| Miniproposals | ||||||||||
| ||||||||||
|
| Operators | |
| Session leader(s): | Steve Wolfe,Jin-Seok Ko |
| Physics operator(s): | Steve Wolfe |
| Engineering operator(s): | Bill Cochran,Ed Fitzgerald |
| Engineering Operator Run Comment |
| MP#461 Zero Loop Voltage Development, MP#394 MSE Calibration, PO;Steve W |
| Session Leader Plans |
| Entered: Apr 24 2007 04:18:25:317PM |
| Author: Steve Wolfe |
Session Leader Plan for Tuesday, 10-Apr-2007 MP#461 Zero Loop Voltage development SL: Wolfe PO: Wolfe The goal of this experiment is to develop the control procedure for clamping the loop voltage to zero for non-inductive current experiments, without affecting the plasma shape. In the absence of non-inductive current the result should be a rampdown at L/R time. The previous run for this MP had one baseline shot and one shot with Ip control off, 1070410005. The run was then terminated due to camac problems. Based on the results on shot 1070410005, this target was too close to vertical instability for convenient use on this experiment. The issue is that li tends to rise significantly during the ramp-down, decreasing ncrit and resulting in growing vertical instability. This tendency is confirmed by Alcasim simulations. The other effect observed on 1070410005 is a thrash in RCUR and in CLEARIN at the time the Ip gain is switched off and the feed-forward voltages on the OHs and EF3 were set to zero (1sec). This thrash is also observed in the Alcasim simulations, although the sign of the step in CLEARIN is not always consistent. This thrash is unexpected, and may indicate that our orthogonal controllers are not as orthogonal as we think. Finally, the evidence from 1070410005 is that the effective derivative gain of PHYS_PID D= 0.04 used on this shot is effectively zero. Alcasim simulations indicate that a PHYS_PID derivative value of ~15 (factor of ~400 larger) is required, possibly in conjunction with P~100/sec, in order to get to a good zero volt condition. Shot 5 got down to ~0.6V, basically because the current control was turned off and the feedforward on the OH coils was zeroed. Simulations also indicate that it is not necessary to turn off the feedforward voltages if the gain is adequate. The simulations indicate that the rampdown times at zero volts are surprisingly short, <300msec to get almost to zero current; I expected L/R to be more like 0.5sec, but we'll see what the plasma thinks. The goals of the experiment on Wednesday are to reduce the thrash at the transition and to increase the D (and possibly P) gains on the PSI0 wire to reach a zero volt rampdown. It may be useful or necessary to increase gains on RCUR, CLEARIN and perhaps the xpoint locations to keep a constant shape. Based on the above considerations, the plan for Wednesday morning (figuring on a total of 10 good shots ) is as follows: 1) Start from a less demanding equilibrium. One possibility is 1070404023. Another is to increase clearin on 1070410001. It also wouldn't hurt to reduce the density from 1e20 to 0.6e20 or so. Load PSI0 with zero gains on Wire 12, as on simulation shot 1070424301 (1/tau_C=2000, factor=0.01, no tauopt). Load observer pro dpcs_lowpass_obs with tavg=0.03 for wire 12, also as on 1070424301. Observe time of PSI0=0, expected to be around 0.9sec. [1-2 Shots] 2) At a time shortly before the PSI0=0 point, switch off Ip gain and switch on PSI0 D gain. Try staircasing D up and Ip down over 100msec to reduce thrash effect, as on 1070424306 . Initial top step D value of 5 is predicted to give a loop voltage of 0.25V . Check on thrash, shape evolution, rampdown time, stability, as well as voltage. [1-2 shots] 3) Increase D gain to push toward zero voltage, if indicated. Increase gain on shape controls if indicated. Reduce V_OHx and V_EF3 to zero gradually after the gain switch. [3-5 shots] 4) Possibly along with step 3, add in some P gain (effective integral gain on the voltage). Need to be careful with the timing relative to the time psi0 gets to zero. Nominal starting gain value should be PHYS_PID P~100/sec (Hyb gain =.05). Probably want to staircase this up as well. If I need to use this, I should probably re-think using an actual voltage observer with P and I gains, which would eliminate the integration constant. [3 shots] 5) The other variable (well, one of them) is the filter time constant. If there is time, vary down to .01sec and up to .05 sec, at constant D gain. The simulations indicate that shorter tavg is unstable (growing 60Hz), while longer values may have less wobble. [time remaining before 13:00] |
| Entered: Apr 25 2007 11:21:43:733AM |
| Author: Jin-Seok Ko |
