tokamak changes quickly. There is no significant noise induced when the power supply output is at a steady voltage. This is a good sign -- there's no evidence of switching noise caused by the power supply's internal electronics. A good pair of shots that show this is comparing diagnostic signals for 114167-8 during 2-10ms, where the the average 8520 output current was the same (~300A) for each, but 114168 was run with a steady output voltage while 114167 was run with a pulse-width-modulated waveform that give the same average current but with large output voltage swings in 10us intervals. Noise scales like dV/dt instead of dI/dt throughout all tests, suggesting a capacitive coupling between the load and the diagnostic systems in question. Noise can be seen in the EUV systems and to some extent in raw data from the magnetic sensors, but noise in the EUV systems is much more significant, especially the tangential EUV. This is likely due to the EUV systems' higher gains, lesser filtering, and single-ended circuit features. Running the power supply into a load that is located farther from the diagnostic systems does not introduce diagnostic noise, meaning that the noise is not just coming from power surges from the 8520's input power.

Monday June 6 2022 6:14 pm Li 114231-114296 Sawtooth study

Today's run goal is to see if long and stable shots with flat-top currents which are more likely to lead to sawteeth and quiescent mode amplitude can be obtained.

With flat-top currents, the shots do not show very strong sawteeth and the overall temperature is not very hot. Shots to look at are 114251, 114263, 114269, 114273, 114274, 114277, 114278.

A few previous shots 112300, 112303, and 112366 have flat-top currents and very strong sawteeth activities, so starting at 114280, the target shot is 112300. Similar shots are achieved, but still, the plasma could not get as hot, which reduces the signal amplitude. From 114286-114296, the shots are similar and stable, which would be useful to inspect further.

Friday June 17 2022 1:37 pm Levesque/Wei/DaSilva 114297-114320 Biased H-mode with tangential EUV measurements

Ran biased H-mode plasmas using the double-faced probe in Section 2 (DP2) inserted to 103cm. The main goal was to observe the time evolution of electron temperature using the two-filter tangential EUV system, comparing H-mode to non-H-mode shots. There were indications that biased H-modes were achieved during the applied biasing waveforms. EUV diagnostics showed an increase in lower-energy EUV emission during H-modes, while the higher energy EUV emission did not rise proportionally as much, or perhaps stayed steady or fell in some cases. Temperature analysis from the tangential EUV (not yet calibrated to Thomson scattering) showed that the core temperature was lower in H-mode. Triple probe data (at 104cm) also had indications of temperature dropping and density rising during the H-modes in most cases. When shots came out of H-mode, either from stepping down the amplifier voltage or due to some back-transition while the amplifier was still strong, electron temperature rose and density dropped. Star shots to consider are 114313, 114315, 114316, 114317, 114320.

Hardware setup: DP2 was connected to a QSC PL380 amplifier and transformer. The Section 9 triple probe was at 104cm for all plasmas, but was only powered for shots from 114310 onward. 114297 through 114304 were crates-only tests while setting up. Grounding jumpers were change to the “low poloidal EUV noise” setup prior to the run; jumper 3.5 is disconnected, and Sections 4-5 are grounded via the poloidal EUV diagnostic ground to the South rack. Poloidal EUV noise was dramatically reduced relative to recent data as expected. The bias probe voltage and current diagnostics were connected to CPCI 12, inputs 33 and 34 respectively. Bias currents were exceeding 100A in some cases, which was saturating the 10V CPCI input, so shot 114310 onward included a 50ohm termination at that CPCI channel in order to drop the Pearson gain by a factor of 2 (per the Pearson specifications). The amplifier reached up to around 13.5kW delivered through the plasma. The fast cameras were running at 250kfps before shot 114310, then at 100kfps for the rest of the run.

The D2 bottle and a regulator valve were left open during the previous run(s), resulting in the bottle being fully depleted before this run campaign. We ran using only the D2 that remained in the closed portion of the puff line closest to the puff valve. Over the course of the 13 plasma shots taken, the fill pressures gradually dropped from 63uT to 60uT. There may be enough gas in the line for another day of running before using a new gas bottle. For now we don’t have any pure D2 reserves, but a 90% D2 + 10% He bottle has been connected into the puff manifold in case we choose to use that before receiving more pure D2.

Friday July 1 2022 5:40 pm Wei 114322-114496 Fast camera-based mode tracking using machine learning method, third run campaign

This run report covers 3 run days from 06/29 to 07/01. Please refer to the run proposal under “Fast Camera - Fast Camera Run Campaign Memos” for experiment goals, hardwares setup and experiment plan.

On 06/29 (114322-114349) we tried to reproduce the outboard-limited, q=4->2 shot style from the previous 2 run campaigns. MR & edge-q trajectories appeared to bifurcate very early (before 2 ms) during the shot, making the target shot style difficult to reproduce even by adjusting the VFE setting. Ringings on the OHC trajectory were observed during successive shots in the early afternoon so we halted the run day. We suspect the error was caused by CAMAC failing to send the OH bank triggers at the correct timings. Jeff found that the heating lamp over the OHS ignitrons was broken and then replaced it with a new one. A second lamp was installed on 06/30 by Jim before the run day started.

On 06/30 (114350-114425) we started by testing the OH banks and found them working well again. We then tried to reproduce the previous shot style, and 30 good shots (114356-493, 8 bad shots) were taken. The same bifurcating MR & edge-q shots still occurred but happened less frequently than in the previous run day. Afterward we tried to develop the new q=5->3 (MR between 90-92cm) shot style, and 14 good shots were taken (114407-425, 5 bad shots). The fast camera system was left untouched after shutting down.

On 07/01 (114426-114496), another 31 shots (114428-473, 15 bad shots) were taken using the previously developed q=5->3 shot style. Afterward we switched to the q=5->3 shot style, and 8 good shots (114474-488, 7 bad shots) were taken. At the end of the run day we fully retracted Sec 2 T&B shells (in CAM-26730’s view) and took 7 shots (114489-492 with q=4->2, 114493, 94, 96 with q=5->3 shot style).

Tuesday July 12 2022 6:11 pm Levesque/DaSilva 114497-114516 Biased H-mode using 8520 power supply

Used one of the new high power 8520 power supplies through a 1:1 Stangenes transformer to drive the bias probe for an H-mode campaign. This run was primarily meant for commissioning the power supply for this type of use with the bias probe, so the setup was put together carefully and methodically during the first day. The double probe in Section 2 was inserted to 104cm, with both faces shorted together. The power supply had no problems driving the desired voltage waveforms. In most cases, the probe operated around in the range of +100V to +300V and 40A-90A. Dual fast camera videos were taken for most of the plasmas. No concerning arcing was observed on the left face of the probe in the camera videos. The probe was biased during 3-4ms for five shots, then for 3.5-4.0ms after blowing fuses in shot 114510.

There was evidence of the plasmas achieving H-mode, with modes spinning up and the low-energy EUV emission increasing. Temperatures from the tangential EUV system were observed to either decrease or simply not increase as they would have during non-biased shots.

In one shot (114510), much more current was driven (~300V, >200A) for around 0.3-0.4ms such that the probe current diagnostic saturated at the digitizer, and fuses on the primary side of the transformer blew. For this case the delivered current was above 200A (digitizer saturation) and below 800A (power supply limit). We were unable to track where the current went by looking at current diagnostics along possible return paths. This shot had evidence that Pearson measuring the current to the probe also saturated during the pulse, consistent with the I*t rating of the Pearson for a current of ~500A. The 10A slo-blo fuses blowing was consistent with their rating for a current of ~500A; the I²t melt rating of the used fuses (Littelfuse KLDR010) is 132A²s. We do not have protection fuses on hand to sustain currents this high through the biasing diagnostic, however it’s appropriate to keep using the same 10A fuses (or a suitable non-destructive replacement) rather than using higher current fuses in order to help protect other components from surges like this, though the power supply can handle it.

Hardware configuration: Shells in Sections 2, 10, and 4B were retracted. Only the DP2 probe was inserted. See comments with the shot comment metadata for further information about diagnostic channels.

Friday August 19 2022 11:52 am DaSilva/Levesque 114544-114554 Recommissioning TF, checking bias probe condition

Recommissioned the TF and checked status of the bias probe following damaging events. This is the first run after repairing damage to the TF leads grounding connection in the basement that occurred during shot 114517 on July 11. Also, the bias probe was damaged when the wooden HBT-EP sign fell on its bellows last week; this run was meant to see if it's still usable. This is the first time testing the fast cameras with their acquisition computer moved to the screen room, and the diagnostic worked as expected.

Used a high voltage resistance meter around the bias probe to check for arcing while moving the probe, but found no problems. Caliban would not connect to Spitzer in order to use our lower-power control amplifiers, so we used a high-power 8520 amplifier to slowly step up max bias voltage from 120-400V to make sure there were no unexpected currents through the probe setup (shots 114544-547). Signals were good so the probe is not shorted to the vessel.

We also ran several TF only shots (11548-552) upping the TF bank voltage from 1 - 6.5kV to ensure the replacement TF connector was functioning properly. All signal traces were acceptable so we concluded the connector is functioning properly. Note that the TF measurement is not working properly especially at low bank settings, due to digitizer problems.

We took a plasma shot with the bias probe floating at 103cm (114554), and noticed some material ejection from the probe in a fast camera video. Upon inspection through a port we noticed the ceramic insulating sheath had separated from the probe body by ~5mm, exposing the copper rod. After wiggling the probe around a bit more via the bellows, the two-rod shaft and head of the probe became unexpectedly rotated by around 80 degrees, meaning there is a significant dislocation in the probe support structure. The probe is out of commission until it can be repaired.

North rack digitizer (CPCI address 13, records high resolution sensors and OH/VF bank traces) seems to have not worked properly; looks like it triggered at the wrong time. The reason for this is not yet known.

Thursday September 1 2022 6:30 pm DaSilva/Levesque/Chandra 114583-114607 Preparation for Biased H-Mode

The first few shots of the day were spent diagnosing and solving the timing triggering issue in the North rack (CPCI 13). The solution was found to be: One of the CAMACs appears to be putting out a power level that is too low to trigger CPCI 13, but still sufficient to trigger CPCI 40 (a newer CPCI) correctly. The CAMAC is now driving an Eagle Harbor optoisolator input-output chain, which on the output side produces sufficient signal to trigger the CPCI.

The next few shots were spent developing a repeatable shot style with D_{tIp & DtRm ≈0, improving upon shot 99267 (Ian's negative biased shot with clear frequency separation of the modes).}

The hemispherical probe was inserted, and it was found that the probe metal tube support bends either up, or parabolically up and down, and it had to be pressed gently downward to get it through the gate valve. Subsequently, it was noticed that the voltage signals being digitized seemed wrong. It is unclear if this problem is persisting, but it may not impede biasing.

Star shots: 114593, 114602

Tuesday September 6 2022 7:51 pm Chandra 114646-114671 RMP Frequency Sweep

The goal of this run day was to see if the tomographically-established internal m=2 island-like emissivity perturbation behaves different than the magnetic edge m=3 kink-like perturbation.

We had low success coupling to the mode in it's saturated phase (natural mode frequency 7-8kHz) with an either m=2 or m=3 C-Coil imposed RMP of linearly increasing frequency from 5-15kHz. Plots will be attached to show the lack of frequency change, the m=2 perturbation phase-locked to the m=3 perturbation, the control coil frequency ramp and verification that the C-Coil flux shape was indeed 2/1 or 3/1, respectively.

Future work should repeat this. Some indication of mode locking was achieved at the end of the day, but results are unclear.

The target shot was 103590, and a good comparison shot is 114668 (m=2 RMP applied).

Tuesday September 20 2022 11:43 am Chandra 114702-114765 RMP Frequency Sweep

The goal of these shots were to establish if the RMP could couple to the plasma for m=2, m=3, and adaptive m applied flux, with a 5-20kHz sweep from 3.5-6ms, and secondarily to establish whether or not the different flux shapes coupled and controlled the m=2 internal or m=3 external component of the mode.

Results were largely unsuccessful, analysis will be appended. Control shots are only usable between 114738-114765, before that the applied flux vectors were not rotating properly.

Star baseline shots include 114723, 114732, 114760.

Example RMP shots include 114746 (empirical m), 114756 (m=3), 114765 (m=2)

Tuesday September 20 2022 12:02 pm Chandra 114795-114832 Frequency Target Feedback

The goal of this experiment was to explicitly control n=1 frequency, using the magnetic feedback system as the input, a fixed input phase offset of 0deg, variable output flux amplitude, and an empirical output flux surface matched to a baseline shot. The target function was holding at 10kHz, then ±5kHz, then back.

Averaged results for at least two shots did not show any difference from the baseline shot, corresponding to the lack of result from the previous RMP frequency chirp experiment. The n=1 frequency as calculated by the GPU did, however, match that as calculated from the offline sensors, and the output currents did appear to match the FB sensor pickup, up to the fixed phase offset.

Future work will redo this experiment with a nonzero phase offset, as this may have been inaccurate for mode acceleration/deceleration. Analysis will be attached.

Star baseline shot: 114816. Negative frequency bump: 114828, Positive frequency bump: 114831

Tuesday September 27 2022 11:14 am DaSilva 114878 – 114905 H-mode Operation Space

The purpose of this run campaign as to establish an “operation space” for achieving H-mode using the low- and high-power amplifiers using the shot style of 114891. Fast cameras were on @400kHz on shots 114890-114900 and @250kHz on 114901-114905.

Classification of a mode as being strongly in H-mode is determined by a combination of: increased SXR Fan Array activity, an extreme drop in the n=1 mode frequency (<-20 kHz), a reduction in the mode amplitude, and tight striation of the magnetic sensor data.

We confirmed that both amplifiers can reach H-mode (ex shots: 114883 on the low power amp and 114891 on the high-power amp) but the low power amplifier cannot sustain the necessary power and cannot last much longer than half a millisecond. Performing a voltage sweep with the bias probe inserted in at 102 cm, H-mode can be robustly achieved when the bias is 280V. Biasing 240V and below produced the same characteristics of an H-mode, but were not extreme enough to warrant classification (e.g. compare shot 114885 biased at 240V to shot 114891 biased to 320V).

Probe distance to plasma as also examined briefly, and it was found that for biasing at 320V, the distance is not extremely sensitive up to a few cm away—specifically for plasmas with magnetic major radii that fluctuate between 92-93cm the probe could be anywhere between 102 and 103cm and still achieve a strong H-mode.

Wednesday November 2 2022 5:51 pm Chandra 114906-114921 Bias Probe Voltage Ramp // Tree Testing

The goal of this campaign was to ramp the bias probe voltage over a single shot from 0 to 150V, and observe if the internal mode spins up at all.

The first few shots of the day had triggering issues, due to the master clock signal at the North rack being disconnected from one digitizer and connected into a fiber connection to the new D-TAQs in the server room. This was disconnected and reset to the normal configuration.

There were no successful shots with the bias probe in the intended configuration.

The second half of the day was spent diagnosing an issue with the tree. It was discovered that spitzer had run out of space in folder where the tree data is stored. The was expanded, and the final shot of the day (114921, crates only) was successful.

Saturday November 5 2022 8:02 pm Chandra 114961-114986 Bias Probe Voltage/Position Scan

The goal of the start of this campaign was to explore the torque magnitude and location necessary to induce locked or differential rotation of the q=2 surface vs edge mode (in this case m ≂ 4 kink mode).

Two positions were scanned: 105cm and 103cm @ section 5.

Voltage was ramped from 2-4.5ms, from floating to roughly +400V with BP> ≂ 50A.

