Alcator C-Mod Run 930520 Information

Back to runs

Miniproposals
Miniproposal:
Date Filed:
Title:
First Author:
Session Leader:Steve Wolfe (shots 1-34)

Operators
Session leader(s):Steve Wolfe
Physics operator(s):John Goetz
Engineering operator(s):Vinny Bertolino

Engineering Operator Run Comment
Continue plasma shots >100KA

Session Leader Plans

Physics Operators Plans

Session Leader Summaries
Entered: Jul 7 2004 03:49:07:273PM
Author: To Be Determined
Summary of Run 930520
Session Leader/Physics Operator: Wolfe
Assistant Physics Operator: Goetz
Engineering Operator: Bertolini

The main goal of today's run was to establish plasma position control by
feedback on the current centroid. This goal was largely achieved, although
some refinement is still required.

Radial position control was achieved using the EF3 coils. We used an M
value of +0.2 and a PID proportional gain of 5 (10 also worked, but tended
to produce some substantial oscillations). The PID gain was turned on at
40msec.

Vertical position control using the EF2 supplies was unsuccessful. The
EF2's were being driven too hard by the EF3's, and EF2U did not seem to be
responding properly to voltage demands. Even current feedback control on
the EF2U was not working well after commutation, although it was possible
to keep positive current in the coils for the duration of the shot. It
probably would have been useful to edit the voltage pre-programming on the
EF2's based on the shots we were getting, but this was not tried. The
response of the EF2 supplies, particularly the EF2U, needs to be
investigated on a maintenance day.

Vertical position control was achieved using the OH2 supplies. The M values
used were +/-0.01 on OH2U&L, respectively. The PID gain was set to 1.0.
This combination was successful in lowering the plasma by 0.5 to 1.0 cm on
shots 930520031 and 032, which were the most successful shots of the day.
Note that operation with a PID proportional gain 5.0 resulted in large
anti-symmetric currents in the OH2's after the fizzle on shot 25.

An attempt to control the plasma current using a controller consisting of
OH1 only was made on shot 32. The M value was set to -5.0e-4 with a PID
proportional gain of 1.0. The feedback signal indicated a response of the
correct behavior, but small magnitude. The change in the plasma current
evolution relative to shot 31 was negligible. The gain was raised to 5 on
shot 33, but the plasma fizzled early.

We found that there is an error in the magnitude of the signal produced by
the Ip-referenced channels of the Wavegen (ie IP_RCUR and IP_ZCUR) of a
factor of ten. This will be remedied when we decide for sure whether the
appropriate fix is in hardware (a jumper controlling gain in the reference
leg) or software (in the expression for WAVEGEN OUTPUT or in the HYBEXE
GET_P_TO_V routine). For now, it is necessary to draw a displacement of 5cm
to get a P_out corresponding to 0.5cm. Note that larger displacements may
require a resetting of FACTOR (using the SCALE BIT option in PCS) to avoid
overflowing the wavegen DAC's.

The secondary goal of the run was to improve the robustness of the startup.
This was not done. The shots alternated almost exactly between good
discharges of 180kA or more and fizzles which died by 30msec with currents
less than 50kA. Of the 30 attempted plasma shots, 17 were "good", with
currents over 180kA and durations of at least 100msec, and 13 were fizzles.
All at least broke down, and even the fizzles would have been considered
good results two days ago; how our perceptions do change! The most good
shots in a row was two, and the most fizzles in a row was also two. Under
these circumstances, it was not practical to try to "fix" the problem. My
best guess as to the cause it that the successful shots, all of which
eventually disrupted, generated enough dirt to compromise the next shot.
There were no obvious differences in the magnetics on the good and bad
shots, although this was not studied in detail.

For most of the day, a minor disruption occurred at 70msec. Efforts to
remedy this by playing with the gas puff, the triangularity, and the radial
position all went for naught. The last two successful shots (31 & 32)
avoided this disruption, or at least delayed it by 50msec or so. The change
for these shots was to lower the plasma by 5 to 10mm. I have no idea why
this improved things, but it seems to have done so.