| Session Leader Plan for Wednesday afternoon, 25-Apr-2007 Title: MP#394 MSE Calibration The purpose: 1. Standard MSE beam-into-gas calibration : For several known pitch angles, the MSE responses are measured. 2. Testing Ian Hutchinson's recent conjecture on Yuh effect: Do fast ions leaving the beam area but still remaining the MSE views are leaving the sightline quicker when the torus pressure becomes lower? 3. MSE filter scan: observe whether the polarization fraction changes as we expect The run plan: Resources: DNB, He4-filled gas torus Set up like 1070410007 with modified BT and EF (1) Purposes 1 and 2 are done simultaneously With the fixed BT = 5.4T (for 0 ~ 1.6 sec) ========================================================== ID EF4 (kA) EF3 (kA) p (mTorr) ========================================================== A1 4 EF3-S 1 A2 0 EF3-S 1 A3 EF4-S 0 1 Repeat A1 ~ A3 for p = 0.3, 0.1 mTorr (ID's: B1~B3, C1~C3) D 4 10 0.05 E 4 10 0.03 ========================================================== 11 shots + 2 shots (for A1 for contingency) (2) Purpose 3 With the fixed EF4 = 3kA and the fixed p = 1 mTorr =================================================================== Id TF (T) EF3 (kA) Filter set (BT) =================================================================== F1 TF-L EF3-L 5.4 F2 TF-H EF3-H 5.4 G1 TF-H EF3-H off (cooling) G2 TF-H EF3-H off (more cooling) G3 TF-L EF3-L 7 (or maximum tuning range) =================================================================== 5 shots (3) Waveforms for EF3-S, EF4-S, TF-L, EF3-L, TF-H, and EF3-H EF3-S flat1 : 0 kA (0.00 ~ 0.25 sec) flat2 : 6 kA (0.65 ~ 0.90 sec) flat3 : 12 kA (1.30 ~ 1.55 sec) EF4-S flat1 : -2 kA (0.00 ~ 0.50 sec) flat2 : -5 kA (0.10 ~ 1.50 sec) TF-L flat1 : 2.80 T (0.00 ~ 0.25 sec) flat2 : 3.45 T (0.65 ~ 0.90 sec) flat3 : 4.20 T (1.30 ~ 1.55 sec) EF3-L flat1 : 1.81 kA (0.00 ~ 0.25 sec) flat2 : 3.37 kA (0.65 ~ 0.90 sec) flat3 : 5.18 kA (1.30 ~ 1.55 sec) TF-H flat1 : 4.90 T (0.00 ~ 0.25 sec) flat2 : 5.55 T (0.65 ~ 0.90 sec) flat3 : 6.20 T (1.30 ~ 1.55 sec) EF3-L flat1 : 6.86 kA (0.00 ~ 0.25 sec) flat2 : 8.43 kA (0.65 ~ 0.90 sec) flat3 : 9.99 kA (1.30 ~ 1.55 sec) |
| Physics Operators Plans |
| Entered: Apr 25 2007 06:50:20:910AM |
| Author: Steve Wolfe |
Physics Operator plan for Wednesday 25-Apr-2007 MP#461 Zero loop voltage development (09:00 to 13:00) and MP#394 MSE Calibration III (13:00 to 17:00) Engineering Setup: ================= Overnight ECDC in D2, P=2e-4, sweep: Rmin=44, Rdwell=45, Rmax=103 Run begins at 09:00 and ends at 17:00. There will be a change of experiment at 13:00. Setup for Morning Experiments (MP#361): --------------------------------------- Power systems as on 1070424007 Acoil: +Dtop -Dbot -Jtop +Jbot Enabled, Hybrid Control. ***** DPCS Type 2 setting in Auto mode **** *** Note: This will be standard from now on **** Gas Setup: --------- Fill B-Top with 6 psi D2 Hybrid enabled (PG4) fill B-side lower with 1 psi Ar Hybrid DISABLED (PG1) leave B-side upper as is Hybrid DISABLED (PG2) fill B-main (C-side) with 40 psi D2 Hybrid enabled (PG3) leave NINJA as is, disabled Enable the following gate valves and shutters, assuming no vacuum problems: ECE, VUV, Z-bolo shutters Torvac setup: Pumpstack gatevalves OPEN ICRF: Not required, allowed if ICRF Group wants (morning only). Check with ICRF for turn on instructions. LH: Not required, allowed if LH Group wants (morning only) **************************************************************** Setup for Afternoon Experiments (MP#394): Power systems as on 1070410007, Except TF Current limit to 180kA. Acoils: Disabled Turn off all supplies except TF, EF3, EF4 Gas setup: ---------- Pump out B-top and replace with 6psi HELIUM; Hybrid Enabled All other valves DISABLED. Torus pump valves CLOSED during pulse. ECE, VUV, Zbolo Shutters and Diagnostic valves closed. DNB permissive enabled. *************************************************************** PO Plan for Morning experiment (MP#461) ***************************************************************************** Before run, load the LIVE parameters for the DPCS_TYPE2_FAULT procedure from the saveset /home/wolfe/cvswork/dpcs/user_algs/output_pros/dpcs_type2_fault_settings1.sav using the structure DPCS_TYPE2_FAULT_DEFAULT_SETTINGS These settings should remain in place from now until Bill Burke's Analog protection circuit is installed. ****************************************************************************** For MP#461 (Zero loop voltage): Load seg2 from 1070404023, reduce ZXU from .06 to .04, Drop CLEARIN from .0045 to .0035 and modify wire 12 and loadable OBSERVER procedure as per step 1 in SL_PLAN Follow the steps under SL_PLAN for MP#361 ------------------------------------------------------------------------------ PO_PLAN for Afternoon Experiments (MP#394 Beam into gas, no plasma) Load from 1070410007; make sure seg1 is on and 2,3,4 are off. Set PG4 for 1mTorr of He (0.5sec PG4 puff, but start it earlier than -.5) Draw crazy TF and/or EF3,EF4 waveforms specified by SL. See SL_PLAN for shot-to-shot changes. |
| Session Leader Summaries |
| Entered: Apr 25 2007 05:12:44:257PM |
| Author: Jin-Seok Ko |