The targeted shot style was very flat R_m ≂ 92cm during the the bias period, future work will also target flat q as well, to better establish the position of the rational surfaces w.r.t. the bias probe. (Example baseline shots: 114967, 114969, 114986)

The 105cm biasing was able to slow or lock the mode rotation, but not reverse it and spin it up (Example: 114978). The 103cm biasing was able to lock the mode, and then abruptly reverse the direction and spin it up (Example: 114979).

There is no indication of differential rotation when the mode locks, but some indication of a low frequency persisting mode during the higher frequency spin up with the probe at 103cm, as seen in the stripey plots, Mirnov probe signals raw data, spectrogram, and potentially in the pEUV as well.

Future work will attempt to directly measure the E_r field and vƟ with the rake and Mach probes.

Friday November 11 2022 5:54 pm Chandra 115014-115028 Rake probe measurements of Bias Er

The purpose of today's run was to try to measure the E_r well created by the bias probe.

We were for some reason unable to apply a voltage to the bias probe, although it did seem to be outputting a current ramp on shots 115027 and 115028.

Machine configuration: BP5 @ 103cm, Rake Probe S4 @107cm, all shells in, BP5 connected to low power amplifier.

Good baseline shot: 115022

Monday November 14 2022 9:19 am Saperstein 115003-115013 D-TACQ triggering testing

The purpose of this period of the run day was to troubleshoot what was causing the N rack CPCIs to fail to store after sending an extra N rack clock output to a pFO transmitter. The tests performed disconnected various segments of the extra clock output until we found two separate problems. First, the Limo->BNC cable we were using was bad and had to be replaced. Second, following that replacement, storing seemed to fail after connecting the transmitter w/o it being powered, which is likely due to the very low input impedance of the transmitter (~50 Ohms) syphoning the signal from the CPCI clock inputs with much higher impedance. We eventually replaced the clock->transmitter setup with a clock->sig-gen->transmitter setup, where the input impedance of the sig-gen was high enough to match that of the CPCIs, and then the sig-gen would transmit the clock to the transmitter. This seemed to resolve the storing issue, but we have not tested if it transmits the clock correctly. This change was also made sometime after the end of this period of the run day, during Rian’s half of the run day in the afternoon (11/11/2022), but what shot the change was added was unfortunately not documented. So the change occurred sometime between 115014-115028.

Additional note: The sig-gen introduces a ~1/2 us (650 ns) delay between the master clock and the clock output by the sig-gen.

Tuesday November 15 2022 1:22 pm DaSilva 114992-115002 Testing a Ramping Bias on the High Power Amplifiers

The purpose of this campaign was to examine the timescale of the L- to H- transition when subjected to a ramping bias rather than a quick turn on. Unfortunately no L-H transition was observed today, as there were various hardware problems affecting the biasing such as fuses blowing during certain ramping schemes, and the high power amplifier sometimes not triggering correctly. I don't think there is a hardware problem with the amplifier that needs to be fixed but rather I'm still inexperienced using it and had trouble triggering it all day. Although the purpose of the campaign was not achieved today, we were able to get a handful of really good, consistent L-mode shots to use as a control against comparable H-mode shots to more quantitatively examine the transition.

Tuesday November 15 2022 5:49 pm Chandra 115074-115094 Feedback shot style development

The goal of today was to evaluate Shiraki's baseline shot style (69642) as an RMP feedback candidate.

The RMP is confirmed to work (115076) with the prep_awg format on Caliban.

The shot style was not successfully reproduced.

Machine Changes: The blade limiters were inserted 2cm to their stopper limits, to match older machine conditions.

Tuesday November 15 2022 10:50 pm Levesque 115029-115066 Runaway electrons

Generated plasmas at low fill pressure to produce runaway/energetic electrons (REs). The intent was to look for large hard x-ray (HXR) bursts during the disruptions. The target shots were 103084 and 103061, which were able to achieve a runaway population on top of a relatively normal plasma. The reference shot campaign had success around 47uT fill pressure, while today's sweet spot was at 57uT. Had difficultly maintaining the desired fill pressure throughout the run. Generally, 56uT fill or a little below would produce plsamas that looked like just an erratic runaway beam without a healthy background plasma, and 58uT fill or above would produce a standard plasma without evidence of runaways. Was only able to get a couple of shots that had both a reasonable plasma and evidence of runaways, yet these still didn't have large HXR production. The erratic runaway-only shots had very large HXR production, regularly saturating the detectors.

Looking at the time history of strong HXR-producing shots from today and earlier target shots, the largest HXR burst always occurs before the disruption current spike (or before where the current spike would be expected if it was present). This is a time when the plasma is already crashing in to the HFS flanges, but the current quench has not yet begun. There was no clear evidence of additional significant RE or HXR production during the current quench, possibly since the entire RE population had already been dumped. Having position control could alleviate this, since the VF now remains too strong at the end of the shot, at a time when we would want to confine REs for disruption studies. However, note that our plasma currents are too low to have the significant knock-on RE avalanche that occurs in the ~MA-class tokamaks.

This discharge style is very sensitive to fill pressure, and presumably other machine conditions such as vacuum quality and possibly shell configuration. Today all shells were inserted to their stoppers except for Section 5 which was fully retracted, while the reference discharge had Section 10 shells retracted for fast camera imaging. Now the HXR array is recorded in digitizer channels ".devices.north_rack:cpci_40:input_65" through input_69, while the reference campaign from August 2019 had an HXR detector recorded at ".devices.north_rack:cpci:input_96". HXR detector #4 in the array was off the machine for the first half of the run, and was added for shots 115052 onward, but it was not uniformly arranged with the other HXR detectors. All HXR detectors are set to their highest gain setting. All probes were retracted behind the walls, except for the rake probe at Section 4, which was left at 107cm.

Thursday November 17 2022 12:38 am Levesque 115095-115126 Feedback shot style development

Tried to reproduce a shot style from Peng's feedback run campaign (target 86061) for the start of this run, and a shot style from Rath's feedback campaigns (target 75311) at the end of today's run. Both types involve approaching edge q=3 from below for the period of interest. I was not successful at getting the target plasmas for either case. The Peng-style shots (115096-115118) all resulted in a disruptive event early in the shot, in the range of 1.6-2.1ms, followed by a relatively long current quench. Examples: shots 115107-113. The Rath-style shots (few shots, 115119-115126) had difficulty getting to the requisite low edge q early in the shot, apparently due to strong 3/1 modes rotating at ~20-25kHz which clamped the edge q above 3 for 1-2ms by stalling the major radius evolution.

Shell configuration: Sections 3, 5, 7, and 9 are fully inserted, while other sections (i.e. shells with LFS SOL tiles) are retracted 1cm from their stoppers. This is so that the undiagnosed blade limiters in two sections are the only outboard/top/bottom limiters, similar to control experiments before the SOL tiles were installed.

Other hardware concerns: 1) the shell-moving motor was removed from Section 5T due to the coupling mechanism failing; 2) the TF charging water resistor will need more water soon; 3) I found the resistor dislodged in the shorting stick for the VF start bank, and re-seated it in position -- the clamp holds the resistor loosely, and should be checked occasionally.

Thursday November 17 2022 4:47 pm Chandra 115129-115160 RMP Shot Development

The purpose of this run was to continue evaluating the Shiraki baseline shot style (69642) as an RMP candidate.

The first part of the day was spent diagnosing the failure to trigger of BP₅. It was discovered that one of the fuses between the QCS amplifier and the Stangenes transformer had blown, but this was not previously detected as the amplifier had been on when the fuse was checked. The probe now outputs the correct voltage waveform.

There was moderate success in replicating the shot style (115156-157).

There was some evidence of coupling: RMP of 15A and 25A fired 2ms, flipped @ 2.5ms, ended 3ms (115158, 159) Signs of mode locking are observed in the mode amplitude and frequency.

The VF rang on shots 115155 and 115140 due to a VF start time that the CPCI didn't like.

Separately, 115147, 148 had good centered disruptions.

Tuesday November 22 2022 10:44 am Levesque 115161-115191 Runaway electron shots with RMPs

Applied magnetic perturbations to shots with runaway electron (RE) populations. The intent was to look for signatures of reduced RE confinement due to the non-axisymmetric fields, mainly focusing on bursts of hard x-ray (HXR) emission. Ran phase-flips with the applied field spectrum dominated by either m/n=3/1 (shots 115166-171), or 2/1 (115173 & 174), or 1/1 (115175-183). Most shots had high currents in the control coils that were located at the mode peaks -- up to 40A output current, which near the maximum applied fields.

Was able to reliably get good shots with a RE population on top of a "standard" plasma for most of today. The successful fill pressure was 57 or 58uT for this discharge style. HXR emission peaked mainly during the 3-4ms period for shots where there was no strong "event" that dumped runaways. Long shots with HXR emission in the 3-4ms period did not have HXR bursts at the disruption, suggesting that the RE population had entirely damped away by ~6ms (either via earlier losses or e.g. damping on background plasma). HXR emission strongly peaked at the blade limiter(s) instead of throughout the HXR array detectors (distribution around the toroidal array was seen in prior HXR shots); this is presumably because the non-SOLC-diagnostic blade limiters were reinserted last week and all 6 LFS SOLC tile sections are retracted 1cm.

The 3/1 RMP produced a very clear phase dependence of large HXR bursts on the HXR array, and therefore indication of preferred location where runaways were being lost. Shot 115170 had the 3/1 RMP at a reference phase of 0 degrees for the first period of the phase-flip, while shot 115171 was set to 180 degrees. The HXR emission due to the RMP for the second shot appeared on the opposite side of the machine as the first shot. Reference shot 115172 had no applied fields from the control coils. q* was around 3.5 at the RMP time for these shots. The RMP caused the major radius to move inward at or slightly after the HXR burst, suggesting the loss of pressure from REs was significant.

Response from the 2/1 RMPs in the HXR detectors was not nearly as strong. A clear stationary response could be seen in the magnetic sensors for this case, though the response might not have been strongly m=2. q* was in the range of 3.3-3.7 for these. The 2/1 shots did not appear to have as strong of a RE population to begin with compared to the 3/1 shots, so it's possible that the weaker RMP response on the HXR detectors is due to lack of sufficient REs.

Response from 1/1 RMPs was not clear. This could be in part due to 1) drifting of shot quality away from what was good for the 3/1 shots, 2) the q was too high during the period of RMP application. Applied fields were attempted during 3-4ms and separately during 4-5ms, per the sawtooth note below. After quite a few shots with lack of a clear causal response, I tried raising the plasma current by ~30% in order to lower the q profile appropriately. This did not produce successful RE+normal plasma shots in the 7 attempted shots -- there was either no breakdown or a low quality short-lived "normal" plasma without REs. Further development of this case is suggested.

Many shots had clear sawteeth later in the discharge. For the most part, this was after the 3-4ms period where there was good HXR emission. I would think the 1/1 applied fields should have the strongest effect when there is a 1/1 surface present, i.e. when there is evidence of sawtoothing, thus the 1/1 applied fields were moved a little later in the discharge compared to the other cases (also moved later since q was dropping due to Ip rising). There is overlap of sawtoothing and small HXR emission in some shots, but the overlap might not have been sufficient to use for conclusive 1/1 RMP studies in these shots. An example shot that had simultaneous sawteeth and HXR emission was 115174, which had HXR emission associated with sawtooth events at 4.3ms and 5.05ms. To me, these events don't show evidence of the sawtooth reconnection event accelerating the electrons, rather it looks like a deconfinement of already-present energetic electrons.

Machine changes before the start of this run: HXR array detector #4 was repositioned on the machine to return to a uniform toroidal array for detectors 1-5. Section 7's feedback sensor amplifier box was remounted on the local TF magnet; this had been sitting on the base pad for a long time (>1 year) to allow access to tangential EUV amplifier boards, but the tEUV diagnostic hasn't needed modification lately.

Monday December 5 2022 6:21 pm Li 115259-115299 Sawtooth development

The goal of today's run is to recover the sawtooth shot style and try to move the major radius to see if the sawtooth will disappear.

Part of the run day is spent to adjust the parameters to get close to the reference shots. From shot 115270, all shells are all the way in, all probes are out. During some shots (115289 and 115290), the tangEUV sysem was not working properly, this might cause further problems and we should keep an eye on it.

ST-free shots: 115277, 115278, 115293, 115296, 115297, 115299. ST shots: 115288, 115294, 115295. The major radius is not outboard enough, so not many ST shots are obtained. The correlation between sawtooth and low mode amplitude / sawtooth-free and large mode amplitude is still valid, which is good. Tomorrow I will try to decrease VFE or increase OHE to push the plasma further out to see if sawtooth shots can be developed more regularly.

Monday December 5 2022 7:19 pm Chandra 115195-115258 RMP Amplitude Scan

The goal of this campaign was to reproduce the RMP coupling results based on Shiraki baseline shot style (69642) over shots ~69692-69738.

The target RMP period was just after q crosses 3, with high outboard R_m. We were unable to reproduce the time of the q crossing featured in the Shiraki shots (q=3 @ t=2.0ms). We saw variability of roughly t_q=3≈2.5ms±0.25ms, 3≥q&ge2.4;, 93.2≤R_m≤94.2cm.

The coupling response is defined as the convolution integral between the strongest Control Coil and all of the perturbed poloidal fluxes as detected by detrended poloidal Mirnov signals. A -3/1 sinusoidal function (matching the applied flux geometry) is fit to the results, and the amplitude is defined as the coupling.

Results indicate an almost identical response of RMP amplitude vs plasma response.

Plots appended.

Star baseline shot: 115239

No significant machine changes.

Tuesday December 6 2022 8:22 pm Li 115300-115368 ST and ST-free shot development

Today's run focuses on swiching between ST and ST-free shots under ramp-up current, targeting 113791. The shots are reaptable. The correlations that sawtooth & low mode amplitude & outboard major radius and sawtooth free & large mode amplitude & inboard major radius are verified.

ST-free shots: 115301, 115304, 115311, 115314, 115317, 115326, 115330, 115334, 115338, 115339, 115340, 115342, 115346.

ST shots: 115302*, 115306*, 115307*, 115309*, 115315, 115318*, 115320*, 115321*, 115323*, 115324*, 115325, 115327*, 115331, 115332, 115337, 115341, 115344, 115345. Some of the ST shots have major radii that went too ourboard and caused minor disruption leading to large mode amplitude and temperature drop in the middle of the shots (markes with *). Shot 115329 is interesting since the sawtooth becomes weaker in the later stage. Its major radius is similar to 115326 and 115330. Further comparison needs to be done. Pay attention to Al Ch7 in future runs. Might need to replace the board. Tomorrow's run will focus on targeting 114215 (flat top current) and trying to switch between ST and ST-free shots under this current profile.

Wednesday December 7 2022 5:43 pm Li 115369-115435 Sawtooth study

The shot development of flat-top current was not successful. So the plan to insert bias probe under flat-top current might be cancelled.

From shot 115402, switched to ramp up current. For the ST shots, the Rm is different from yesterday. It goes inboard first, then goes outboard, so the sawtooth starts later than yesterday’s shots. Mode amp dropped pretty late, but match with the beginning of the sawtooth. Also, these shots last longer. Compare 115406, 115407, 115408, 115412, 115413, 115415, 115416, 115417, 115418, 115419, 115420, 115425.

ST-free shots are 115404, 115405, 115406, 115411, 115414, 115422, 115423, 115424, 115426, 115430, 115431, 115434.

Other shots that are similar and might be worth comparing are 115425 & 115427 (both ST shots),115434 & 115435 (one ST and one ST-free, but very similar Rm).

