Miniproposals

Miniproposal: Date Filed: Title: First Author: Session Leader: Robert Granetz (shots 1-31) Session Leader: Robert Granetz (shots 1-31)

Engineering Operator Run Comment

Plasma Current feedback; low-voltage EF3

Session Leader Plans

Physics Operators Plans

Session Leader Summaries

Entered: Jul 7 2004 03:49:08:023PM

Author: To Be Determined

Summary of Run 930630 Physics Operator: Goetz Asst. Operator: Granetz Eng. Operator: Fairfax The goal of this run was to demonstrate feedback of the plasma current using OH1. Difficulties were to be expected due to the fact that OH1 is also part of a R_CUR position controller and the demands could be in conflict. Secondary goals developed during the run. They consisted of trying different controllers for IP, consisting of combinations of OH1, EF1U, EF1L, OH2L, OH2U, and EF3, as calculated by Wolfe's orthogonal controller generator code. All these controllers are prone to the same difficulties as using OH1 alone. Four different types of IP controllers were tried out during the run. Controller #1: OH1 only; #2: OH1, OH2U, OH2L, and EF3; #3: OH1, OH2U, OH2L, EF1U, EF1L, and EF3; and #4: OH1, EF1U, EF1L, and EF3. Controller #1 showed no indications of being obeyed. The suspicion is that the R_CUR controller is winning the battle for OH1 control. Controller #2 was obeyed when the demand current was above 300 kA. A record current of 387 kA was achieved using this controller. This conroller has trouble with R & Z position when it is turned on early. Controller #3 seemed to control the flat top of the current although the magnitude was wrong. R and Z position control with this controller suffer greatly. Controller #4 also worked for current control. As the drawn waveform increased, so did the peak current achieved although there was no flat top. R and Z position control with this controller was good. For all controllers, power supplies running out of current was the limiting factor in plasma performance. Argon was puffed as a piggyback run. Beginning on shot 13 argon was puffed into the machine for HIREX measurements. These plasmas had high radiated power, high Zeff, and saturated SXR channels. Beginning on shot 20 the argon puff was decreased with Zeff and radiated power decreasing also. After shot 29 there was no argon puffed. Radiated power, Zeff, and SXR signals all decreased to their pre-argon levels. Problems: --------- No operational hold-ups today. The engineering staff did not have to go through the door after the EF3 tests were completed on shots 1-3. Access control was limited to one run to the diagnostic labs to remove condensation from the Michelson. The unexplained oscillation in the EF3 current came and went throughout the day EF3 turn-on timing drifted during the run. This was solved by adding a 'reverse front-porch' to the EF3 voltage (see shot 6) to ensure that there will be current flowing at commutation. This change in EF3 necessitates a compensating change in the EF4 programmed current to keep Bz in the correct range for breakdown. Care must be taken during the day to watch this relationship. Wall Conditioning/Vacuum: ------------------------- ECDC was run for one and one-half hours prior to the run, ending at 08:00. The base pressure was 4.E-6 when the run began at 09:00. Detailed Summary: ----------------- Shots 1-3 were engineering tests to check the board change that occured last night. Tests were o.k. under PLC control but the mystery relay problem nullified the hybrid test shot (3). The day began with programming identical to shot 930629009. This shot was a LEVP shot (low elongation, vanilla shot) with reduced EF3 voltage and a moderate current ramp rate. Shots 4-11 were attempts to use OH1 as the IP controller. This set of shots was plagued by EF3 jitter and Bz changes and operator error in the tweaking of fill pressure. Throughout these shots the programmed current level was 250kA at 100 msec, which is less current than what the ramp rate would have achieved at 100 msec. This led the operators into the land of confusion as the voltage demand to OH1 did not seem to be affected by the IP controller output. Shots 8 and 10 had flat-tops of aprrox. 300kA and the M-out was constant at -200V throughout this period from 100 to 300 msec. Shots 12-23 were attempts to use OH1, OH2U, OH2L, and EF3 as the IP controller. During this set of shots the drawn waveform was upped to 350kA to see if indeed the total M out to OH1 would change. It did, proving that the hybrid was indeed calculating the error in the correct manner. Starting with shot 18 the achieved current does track the drawn waveform in magnitude and change in shape at the switch times. Switch times were moved from 80 to 126 msec. Moving the switch times early causes bad things to happen to the R and Z control. As expected, the supplies ran out of current during the attempted feedback and muddy the picture. Shots 24-27 were attempts to use OH1, OH2U, OH2L, EF1U, EF1L, and EF3 as the IP controller. This controller had a deleterious effect on the R and Z position control and I would not recommend its use. The PID gain was increased to 10 for this set of shots because the calculated M values were too big for the controller and were divided by 10. The achieved current scaled with the drawn demand but at no time did the current match the drawn waveform. Shots 28-31 were attempts to use OH1, EF1U, EF1L, and EF3 as the IP controller. This controller was used because the EF1's still had plenty of volt-secs to provide to the plasma. This controller did well with R and Z position control. The drawn demand grew from 350 -> 400 -> 450 kA on shots 29-31 (shot 28 had too much pulse gas, density rose too early) and the current achieved went from 330 -> 350 -> 366 kA. All of these plasma currents rolled over when OH1 ran out of current. They died peaceful deaths. Inflections in the current waveform corresponded to the switch times of the controller and the switch off of various supplies. Recommendations: If I were to run this scan of IP controllers again I would start with a LEVP with a slower current rise (see 930629013). This may allow for the use of more volt-secs later in time and discriminate between the current rise rate and the feedback current control. Score-card: ----------- 3 engineering shots 22 "good" shots 5 fizzles 1 no-breakdowns -- 31 total shots