| Session Leader Summary for Wednesday afternoon, 25-Apr-2007 Title: MP#394 MSE Calibration The purpose: 1. Standard MSE beam-into-gas calibration : For several known pitch angles, the MSE responses are measured. 2. Testing Ian Hutchinson's recent conjecture on Yuh effect: Do fast ions leaving the beam area but still remaining the MSE views are leaving the sightline quicker when the torus pressure becomes lower? 3. MSE filter scan: observe whether the polarization fraction changes as we expect The run summary: We had 12 useful shots. Unfortunately, we didn't have all we planned. There were some workstation delay and shot cycle problems which lengthed the interval between shots. The overall results are For the purpose 1 and 2, we accomplished 100% we want. For the purpose 3, we only had 1 shot Resources: DNB, He4-filled torus Set up like 10704100007 with modified BT and EF ================================================================================================ Shot ID EF4 (kA) EF3 (kA) mks* rat** DNB HV BT/Eb Comments ================================================================================================ 15 A1 4 EF3-S 1.3 0.50-0.53-0.40 good 1800 5.4/50 16 A2 0 EF3-S 1.3 0.49-0.58-0.50 2nd*** 1820 5.4/50 17 A3 EF4-S 0 1.3 0.46-0.49-0.41 good 1870 5.4/50 18 B1 4 EF3-S 0.5 0.13-0.18-0.17 good 1870 5.4/50 19 B2 0 EF3-S 0.5 0.13-0.21-0.20 good 1870 5.4/50 20 B3 EF4-S 0 0.5 0.13-0.17-0.17 good 1870 5.4/50 21 C1 4 EF3-S 0.2 0.03-0.08-0.09 good 1890 5.4/50 22 C2 0 EF3-S 0.2 0.03-0.10-0.10 good 1890 5.4/50 23 C3 EF4-S 0 0.3 0.03-0.09-0.09 1st 1890 5.4/50 24 D 4 10 0.15 0.02-0.07-0.07 good 1894 5.4/50 25 E 4 10 0.1 0.01-0.06-0.06 1st 1894 5.4/40 not a big fault 26 G2 TF-H EF3-H 1.3 0.46-0.49-0.43 1st 1870 5.4/40 filter off 20 min before this shot ================================================================================================ *rough average around 1sec **at 5s - max - at 1.5s ***pulse that has faults |
| Entered: Apr 26 2007 08:30:41:017AM |
| Author: Steve Wolfe |
Session Leader summary for Wednesday 25-Apr-2007 (Morning) MP#461 "Zero loop voltage development" SL: Wolfe PO: " Considerable progress was made on the development of zero voltage control for use in non-inductive current experiments, but the technique is not yet ready for application. The basic concept of sensing the surface voltage via the derivative of an observer for the flux at the nominal x-point location was validated. Use of a controller (vector of EF and OH coil voltages) based on providing a nearly flat flux pattern over the plasma volume (PSI0 controller) was moderately successful, although the orthogonality of this controller to the equilibrium shape control was less robust than expected. It was found that increasing the Integral gain on CLEARIN was useful in reducing the excursion and speeding the recovery of this parameter following the transition from Ip (current) control to Voltage (PSI0) control. The main problems identified had to do with the stability of the control system with large values of derivative (D) gain applied to the PSI0 observer (which corresponds to proportional (P) gain on the Voltage) and the difficulty of applying proportional gain for PSI0 (which corresponds to integral (I) gain for the Voltage) without having prior knowledge of the actual target value of the absolute flux. Use of effective Integral gain with respect to the voltage was demonstrated to be very helpful in stabilizing the control loop and achieving the zero voltage condtion, so the algorithm needs to be modified to accommodate this feature in a robust manner. The application of a low-pass filter to the observer prior to application of the derivative gain was successful. However, the full interaction of varying the D gain and the filter time constant was not explored, particularly with respect to stability and overshoot issues. Particularly with zero or low proportional gain (on PSI0), values of derivative gain sufficient to reduce the loop voltage below ~0.4V resulted in the appearance of a growing (unstable) oscillation with a period of about 30msec, usually terminating in a disruption. In at least one case (#12) a relatively benign marginally damped oscillation was driven unstable during the ICRF pulse beginning at 1.1sec; whether this was due to a change in the plasma equilibrium by ICRF heating or by noise pickup on the magnetics was not immediately clear. The best conditions were achieved on shot#14, using moderate value of D (stepped up to 8) and fairly strong P (100/sec), lowpass filter time constant of 0.05sec, and with the feedforward voltages on the OH and EF supplies reduced smoothly to zero well after the transition to PSI0 control. On this shot the surface voltage was reduced to zero and maintained there from about .25 sec after the transition. The low frequency oscillation appeared to be strongly damped once the P gain was switched on at 1.0 sec. Earlier application of this effective Integral gain on the voltage might have improved the control even more. Further details, including the gain values and timing applied, can be found in the PO_SUMMARY and in the PHYSICS_OPERATOR logbook entries. have improved the feedforward voltages on the OH and |
| Physics Operator Summaries |
| Entered: Apr 25 2007 06:23:09:220PM |
| Author: Steve Wolfe |
Physop Summary for Wednesday, 25-Apr-2007
PO: wolfe
EO: Cochran,Pfeiffer,Fitzgerald
MP#461 Zero loop voltage alg. development (0900 to 1300)
SL: wolfe
MP#394 MSE Calibration III (1300 to 1700)
SL: Jinseok Ko
The morning run was my Zero loop voltage algorithm development,
wearing both SL and PO hats. Progress was made on this development,
but the technique is NOT ready for prime time, and the algorithm
needs to be re-thought (or at least re-typed) to some extent.