Friday December 9 2022 6:19 pm Li 115436-115499 Sawtooth study

Today's run is to check if the floating bias probe will affect the exitence (generation) of sawtooth events. The bias probe is inserted to 103 cm. All other probes are retracted. All shells are fully inserted.

Before 115471, no target shot was referred to. ST shots are 115437, 115449, 115450, 115454, 115456, 115464. The sawtooth periods are small and the shots switch to ST-free when Rm goes inboard, which is interesting and needs more analysis.The ST-free shots are 115441, 115443, 115444, 115445, 115455, 115458, 115458, 115459, 115460, 115461, 115463, 115466, 115469, 115470. The Rm of 115467 and 115466 bifurcate around 6 ms, and they show different Te profiles.

After 115471, the target shot is 111533. A few good recoveries are 115471, 115485, 115491, 115495, 115496, 115498. The parameters (Ip, Rm, q) are very similar to 115333 but these shots seem to have mitigated sawtooth at latter stage comparing to 115333. 115495, 115496, 115497, 115498 are very similar, but the Te profile of 115497 is very different from the others. The ST-free shots are 115473, 115486, 115488.

Generally, the correlation among the existence of sawtooth, major radius and the mode amplitude is valid.

Thursday December 15 2022 7:23 pm Chandra 115504-115565 RMP Amplitude Scan - Shells In

The goal of today was to repeat the results from Dec 2^nd (RMP scan, Shiraki shotstyle, with limiters in), but with the shells in and limiters out (a more modern shotstyle).

Results were not promising. The shotstyle was harder to reproduce and less stable (baseline: 115533, 115543 (used), 115530 (unused, very good shot, but incorrect q evolution) ). Three RMP amplitudes were tested (-3/1 helicity), with a goal of at least two shots per amplitude. This was not achieved. We managed to obtain four good baseline shots, two shots at 25A, and one at 37A.

Result plots are appended. Note that there is visible coupling in the RMP on shots, but it appears barely above the natural mode amplitude. It appears that the shells being in does negatively impact the ability of the RMP to effect the plasma.

The first shot of the day was a VF-only vacuum pulse to observe eddy currents in the vacuum vessel conducting structures.

The OH rang on due to an "un-magic" VF start time on the following shots: 115551, 115552.

The feedback code appears to output current correctly, but no good shots were tested with it.

Friday December 16 2022 5:56 pm Chandra -115566115607 q-Scan with Triple Probe

The goal of this run campaign was to sweep the q=2 and/or q=3 surface over the location of the triple probe (TP_09) to observe the existence or non-existence of an inverse in temperature, indicative of an island.

It took quite some time to develop a repeatable shot modeled after 115530, but several shots stand out, namely: 115600, 115602, 115603, and to some degree 115607.

These will be analyzed to see if the q=2 or 3 surface swept the tip of the probe.

Shot 115569 Rang due to an un-magic VF start time (+915uS).

Machine changes: The triple probe is inserted at 104cm. Note that two of the grounding sticks in the basement have resistors which have fallen out.

Tuesday January 10 2023 6:19 pm Chandra 115608-115634 Centered Disruption Search

The goal of this run campaign was to reproduce Alex's shot 110428 but with the pEUV in the lower noise configuration and with less R_m motion during the disruption.

We were unable to achieve these goals with an exact match of the target shot, but a similar enough shot style was produced which shows strong centered situations and the characteristic ramp up of the mode amplitude.

Plots will be attached showing the reproduced shots, and analyzing if we can see internal activity on the two functioning pEUV arrays (0deg, 270deg).

Star shots include: 115630, 115632, 115633.

This shot style was also reasonably repeatable and controllable.

Machine changes: Triple Probe retracted to 120cm, all shells in, mineral oil refilled.

Thursday January 19 2023 2:05 pm DaSilva 115635-115661 HPA testing and robust H-mode

Shots 115635-43 were shots aimed at testing the triggering for the high-power amplifier system as it had not been fully set up since moving to the basement. Alex and I found a bad BNC cable connecting the fiber optic transmitter to the North rack, which was thrown out and replaced. Since then the triggering is working exactly as it had been when the HPA was in the linear machine room. Once the triggering was good, we moved on to looking for a robust H-mode criteria. For this run campaign I used Alex's shot style rather than my own, which has a few advantages for this testing. First it is very repeatable in terms of the major radius, plasma current and q profile, and second the major radius is a few centimeters further in than mine (about 91 cm compared to my 92.5-93 cm), which should make access to H-mode slightly more difficult. Ultimately, we did not find a robust H-mode criterion yet but significant progress was made. Shot 115650 is one of note, as it is an exemplary H-mode shot in this shot style. Attempts to recreate it failed, however, as the bias probe current would rise until it clipped out at over 100 A. Alex and Jeff installed a Pearson current monitor (digitized at ch 87 CPCI 13 N rack - gain x40) on the input to the bias probe to check if the current was leaking through the probe stand, which is looks like it was. For shot 115656 and later, the circuit box driving BP5 was replaced by a white one that Ian had previously used that can handle higher current readings. This changes the BP5 gain settings to: current = 100/3x gain and voltage = 200x gain. A major (unintended) theme of this run is bias probe current. Changing the BP5 box seems to have helped if not fixed the runaway current issue, as there must have been some short to ground in the previous one used. This primarily affected strong (>280V) biasings, so much of the testing was done at 200V biases, though this made access to H-mode impossible. The last two shots of the day (115660 and 61) revamped the bias voltage back to 320V and everything seemed to work. Additionally, we are having a reoccurring problem of blowing fuses so for future tests we will try using the transformer in a 2:1 configuration instead of the 1:1 config it is in now. This should also allow us to run longer bias pulses and avoid transformer saturation which affected a few shots during this campaign. Finally, an important note from Jeff regarding the system pressure: "The Granville-Phillips Vacuum Gauge Controller was replaced with the one that arrived in August (teaching lab controller was borrowed and in use in the meantime). The base pressure (~1.8e-8 torr) and fill pressure (~68uT) readings are now close to what they were with the borrowed controller, but the settings were incorrect for the first few shots of the day; fill pressure recorded as roughly double the the actual pressure at first. This new gauge controller should be calibrated against the RGA for a range of static fill pressures at some opportune time, though the readings are close enough now for general use, matching expectations from the old readings within ~2uT."

Friday January 20 2023 7:11 pm Li 115662-115755 Sawtooth with RMP

The goal of these shots is to develop shots that have sawteeth under natural circumstance and apply RMP on them to manuelly increase the mode amplitude. All shells are fully inserted. All probes are retracted.

Static RMP was applied on 115678-115686. Both 3/1 and 2/1 modes were tried and 3/1 mode seemed to couple stronger with the plasma.

Rotating RMP shots were not very successful. Waves with amplitude = 7, t = 3~6 ms, f = 10 kHz were applied. For m/n = 3/1 mode, a few relatively good shots are 115706, 115715, 115716, 115730, 115732. For m/n = 2/1 mode, take a look at 115736, 115740, 115747. 2/1 RMP seems to be likely to cause disruption.

The mode amplitude in shots without RMP can be used as reference. They generally have quite low mode ampltude: 115695, 115696, 115699, 115702, 115704, 115721, 115722, 115723, 115724.

For further analysis, compare the mode amplitude of the 2/1 RMP shots and RMP off shots when q is around 2, 3/1 RMP shots and RMP off shots when q is around 3.

The shots were very unstable. See if it's still like this on next run day.

Tuesday January 24 2023 4:56 pm Li 115756-115814 ST+RMP

Today's run goal is to develope ST shots and apply RMP to them.

Reference ST shots without RMP are 115760, 115770, 115775, 115780, 115805.

RMP shots with amp = 5 3/1 mode are 115773, 115774, 115776, 155777, 115781, 115789.

RMP shots with amp = 7 3/1 mode are 115793, 115797, 115800, 115802, 115804, 115810, 115811.

Further analysis is to compare the ST starting time and mode amplitude between RMP shots and ST shots.

Wednesday January 25 2023 5:35 pm Li 115815-115859 ST+RMP

Today I tried to apply 2/1 RMP to ST shots.

It seems that the major radius is pushed inboard when RMP is applied. The only difference between 115818 and 115819 is that RMP was on during 115819, and the major radius is more inboard.

Some good RMP shots are 115827, 115828, 115829, 115830, 115836, 115841, 114852, 115855.

Towards the end of the day, the plasma current became unstable. It would suddenly drop even when the bank settings have not been changed.

Friday January 27 2023 2:56 pm Saperstein 115888-115899 Testing triggering of new clock

The purpose of this "run day" was just to test whether or not we could get the new BNC 577 pulse generator (that we plan as using as the new clock for the triggering upgrade) to receive a software trigger from take_shot by making changes to the tree. By the end of the day we were successful in both arming and triggering the clock. The tree was re-made with almost every shot until we found the right settings to get the new system to work. We added a new INIT action to arm the clock (through TCP commands) and replaced the old "start_shot" PULSE_ON action with a new one that sends a TCP software trigger to the new clock instead of triggering the basement j221. The "SPAWN" function was used in each node to have the MDSplus call functions that sent the TCP commands.

Star shots include: (115897) we got the arming action to successfully work for the first time, (115899) we got the triggering action to successfully work for the fist time.

The old tree was backed-up before the start of the day. At the end of the run day, we reverted back to the old tree.

Friday January 27 2023 3:46 pm DaSilva 115860-115887 Biased H-mode with a 1:2 transformer configuration

The primary theme of this run campaign was to investigate using the transformers in a 1:2 step up configuration to achieve H-mode rather than the 1:1 setup that has been previously used. We also tested longer biasing times (1.5 and 2 ms) using this configuration. We were ultimately successful in getting a robust H-mode on Alex's shot style (which should be more difficult than the one I had been toying with in the summer) provided we can keep the major radius between 91 and 92 cm, and the plasma does not disrupt early, i.e., before biasing. In this summary we'll go through key developments throughout the day and discuss a few star shots.

The first several shots (115860-67) of the day were testing the transformer and crates to make sure everything was functioning properly (there were a few CPCIs failing early on). Shots 868-872 had the HPA programmed to 160V with a theoretical output of 320, though the probe was only reading 150V and was extremely noisy. Jeff and I investigated if voltage losses were occurring or if the grounds on the probe stand were interfering with each other. We reasoned that based on the reported probe current and voltage, the plasma resistance must be around 3 Ohms, which due to the 1:2 configuration the primary coil only sees 0.75 Ohms. Because the primary resistance is so low, the resistances from the wires, fuses, and all the connections must be comparable leading to only half the voltage dropping across the transformer. Thus, we changed out the Jensen transformer connecting the probe output to the crates with a single channel DIN-PB Jensen that had more capacitive shielding (significantly reducing noise and stopping the V and I measurements from bottoming out at 0), and programmed the HPA to input 320V, hoping it would drop to 160V across the primary so the probe read 320V.

Shot 115873 is a star shot and performed much better than expected. The probe voltage read ~500V for the entire millisecond, and the plasma showed evidence of a strong H-mode for the entire duration. The next few shots were repeats of this plasma configuration, though we ran into problems with early disruptions and evidence of the probe detaching from the plasma. Shot 115874 shows evidence of disconnection in the probe current scope, where there were increases trying to recover back to their equilibrium value of approximately 70A interrupted by large spikes in current likely due to arcing. These arcs were concurrent with the plasma’s major radius becoming extremely noisy and pulling further in to the tokamak. Thus the next few shots were pulling in the major radius and also testing an extended pulse time of 1.5 ms to check if the transformer will continue to saturate.

Shot 115881 is another star shot for a 1.5 ms pulse. Once the major radius was adjusted correctly, the H-mode lasted the entire bias window with no evidence of transformer saturation. Finally, we tested a 2ms pulse which produced 3 strong, consistent H-modes in shots 115882, 84, and 85. It is worth noting that for longer pulses, the 3A KLDR fuses will blow in between shots due to a lingering current after the plasma disrupts, but this issue did not persist once we switched to 10A fuses.

Thus we have demonstrated robust H-mode access on Alex’s shot style, and plan to do another similar run on the shot style I had experimented with previously. It seems the only restriction for H-mode access at this voltage is that the major radius cannot drop below 91cm when the probe is at 103cm or else it will detach from the plasma. This is especially worrisome in the 1:2 transformer configuration, as if the plasma resistance suddenly increases, the primary’s resistance will jump and the full 320V driving the system from the HPA will drop across the transformer, causing a >600V surge in the bias probe. The MOV-14D291K in the box running the bias probe cannot run at voltages that high and will likely explode if we are not careful. Thus for future use it’s advisable to run the HPA at a lower voltage (>=200V) for the shot development, and only increase the voltage when the plasma is behaving consistently.

Wednesday February 1 2023 11:53 am Li 115901-115945 ST+Wall

To start with, all shells are fully inserted. Target shot is 115333. A few shots were good recovery of the target shot: 115908, 115909, 115910, 115911. A few ST-free shots are 115904, 115907, 115912.

From 115935, all shells are fully retracted. Had problem getting breakdown. 115943 is a hard x-ray shot.

Plan to retract the shells by 1.5 cm instead of fully retracted to get easier breakdown and better analogy of the plasma-wall distance difference between natural ST and ST-free shots.

Wednesday February 1 2023 5:04 pm Li 115946-115991 ST+Wall

All shells are retracted 1.5 cm from the inner stopper, which is basically the plasma-wall distance difference between ST and ST-free shots when all shells are fully inserted. Target shot is 115333, try to develop shots with Ip and Rm similar to the target shot, and see what the mode amplitude look like and whether sawteeth exist. From today's shots, it can be concluded that the plasma-wall interaction plays important roll in the formation of sawteeth.

ST-free shots with outboard major radius and large mode amplitude: 115950, 115962, 115963, 115966, 115971, 115975, 115976, 115977, 115984, 115985, 115988, 115989. Note that 115977 and 115985 have very outboard major radius.

ST-free shots with inboard major radius and large mode amplitude: 115956, 115957, 115961, 115964, 115965, 115970, 115979, 115981, 115982, 115983, 115986, 115987.

Wednesday February 8 2023 10:52 pm Li 116022-116066 ST+Wall

Today's run continued to study the correlation between sawtooth formation and plasma-wall interaction. Starting from 116022, the shells are retracted 1.5 cm from the inner side stopper. Starting from 116052, the shells are retracted 1 cm from the inner side stopper. With the 1 cm retracted walls, the plasmas still have large mode amplitude and no sawteeth.

Shots with inboard major radius and 1.5 cm retracted wall are: 116030, 116031, 116032, 116037, 116044, 116048. Shot with inboard major radius and 1 cm retracted wll are: 116053, 116055, 116056, 116058, 116059, 116061, 116064.

Shots with outboard major radius and 1.5 cm retracted wall are: 116025, 116026, 116027, 116028, 116033, 116036, 116041. Shots with outboard major radius and 1 cm retracted wall are: 116057, 116062, 116063, 116065, 116066.

Friday February 10 2023 12:01 pm DaSilva 115996-116021 Completing Robust H-mode

This run had two primary goals: first to recreate the bias voltage/current bifurcation observed in shot 115878 (and perhaps get some more quantitative information about it), and second to test the new bias configuration on the previous shot style to confirm that it can produce a robust H-mode. Note that after the initial setup and tests every shot of this campaign was biased from 3-4.5ms on an output voltage of 320V.

Shots 11600-116009 recreated the Ip, MR, and q of shot 115878 but were unable to get any bifurcation, even when the MR went out to 94 cm in shot 116004. Thus the abrupt voltage drop in 115878 likely depended on the density (which we couldn't measure because the ion gauge isn't working) or more likely the shell positions which for this run day were all out 1.5cm for Boting's experiments. When precise shell positions are not needed for anyone else's experiments we will return to this since it's important to understand wild HPA changes for safety reasons.