After shot 31, it was noted that the hybrid predictor for Ip was reading
lower than the official Ip by about 20%. Since the hybrid predictor is
simply the ROG_CD rogowski, this is very strange. Investigation after the
run indicates that the ROG_CD input received at the hybrid is indeed low.
Suspicion falls on the iso-amp, but this was previously calibrated and
checked out OK. If there is a problem with the rogowski, we need to worry
about other signals as well. However, the other predictors seem to be ok.
This puzzle needs to be addressed on the next maintenance day.

Physics Operator Summaries

Session Leader Comments
May 20 1993 07:24:43:240AMSteve WolfeRun Plan for 5/20/93

Primary topic: Establish current centroid position control
Secondary topic:Investigate lack of reproducibility of discharge startup,
and attempt to improve robustness of startup program.
Addt'l topics: Control plasma current, try increasing density, vary
plasma shape.

Sequence: Start from setup of 930519027.

May 20 1993 07:25:45:350AMSteve WolfeDetail of run plan on primary topic:
Addressing control of centroids:
After obtaining discharges with Ip>200kA, pulse length >75ms:
1) Program 1-2cm steps in RCUR and ZCUR positions; observe
Pout and Aout with loop open
2) Close loop on RCUR using EF3 controller; increase PID
prop. gain and observe result
3) Close loop gain on ZCUR using EF2's ; need to hold
EF2's on longer than on 930519.
Also try using OH2's.
4) Remove steps and tune position for best plasma, least
interactions with RF limiter.


May 20 1993 07:26:33:570AMSteve WolfeDetail of run plan on secondary topic:
Addressing startup:
1) Try scanning initial Bz
a) first try EF4 scan
b) if indicated, try BZ control of EF4 in pre-blip
2) Vary OH2 programming and feedback
3) Vary OH1 programming
4) Delay inital gas pulse to 15:20 msec


May 20 1993 07:27:23:990AMSteve WolfeDetail of run plan for additional topics:
Other topics:
1) If time permits, try OH1 controller for Ip; requires removing
something else from a wire; dBZ_dR is a good candidate.
2) If it doesn't seem to cause problems, try a big, short gas pulse
at ~50msec; observe on TCI and Z-meter.
3) If time permits, try driving up EF1 currents (by IC f.b. control)
after 50msec

May 20 1993 09:00:32:350AM930520003John Goetz
Zero power test of OH1. No apparent output on RCUR and ZCUR.
Error Message communicating with node 0.
Coldstart done. BCS search o.k.

May 20 1993 09:10:53:820AM930520004Steve WolfeSet up as for shot 930519027.
Start Close Open Crow Invert Resistor
EF1 U&L -1.6 -1.55 0. .04 .25 .180
OH1 -1.45 -1.4 0. .04 .25 .100
OH2U&L -1.6 -1.55 0. .05 .25 .025
EF2U&L -1.0 -1.0 .25 .0455
EF3 -0.1 .25
EF4 -1.6 .25

PCS programmed as for 930519027, except 2cm step in Rcur and 1cm in ZCur
Pressure = 6e-5
Good flash: 120msec long 250kA; disruption at end
See clear output on RCUR and Zcur Pout values

May 20 1993 09:28:46:430AM930520004Steve WolfeClose loop on RCUR:
M=0.2 from 0.05 to 0.08
PID = 1. from .04 to .09
BZ pID off from 0.048 on
Everything else the same.
Good flash.
SAme shot, pretty much.
A hint of output on the feedback signal, pretty negligible. No
noticable response on position.

May 20 1993 09:41:13:940AM930520006Steve WolfeINcrease PID to 5. Leave M at 0.2 from -3 to 3
Extend P_RCUR step to go from 50 - 80 msec
Wimpy. Dies around 30msec

May 20 1993 10:08:30:990AM930520006Steve Wolfechange OH2 voltages from 30V to 10V between 4 and 48msec to try to give margin
wrt DBz/DR going negative.

Result: still no good. Gradient did come up ~10mT/m.

May 20 1993 10:09:32:160AM930520007Steve Wolfechange OH2 voltages from 30V to 10V between 4 and 48msec to try to give margin
wrt DBz/DR going negative.

Result: still no good. Gradient did come up ~10mT/m.
Note: last shot comment was for this shot. Will Void it out later.

May 20 1993 10:29:12:450AM930520008Steve WolfeNo changes, see if it will come back on its own.
It did.
Got visible change in EF3 current (~500A)
Small position change in right(?) direction.