I decided to use the 1MA LSN discharges from Wukie's run yesterday
as a target, with the density dropped from 9 to 6e19, and a slightly
tweaked startup. The startup worked pretty well (13/14 despite a number
of disruptions). Runaways appeared on a couple of early shots, but
were not a persistent problem (RCUR offset and gas tweak at shot#3).
The switchover from Ip control to "deriv(PSI0)" control was started
at 0.8sec, near the time the PSI0 predictor value crossed zero. I started
with a relatively fast staircase on both Ip (P) and PSI0 (D) gains, which
seemed to result in instability and disruption. By shot 4 I found that
turning Ip P gain off in one step worked better. Shot #4 ramped all the
way down with a surface voltage around -0.4V, with some thrashes around
the two-step PSI0 D gain switches. Total current decay took ~.75sec.
The maximum value was D=2, a factor of 7.5 less than seemed to work in my simulations (for a different equilibrium and starting current).
Raising the D gain resulted in growing oscillations with about a 30ms period, and disruptions. These seemed possibly related to conflict with CLEARin
which developed significant (~5000a-m) error and had a slow recovery (in
cases where the plasma lasted long enough to recover). I tried increasing
the INT gain on CLEARIn before the switch-over, and it may have helped
a little. The EF2's seemed to be major players in the instability, since
their actual voltages were lagging their demand by >90 degrees, getting
close to 180. Turning D above 4 or moving the step up earlier seemed
to exacerbate the instability.
Turning on proportional gain on PSI0 (which amounts to integral gain
on Vloop) did have a beneficial effect. Doing it was only slightly
complicated by the fact that I didn't want to turn it on too early,
so I needed to redraw the target from 0 to +10000 to try to match
the value at ~0.90sec, after the D gain seemed to be taking effect
and the thrashes were cleared. PCS software still squawks about
out of range Wavegen values, but loads them anyway. I know what
routines to turn off to get rid of that, I think. However, the real
problem is that if the actual psi0 is significant when the P gain
comes on then the resulting thrash is fatal. I need to eliminate
the constant and set up to use a true integral gain on the voltage.
An example of the badness is shot 9, with target=0, compared to
shot 10, with target=10000.
Increasing the lowpass filter timeconstant from .03 to .05 seemed
to damp the oscillations a little (shot#11), however raising the gains
still ended up with growing oscillations, especially during the RF phase.
The final shot of this part of the run, #14, was the best. This had
D rising to 8 (at 0.9sec), and P=100/sec at 1.0sec. It also had the
feedforward voltages on the OH's and EF3 ramped to zero between 1.0 and 1.3sec.
This shot ramped all the way down and had essentially zero Vsurf (+/- 75mV
or so) after 1.06sec. There were remnants of the 33Hz oscillations between
0.8 and 0.9sec, but these were completely damped after the switch to P=100/s
at 1.0sec. This would be a fine place to start a continuation, except that
the right thing to do is to come up with a clean way to get rid of the
target value.
The afternoon run for the MSE calibration went smoothly until we started
getting data system problems, which ended up costing us over an hour in
lost shots (really lost! we had no power cycles for #21 and #26 that
left no record at all, and then had subsequent shots with the same numbers).
We ended up getting the first 11 of the cases in the run plan (A1-E),
although the pressure scan may not have gotten to the lowest value requested
in D and E. There was some problem with the MKS gauge and the ratiomatic
is clearly being affected by fields. It may be possible to back out a
value using the end of the ratiomatic signals, after 5sec, since the
pumpstack gate valves were closed in puslse and the only pumping was
throught the beam duct. The final shot of the sequence,#26 (second try),
used the stepped TF and EF3 waveforms for case F2, but the filters were
started cooling early and we delayed taking the shot, so the effect
is that this was more like G2 (without an F2 baseline); hopefully Steve
and Jinseok can make sense of the result.
Machine systems performed pretty well today. No issues with power
systems and cooling was generally ok. There were no vacuum or gas
system problems.
The new DPCS coil protection scheme (dpcs_type2_fault) was enabled
for the first time and did not cause any problems. No bad inputs of
the sort the routine is designed to catch were observed.
There was one magnetics warning for a .027V baseline error on F10
on shot#5 (disruption). This did not recur.
There were (apparently unrelated) data system problems during both
parts of the run. After shot#6 Josh reverted the camac software to
an earlier incarnation, and that seemed to fix a storage problem with
several INCAA modules; around the same time Yijun was experiencing a
problem with the FFT realtime system that was apparently fixed by power
cycling pcdaqdpcs2. In the afternoon session, the cmod cycle/state
application blew itself away at least three times, costing shots and
significant lost time. Josh thinks the problem is caused by the Xserver
on cmodws44, and has recommended that the cycle be run from the other
Engineering workstation tomorrow. It was also noted that for most of the
morning data did not begin to appear until 20-30seconds into recool. For
shot#13, and the rest of the day, the data began appearing at its usual
time, with bus rogs and magnetics showing up almost immediately at the
start of RECOOL; no idea what caused it, but maybe related to LH data,
since they turned off around that time.
Steve Wukitch pointed out that the magnetics were seeing pickup from the
ICRF, including noise on BTOR and Ip. This may have contributed to the
extra thrashing of the control system during the RF pulse (after 1.1sec).
This needs to be examined more closely.