Shots 116010-116021 were mostly shot development trying to match the plasma conditions of shot 114891, which was the archetypical H-mode shot from my experimental campaign leading into APS. We were unable to perfectly match the Ip, q, and MR simultaneously (likely due to no pressure readings and differing shall positions), but every shot taken was strongly in H-mode. Thus a robust H-mode was successfully demonstrated over a wide variety of conditions. It should be noted that with an output voltage of 320V from the HPA, the probe was seeing ~600V in this shot style, compare to ~500-550V in Alex's shot style which is something to keep in mind for further H-mode studies.

Wednesday February 15 2023 10:06 am Li 116075-116110 ST+Wall

At the begining of the day, one of the "take_shot_hbt.py" windows could not open. The issue was that one wire end on the phidget circuit board was off.

All shells were retracted 1 cm from the inner side stopper. Target shot 115333. Still no sawtooth in shots with outboard major radius.

Shots with inboard major radius: 115078, 116086, 116087, 116088, 116097, 116099, 116103, 116107, 116108, 116109, 116110.

Shots with outboard major radius: 116080, 116081, 116082, 116083, 116084, 116085, 116089, 116091, 116093, 116094, 116096, 116098, 116104, 116105.

Wednesday February 15 2023 5:40 pm Li 116111-116155 ST+Wall

All shells are retrated 0.5 cm from the inner stopper. At this shell location, the edge mode seems to be suppressed by the shell and sawtooth starts to show up. The mode amplitude decreases to zero at time point later than when the shells are fully inserted. But large mode pops up when q is too low and near 2. So tomorrow I will continue on shot development to get better ST shots. A few good ST shots are 116126, 116131, 116134.

Since no sawtooth existed with shells retracted 1 cm, the next step is to try retracting the shells 0.7 cm and 0.3 cm.

Thursday February 16 2023 5:07 pm Li 116156-116197 ST+Wall

All shells are retracted 5 mm from the inner stopper. A few good sawtooth shots are 116159, 116160, 116162, 116165. A few ST-free shots that have quite outboard major radius are 116161, 116175, 116189. Among these, 116159, 116160, 116161, 116165 should be further compared.

From 116182 to 116186, the OHS did not fire correctly. We turned off the rack power and the crates, waited 30 min and then turn everything on, it started to work again. The possible reason is that the weather is too warm that the circuits are overheated.

Tuesday February 21 2023 8:49 pm Li 116198-116249 ST+Wall

The shells at section 7 are 15 mm away from the inner stopper. From 116198-116242, the shells are 5 mm away from the inner stopper. Outboard and ST shots are 116205, 116206, 116213, 116225, 116230. Inboard and ST-free shots are 116208, 116218, 116220, 116229. Under the 5 mm setup, 116133, 116135, 116139, 116142, 116161 and 116203 are great examples of how slight difference in major radius (plasma-wall interaction) can affect the existence of sawtooth.

From 116243-116249, the shells are 7.5 mm away from the inner stopper. 116246 and 116249 have outboard major radius but no sawtooth.

During 116201, there was some storing issue in south rack and the tangEUV signals were not stored correctly. Druing 116215 and 116216, the banks were ringing really bad. The crates and rack power were turned off for half an hour and a more powerful fan was moved to the basement to cool down the rack.

Wednesday February 22 2023 4:07 pm Li 116250-116278 ST+Wall

From 116250 to 116275, the shells are 7.5 mm away from the inner stopper. Only one shot 116267 had sawteeth, which is a very interesting shot. Compare 116267, 116268 and 115333: 116267's major radius and mode amplitude is similar to 116268 before 6 ms, but similar to 11533 after 6 ms. Sawteeth happened after 6 ms. Why did sawteeth happen?

From 116276, the shells are 2.5 mm away from the inner stopper.

During the run day, the banks have been ringing a lot. 116271, 116272, 116277, 116278 all had bank issues.

Friday February 24 2023 3:00 pm Saperstein 116279-116281 new triggering system test

The purpose of these crates-only shots was to verify that the new triggering system to correctly trigger ONLY CPCI 13 in the N rack and have it record a pulse correctly. After quite a bit of troubleshooting these tests were a success!!

To do these tests, we remade the tree using the test_make_tree_alex.py script, which switches the start_shot trigger from the basement to the new BNC clock and adds init actions to setup the clock. Arming the D-TACQ module had to be done manually from Landau, since the module has not been moved over to Spitzer. The new clock sent a clock signal to CPCI 13 in the N rack as well as a trigger to the D-TACQ module. The D-TACQ module then sent a trigger to the CPCI to start recording and then another 20us pulse to ch 76 of the CPCI 20us later. The master clock, CPCI trigger, and 20us pulse were all T'd off to a scope by the N rack to verify the timing of the signals. Since the basement jorway was never triggered, nothing other the CPCI 13 in N rack actually triggered (or even received a clock). This meant that every other digitizer failed to store.

Most of the testing was done using PULSE_ON to make sure the signals were being sent correctly, and the first two shots were duds because I forgot to turn on the crates and arm the D-TACQ module. On shot 116281, however, all the signals went through correctly and we were able to correctly trigger the CPCI and have it record the 20us pulse at the correct time (as verified using a jScope). The CPCI channel also recorded for the full 30ms it was set to.

More details on this test can be found on the wiki at http://sites.apam.columbia.edu/HBT-EP/wiki/doku.php/tsu_-_testing_n_rack_solo_triggering#test_setup

Monday March 6 2023 5:10 pm Saperstein 116307-116326 Triggering banks w/ new system

The purpose of today’s run day was to test the triggering of the capacitor banks using the new triggering system, and if that worked, move on to generating plasmas. Fortunately, both tests were a success! We were able to verify that the banks/puff fired correctly and we got normal plasmas after turning on the e-gun. However, running required arming the D-TACQ module manually from Landau before every shot.

We first tested each bank individually, varying the voltage settings. Starting from low voltage and moving up to normal voltage settings. After verifying that the individual banks worked, we then fired all banks at the same time and found that there was still no issue.

We generated our first plasma shot on the first attempt with 116317, but it was a wimpy plasma. We found that increasing the puff time improved breakdown, implying that the gated pulse we’re sending directly to the puff circuit may not be the same as the pulse sent using the CAMAC Jorway. We found that we got good breakdown using a puff time (sent to circuit) of 1050us. The rest of the short run day was just devoted to some shot development to see if we could recover something like the shot style of 110428, which we were able to with 116326.

Tuesday March 7 2023 4:54 pm Li 116337-116352 ST+Wall

Everything works fine after the triggering system replacement. The shells are all 2.5 mm retracted from the inner stopper. I was able to get the shots to be similar to previous shots, and got a few ST (116345, 166350) and ST-free (116346, 116348, 116351) shots. This run campaign will continue tomorrow.

Wednesday March 8 2023 4:48 pm Li 116353-116406 ST+Wall

All shells are retracted 2.5 mm from the inner stopper. ST shots: 116356, 116363, 116370, 116374, 116376, 116382, 116384, 116385, 116387, 116388, 116397, 116400. ST-free shots: 116354, 116369, 116371, 116383, 116398. Look further into 116369, 116384, 116388, 116397, 116398, 116400: Similar major radius, but different shot styles.

Tuesday March 14 2023 6:18 pm Chandra/Saperstein/DaSilva 116440-116474 High Power Amplifier Test/Biased Differential Rotation/SOLC Mapping

The goal of today was to verify that the high power amplifier can be controlled directly from the CPCI, to use it to induced differential rotation on a edge 3/1-core 2/1 shot, and to map the current from the probe through the SOLC.

The probe is verified to trigger and output an arbitrary waveform, programmed directly from the CPCI (similarly to the RMP and older bias amplifiers).

Only negative biasing was possible, given that the probe is insufficiently ``dirty''. The shot style (114029) was chosen as it has a clear inversion in the pEUV (associated with a clear m=2 tomogram, locked to the edge magnetic 3/1 kink structure). We do not yet have a clear shot that both shows the inversion radius, and that the m=2 structure can be decoupled from the edge mode.

This was unsuccessful. Attached analysis will show that the q=2 surface should have been radially inward of the probe (r_q=2 ≲ r_bp5 ≈ 105cm), and therefore should not have been directly affected by the E_r well. Multidiagnostic SVD analysis and analysis of the raw data does show the inversion radius in the pEUV signal, but implies that the inner m=2 and edge m=3 components remain coupled during the bias period, during which f_n=1 shifts from 10kHz to ≲ -20kHz

Analysis of the probe current path into the SOLC tiles will be discussed later.

Wednesday March 15 2023 6:37 pm Cha

Use "

" to force a line break. Carriage returns are ignored.

Wednesday March 15 2023 7:27 pm Chandra 116476-116513 Bias Probe Differential Rotation

The goal of today was to better match the q evolution of Ian's successful negative biasing-induced differential rotation (2 ≤ q_BP2 ≤ 3, shot 99267), with the q=3 surface still inside the plasma for the majority of the shot.

Results indicate with reasonable robustness that positive biasing (the only type achievable with a clean probe) cannot differentially rotate the m=2 (potentially a 2/1 tearing mode) and the edge kink components.

The shot style was reproduced reasonably well (116504, a better match to the target than the previously successful positive bias baseline: 114787). Tests at +500V at 103cm (116506) and 104cm (116503) produced robust H-Mode using the new High Power Amplifier (HPA). A periodic "bursting" mode amplitude is observed, correlated with a decreasing ||f_n=1||, and an increase in pEUV and SXR signal, but no clear "slow" internal mode can be extracted by the multidiagnostic SVD. Notice the corresponding jumps in loop voltage, potentially indicative of sawteeth. Also notice the decrease in SXR signal after the bias is removed, once the plasma is again in L-Mode.

Psi-Tri indicates that the probe is between q=2 & q=3 for the majority of the bias period (2.0-4.0 ms), with q=3 inside the plasma for the majority of this time as well.

These results indicate that the specific shot desired (m=2 core with an inversion in the pEUV coupled to an m=3 magnetic edge, with the m=3 spinning up separately during the bias period) should be re-attempted after the up to air for the pEUV filter replacement and HFS SOL tile install.

Fuses blew for the bias probe on shots 116497 and 116510 and were replaced. The HPA is functioning as desired.

Monday March 20 2023 1:38 pm DaSilva 116407-116454 Rewiring the HPA

For this campaign, we successfully rewired the HPA so it can now be triggered remotely through ./do_awg code just like the control coils and the low power amplifier. The signal comes from CPCI port 51 on the South Rack, goes through a Jensen transformer (also on the south rack), and into a T-connector that sends the signal into the HPA, an oscilloscope, and a 1kohm terminator (likely to be changed to better match the impedance on both sides of the transformer).

To use the HPA, make sure you are using the most updated jScope file in my directory (labelled "BP5_V_and_I_Ibox-1-26-23.jscp") for the correct gains when monitoring the signal (to test, make sure shot 116448 goes from floating (-40V) to 80V then back to floating). Next copy the file /home/dasilva/hbt/control/prep_awg_HPA_bias.py to your working directory where you can edit the waveforms. It should be decently documented (at least the parts that I've used) but if you need help with editing just ask!

To run the code, first do the command `./init_rtm-t.sh` ON CALIBAN (everything is on Caliban) at the start of the day before doing anything else to initialize everything. Then before taking a shot, run `./prep_awg_HPA_bias.py` which will either do nothing or give you a 1-2 line warning, both are fine and Caliban should fire. Finally run `./do_awg` and a bunch of stuff will pop up. Make sure the last line says "Running..." before hitting fire, then once HBT fires make sure that Caliban isn't stuck on "Running...". If so then the HPA did not fire.

Thursday April 13 2023 5:22 pm Li 116555-116600 ST-free+biasing

Today's run goal is to use the bias probe to stabilize the edge mode and see if the internal and external modes decouple. The bias probe only affects the modes outside the tip. So, there is no point putting it at the q=2 surface since all modes will be affected. Putting it at 105 cm or even further out which is outside the q=2 surface makes it only affect 3/1 component.

1. If the coupling disappears, then there are 3/1 kink and another mode (possible 2/1 or 1/1)This might lead to something interesting. 2. If the coupling still exists, then there is only 3/1, the internal mode structure might come from the 2/1 component of the 3/1, which rotates rigidly together.

The bias probe is at 105 cm starting from 115592. A few shots to look further into are 116592, 116593, 116595, 116596, 116599. Tomorrow I will try to vary the biasing amplitude and probe location and see what happens.

Friday April 14 2023 5:10 pm LI 116601-116648 ST-free+biasing

All shells are fully inserted.The biasing probe was first placed at 105 cm, which did not affect the mode as much as we want. So we inserted it further to 104 cm. A few good shots are 116632, 116637, 116638, 116639, 116640, 116642, 116643, 116647.116646 and 166647 have almost the same plasma parameters while one is biasing on and one is biasing off, which makes them a good comparison pair.

An interesting observation is that 116646 has quite outboard major radius, while is still sawtooth-free. The plasma current and q profile in this shot are slight different from the standard shot style used in the previous run campaign (was not able to reach the same current curve since increasing OHE above 220 caused disruption, which might be casued by different plasma condition from previous shots).This suggests that the threshold major radius for sawtooth to occur (edge mode to be fully stabilized) is dependent of the plasma parameters and different under each shot style.

Monday April 17 2023 5:03 pm LI 116649-116689 ST-free+biasing

Today's run did not go very well since the plasma was quite unstable and the current was not generally lower than the previous shots. A few shots that might worth looking into are 116664, 116666, 116667, 116668, 116671, 116674, 116675, 116676.

The switch of the heat lamp in the back of the basement is broken. I unplugged that power strip for now. A few of the shorting sticks need to be fixed since the cores have fallen out.

Monday April 24 2023 5:19 pm Li 116690-116724 ST-free+biasing

Today's run continued to explore if biasing can separate the edge and internal modes. However, the target shot style (115336) seems to be unstable with the biasing probed inserted, and the biasing probe was not working.

To solve the stability issue, there are two further steps. First, try retracting the probe and see if the shot can run stably. Then, try a different shot style with slightly more outboard major radius and lower current ramp which may be more stable.

One hardware issue is the switch for the heat lamp in the back of the basement is stuck and the lamp has to be turned off by unplugging the whole power strip from the transformer. This needs to be replaced.

Tuesday May 2 2023 5:21 pm Li 116725-116781 ST-free+biasing

During today's run, the first goal is to check if the targte shot style 115336 can be restored with the bias probe fully retracted, which was achieved (see 116741).

Then the bias probe was inserted to 104 cm with v2 = 4.5, a few good shots during which the edge mode slowed down are 116745, 116749, 116752, 116753, 116754, 116755, 116757, 116758, 116759. In 116758, the mode amplitude was suppressed to a very low level.

After this, the bias probe was further inserted to 103 cm with v2 = 4.5. A few good shots are 116761, 116762, 116763, 116764, 116767, 116768, 116769. However, the bias probe tends to push the plasma inward and makes it hard to match the major radius with the target shot. During 116762 and 116763, the mode amplitude was suppressed.

To have a better matching of the major radius, the bias probe was pulled out again to 104 cm, but with a higher biasing volatge v2 = 5. Some good shots are 116778, 116780, 116781.

The current curve was maintained to be very close to the target shot during today's run. 104 cm seems to be a good location of the bias probe. Tomorrow's run will continue with the setup of bias probe at 104 cm and v2 = 5 (subject to change).