May 20 1993 10:43:07:780AM930520009Steve WolfeDouble the PID gain (10)
Also extend gas pulse to 150msec

Wiggles on the position. Got > 1000 Amp jump out of the EF3.
Disrupts beginning at 95msec, but it's in trouble before that.
Bumps on outside limiter at 70msec.

It's working, I think!

May 20 1993 11:07:33:420AM930520010Steve WolfeTry moving it inside 1cm. No other changes.

Result: Another wimpy failure. The irreproducibility strikes again.

May 20 1993 11:12:52:220AM930520011Steve WolfeThis one went OK. Seems to hold position pretty well, but can't see the
step. It may have been near 65.5 anyway.

Still disrupting like crazy.

May 20 1993 11:31:37:670AM930520012Steve WolfeExtend IC_EF2 programming to hold 500 amps through end of shot
2cm inward step on rcur
Changed EF2 voltage to -100V (from -250) from 50msec
PID on RCUR extended to 150msec

Bad shot- dies early again.

May 20 1993 11:38:41:450AM930520013Steve WolfeExtend IC_EF2 programming to hold 500 amps through end of shot
2cm inward step on rcur
Changed EF2 voltage to -100V (from -250) from 50msec
PID on RCUR extended to 150msec

Bad shot- dies early again.

May 20 1993 11:40:26:460AM930520014Steve WolfeThis one went ok . Lasts till 140msec, with disruptions.
Error on position is decreasing the whole shot.
No obvious inward step.
EF2's still come down too fast.
Rice says this is hottest shot yet : 1.3keV
Lot's of Mo.

May 20 1993 11:42:16:270AM930520014Steve WolfeAdjusted EF2 current trace.

Short one again. It's alternating!

May 20 1993 12:26:38:870PM930520015Steve WolfeEF2 U is still going to zero. It doesn't seem to be responding to
program signal. Also, the bus voltages may be interchanged??

EF2U goes to zero around 70ms.

May 20 1993 12:33:45:950PM930520016Steve WolfeAnother dud!
EF2 voltages don't seem to be balanced, although it seems to regulate close to where
we ask on the dud shots.

May 20 1993 12:34:51:090PM930520016Steve WolfeHad raised program current to 750 a from 500 between 50msec and 150msec

May 20 1993 12:53:56:250PM930520017Steve WolfeHad raised program current to 750 a from 500 between 50msec and 150msec
This one went (every other fucking shot!)
Seems to regulate EF2's OK this time, except that they're driven by EF3,
which is trying (and succeeding) to keep the RCUR position constant
during the disruptions.

May 20 1993 12:55:00:200PM930520018Steve WolfeEliminated step in RCUR position.
Ramp gas puff down from 70msec

Result: It's an even numbered shot, so it's a dud.

May 20 1993 01:16:05:730PM930520019Steve WolfeStill disrupting at the same time. It may be a density limit disruption.
TCI says ~1e20, [G] limit is 1.3e20 or so, depending on size
Position looks good.

May 20 1993 01:19:56:950PM930520020Steve WolfeTry for ZCUR control using EF2's.:
M= +/- 0.05 on EF2U&L
PID = 1.0 at 30msec
Step 1cm positive at 50:70msec
shorten gas pulse (down at 50msec instead of 70msec)

Result: another dud. Try it again next shot.

May 20 1993 01:36:15:300PM930520021Steve WolfeTry for ZCUR control using EF2's.:
M= +/- 0.05 on EF2U&L
PID = 1.0 at 30msec
Step 1cm positive at 50:70msec
shorten gas pulse (down at 50msec instead of 70msec)

Shot still disrupting at the same time.
Density is lower: .25 fringe compared to .35
Can't see any step in Z position.

May 20 1993 01:41:20:840PM930520022Steve WolfeTry for ZCUR control using EF2's.:
M= +/- 0.05 on EF2U&L
PID = 5.0 at 30msec
Step 1cm positive at 50:70msec
Turn off Br PID at 30msec

Another dud

May 20 1993 02:15:21:300PM930520023Steve WolfeTry for ZCUR control using EF2's.:
M= +/- 0.05 on EF2U&L
PID = 5.0 at 30msec
Step 1cm positive at 50:70msec
Turn off Br PID at 30msec

Delay crowbar on EF1U&L from 40msec to 50msec-
try to get rid of plasma diverting around 70msec.

result: EF2U is not responding to voltage commands properly.
It ended up going to zero at 70 msec
The feedback on Z isn't likely to work like this!