Scorecard:
----------
MSE cal shots 12
Duds 0
Fizzles 1
Plasmas 13
============= ==
Shots 26
Non shots 2 (partial data cycles, no power)
|
| Session Leader Comments | |||
| Apr 25 2007 01:46:25:300PM | 1070425015 | Jin-Seok Ko | Good beam, a little faulty on the begining of the last pulse MFLUX angle (due to EF3 + EF4) shows a little bit of ramp (~150ms) at the beginnings of the last two flattops. We tried to lengthen the DNB off time to compensate this, but Dexter doesn't want to do so. We will leave it as it is. Signal strength is fine. Next shot will have 1820V. |
| Apr 25 2007 02:24:45:847PM | 1070425016 | Jin-Seok Ko | Shot 15 was A1. This shot A2. 2nd beam pulse had some faults but fine with MSE signal. APD was 1820V. |
| Apr 25 2007 02:25:08:957PM | 1070425017 | Jin-Seok Ko | A3 good beam, good signal |
| Apr 25 2007 02:25:31:503PM | 1070425018 | Jin-Seok Ko | B1 APD HV = 1870V good beam, good signal |
| Apr 25 2007 02:35:49:337PM | 1070425019 | Jin-Seok Ko | B2 good magnetics, good beam, good signal |
| Apr 25 2007 03:01:16:550PM | 1070425020 | Jin-Seok Ko | B3 good magnetics, good beam, good signal Moving to C1 where p = 0.1 mTorr APD HV to 1890V for next shots. |
| Apr 25 2007 03:40:28:457PM | 1070425021 | Jin-Seok Ko | C1 good magnetics, good beam, good signal Moving to C2 APD HV is still 1890. |
| Apr 25 2007 03:55:42:273PM | 1070425022 | Jin-Seok Ko | C2 good magnetics, good beam, good signal |
| Apr 25 2007 04:08:18:763PM | 1070425023 | Jin-Seok Ko | C3 good magnetics, good beam, good signal small faults in the fist beam pulse. For next shots, HV will be 1894. |
| Apr 25 2007 04:22:45:627PM | 1070425024 | Jin-Seok Ko | D APD HV : 1894V good magnetics, good beam, good signal |
| Apr 25 2007 04:46:47:590PM | 1070425025 | Jin-Seok Ko | E APD HV : 1894V good magnetics, good beam, good signal |
| Apr 25 2007 06:25:51:410PM | 1070425026 | Jin-Seok Ko | ~ G1 Filter switches were off at 4:40PM. APD HV is set to 1870V for this shot. Shot was taken at 5:01PM and at this momnet we had the following: ------------------------------------------------ Ch Demanded Temp/CWL Current Temp/CWL (5.4/50) ------------------------------------------------ 0 40.2/659.9 28.5/659.7 1 40.1/660.2 29.5/660.0 2 40.3/660.5 29.5/660.4 3 48.0/660.8 32.5/660.5 4 44.9/661.3 29.4/661.0 5 43.2/661.6 30.6/661.4 6 49.1/661.8 32.6/661.6 7 45.8/662.2 31.1/661.9 8 27.7/662.4 24.5/662.4 9 43.9/661.0 30.4/660.8 ------------------------------------------------ good magnetics, good beam, good signal |
| Apr 25 2007 06:29:56:410PM | 1070425026 | Jin-Seok Ko | This is the last shot of the day. Configurations: Shutter OPEN Linear polarizre IN PEM OFF Filter heater OFF Bertain OFF Chiller OFF |
| Physics Operator Comments | |||
| Apr 25 2007 07:15:26:890AM | Steve Wolfe |
Setting up for MP#461 (Zero loop voltage development)
Changed my mind about the target shot, I'll just use the
1MA equilibrium from yesterday's ICRF run. The gaps both
grow in rampdown, but I won't be getting past 1.5sec anyway
and they all ramped down.
Turn off dpcs_ftae2 (TEST PROS)
Seg#1: Br0 from -.001 to 0.007
Pg4 from 13ms to 20ms
RCUR offset from 1800 to 1500
Leave IC_EF4U at -1335 for now
Seg#2: Load Wire 12 from 1070424301 (test shot) and zero D gains
OBSERVERS: Add dpcs_lowpass_obs: Parameters={tavg:0.03},Segments=2,Wires=[12]
And something in the interface is broken. It keeps coming up with quotes on
the parameters, and staying as MDS expression. Took several tries,but it
finally seems to take it. Had to click IDL Value, then get rid of the
quotes and square brackets, then hit return in the box. OK, that's there.