Wednesday May 3 2023 5:05 pm Li 116796-116818 ST-free+biasing

Note: Puff time has been 1050 since 116329, not 850 as recorded.

Bias probe was at 104 cm during the run day. v2 has been varying. v2=5: 116803, 116804. v2=5.2: 116806. v2=5.5: 116807, 116808, 116809, 116810. v2=5.7: 116811, 116814.

v2=5.7 seems to be too high and the fuse died two times.

Friday May 5 2023 7:14 pm Chandra 116819-116866 RWTM Wall Scan

The goal of today's run was to evaluate the the predictions of the RWTM model for pre-disruptive 2/1 TM growth vs wall position.

The target shot was 115630. It was difficult to perfectly reproduce this shot.

Shots 116850-857 had the shells all 1.9cm from the nominal plasma surface, were very repoduceable, and matched the target shot style quite well. Shots 116858-859 disrupted at similar value of R_m but a higher value of q_a, which is worth investigating as well. At the end of the day, we explored a shot style with slightly more outboard R_m. Shots 116863-866 are reasonably reproducible.

The goal of next week's run will be to pull the shells further back, and explore the entire range of r_w. We predict a 25% reduction in the 2/1 TM growth rate.

Determination of the growth rate is complicated by uncertainty in selecting the proper start/end time. We observe some decrease in temperature during the lead up to the current spike, which would be expected by a Thermal Quench preceding the Current Quench.

An appended plot shows the magnetic and SVD 2/1 evolution leading into the disruption, with a growth rate roughly x2 higher than what the model would predict, but further analysis is needed here.

Wednesday May 10 2023 6:17 pm Chandra 116867-117000 RWTM Wall Scan

These two run days are a continuation of the RWTM wall scan. Shots 116867-116887 are at a plasma distance of +1.9cm, shots 116888-116965 are at a distance of +3.6cm, shots 116966 are all in.

Disruption groups of note:

     - 850, 851, 855 (+1.9cm, low R_m).

     - 868, 878, 880, 882, 883, 884, 885, 886, (+1.9cm, centered R_m)

     - 891, 892, 894, 902, 903, 908, 912 (+3.9cm, high R_m)

     - 896, 897, 907, 909, 910, 913 (+3.9cm, centered R_m)

)

Notice that all these shots disrupt around q_a of 2.0-2.2. We were able to reproduce some of the previously observed low mode amplitudes with the walls out. We were not able to successfully achieve a high R_m disruption with the walls all in. This will be worked on further. This dataset could also be used to confirm the sawtooth trend with r_w observed previously.

Mode growth rates have not yet been fitted to the disruption, although some evidence of thermal quench during the current spike is observed.

Friday May 12 2023 5:15 pm Chandra 117001-117098 RWTM Wall Scan

The goal of these runs was to finish the preliminary RWTM wall scan campaign. Shots 117001-061 focused on collecting a database of walls-in sharp disruptions at various values of R_m, Shots 062- focused on testing the effect of the bias probe to lock the plasma, to remove the effect of rotation from γ_2/1.

Using the control coils as a δV_f pulse to get over the q=3 3/1 kink early disruption, we successfully obtained a useable set of sharp 2/1 disruptions to analyze.

The walls-in no-mode shot style observed by Boting was confirmed.

Finally, the bias probe shots were attempted, but the plasmas were not well-repeatable. Shot 117093 shows mode deceleration due to the bias probe from 5-7ms, but this was the only usable shot we obtained.

Monday July 3 2023 12:37 pm Levesque 117122-117192 Commissioning new HFS SOL tiles, replacement pEUV filters.

This was the first 2 run days after an up-to-air for diagnostic upgrades and repairs. The tokamak was vented for about one month, ending on June 15th, followed by a bake for 12 days. Continuing the bake longer could have further helped the base pressure, but didn't seem necessary; water was still the dominant partial pressure, but we decided the partial pressures were good enough. The bake temperature was a little lower than our usual temperature. The thermocouple readings were more erratic than usual, with inconsistent temperature variations around the machine with restarts of the bake control code, so I set the target temperature a little lower. The bake control code or thermocouple connections should be debugged/improved, but this cannot be done during tokamak operations due to destructive ground loops when the system is connected and the OH is pulsed.

Machine changes during the up-to-air:

• High field side scrape off layer (HFS SOL) tiles were installed in the left side of Section 2 and the right side of Section 4, with 8 independent tiles at each location. The naming convention is that tile 1 is at the bottom, 8 is at the top, and numbers increase from left to right when looking toward the HFS. Tree nodes are at .SENSORS.SOL.HFS##_S#, with the first number for the Section and the second number for the tile. More detail will be posted on the wiki when I'm able. The installed structures change the HFS limiting locations of the plasma, so equilibrium calculation codes must be updated with the new geometry. The nominal HFS limiting surface is now more complicated, rather than the prior HFS-limiting flanges at roughly R=75.4cm, Z=0cm. The new geometry can be approximated as the plasma limiting at R=77.2cm, Z=+-3.3cm, but more accurate numbers and the stepped+curved geometry can be used in codes compared to what I'm estimating here. Initial data from the tiles shows that they are operating well, except for HFS04_S6 which probably has a broken amplifier channel (will be repaired soon). Calibration is roughly correct for the tiles that have been checked so far, though overall system polarity needs to be confirmed.
• Two of the poloidal EUV filters were replaced: the upper-midplane filter (for DET025) and the topmost filter (for DET090). The system now has fully intact filters on each of the 4 arrays. Previously the upper-midplane filter had been roughly half open ever since the separate aperature/filter installation in December 2021 (that filter was damaged during install, and there were no remaining spares at the time). We found that the previous topmost filter was almost completely eroded away from tokamak operations -- only a very small amount of aluminum remained on the fully-intact copper mesh. The timing of this erosion/opening can be inferred by looking at historical signal levels. Note that during installation, Jim found that the newly installed filters were manufactured wrong -- the filter is on the wrong side of its mounting piece. This presumably makes the filter more prone to damage during installation, but it is now shielded a little better from plasma bombardment. The as-built filter geometry is acceptable for the locations where we installed them, but would not be acceptable for the bottom-most array since filter debris would be much more likely to fall onto that diode array and cause permanent damage. Initial plasma data shows that this system is now working well.
• The damaged two-sided bias probe at Section 2 was removed. The 10" conflat attachment for this section has been changed so that there is still a gate valve, but the CXA ports are no longer there (CXA-specific ports were unused for many years). The hemispherical bias probe previously in Section 5 was repaired and moved to Section 2. It was unclear when the damage to the hemispherical probe had occurred.
• Sections 4 and 5 no longer have probes installed since access was needed during the up-to-air. Probes will be installed when they are needed; these sections still have standard gate valves.
• A replacement ion gauge was installed in place of the broken gauge, including a new insulating break. The insulating break gets hot due to the ion gauge filament. Use caution around that area of the machine. Jim will check into how acceptable this configuration is.
• The shunt-based LFS SOL tiles were plugged back in at the digitizer, at hardware location .DEVICES.SOUTH_RACK:CPCI_12:INPUT_49 through 60 . These had been unplugged since 6/14/2022 due to using biasing diagnostics instead in the desired CPCI input connector. The tree has been updated so that raw sensor subnodes point to the right locations, but some existing scripts might use an incorrect older set of hardware nodes instead.

Plasma results during cleanup shots have been good so far. Currents collected by the HFS SOL tiles can be seen rotating poloidally from tile-to-tile during disruptions. Peak currents observed in the HFS tiles so far have been up to about 120A. The fast cameras have had some triggering issues during the initial setup, which are being resolved as we improve the temporary triggering scheme. The cameras are set to view the HFS SOL tiles during the commissioning. Evidence of a small amount of material leaving the tile structure has been seen for one shot so far, based on bright specks moving in a video, but this is not yet concerning. One of the broadband USB spectrometers was taking data. This presently has a timing/triggering issue that needs to be resolved.

Friday July 7 2023 11:59 pm Levesque/DaSilva/Notis 117193-117250 Cleanup shots

Two more days of running cleanup shots. There were triggering problems for 5 shots yesterday and 1 shot today. We spent some time troubleshooting the triggering issues. In each failed shot yesterday, it's possible that none of the triggers/gates were sent around the experiment for the banks or digitizers. For the present stage of the triggering upgrade, a BNC 577 pulse generator normally sends out 1) the master clock and 2) the single initiating hardware trigger for the D-tAcq to initiate the distributed pulse sequence. For the shot that failed today, an oscilloscope in the triggering path showed that the 577 unit did not send out its master hardware trigger. This suggests a problem between spitzer sending the software trigger to initiate the shot and the BNC 577 receiving the software trigger, or with the output of the 577 unit. A further debug stage was set up for some subsequent shots today, but no additional shots failed. This should be investigated if problems continue.

Successful shots showed good evidence of plasmas cleaning up and getting better. The final shot (117250) was the longest, lasting until around 5.5ms. There were good modes in many shots. HFS SOL tiles showed oscillating currents related to mode activity during the main discharge, with current magnitudes being larger for plasmas that were closer to the HFS. HFS SOL tiles had interesting features during disruptions. Most of the shots today had strong negative currents in at least 1 tile, suggestive of electron emission (convention is that negative current is collecting ions or emitting electrons). At least one shot had a tile exceeding -300A, saturating the digitizer with the present circuit gain.

Fast camera videos were saturated during the disruption for most of the shots, so for the final two shots the exposure was reduced from 3us to 1us. The video for the last shot shows clear evidence of cathode spots on the two tiles that make up HFS02_S1 during the disruption, consistent with strong electron emission. Most cases of strong negative currents was from tiles on the lower-phi side of the arrays, meaning that electron emission would preferentially be in the counter-Ip direction (consistent with the loop voltage).

Tuesday July 11 2023 4:44 pm Levesque/Wei 117251-117266 Cleanup shots and studying cathode spots

Continued cleanup shots while also looking for cathode spots on the HFS SOL tiles. Moved the camera on the tripod at Section 3 to view the top HFS SOL tiles in Section 4, instead of the bottom tiles (which it was set up for the previous run days). The exposure was at 1us all day, and the apertures were reduced for the last 3 shots. Also temporarily connected External Rogowski D's jumper in series with the existing grounding for Sections 4 & 5, which are grounded through the pEUV system conduit (low pEUV noise configuration). This diagnostic jumper has been off the machine for several years. This allows us to measure currents from Sections 4 & 5 to the south rack. Positive values here mean conventional current flowing from the rack to the vessel. Disruption currents through this ground got up to around -50 amps at most.

Cathode spots were apparent again during disruptions, and good videos were taken. A star shot with many cathode spots on Section 2 tiles 6 and 7 was shot 117264. Negative current saturated the digitizer at ~-300A for a couple shots.

Recent shots have had a strange feature at around 2.5ms in the tangential EUV system around chord 12, especially in the Ti array (see 117256 for example). For the last 2 shots (and also the next run day), the filter wheel was set to block the tEUV arrays in order to check for pickup. The blocked shots did not have the ~2.5ms feature, meaning it's not just an effect of a strange magnetic pickup or current path through the tEUV system.

The triggering system was again not triggering shots at the start of the run. This was also the case for crates-only pulses and manual attempts at triggering the 577 unit. It appeared that the output voltage for the master trigger was too low, so we increased the programmed voltage. This resolved the problem for now. The programmed output pulse voltage was raised from 5V to 6V, though the actual output is only going up to around 3V for the tests. The lower output voltage is not yet explained since there's no obvious low impedance in the output path for this trigger. This should be investigated to improve triggering robustness.

Thursday July 13 2023 11:45 am Notis 117267 117309

The run theme was cleanup shots. Throughout the day, plasmas showed evidence of cleaning up, with the longest shots lasting longer than 6 ms. All of the shots triggered properly, so the solution to the triggering problem of the previous two run days is holding up. The parameters determining the major radius (e.g. VFE, VFS, and VF start time) were adjusted to see their effects on the length of the shot as well as effects on the major radius.

The longest shots showed at least 3 different periods of mode activity during the main discharge, with several, such as 117294, showing a mode with frequency ~30 kHz around 3ms. This mode is correlated with interesting behavior in the HFS tiles: a current offset in addition to the high frequency oscillations in some of the tiles, notably in tiles 4 and 5, which are radially farther from the plasma and rarely see currents of this magnitude (~10 A).

Wednesday July 19 2023 10:43 pm Chandra 117310-117430 pEUV Mode Amplitude Studies

The goal of this campaign was to verify the USB spectrometer timings and to investigate the following mode behaviors following sawteeth crashes: Mode bursts from low mode amplitude, the jump to saturated amplitude, bursts from saturated, and the burst preceding the disruption.

Shots 117310-117353 were crates only testing for the USB spectrometer. Currently, it takes data on frame 3 for 10ms.

The primary target was 103590 (low mode amplitude followed by a saturated level for at least 2ms). It was found very difficult in the post-upgrade configuration to achieve saturated mode amplitudes. This will be attempted further for the rest of the week.

The target shot style was reasonably reproduced, particularly in terms of I_p and q. Mode bursts from low amplitude, and sawteeth crashes preceding the disruption were observed very well. Preliminary analysis will be attached.

Finally, It is noticed that that gain-corrected pEUV profiles are changed in the two arrays which were replaced (25^o and 90^o>). This may support the hypothesis that the "supplementary" correction of the signal profile to something self-consisitent is due to defects in the filters themselves. The signals have been corrected. Notably, the plasma profile appears to be smaller in the horizontal direction than in the vertical direction, consistent with intuition given the placement of the HFS-SOLC tiles.

Star Shots: 117407, 117428

Friday July 21 2023 3:16 pm Chandra/DaSilva 117431-117471 Bias Probe Differential Rotation

The goal of today was to trigger differential rotation with the negative biasing.

As has been previously observed, the probe is only able to create a finite negative E_r well when it can drive finite negative current (otherwise, the sheath produced does not penetrate into the plasma). With the probe in the "Dirty" configuration post up-to-air, we were able to achieve this result once more.

Much of the start of this campaign was devoted to verifying that the High Power Amplifier was working properly after the up to air, and after being moved from section 5 to section 2. This was verified.

Shot 117464 had excellent, clear, and sustained negative biasing, with a clear 2^nd mode observed. In the Multidiagnostic SVD, we clearly separte the 3/1 and m=2 structures, which is clear evidence that the m=2 is not simply part of the 3/1 kink eigenfunction.

Analysis will be attached.

Star Shots: (differential rotation) 117464, 116466, (saturated modes)117441, 117446, 117448

Machine Configuration Changes: BP₂ @ 104cm

Thursday August 10 2023 5:23 pm DaSilva 117512 - 117546 Setting up Rake Probe & High Frequency Magnetic Data

This run campaign consisted of 3 major goals running in parallel: looking at spectrometer data before and after the S4 gate valve is opened lead by Jeff and Jim, setting up the rake probe lead by me, and getting high-frequency magnetic data on an oscilloscope via an external coil lead by Jeff and Arian. The campaign spanned 3 run days and I will organize this report by first discussing the 3 goals in the order listed above.

Spectrometer The run campaign began with the rake probe behind the gate valve still being pumped down to ~1mTorr. Jeff and Jim used the spectrometer & RGA to look at impurities before and after the gate valve was opened for shot 117523. Note that pre-gate valve the spectrometer was on for shots 117517, -20, -21, -22 and worked post-gate valve shots -23 and on (the spectrometer was finicky early on due to a few settings that should’ve been on but defaulted off and the screen room computer falling asleep). The RGA was run before the run day and immediately after the gate valve opened where it saw a ton of water enter the vessel. Overall, the amount of impurities did not change by a substantial margin as neither the spectrometer nor RGA saw much change (besides an increase in water seen in the RGA immediately after opening the valve), and the lack of an impact on the plasma performance.