May 20 1993 02:52:14:160PM930520024Steve WolfeTry for ZCUR control using OH2's.:
M= +/- 0.01 on OH2U&L
PID = 5.0 at 30msec
Step 1cm positive at 50:70msec
Leave Br PID at 30msec

Ran OK - no sign of the step.
The wavegen output is only 40mV!!
I think there is a factor of 10 missing somewhere
Looking back at shot #4, we only got .25 V on RCUR.
It should have been ~2.5V I think.

May 20 1993 03:18:15:550PM930520025Steve WolfeTry for ZCUR control using OH2's.:
M= +/- 0.01 on OH2U&L
PID = 5.0 at 30msec
Step 1cm positive at 50:70msec
Leave Br PID at 30msec

Gain to 5 on PID (last was at 1.0)
Horne is chasing down the factor of ten.
Run this shot anyway.

Commutation failure in OH2L supply .
We got +/- 6V of demand out of hybrid.
Gain was too high, we have an offset of
-.3V out of A-out.

May 20 1993 03:37:48:560PM930520026Steve WolfeTry for ZCUR control using OH2's.:
M= +/- 0.01 on OH2U&L
PID = 1.0 at 30msec
Step 5cm positive at 50:70msec (we think this will give 0.5cm really)
Leave Br PID at 30msec

Get a +/-10V output from OH2's at step, smaller than step at 30msec
when the switch time occurs. No visible response on position.

May 20 1993 03:53:51:060PM930520027Steve WolfeTry for ZCUR control using OH2's.:
M= +/- 0.01 on OH2U&L
PID = 2.0 from 30msec to 80msec, then back to 1.0
Step 5cm positive at 50:70msec (we think this will give 0.5cm really)
Leave Br PID at 30msec

Get a +/-20V output from OH2's at step, smaller than step at 30msec
when the switch time occurs.
Plasma disrupts during step time at 200kA. I think we did it, but not sure.
A good imagination could see a few mm step on position.

May 20 1993 04:00:47:630PM930520028Steve WolfeTry for ZCUR control using OH2's.:
M= +/- 0.01 on OH2U&L
PID = 2.0 from 30msec to 80msec, then back to 1.0
Step 5cm positive at 50:70msec (we think this will give 0.5cm really)
Leave Br PID at 30msec

Same programming as last time. Got another dud.

May 20 1993 04:16:39:680PM930520029Steve WolfeTry for ZCUR control using OH2's.:
M= +/- 0.01 on OH2U&L
PID = 1.0 from 30msec to 80msec, then back to 1.0
Step 5cm positive at 50:150msec (we think this will give 0.5cm really)
Leave Br PID off at 30msec

Runs raggedly to 180msec. Z position is fairly stable, but no sign of any
step.
We still have that disruption at 68msec.

May 20 1993 04:32:33:640PM930520030Steve WolfeTry for ZCUR control using OH2's.:
M= +/- 0.01 on OH2U&L
PID = 1.0 from 30msec to 80msec, then back to 1.0
Step 5cm negative at 50:150msec (we think this will give 0.5cm really)
Lower RCUR gain to 5, and extend to 200msec

A dud!!!$%^&*

May 20 1993 04:41:45:900PM930520031Steve WolfeTry for ZCUR control using OH2's.:
M= +/- 0.01 on OH2U&L
PID = 1.0 from 30msec to 80msec, then back to 1.0
Step 5cm negative at 50:150msec (we think this will give 0.5cm really)
Lower RCUR gain to 5, and extend to 200msec

Set up feedback control on IP with OH1, but don't close the loop!
M=-5.e-4 from 45 msec on

Shot ran until ~150. No disruption until 120msec
JT reports it did look low. FFIT seems to agree, by maybe 1cm.
This was a good one.

May 20 1993 04:57:50:490PM930520032Steve WolfeTry for ZCUR control using OH2's.:
M= +/- 0.01 on OH2U&L
PID = 1.0 from 30msec to 80msec, then back to 1.0
Step 5cm negative at 50:150msec (we think this will give 0.5cm really)
Lower RCUR gain to 5, and extend to 200msec

Set up feedback control on IP with OH1,
M=-5.e-4 from 45 msec on
PID at 1.0 from 44msec

Shot ran until ~150. No disruption until 120msec
looks a lot like the last one, but did get some response from OH1 to try to flattop.