Load it
Load at 25-Apr-2007 07:14:53.00
Open tree /home/wolfe/pcs_scratch -1
Open tree done
| |
| Apr 25 2007 07:19:00:640AM | Steve Wolfe | The dpcs_type2_fault trigger procedure was set with its default parameters last night after the conclusion of the run. See long logbook enty under DPCS for details. This routine should now be on and enabled until further notice. Note that some combinations of callable procedures in DPCS may now exceed the available DPCS cycle time. | |
| Apr 25 2007 07:44:14:947AM | Steve Wolfe | Start setting up a segment to use for Steve and Jinseok's run this afternoon. According to the SL Plan I think they want to start with a stepped EF3 waveform with fixed TF and EF4 at +4kA. Import into seg4 from seg1 0f 1070410007: Move gas puff from [-.5,0] to [-.6,-.1] Draw EF4 at +4kA, EF3 waveform with two steps at 6kA,12kA Make sure seg4 is OFF; I'll import it into seg1 when we get to it. Go ahead and load this one up. Load at 25-Apr-2007 07:43:30.00 Open tree /home/wolfe/pcs_scratch -1 Open tree done | |
| Apr 25 2007 08:11:59:033AM | Steve Wolfe | For shot#1, lower the nl04 demand in seg2 from 0.9e20 to 0.6e20. Load it Load at 25-Apr-2007 08:11:27.00 Open tree /home/wolfe/pcs_scratch -1 Open tree done I think that's enough changes for now. | |
| Apr 25 2007 09:12:50:510AM | 1070425001 | Steve Wolfe | shot#1: PSI0 observer on wire 12, with gains=0 Program 1MA, 6e19, LSN like 1070424030 plasma full length; slight bounce on startup. Magnetics data came in late. Looks like we're still not getting the piezo drivers. RF fired at 1.1sec Psi0 crosses zero at 0.827 sec on this one. Vsurf (EFIT) =-1.1V, d/dt(PSI0)=-1.1V Next: put in gain switch at 0.8sec. |
| Apr 25 2007 09:34:17:740AM | 1070425002 | Steve Wolfe | shot#2: Staircase gains from 0.8 to 0.9sec, Dmax(12)=5; Ip staircase down from 9 to 0 over same time. No seg 1 changes; DNB firing this shot. Runaways early, disrupted at 0.95sec. Not good. It had ramped down to 0.75MA only. Lots of early bounces. Not much of an initial rcur thrash this time, but clearin goes positive, There are big slow (40msec period) oscillations on all the shape errors before the disruption. Could be the staircase but it looks like it's unstable. The simulations did not do this. |
| Apr 25 2007 09:52:03:620AM | 1070425003 | Steve Wolfe | shot#3: Seg1: PG4 to 24msec, RCUR offset from 1500 to 1800 Reduce the PSI0 max gain from 5 to 2, do it in two steps. Still runaways, still disrupting. This one had a big marfe early, almost didn't go. The disruption is prompt at .86, hardly any rampdown. No bounce visible on the startup. Where are the runaways coming from? This shot was at psi0=0 by .35sec, the loop voltage is high throughout. Not really a fair test, but it still looks unstable Not that it matters, but the xpoint on this is closer to normal spot, change Rc,Zc in psi0 predictor to .55,-.395 Try dropping the Ip gain to zero in one step at 0.8sec; I'd expect this to make it worse. |
| Apr 25 2007 10:00:48:547AM | 1070425004 | Steve Wolfe | shot#4:change Rc,Zc in psi0 predictor to .55,-.395 Try dropping the Ip gain to zero in one step at 0.8sec; I'd expect this to make it worse. Plasma, long decay, this looks like it should! termination at 1.55sec Loop voltage is now at -.4V. Ther are thrashes, but it recovers. both gaps jump up (bigger clearin) take a long time to come back. next: raise D gain, try for lower voltage |
| Apr 25 2007 10:19:01:897AM | 1070425005 | Steve Wolfe | shot#5: Increase PSI0 D from 2 to 5 at 0.9sec This gives 100msec at D=2. Note that on shot#4 the error signal ramps to -.5 by 0.86, and ekes down to -.4V by 0.9sec Disrupts at 0.978sec, oscillating demands after 0.9sec, 30msec period. Now seeing a baseline offset on f10, -27mV. Doesn't look terrible. Next: back D down from 5 to 4 (phys), incresae clearin int gain from 125 to 200/s^2 at 0.75sec, before the switch |
| Apr 25 2007 10:26:20:337AM | 1070425006 | Steve Wolfe | shot#6: back D down from 5 to 4 (phys), incresae clearin int gain from 125 to 200/s^2 at 0.75sec, before the switch fizzle. One disruption too many? Actually, it looks like its rising too fast still, dies outside. next: increase pg4 by 1msec |
| Apr 25 2007 11:29:13:277AM | 1070425007 | Steve Wolfe | shot#7:PG4 from 24 to 25msec back D down from 5 to 4 (phys), incresae clearin int gain from 125 to 200/s^2 at 0.75sec, before the switch. The EF2's have a big phase lag in this oscillation, could be the driver of the instability. Note that the Ip controller (tauopt=0.3) has a much bigger relative gain in ef2's than the Psi0 controller I'm using. Before this shot: Cell access for TCI, Josh reverts camac software to see if it will cure problems with INCAA stores (Blip and Piezo drivers) About an hour delay. plasma and it ramps down. Gets down to around .3V Clearin comes back faster, Rcur starts to lose it at 1.1sec? That's the RF. Otherwise, this one stays under control. I can either push the D gain or start adding some P gain. |