Rake Probe The rake probe setup went rather smoothly, with no issues with getting it in the vessel or plasma. It is currently (temporarily) digitized to north rack CPCI_40:inputs 33-64, hijacking the SCSI cord that the last few control coils are digitized to. Since only 18 channels need to be used, I will make a more permanent cord in the next week so both can be digitized simultaneously.

During the first several plasma shots (117513, -14, -16-24) the rake probe was exhibiting a concerning amount of noise (all rake probe comparisons are made to Ian’s old data, specifically shots 103254 [vacuum], 105941 [RP behind shells], and 105957 [in plasma @109cm]), fluctuating between up to +-20V, most of the noise being high frequency. Vacuum shots were reasonable. However, we noticed that the RP was picking up MHD activity (even though it’s in the bellows and behind a gate valve) that was also seen by Arian’s differential oscilloscope connected to a PA1 sensor (CPCI:50), leading us to believe the RP had a bad grounding connection due to a ground loop: NR-->RP CPCI & breakout board-->probe stand-->vessel-->NR through S3 jumper (north rack grounding config). After some troubleshooting (i.e. continuity tests in the RP box, considering differential voltage dividers, changing plugs to eliminate ground loops as best as possible) we found the following configuration produced the least noise:

The grounding cable that runs out of the RP box (the original – terminal that all the pins shared) was taken and run into the positive terminal of a different CPCI on the breakout board (CPCI_40:64). Both the shared negative for the pins and the negative for the new channel was attached to a screw on the front of the CPCI, essentially digitizing the RP in a differential condition. In shot analysis, we expect then the true signal to be (signal on pin – ground reading)*gain. Through some experimentation in python we found that the best config is (pin – 0.96*ground)*gain, which is applied to the top 12 pins in the test_rake2.jscp file in my config folder, though this will be renamed to something more permanent when the digitization is permanent. Doing this produced very clean data, only fluctuating about +-1.2V when the rake probe was in the bellows (shots 117534 & -35) as opposed to the ~15-20V from before. The high frequency noise appears low enough compared to Ian’s that turbulent data can be collected.

Starting with shot 117536 the rake probe entered the plasma 1cm at a time, each time producing reasonable data so long as it was stored (CPCI was unplugged for -38). The bias probe was inserted to 105cm starting shot -40, getting H-mode shots for shots -41 through -43. All the H-modes were exceptionally strong, where shots 117541 and -43 are star shots for analysis. It’s notable that these were the first H-modes since the up to air, and we are seeing some new behavior not seen before. Specifically, the SXR midplane data climbed rapidly and saturated for shot -43, -42 saw a back transition halfway through the biasing followed immediately by another H-mode despite the probe being at 400V, and we saw the bias current staying uncharacteristically constant later in the biasing period. Overall, we observed the behavior we wanted and got data on par if not cleaner than Ian’s.

High Frequency Probe Arian’s campaign began by looking at PA1 signals on an oscilloscope using a differential setup, though this setup only lasted a day in favor of an external coil. Arian and I constructed an external coil ~8cm in diameter with 10 turns, that connects to an oscilloscope in the 50ohm setting through a 1mF capacitor. The coil was zip-tied to the vessel right next to the quartz break between sections 6 and 7 at about the midplane. After encountering some problems with the oscilloscope randomly triggering after a shot and erasing the data, Arian was able to store data with: 1 shot using a no-area coil to measure the background noise, 4 shots with the coil measuring the poloidal fields, and 3 shots each measuring the toroidal and radial fields. The preliminary data looks promising, and should inform the construction of an in-vessel high frequency magnetic probe.

Tuesday August 15 2023 4:01 pm DaSilva 117547-117561 Permanent RP digitization

The goal of this run was to finalize the rake probe digitization such that the control coils and RP can share the same cable going into CPCI_40 inputs 32-64 in the north rack.

The RP pins are now digitized at CPCI 45-62, a diagram of which is which will be put on the RP page on the wiki shortly. I have set up a jScope in my config folder called "RP4-2023..." that views all 17 pins plus the BP2 V and I traces.

We saw severl Er wells (see attached files in the follow-up email) so today was overall a success!

Wednesday September 6 2023 7:42 pm Chandra 117562-117582 USB Spectrometer

The goal of this run day was benchmarking the triggering for the FX spectrometer.

Shots 117562-117569 were crates only.

For shots 117570-582 the fast camera was on. We are now reliably triggering the FX spectrometer on every shot. The maximum rep rate is 4.5kHz, with a minimum integration time of 10us. The minimum usable integration time appears to be roughly 300us (plots will be attached), for a SNR of 2.

Star shots are 117580, 581, 582

Monday September 11 2023 9:30 pm Chandra 117583-117624 2/1 Island Shape Wall Scan

The goal of this run day was to observe the effect of wall position on the structure of the 2/1 mode (inferred by the q_a value to be a TM).

Shots 117598-604 were repeat shots with the walls all in to the stoppers (average wall position +0.5cm.

Shots 117608-614 and 616 had the walls all out their outer stoppers (average wall position unclear, something between 4-5cm).

Shots 117618-624 had the walls at a uniform distance of +2.5cm.

While it was difficult to get a sustained period of saturated modes, there was enough time to extract both the 2/1 sin/cos mode and either sawtooth crash or 1/1 sin/cos modes. these three or four modes made up almost 90% of the mode energy in these shots, with the next largest mode being a 3/2 at double the dominant mode frequency.

Strong sawtooth activity was seen in all shots.

Analysis will be attached showing the viability of this shot database for analysis.

The fast camera and FX spectrometer were operating for all shots taken today.

Machine changes: The shells are inserted to a uniform distance of +2.5cm from the nominal plasma surface. This surface is set by the blade limiters, which remain all the way in. The pEUV is in the low noise configuration (for the walls-in shots, it was in the high-noise configuration). The front and back doors of the north rack were closed and locked for thermal as well as electrical efficiency. The HPA trigger and tEUV power cable were re-routed in through the side of the west rack case to their respective destinations, in anticipation of sealing that rack. A heavy gauge 240v cable must be found to extend the old bias amplifier rack power cable (we could discard this if we are only going to use the HPA going forward. One of the mercury lamps in the basement took some jiggling to turn on this morning.

Wednesday September 13 2023 9:07 pm Chandra 117625-117667 2/1 Island Shape Wall Scan

The goal of today was to continue investigating the 2/1 island vs wall coupling, with the Rake Probe withdrawn (to 120cm), and later in the run with a less steep current ramp following Boting's shots 115330 (wall out) and 115331 (wall in).

It was difficult to get sustained 2/1 modes with the wall at +2.5cm uniformly. Note that perfectly matching the target shots is hindered by inaccurate determination of q in hbtplot due to the new HFS-SOLC tiles. It is likely that q was closer to 2 than in the target shot.

Shots 117558-662 were a repeatable shotstyle, with a 2/1 dominated disruption around 4.5-5ms.

To avoid the early disruption, shots 117665-667 had a small V_f pulse from the CCoils. This did not appear to extend the shot. Shots 666 and 668 had opposite sign pulses (visible in the direction that R_m is perturbed).

For some unclear reason, the pEUV was not saving data from shots 117625-117652. Restarting the racks solved the issue.

The fast camera was on starting at around shot 117650.

We verify that the FX spectrometer is triggering at times and integration periods which can be set arbitrarily in the tree. The new integration time node can be found/set at TOP.SENSORS.USB_SPECT_FX.INTEGRATION. It is verified that the spectrometer is seeing all the triggers. However, there is some timebase shift issue which is causing the pulses to be sent out with some linear shift in time. This will be investigated further with the fancy oscilloscope.

Further work will continue the wall scan.

Friday September 15 2023 3:56 pm Chandra 117670-117721 2

Use "

" to force a line break. Carriage returns are ignored.

Friday September 15 2023 4:14 pm Chandra 117670-117721 2/1 Island Shape Wall Scan

The goal of this run day was to continue exploring the effect of the wall on the shape of the 2/1 (presumed) island at r_w=+2.5cm and +4.5cm.

Star shots from +2.5cm are: 117676-678

Star shots for +4.5cm are: 117686-688, 707-709, 717.

It was difficult to get stable, saturated modes with the walls fully retracted.

There is enough usable pEUV signal to hopefully be able to measure the width and position of the m=2 internal perturbation.

Preliminary analysis suggests that while the amplitude of the m=2 internal perturbation tracks that of the edge mode very well (both when the edge mode is m=3 dominanted and when it is m=2 dominated), the location and width of the perturbation do not change very much over time (as measured by a rolling SVD window, if at all, unlike the q₂ and w_2/1 predictions from the edge Mirnove mesaurement + PsiTri. It is at present unclear how to interpret this discrepancy.

The FX spectrometer was active for most shots. It is noticed that there is some relatively large dead time necessary between integration windows to capture data, and that the max frequency is slightly below the spec'd frequency (4.25kHz vs 4.5kHz). At an integration period of 20us, we capture D-alpha for the entire shot, and D-beta for the breakdown and disruption. At 300us, we capture some impurity line radiation during the disruption. Even with binning and summing of frames, it may not be possible to detect impurity radiation at a sub-MHD frequency, as was desired.

The fast camera was on for most of the described shots.

Analysis will be appended.

At around shot 117691 the tree had to be rebuilt, to allow take_shot_hbt to launch.

Monday September 18 2023 9:16 pm Chandra 117723-117760 2/1 Island Shape Wall Scan (r_w=+4.5cm)

The goal of today was to finish the "walls-out" component of the lower ∂_tI_p 2/1 inferred TM shape and dynamics scan.

The goal was to match as closely as possible one of the star r_w=+2.5cm shots from the previous run (117676). This was achieved better than on the previous run day, although it was not possible to achieve quite as long a period of saturated 2/1 modes as in 117676. A comparison with previous walls-out shot 117623 (higher ∂_tI_p) shows qualitatively similar dynamics.

Star shots: 117744, 746, 748, 749, 752, 754, 758.

Note: shot 117750 had the control coils activated in a ±30A V_f pulse as a demonstration of the GPU control system for Javier.

Tuesday September 19 2023 7:22 pm Chandra 117761-117780 2/1 Island Shape Wall Scan (r_w=+0.5cm)

The goal of this run day was to complete the "lower ∂_tI_p ramp" 2/1 shape wall scan.

The run was quite successful. All shots in the range 117768-117780 except 769, 777, and slightly 780 were very good matches to the target (117676, r_w=+2.5cm), and also to the earlier walls-in target shot of 115330/331 (See attached plots). It was very easy to get to a stable shot style. The shot style featured a ∼1ms 7G saturated mode period.

It is interesting that this saturated period was so inaccessible in the other two configurations, particularly the walls out (r_w≃+4.5cm) configuration.

Next Steps: The same analysis applied to the previous wall scan will be applied to this one, to attempt to detect any shape changes ≃r_w.

Machine changes: The walls are all in, except for section 10 (camera section). All probes are withdrawn. The pEUV is in the "low noise" grounding configuration. The FX spectrometer is at 20μs.

Further notes: There is a larger than average amount of oil around the reclamation barrel in the basement. Some of the oil in that barrel was transferred to the reservoir barrel.

Sunday October 15 2023 6:26 am Notis 117835-117876 HFS Tile Bias

The goal of this run campaign was to bias tile 6, both to measure I_{sat} in the SOL and to see the effect of tile biasing during the plasma disruption. Following upgrades to the section 4 HFS tile array, tile 6 was biased both positive and negative, with varying amplitude, and was biased both during the disruptive and non-disruptive portions of the plasma shot. The first day of the campaign (10/12), we had trouble connecting to Caliban, the issue was that the external connection was disconnected in the server room.

The only shot that was biased positive was 117845. When biased positive, it measured increased positive current. From 117846 and on, the tile was biased negatively. During shots 117848-117851 the bias was moved from 5-7 ms to see the effect of tile biasing during disruption.

On Day 2 of the run campaign, we tried to find a shot style conducive to I_{sat} measurements. We focused on getting the major radius fairly constant by changing the VF start time. Shots 117859-117863 were biased from 2-4 ms, and Shots 117864-117870 were biased from 3-4 ms. Finally, we returned to disruption biasing using the constant major radius shot style, biasing from 5-7 ms. Shots 117873-117874 were biased from 5-8 ms.

We found that the biasing system did not allow us to bias the tile further negative than -130 V. In the future, we would like to ramp the current negative to measure I_{sat}. A useful jScope file for seeing the results of this run is hfs_bias.jscp in my folder on Spitzer.

Star shots from the non-disruptive biasing for I_{sat} are shots 117868-117870. Star shots for disruption biasing are 117871-117874.

Tuesday November 21 2023 12:06 pm Notis 117877-117934 HFS Tile Bias

There were two goals of this run campaign: continuing measurements of I_{sat} in the SOL as well as seeing how positive tile biasing influenced the collected current on tile 6 and other tiles. Along the way, I noticed that the shot style in use led to fairly long 2/1 modes, so I tried biasing for a short time pulse in the middle of the mode to see if there was any effect. There did not seem to be any effect on the mode or the other tile currents. It would be interesting to bias with a larger current to see when and how the biasing begins to influence other parts of the plasma. Further, it would be interesting to see if there is any internal plasma effect of the biasing.

For the two run goals, a shot style was used with fairly constant major radius, and ramped biasing was used. Shots with negative bias: 117892-117896. Shots with positive bias: 117897-899, 901, 903, 918-927, and 931. Shots with a short positive burst during the 2/1 mode: 117933-934. Shots with 2/1 mode but no bias: 117928-930. Using the results from the run, I hope to build an I-V curve for the tile using similar plasma shots.

A useful jScope file for seeing the tile voltage and current is hfs_bias.jscp in my folder on Spitzer. A useful file for seeing the tile arrays is hfs_sol_new.jscp.

Monday December 4 2023 12:14 pm Notis 117935-118019 HFS Tile Biasing (cont.)

The run took place on 11/27 and 11/30. On 11/27 similar biasing was done as in previous runs, except that the transformer was in the 1 to 2 voltage configuration, enabling tile 6 to be biased to a higher voltage. On 11/30, two new tests were tried. First, the double probe configuration was tested with tiles 6 and 7, and two tiles which for certain q values lie on the same field line were both biased positive. For the double probe configuration, the grounding scheme had to be changed on the biasing box, instead of one of the transformer outputs being grounded, the output was sent to the other tile. For the same field line positive bias configuration, the grounded output was left grounded, but a Y-connector was made to split the positive voltage to two separate tiles.

Sinusoidal biasing was tested on tile 6, with the transformer in the 1 to 2 voltage configuration for shots 11949-117975, excluding shots 117972 and 73 which were no bias reference shots. Both the frequency and time period of the biasing were varied. 10 kHz biasing was tried from 5-5.5 ms from shots 117949-117954, and from 2.5-3.5 ms from shots 117955-117959. 5 kHz biasing was tested on shots 11960-117968 from 2.5-3.5 ms. 15 kHz biasing was tested on shots 117969-117971, and 20 kHz for 117974-75. These different frequencies were tested to see if they could be detected on any other tiles or magnetics above the other MHD oscillations, which are mostly around a 10 kHz frequency.

Although biasing Array 4 Tile 6 did not produce any noticeable effect on any Array 2 tiles, it may have produced an effect on the magnetics, specifically PA1 S39/30/31. For example, in shots 117969 and 117970, there does seem to be correlated fluctuations of the correct frequency on PA1S30 when there is 15 kHz biasing on Array 4 Tile 6. The second interesting observation is the presence of simultaneous noise on the tiles and magnetics during high amplitude, high frequency biasing. This phenomenon requires further investigation.