May 20 1993 05:03:51:610PM930520033Steve WolfeTry for ZCUR control using OH2's.:
M= +/- 0.01 on OH2U&L
PID = 1.0 from 30msec to 80msec, then back to 1.0
Step 5cm negative at 50:150msec (we think this will give 0.5cm really)
Lower RCUR gain to 5, and extend to 200msec

Set up feedback control on IP with OH1,
M=-5.e-4 from 44 msec on
PID at 5.0 from 44msec

A dud! Oh well, SH can take up here tomorrow.

May 21 1993 08:49:16:800AM930520034Martin GreenwaldPower Supply settings:
Supply Start Close Open Crowbar Invert
EF1L -1.6 -1.55 0.0 .05 .25
EF1U -1.6 -1.55 0.0 .05 .25
OH1 -1.45 -1.40 0.0 .04 .25
OH2L -1.6 -1.55 0.0 .05 .25
OH2U -1.6 -1.55 0.0 .05 .25
EF2L -1.0 .25
EF2U -1.0 .25
EF3 -0.1 .25
EF4 -1.6 .25
EFC -0.1 .30
TF -1.65 .10
Bus Diagnositcs integrator start: -2 sec
end +6 sec
Dig time = -1.7

Physics Operator Comments

Engineering Operator Comments
ShotTimeTypeStatusComment
107:45:17:620AMTestOktesting oh1 camac enable
208:08:32:300AMTestOktest of oh1 camac enable command
308:46:29:760AMTestOkoh1 camac pulse test
409:02:10:530AMTestOk240KA PLASMA FIRST SHOT
509:20:05:480AMPlasmaOk240KA PLASMA,
609:37:39:400AMPlasmaOk70KA PLASMA POWER SUPPLIES OK
709:56:39:730AMPlasmaOk20KA PLASMA, OH1 1SCR NOT CONDUCTING, OH2'S CURRENT RIPPLE
810:09:58:330AMPlasmaOk260KA PLASMA, LESS RIPPLE ON THE OH2'S--300A
910:29:54:680AMPlasmaOk180KA PLASMA
1010:45:39:450AMPlasmaOk20KA PLASMA
1110:55:27:920AMPlasmaOk200KA PLASMA
1211:14:15:350AMPlasmaOk30KA PLASMA, TF 139KA--THIS HAS BEEN DECREASING WITH EACH RUN FROM 149KA
1311:24:48:120AMPlasmaOk220KA PLASMA, EVERY OTHER SHOT SEEMS TO PRODUCE HIGH PLASMA CURRENTS-- P
1411:40:07:710AMPlasmaOk20KA
1511:50:39:230AMPlasmaOk200KA
1612:25:12:960PMPlasmaOk20KA
1712:35:33:650PMPlasmaOk210KA
1812:51:11:360PMPlasmaOk20KA
1901:00:48:800PMPlasmaOk190KA
2001:16:39:490PMPlasmaOk20KA, OH1 1SCR NOT CONDUCTING - TYPE 3 FAULT
2101:26:16:720PMPlasmaOk200KA PLASMA
2201:38:37:190PMPlasmaOk30KA PLASMA
2301:50:10:150PMPlasmaOk190KA PLASMA
2402:22:44:110PMPlasmaOk190KA PLASMA -- OH1 1SCR NOT CONDUCTING
2502:57:36:120PMPlasmaOk20KA PLASMA -- OH2L COMM FAULT
2603:24:33:010PMPlasmaOk180KA PLASMA -- OH1 ONE SCR NOT CONDUCTING
2703:40:11:600PMPlasmaOk210KA PLASMA CURRENT, NO FAULTS
2803:53:58:270PMPlasmaOk30KA
2904:04:40:200PMPlasmaOk200KA PLASMA
3004:19:55:340PMPlasmaOk50KA PLASMA
3104:34:02:980PMPlasmaOk210KA PLASMA
3204:45:57:030PMPlasmaOk210KA PLASMA
3304:58:30:850PMPlasmaOkALMOST 3 IN A ROW
3405:20:52:570PMTestBad