| Apr 25 2007 11:43:18:600AM | 1070425008 | Steve Wolfe | shot#8: Move D=4 to 0.86, D=8 to 0.9sec; hold off on P for now. That one disrupts at 0.926, Looks like the oscillation starts again near 0.86 sec, then worse at 0.9 for about a half-cycle before it goes boom. It looks like the clearin and the psi0 are fighting it out. Next, back to max D of 4, add some P gain |
| Apr 25 2007 11:54:23:057AM | 1070425009 | Steve Wolfe | shot#9: back to max D of 4 at 0.85, add some P gain=50/sec at 0.95sec (if it makes it that far). Disrupted, but later about 1sec Gets to the P gain time, and takes a big thrash. That seems to be too strong a P gain, especially since by then the error in PSI0 is 10000. Next: put target at 10000 (const value) and drop P |
| Apr 25 2007 12:09:15:407PM | 1070425010 | Steve Wolfe | shot#10: put PSI0 target at 10000 (const value) and drop P to 25/sec at 0.9sec. D jump from 2 to 4 still at 0.86sec. It won't let me draw 10000 with this factor, but I'll do it anyway. Apply it, ok the error screen, cancel and load. Have to fix PCS later. First see if this works (and if I got the sign right). Plasma to 1.4sec, down to 130kA. deriv(psi0) error to ~0 by 1.1sec, psi0 error around -5000. This doesn't look too bad. I've three shots to tune it better. |
| Apr 25 2007 12:18:23:253PM | 1070425011 | Steve Wolfe | shot#11: Raise filter time constant from .03 to .05sec; if that calms down the residual oscillation I can increase D. Ramps down to 1.4sec. That seems to damp the oscillations, I still see every gain switch Vsurf is close to zero, but not quite there. Next: increase D and P |
| Apr 25 2007 12:35:51:570PM | 1070425012 | Steve Wolfe | shot#12: increase D to 8 at 0.9sec, P to 50 at 1.0sec These are each about half what I had in the best simulation, twice previous shot. It should either work or go unstable again. Ramped down to 285kA at 1.29sec, during the RF. Steve Wukitch points out that we're seeing RF pickup on the magnetics. This gets to 0.2V Vsurf at 1.0sec, takes a step down, then breaks into oscillations around 1.1sec. Actually, the oscillation just gets bigger there, it's present the whole time; seems to get big and unstable during the RF pulse. I'm not putting in any gain switch at 1.1 Next: push the gains on up |
| Apr 25 2007 12:45:45:857PM | 1070425013 | Steve Wolfe | shot#13: increase D to 8 at 0.86sec, 15 at .9, P to 100 at 1.0sec. These values were working in simulation, but may be too much. I don't understand the instability setting in at 1.1sec, seems to be driven by RF. Disrupted earlier, at 0.9sec. Looks like the gain step at 0.86 may have been too much for it, didn't really make it to the ones at 0.9; looks like the D gain has to stagger in, may not be able to deal with D>8. Next: Back off, remove feedforwards. Note: whatever stopped running on shot#13, the data came through at the usual time, instead of delayed by 30sec, as it has been all day. |
| Apr 25 2007 01:04:22:443PM | 1070425014 | Steve Wolfe | shot#14: decrease D to 4 at 0.86sec, 8 at .9, leave P at 100 at 1.0sec. Ramp OH and EF3 feedforward down from 1.0 to 1.3sec Last shot for this MP today. Rampded down. |Vsurf| is <0.1V from 1.06sec to 1.3sec RF was faulting. Oscillations between 0.8 and 0.9sec, damped completely after 1sec. The high P gain is a good thing! Should have run it up there earlier. The psi0 error ends up around -1500 with the offset of 10000. Next: Move on to Scott/Ko experiment |
| Apr 25 2007 01:41:24:893PM | 1070425015 | Steve Wolfe | shot#15: Move on to Scott/Ko experiment Import seg1 from Seg4 Seg 2 off. Turn off dpcs_lowpass_obs. First shot: TF at 5.4, EF3 stepped, EF4 +4kA, PG4 [-.6,-.1] Load at 25-Apr-2007 13:11:41.00 Open tree /home/wolfe/pcs_scratch -1 Open tree done Got fields, three levels. Got gas, there is a glitch on the pressure at 1.5sec. Looks like the field affecting the ratiomatic. I think we need a little longer flattops on the EF3, the last one doesn't seem to be quite flat at end of pulse. Adjust EF3 pulses to end 50ms later each, and add 100msec to TF |
| Apr 25 2007 01:51:08:897PM | 1070425016 | Steve Wolfe | shot#16: No adjustment on EF3 after all. Move on to A2, EF4 off, all else same. Got EF3 and TF, no EF4 as wanted. Next: run the negative EF4 waveform, EF3 off |
| Apr 25 2007 02:06:28:060PM | 1070425017 | Steve Wolfe | shot#17: Negative EF4 waveform EF3 off, aux power EF3 off, EF3 breaker open Load at 25-Apr-2007 13:52:26.00 Open tree /home/wolfe/pcs_scratch -1 Open tree done Got two negative field values. Looks ok to me Waiting for the ok to move on to B1. |
| Apr 25 2007 02:21:16:987PM | 1070425018 | Steve Wolfe | shot#18: Reload shot 15, reduce PG4 aiming for 0.3mTorr PG4 pulse to [-.27,-.1] (linear would have been -.25, but I'm going to guess there's an opening time. OK, on to B2 (EF4 off) |