The double probe configuration was tested on shots 117986-117991. As expected, the current was quite low, not exceeding 1 A during steady state and several amps during the disruption. A square wave was used for the double probe configuration, with 10 pulses each 100 us long.

The two-tile positive biasing was tested on tiles 6 and 2 from 117994-117999 and on tiles 6 and 3 from tiles 180003-180008. Based on field line tracing sent to me from Jeff, tiles 6 and 2 should be on the same field line at q ~ 3.5, and 6 and 3 should be on the same field line for q ~ 3.3. For these tests, a square wave with 5 pulses, each 100 us long, starting at 2.5 ms was used. Sometimes, as in shots 117994, 117995 there was an extreme response such that two pulses became one long extended pulse. Other times, as in 117997, the tile responded as it always does, and exhibited clean, nonoverlapping pulses. For the 6 and 3 biasing, shot 180003 saw the pulse overlap later in the shot, which is expected due to the field line overlap being at lower q. However, in shot 180006, 6 and 3 biasing also saw overlap earlier in the shot, which is not consistent with the expectation of overlapping pulse response being due to field line connection between 6 and 3.

Finally, after the two-tile positive biasing was performed, tiles 6,3, and 2 were biased separately for reference. Tile 6 only square wave bias: 180009-10. Tile 3 only square wave bias: 180011-13. Tile 2 only square wave bias: 180014-180016. For similar shots, tiles 3 and 2 produced much more current than biasing tile 6. Biasing Array 4 Tile 2 in shot 118014 even produced the pulse overlap seen in the two-tile positive biasing tests. A note about the pinging noise: Section 2 Tile 8 does not see the pinging noise even when all the other tiles do. For the reference biasing of tiles 6, 3 and 2, tile 6 did not produce the pinging noise, but tiles 3 and 2 did.

Wednesday December 13 2023 5:43 pm Chandra 118020-118121 Bias Probe q2 Differential Rotation and ST Decoupling

The goal of this run campaign was to attempt to differentially rotate the q₂ surface and thereby decouple it magnetically from the 1/1 perturbation of the St crash.

Shots 118054-064 were star shots with BP₂ @ 103cm, 0V, Shots 069, 070, 072 had voltage, with 072 showing mode acceleration.

The second day, star baseline shots include 090 092, 097, 102, with applied voltages on 093 (200V), 095 (160V), 098, 099 (180V), 100,101 (120V). These shots showed evidence of the ST triggering modes, which quickly locked and died down again. The overall shot length was shorter.

Shots 104-107 had BP₂ @ 107cm, with unclear difference between baseline (104,105,113,114,120,121) and shots with voltage (106, 107 (120V), 115,117 (200V), 118,119 (100V)). Some shots showed a strong region of mode spindown for a 3/1-like mode, but little significant change in the final jump to saturated 2/1 level.

Results were similar for BP₂@105cm: baseline 108, 160V: 110, 111.

It is overall not clear if the biased flow is affecting the 2/1 growth, or just the 3/1. It is unclear from PsiTri is the q₂ surface was at or outside the probe or not, although small changes in the equilibrium J,P profiles may be sufficient to change this prediction. Results may indicate that the 2/1 is driven primarily by equilibrium profile changes to Δ', which would be unaffected by the sheared-flow profile, or that the flow-shear applied was simply insufficient to decouple the modes. Further analysis will be appended.

Machine changes: Fuses blew for the HPA on some shots and were replaced. The pEUV is in the low-noise configuration. BP₂@107cm. All shells in. Overall the probe performed well.

Friday December 15 2023 7:58 pm Chandra 118122-118184 Bias Probe q2 Differential Rotation and ST Decoupling

The goal of today was to continue the bias probe scan of differential rotating the edge of the plasma.

Results were positive, with BP₂@104cm the saturated mode region was repeatably suppressed, although the ST crashes still were in some cases able to briefly trigger the 2/1 presumed TM.

Results will be attached.

Machine configuration: BP₂ is at 104cm, the walls are all in.

Tuesday January 9 2024 10:08 am Notis/Levesque 118201-118262 Thomson

The focus of the run on 1/8/2024 was the testing of the TS system after the installation of a new power system and a lengthy hiatus. In short, the system is working with several polychromators (4,5,6,7,8,10) showing reasonable data, with 5, 6, and 10 being the strongest. Other polychromators which one would expect to show stronger signals are not. Reasons for this will be discussed below. One star shot to look at using the ts_scope_10_views2.jscp is 118254, which had Thomson fire at 6.8 ms (Look at next paragraph for more star shots).

The shot style was kept fairly consistent to allow comparison of Thomson data from shot to shot. There were several measurements taken at 5.5 ms (118242, 118243, 118248, 118249, 118250, 118251) and 6.8 ms (118254, 118255, 118258, 118260) along with some at 3 ms and 4ms, but these showed less signal or none at all. When we checked the collection optics, the fibers for polychromator 2 and 8 were loose, so in the middle of the run, they were adjusted (before shot 118242), which led to a stronger signal in those polychromators. Further testing is required to diagnose the issue the signals in polychromators looking at core chords which should be seeing more signal than they are (such as 3 and 8). Comparison of the temperatures to those measured in the TEUV system is forthcoming.

Tuesday February 6 2024 11:29 am Notis 118291-118440 HFS Tile Bias

Short summary: The goal of the run was to produce non-rotating helical current filaments observable on the magnetic arrays by biasing the HFS SOL tiles (i.e. to see if we can reproduce something similar to Chapter 6 in J. W. Brooks’ thesis using HFS tiles instead of a bias probe). Look at shots 118331 (tile 3 bias), 118409 (tile 3 bias), and 118418 (tile 3 bias).

Longer summary: The first run day (shots 118291-118340) was mostly spent on finding a desirable shot style. We want to bias during a period of little or no mode activity, so that the external magnetic perturbations are low enough that the current filaments produce noticeable signal on the magnetic sensors. To do this, we attempted to recreate shot 105304, which had very little mode amplitude from 2-3.5 ms and from 4.5-5.5 ms.. We also want a low major radius (~91 cm) so that the density is high enough near the HFS tiles to support currents > 20/25 A .

On run day 2 (shots 118341 - 118395), tiles 3, 2, 1, 8, and 7 were biased, and on day 3 (shots 118396 - 118440), tiles 3, 2, 1, and 8 were biased. Unbiased shots (e.g. 118355, 118368, 118414, 118419, 118424, 118429) were taken every couple shots as well to provide a baseline scenario with which to compare the biased discharges. During the first two run days, the QSC was used to bias a single tile at a time, and on the third run day, the HPA was used in an attempt to drive more current. The currents reached a maximum around 30 A (occasionally going up to 35 A if Rm was low enough), with added voltage leading to diminishing returns in tile current, suggesting that e-sat has been reached.

Good shots that may have recognizable signal on magnetics due to helical current filaments (organized by tile being biased) are: Tile 3: 118353, 354, 357, 118409, 118416, 118418. Tile 2: 118366, 118368, 118432, 118435. Tile 7: 118374. Tile 8: 118378, 118379, 118384, 118423. Tile 1: 118390, 118392, 118439.

Instructions for using analysis code: Navigate to my directory on Landau, and run “python bias_magnetics.py [shotnum] [start_time] [end_time] [tile number] [“filt_on”/“filt_off”] ‘save_off’ ”. This will produce a figure with the current trace and the magnetics from the two poloidal arrays and the feedback sensors.

Further work includes producing an IV curve using the HPA and analyzing data to see if the stationary perturbations appear with consistent phase structure.

Thursday February 8 2024 10:35 am Notis/Boting 118441-118511 Thomson/ HFS SOL

The goal for the run day was to build a library of shots to allow for comparison of temperature measurements of the TEUV and TS systems. The target shot style was 115330. After a short shot development phase, we took Thomson data at 4, 5, 6, and 7 ms (shots specified later), each with several repeatable shots. The short conclusion is that Thomson seems to give higher temperature measurements by up to 50%. Other important notes- a) Polychromators 5 (97 cm), 6 (101 cm), 10 (99 cm) seem to have the most reliable signal. b) The temperature in the screen room rose significantly throughout the run day, with the thermocouples on the polychromators starting around 65 degrees and ending the day about 10 degrees warmer. This means that it might be difficult to compare Thomson measurements from the beginning and end of the day. For the last three shots of the day (118509-511), the fiber optics were switched between polychromator 4 and 5 to see if the line of sight from poly 4 (95 cm) would be better seen by poly 5. This produced reasonable data for 95 cm. It seems like Poly 4 has an issue with channel B.

Piggybacking off of the Thomson/Teuv run, we gathered data on currents during the disruption by connecting Jumper B and disconnecting the PEUV ground as well as by grounding the Section 4 biasable HFS tiles by grounding them back into the shorting bus inside the box. Note however, that the connections are outside the box, but this does not seem to have affected the signals. From 118463 – 118482, the section 4 tiles are all electrically connected to each other but floating. In this configuration, a cathode spot was seen on section 4 during every single disruption. For great fast camera footage of cathode spots from this run, look at shots 118479 (cathode spot on tile 6) and 118482 (cathode spot on both tiles 6 and 7).

Shots with TS at 4 ms: 118472, 473, 474, 476, 477, 480, 482, 483, 484. 5 ms: 118448, 452, 455, 456, 457, 460, 461. 6 ms: 118486 – 118494. 7 ms: 118496 – 118506. To look at the TS data, use the ts_scope_10_views2.jscp file.

Tuesday February 20 2024 6:29 pm Chandra/Notis 118566-118521 Mode shape study with Thomson scattering

The goal of today was to start a cleaner mode shape scan with repeat shots for use with the multidiagnostic multishot SVD. Today, the goal was to start with a scan of q and T_e.

It was difficult to get a consistent shot style with a clean major radius before and after q crosses 3 (see for example shot 118565/566)

We were however successful in verifying which fibers/polycrometers are likely in working order. We deduced that 95-105cm have good fibers/signal, and polycrometers 2,3,5,6,7,9,10 are in working order, although some of them have noisy/difficult to fit signal. We have not yet tested polycrometer channel #1 or the 107cm fiber bundle.

This should be sufficient to get a clear idea of how the profile width is changing as the plasma heats, but it may make isolating the absolute core temperature difficult.

The rest of this campaign will focus on clarify what in the structure of the mode changes as q evolves, followed by a similar investigation of ∂_t I_p, potentially wall position, and potentially control coil flux.

Thursday February 22 2024 6:45 pm Chandra/Notis 118567-118624 Mode shape study with Thomson scattering

The goal of today was to continue using Thomson scattering to study mode shape changes in two situations: before and after q_a ≈ 3, and before and after the sudden "jump" in ||B_θ||.

Key point to note: on shots 118584-603, every single diagnostic that we own was turned on, including both spectrometers and fast cameras, tEUV/pEUV, all magnetics, current tiles, etc. The walls were all in (except sections 10 and 2 for the cameras), all probes withdrawn to 115cm.

The thompson scattering laser shot times were (on reasonably good shots):

118585-587: 5ms

588-591: 2ms

592-596: 4ms

597-600: 3ms

601-603, 614-24: 5ms

The laser pulse time is measured independently in the tree as well.

The current Thomson channels and fibers are mapped and stored in the tree in terms of polycrometer channel and major radial location.

Machine changes: the TF shorting bar is starting to sag a little bit. We should keep an eye on this.

Saturday February 24 2024 4:53 pm Chandra 118625-118649 Mode shape study with static RMP

The goal of today was to collect data on the effect of a static 3/1 RMP on the shape of the mode (as seen by the pEUV)

Thomson scattering was turned off for these shots, the FastCam was on for shots 632-649, with a particularly good fastcam video for shot 634.

The RMP was set to be a static phase, with a 180^o flip half way though. RMP timings and approximate max current below:

634-34: 3.5-4.5ms, 20A

40-42: 4.5-5.5ms, 20A

43, 46: 4-6ms, 20A

47-49: 4-6ms, 15A

Shots showing particularly strong coupling are: 34 (disruption triggered),40,47 (clear mode "ruffling" in the magnetics")

Results will be appended later showing the effect or lack thereof on the pEUV mode shape. Next steps will be to apply a 2/1 RMP instead to couple with the internal component.

No significant machine changes.

Friday March 1 2024 5:49 pm Chandra 118652-118754 Mode shape study with Thomson scattering

The goal of this two day run campaign was to measure, using the Thomson laser, if there are significant profile changes directly before and after the sudden jump in mode amplitude which could imply that the mode growth is merely due to changes in Δ'

Star shots with Thomson include (with laser times):

Day 1 (118600 series): (4.5ms) 57, 66, (5ms) 75, 77, 79, 82, 85, (4ms) 89, 93, (4.5ms) 94, 96

Day 2 (118700 series): (4.5ms) 10, 16, 27, 32, (5ms) 36, 38, 39, 42, 47, 49, 50, (3ms) 54

Preliminary analysis suggests that there is no significant change pre vs post jump in ||B_θ||

There were TF store errors on shots 741, 751, 753, but the data for those shots is otherwise good.

The fast camera was on for all shots, the walls are all in. No other machine changes were made.

Thursday March 7 2024 5:53 pm Levesque 118755-118764 Zoomed-in videos of HFS SOL boron nitride crack

Ran a few shots with the fast camera view from Section 3 zoomed far in on the Section 2 HFS SOL insulating spacer for tile #3. The camera was looking through a theodolite, which cut down on the light dramatically but allowed extremely zoomed views. Since the emission phenomenon I was looking for (suspected to be from a crack in the material or from a line of deposited material) persists for a long time during the disruption, I was able to run at 25kfps frame rates with exposure of 35us to get sufficient light in the most zoomed-in view tested. Note that standard exposures we use during disruptions have been 1us. Good videos were recorded. The bright feature still looks like a crack in the insulator, running most (or all?) of the length of that boron nitride piece. This should not affect operation of the diagnostic (it's unclear how long this has existed), but it could be a problem for the insulator falling apart if the tile structure is physically manipulated, e.g. if it's hit during in-vessel work while vented.

Monday March 25 2024 2:44 pm Notis 118765-118801 Thomson + HFS SOL

The goal of the run (3/20/2024) was gathering Thomson data early in the shot. Data was taken using the Thomson scattering system for 2 ms (118777-118791), 2.5 ms (118792-118801), and 3 ms (118767-118776). The run also tied up some loose ends, such as using the clip-on Rogowski coils to see if there were currents flowing through the puff lines (there were none), as well as measure the jumper currents using the clip-on Rogowskis (118778, 118783 are not cut off for Jumper B), and to individually float all the tiles in section 4 (this is done after shot 118774) to see the effect on the poloidal field between sections 2 and 4. It did not significant change the TA data, suggesting that current is leaving the plasma column through some other path in that section, perhaps the tile array spine or the flange. While we were running, high speed (400 kHz) fast camera data was taken while looking fully at all of section 2 tiles (as opposed to only the bottom tiles). The camera was on for shots 118785-118801. For shots 118788-118792, the camera aperture was decreased to f = 2. For shots 118793-118801, the camera aperture was changed to f = 1.4.