| Apr 25 2007 02:32:40:960PM | 1070425019 | Steve Wolfe | shot#19: on to B2 (EF4 off) Load this shot with seg4 set to TF-L and EF3-L waveforms (no changes in seg 1, just want to get the seg4 stored). While waiting, reimport seg1 of shot 17 into seg 4, so I can copy over the EF4 negative waveform for the next shot (B3) OK, got fields (EF3,TF) and gas. Waiting for ok on beam and MSE signals. |
| Apr 25 2007 02:49:09:140PM | 1070425020 | Steve Wolfe | shot#20: on to B3 (EF4 negative, EF3 off) Copy the negative EF4 waveform from seg3 to seg1. Load at 25-Apr-2007 14:35:39.00 Open tree /home/wolfe/pcs_scratch -1 Open tree done Now start working on the TF-H,EF3-H waveforms use seg2: Import from -1, seg4 Change title to TF-H,EF3-H Modify waveforms. Note this is not yet loaded. Note also that I need to modify the EF4 to 3kA on these, from 4kA Got fields got gas, waiting for SL ok to proceed to C1 |
| Apr 25 2007 03:16:05:087PM | 1070425021 | Steve Wolfe | shot#21: Import seg1 from 15 and adjust gas for 0.1mTorr (C1) Change pg4 from [-.6,-.1] to [-.16,-.1] (allowing a short opening bias) Load at 25-Apr-2007 14:55:15.00 Open tree /home/wolfe/pcs_scratch -1 Open tree done Now I've got the stepped tf waveforms loaded in segs 2 and 4, and the negative EF4 in seg3. Hung for 10 minutes in INIT, state blown away cmodws44 is incredibly slow No power shot. |
| Apr 25 2007 03:36:50:500PM | 1070425021 | Steve Wolfe | shot#21: (second time for shot#21) Second try at case C1: TF,EF3,EF4, 0.1mTorr Hung up in INIT again. Got fields and gas. Is there beam? pressure may be too high. It's ok, on to C2 (EF4 off) |
| Apr 25 2007 03:54:57:150PM | 1070425022 | Steve Wolfe | shot#22: on to C2 (EF4 off), same gas,same tf Problem with cooling and/or ross relays Got ef3, tf, gas. next c3. Need to find ef5 negative waveform, it's not in seg3 anymore |
| Apr 25 2007 04:05:40:793PM | 1070425023 | Steve Wolfe | shot#23: on to C3 (EF3 off, EF4 negative waveform), same gas,same tf Actually, it probably was, but I imported seg 1 from 20 into seg3 and copied from that. Load at 25-Apr-2007 15:58:04.00 Open tree /home/wolfe/pcs_scratch -1 Open tree done Got Ef4 and TF. If beam is ok on to step D (lower pressure, ef3 and ef4 constant values, half the pressure) |
| Apr 25 2007 04:18:48:610PM | 1070425024 | Steve Wolfe | shot#24: Import shot21 seg 1 EF3 pulse at 10kA, starting at -.5, EF4 at 4ka (as is) PG4 to [-.13,-.10] EF3 back on. Load at 25-Apr-2007 16:10:12.00 Open tree /home/wolfe/pcs_scratch -1 Open tree done Got all the fields, and gas. If they like the data, we'll drop the gas again next shot. |
| Apr 25 2007 04:30:51:820PM | 1070425025 | Steve Wolfe | shot#25: Case E, drop PG4 from 30ms to 15ms All fields same as 24 Load at 25-Apr-2007 16:24:20.00 Open tree /home/wolfe/pcs_scratch -1 Open tree done Got fields got gas. Ready for the tf step case? |
| Apr 25 2007 04:47:30:747PM | 1070425026 | Steve Wolfe | shot#26: Go on to F2 case (high field TF,EF3 steps) 1mTorr pressure EF4=3kA. Skipping the F1 case as being lower interest. I have it in seg 2, could just turn it on, but copy it over into seg1 just in case. Load at 25-Apr-2007 16:40:22.00 Open tree /home/wolfe/pcs_scratch -1 Open tree done Lost cycle. No shot. DPCS never saw check. |
| Apr 25 2007 05:04:27:113PM | 1070425026 | Steve Wolfe | shot#26: Second try. Programming the same, but filters are cooling off, so this is really G2 or something. Will start cycle at 17:00. Got fields got gas. Dexter reports the beam fired twice once in CHECK around the 7 count. Anyhow, we're done. |
| Engineering Operator Comments | ||||
| Shot | Time | Type | Status | Comment |
| 1 | 09:05:47:290AM | Plasma | Ok | |
| 2 | 09:21:55:513AM | Plasma | Ok | |
| 3 | 09:40:01:677AM | Plasma | Ok | |
| 4 | 09:52:35:557AM | Plasma | Ok | |
| 5 | 10:05:48:187AM | Plasma | Ok | |
| 6 | 10:22:18:710AM | Plasma | Ok | |
| 7 | 11:21:49:797AM | Plasma | Ok | |
| 8 | 11:34:28:993AM | Plasma | Ok | |
| 9 | 11:47:32:163AM | Plasma | Ok | |
| 10 | 12:00:15:777PM | Plasma | Ok | |
| 11 | 12:13:02:923PM | Plasma | Ok | |
| 12 | 12:25:56:770PM | Plasma | Ok | |
| 13 | 12:39:52:793PM | Plasma | Ok | |
| 14 | 12:54:00:923PM | Test | Bad | |
| 15 | 01:28:23:297PM | Test | Ok | MP#394 DNB |
| 16 | 01:44:08:770PM | Test | Ok | |
| 17 | 02:03:49:417PM | Test | Ok | |
| 18 | 02:16:13:127PM | Test | Ok | |
| 19 | 02:30:05:070PM | Test | Ok | |
| 20 | 02:46:41:717PM | Test | Ok | |
| 21 | 03:05:03:097PM | Test | Ok | |
| 22 | 03:50:05:770PM | Test | Ok | |
| 23 | 04:02:29:557PM | Test | Ok | |
| 24 | 04:14:56:170PM | Test | Ok | |
| 25 | 04:27:48:147PM | Test | Ok | |
| 26 | 04:40:55:777PM | Test | Ok | |
| System Availability | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Apr 25 2007 09:05:21:243AM | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Apr 25 2007 09:52:13:447AM | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Apr 25 2007 11:21:27:577AM | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||