Tuesday September 24 2024 10:02 pm Levesque/Notis/DaSilva 118931-118942 Venting for installation of REMC and 2nd-gen HFS SOL tiles

This report summarizes hardware changes during the machine venting that lasted from April into September. Physics notes from the first run campaign will appear in a separate run report. Hardware changes included i) installation of HFS SOL tile arrays in Sections 7 and 9 (to complement those already existing in Sections 2 and 4); ii) complete installation of one HFS Runaway Electron Mitigation Coil (REMC) which presently has several names including the Runaway Electron Suppression for [Current] Quenches (RESQ/RESCQ, "rescue") coil, passive coil, or Boozer coil; iii) partial installation of one LFS REMC (~90% completion) with remaining work needed in Sections 1 and 2 , iv) removal of the strike point limiters from Sections 2 and 7; and v) a new window installed at the Section 5 midplane port for viewing the Section 7 HFS SOL array and the Section 6 poloidal leg of the HFS REMC. All 10 midplane ports were open during the venting. Baking was done for 6 days. A longer bake would have further helped vacuum conditions, but the bake was ended in the interest of having more run days before the upcoming APS DPP meeting. The teflon gaskets for the midplane conflats at Sections 7 (tEUV) and 8 (small pump stand) were not faced-off during this venting in order to see if we could re-seal with them without machining; the logic was that machining the Section 7 teflon which would change the tEUV sightlines at least a little. These teflon gaskets resealed without evidence of leaks during leak checking.

The completed REMC is the HFS version of the coil. Most of the LFS coil version was also installed (over 90% completion), but the LFS coil was not finished due to time constraints and issues with mechanical interference in the incomplete region. Finishing the LFS coil can be done at a later date by only opening the midplane ports at Sections 1 and 2, possibly using replacement mounting parts locally. Other sections could be opened if desired to verify mechanical or insulation integrity, but this is likely not needed.

The new HFS SOL tile arrays in Sections 7 and 9 are almost the same as the other sections, except that each new array has 10 diagnostic signals instead of 8. In the older arrays, sensor number 1 is collecting current from 2 physically distinct tiles that are wired together at the tiles, while in the newer arrays these tiles are separately diagnosed as sensors number 0 and 1. Likewise, sensor 8 in the older tile convention has been split into sensors 8 and 9 in the new tile arrays. This convention allows for easier comparison among the arrays, as most sensor numbers are in the same poloidal location. E.g. Sensors #2-7 are the same in all four toroidal locations. The newer tile arrays used Mo tiles that were slightly rounded along the edges, rather than the sharp edges that were present in the original arrays. The amplifier circuit grounding and patch panel is located differently for the new tiles such that the new tiles have a longer electrical path to ground compared to the older tiles, but the amplifier enclosures are much easier to work with. Inspection of the previous arrays showed evidence of erosion on the BN insulating pieces between the tiles, with an estimated uniform erosion of around 0.2-0.5mm (visual/tactile estimate from Jeff in a couple of places).

Several LFS SOL tiles in Section 6 became severely misaligned during work in vessel. These tiles were re-located to positions/orientations that appear good enough by eye, though they are likely a little different from the original install -- be aware of this for any differences between old and new data, or if analyzing tiles in fine detail with respect to metrology/alignment. Tiles that are closest to the HFS edge of the shells were the most affected. Two of the Section 2 LFS SOL tiles have quartz discs that broke; one or both of them broke due to work in-vessel. The broken quartz discs might compromise the insulation between those tiles and their mounts that are supposed to be isolating. This should be checked rigorously if quantitatively analyzing the affected tiles. Section 2 tile #3 quartz is intact; two unspecified tiles of the other three have half-broken quartz (unspecified as of this writing -- I can't find the broken ones in my pictures right now).

360-degree camera images we taken in 30 locations inside the vessel with all shells fully retracted. The camera was located near the midplane in each case, near R=90cm, 3 locations per Section. These images will be posted as a virtual walk-through for the tokamak. None of the midplane ports were installed for these images, though the tEUV conflat was installed for a separate later image.

Almost all (5 out of 6) Tangential EUV filters broke as part of opening ports for the up-to-air. This is suspected to have happened while handling the conflat during removal or while the conflat was on the surface where it was stored. The damage was noticed a couple of weeks after opening the vessel. These filters were replaced before reinstalling the tEUV conflat. The tEUV circuit setup is not quite the same as before the venting -- this needs to be debugged when there is time. One previously-broken pEUV filter (for the top-most diode array) was also replaced; the filter had eroded during plasma operations. The SXR fan array filter broke during this venting; this was not replaced -- replacement can happen at the next vessel venting (Section 6.5 top).

Monday October 14 2024 11:14 am Notis 118927-119067 cleanup + HFSSOL commissioning

The goal of this run was cleanup and commissioning of the new HFS SOL tile arrays in section 7 and 9. There were several “runaway” style shots, long shot length, low Ip (e.g. 118948). Shots began to look more normal as cleanup progressed. However, there still is a sudden collapse in minor radius and sudden mode growth happening prior to the full disruption, making the plasma shots fairly short before the long multi stage disruption begins (e.g. 119119).

HFS SOL tiles were installed in sections 7 and 9. Mapping was not done properly originally, due to confusion based on the mapping of a SCSI shared between the HFS tiles and some control coils. A new cable was used for the HFS tiles, but that means that some of the control coils are now not being digitized. Mapping for the new HFS tiles is correct as of 118975-118927 (except that Section 9 tiles 8 and 9 are switched), and is fully correct after shot 118954. Currently, 7:0-1, and 9:8-9 are digitized into the HXR BNC panel, this was done for testing purposes and will be altered for more permanent setup.

Cathode spots were observed on tiles in all the sections. Some were simultaneously seen on the fast camera as well. A good example of this is the arc on Section 7 Tile 7 in shot 118979. It seems like cathode spots are more prevalent on the upper tiles, but there are often cathode spots on lower tiles as well.

One other difference between the upper and lower tiles may be an offset of opposite sign during disruption – see shot 118981 for a good example. Upper tiles see net negative current while lower tiles see net positive current. The sign of the current switches at approximately the midplane.

Some jumper tests were done for shots 119128-119157, using the same shot parameters but with several different jumper configurations. Shots were very repeatable, look at shot 119130 for reference.

Tuesday October 22 2024 10:25 pm Chandra 119280-119341 1/1 RMP Shot Development

The goal of this run was to develop a consistent, longer lived shot in the style of Boting's longer sawtooth period, lower mode amplitude 115704 shots.

Getting a reliable "typical" shot style after the RESCQ coil install proved somewhat difficult, however, after two days of running a consistent shot style was obtained, with reasonably low mode amplitudes, outboard R_m, and acceptable sawtooth signatures. For the second half of the run campaign, we will attempt to couple a ±1/1 RMP to the q₁ surface to modify τ_st. The RMP was confirmed to rotate in both directions on shots 119339 vs 119340.

Star shots: 119337, 338, 339

Wednesday October 23 2024 5:42 pm Chandra 119342- 1/1 RMP Sawtooth Period Scan

The goal of the run day was to collect data on ±8kHz 1/1 RMP effects on Ƭ_st

Reproducible baseline shots in the style of 115704 were found. Star shots (1193xx):

- Baseline: 54, 56, 63, 64, 76, 78, 79, 81, 83, 91, 100, 101, 102,

- +8kHz: 52, 58, 59, 70, 71, 84, 86, 87, 93, 94, 103, 105

- -8kHz: 61, 62, 72, 74, 75, 89, 90, 95, 96, 97, 99,

Note that shots after 119379 matched the target q evolution better.

Note that north rack CPCI 40 frequently had store errors, rebooting the server appears to fix this.

Monday October 28 2024 4:01 pm Notis 119406-119444 Control Coil Radial Field

This run focused on demonstrating that the control coils can be used to produce a largely radial field that leads to a vertical displacement in the plasma. Star shots are 119440-119441 which show a positive vertical displacement from 3-4 ms, and 119442-119443 which show a negative vertical displacement from 3-4 ms. Shots with no control coil fields for comparison- 119438, 119444. To produce the radial field, an m/n=1/0 control coil signal with the proper phase (-pi in the prep_awg.py code) was used.

The vertical displacement can also be seen on the HFS tiles. With a positive vertical displacement, signal largely moves to the top tiles, while with a negative vertical displacement, signal largely moves to the bottom tiles.

To plot the vertical current moment based on the magnetics of PA1 and PA2, use the file vert_moment_only.py in my Landau directory. To generate a plot of [shotnum] from [start_time] to [end_time], navigate to my directory and type in “python vert_moment_only.py [shotnum] [start_time] [end_time] &” into the command line.

Shots 119433-119437 used the same shot parameters as 119891, with the goal of using the radial field from control coils to study how the opposite current sign effect observed in that shot changes with the vertical position of the plasma. Further work on shot development is needed to study this further.

Other shots worth looking at- 119419 has a radial kick due to an applied vertical field. Compare to 119420 with no applied field.

Important note- On many shots today the north rack cpci and north rack cpci_40 failed to store. Wiggling the connections between the Eagle Harbor and the cpci seemed to temporarily fix this for several shots before the problem returned. Troubleshooting is needed.

Monday October 28 2024 7:11 pm Chandra

Use "

" to force a line break. Carriage returns are ignored.

Monday October 28 2024 10:14 pm Chandra 119445-119511 1/1 & 2/1 RMP Sawtooth Period Scan

The goal of today was to fill in and expand the RMP scan to include different control coil currents and alternative magnetic topologies, to observe the effect on Ʈ_st.

The run was very successful in terms of data collected and shot stability.

Star Shots (1194xx series): RMP 1/1, 1-5ms, 20A, +8kHz: 61, 62, 68, 69, 75, 76, 77, 78

RMP 1/1, 1-5ms, 20A, -8kHz: 64, 65, 72, 74, 80

Baseline: 56, 66, 67, 81, 82, 83

RMP 2/1, 1-5ms, 30A, +8kHz: 87, 88, 95, 97

RMP 2/1, 1-5ms, 30A, -8kHz: 90, 91, 94, 99

Baseline (prev. baselines also applicable): 92, 93, 102

RMP 3/1, 1-5ms, 30A, +8kHz: 104, 105, 106

RMP 3/1, 1-5ms, 30A, -8kHz: 107, 108, 109

Baseline: 102, 110, 111

Mode rotation direction/RMP polarity is visually confirmed on the magnetic contours and mode frequencies of shots 119506, 507.

This is this machine operators last run day on HBT-EP, and it was a good run to finish on.

Monday November 4 2024 12:26 pm DaSilva 119519 - 119571 REMC Shot Development

The purpose of this run was to establish a consistent shot style for HXR bursts, and to better calibrate the HXR detectors' positions and gains. There are two shot styles for HXR production with trade-offs: the one used pre-APS for data (see ~119234), which produced a lot of HXR signals but was inconsistent in its disruption time, and the one developed in this campaign (shot 119533). This new shot style sees less HXR activity, but is extremely consistent in its disruption time ~5ms. The original purpose of this campaign was to find a shot style where there was a consistent difference in disruption time depending on whether or not the REMC was activated (for example, 5ms when the REMC is off and 4ms when it is on). For this style, the REMC seems to have very little effect on the plasma disruption time, but does have a seemingly consistent effect on the production of HXRs. Therefore the second leg of this campaign (to run the CCs to cancel out the error fields causing the early disruption) may not need to be done now since we are able to mitigate the EFs' effects without needing the CCs. Some star shots for comparison are 19533 (REMC off) and 119536 (REMC on). The older shot style with more HXRs was attempted, and a pretty consistent disruption time was achieved (shots 119545 - 547), though the number of HXRs was not saturating the detector to the extent we saw in the pre-APS shots. The REMC activation also changed the Ip evolution to where the CCs can be explored, and occasionally killed off the HXRs completely (shot 119549). Finally, starting shot 119565, the gain of the HXR detectors has been changed to the 10e6 setting (from the 10e7 setting) despite not getting much detector saturation. Still, in shot 1159566 in particular we see the signal exceeding 1V, so the gain settings should be 10e6 when doing REMC disruption experiments. * Note the REMC turned on at 2ms for every "REMC on" shot *

Friday November 15 2024 11:10 am Notis 119607-119638 Radial Field and Halo Currents

Run from 11/13/2024.

The run focused on looking at the effect of a radial field perturbation (= vertical kick) on the non rotating halo currents (NRHC) seen on the HFS tiles during the disruption. Specifically, we were looking to see if a vertical kick spatially moved the inversion point from negative to positive HC on the tile array. Look at figure 5.3 in Alex Saperstein’s thesis for an example of this experiment for the “flattop” NRHC on the LFS tiles.

Two different shot styles were used- (119607-119627) = Shotstyle I was the runaway style low Ip shot, while (119628-119444) = Shotstyle II had a strong Ip ramp. Both positive and negative radial fields were used.

Shot summary - Shotstyle I (+ kick: 119620-21. – kick: 119626-27. No kick: 119622-23.) Shotstyle II (+ kick: 119635-36. - kick: 119632. No kick: 119628,119634). So far it is not clear if the application of a radial field moved the inversion point during the disruption. Further analysis is ongoing.

To generate a plot of vertical moment for [shotnum] from [start_time] to [end_time], navigate to my directory and type in python vert_moment_only.py [shotnum] [start_time] [end_time] into the command line.

Monday November 18 2024 1:27 pm DaSilva 119676 - 119731 REMC Major Radius Scan

The purpose of this run day was to scan the plasma major radius (to the best of my ability) with both the REMC on and off to see the REMC and MR affect of HXR production. We also commissioned 3 new HXR detectors on the machine

Hardware: 3 new HXR detectors were added to the machine, in sections 2, 7, and 8 (digitized at NR:CPCI_40 in channels 6, 7, and 8 on the BNC breakout board, inputs 70-72 in the software). Section 2 detector was placed on the limiter much closer to the vessel wall than the other detectors which are normally placed on the CC amplifier boxes (and saw much higher HXR signals as a result). The new section detector was placed directly on top of HXR4 to benchmark its (sometimes iffy) signal. 7 All HXR detectors were set at the 1e6 gain setting with BW = H.

Software: Crate 109 failed during the 'crates_on' command, which caused 2 NR preparation failures in J221 (which are persisting from previous run days). The NR also failed to store data for 27% of the shots taken today. One shot of note (119725) had one NR CPCI fail but not the other and may be helpful for troubleshooting.

The scan consisted of low plasma current, cold, beam-like discharges where the VF was adjusted to change the major radius trajectory. Starting with shot style from 119530, I modified the MR trying to keep it relatively flat for at least the period of 2-3 ms (according to hbtshot.py). The MR calculation still needs to be updated to take the REMC current and limiters into account. When the REMC was triggered it was always at 2ms.

Some individual shots of note are 119679 and 119720, which both produced a large amount of HXRs, saturating the detectors even at the lower gain setting.

Thomas helped decide many bank settings in shots 119727 and on and is a certified tokamak operator!

Tuesday November 19 2024 10:11 am Chiriboga 119572-119606, 119639-119675 fast camera ML training data collection

The purpose of this campaign was to develop a consistent shot style and record fast camera data for retraining the optical based feedback CNN. The shot style was based on the recent campaign run by Rian Chandra, shots 119445-119511. The main goal of this shot style was to detect strong 3/1 and 2/1 mode activity (mode amplitude of approximately 3 or greater) for the CNN to train on. These mode strengths are meant to be comparable to those used in William’s training set (114449). The shot style typically contained a strong 3/1 mode near the beginning of the shot almost always peaking at 1.5ms and decaying by 2.5ms and a 2/1 mode growing later around 4ms when there was no disruption at that time. Some star shots that illustrate the shot style are 119648 and 119653.

During start up there were unexpected errors with turning crates 103 and 109 on when running “crates_on” on spitzer, and still with crate 109 after running a second time. We power cycled the digitizer racks and restarted Spitzer, but the problem persisted. There are errors during shot preparation with the north rack j221 modules and north rack cpci errors on some shots with inconsistent frequency.