Physics Operator Summaries

Session Leader Comments
Jun 30 1993 08:40:02:410AM		Robert Granetz	Run 930630 Session leader: S. Goetz Lackey: R. Granetz Head Engineer S. Fairfax Primary object of today's run: feedback control of plasma current using OH1. The principle difficulty will be interactions between the demands on OH1 for both the Ip control and the R_CUR position control. Prior to the run we had 1.5 hours of ECDC in hydrogen. Base pressure before the first plasma shot is 1.e-6 (because ECDC was not turned off until 08:00). Commutation for all 5 supplies is set at t=0.000 (yes, OH1 will be commutated today). Supply Resistance Crowbar time OH1 0.060 ohm 0.070 s OH2U&L 0.013 0.120 EF1U&L 0.120 0.070 EF2 0.046 First shot will be copied from 930629009. This is a low elongation, plain vanilla shot, with somewhat reduced EF3 voltage.
Jun 30 1993 08:40:14:960AM	930630001	Robert Granetz	Run 930630 Session leader: S. Goetz Lackey: R. Granetz Head Engineer S. Fairfax Primary object of today's run: feedback control of plasma current using OH1. The principle difficulty will be interactions between the demands on OH1 for both the Ip control and the R_CUR position control. Prior to the run we had 1.5 hours of ECDC in hydrogen. Base pressure before the first plasma shot is 1.e-6 (because ECDC was not turned off until 08:00). Commutation for all 5 supplies is set at t=0.000 (yes, OH1 will be commutated today). Supply Resistance Crowbar time OH1 0.060 ohm 0.070 s OH2U&L 0.013 0.120 EF1U&L 0.120 0.070 EF2 0.046 First shot will be copied from 930629009. This is a low elongation, plain vanilla shot, with somewhat reduced EF3 voltage.
Jun 30 1993 08:40:35:120AM	930630001	Robert Granetz	Test of EF3 only.
Jun 30 1993 08:45:44:980AM	930630002	Robert Granetz	Another EF3 test.
Jun 30 1993 09:22:22:170AM	930630003	Robert Granetz	EF3 test under hybrid control. Test failed.
Jun 30 1993 09:33:28:740AM	930630004	Robert Granetz	Nice shot. Ipmax=0.26 MA, duration=0.322 s. Sawtoothing for the entire shot. Marfe occurs earlier than normal. Current droops during latter half of the shot.
Jun 30 1993 09:36:18:550AM	930630004	Robert Granetz	For the next shot, turn on feedback control of IP by setting the PID proportional gain to 0.5, starting at 0.100 s. The programmed demand is 250 kA until t=0.3, then rampdown in 0.1 s. The M-matrix controller is includes only OH1 (for now). The RCUR controller also contains OH1, along with the OH2's and EF3's.
Jun 30 1993 09:38:24:330AM	930630005	Robert Granetz	Fizzle. We didn't change anything prior to 0.100 s, so I don't think it was due to our programming.
Jun 30 1993 09:44:59:960AM	930630005	Robert Granetz	Bz was -5 mT at breakdown, which is marginal (it was -3 mT on the previous shot). The difference is due to the jitter problem with EF3. For the next shot, try putting a reverse front porch on EF3 voltage programming, starting at -0.015 s.
Jun 30 1993 09:49:35:810AM	930630005	Robert Granetz	Additional note for next shot: the reverse front porch (i.e. -50 volts) on EF3 just prior to breakdown will give some current in EF3. This must be compensated by about +100 amps in EF4. This will be programmed in also for the next shot.
Jun 30 1993 09:57:19:480AM	930630006	Robert Granetz	No breakdown.
Jun 30 1993 10:01:18:230AM	930630006	Robert Granetz	For next shot, raise EF4 current by another 75 amps.
Jun 30 1993 10:14:07:170AM	930630007	Robert Granetz	Fizzle.
Jun 30 1993 10:20:55:620AM	930630007	Robert Granetz	For next shot, back off on EF4 by 35 amps.
Jun 30 1993 10:40:45:110AM	930630008	Robert Granetz	Beauty! Ipmax=0.30 MA, duration=0.38 s. Sawtoothing for most of shot.
Jun 30 1993 10:45:05:540AM	930630008	Robert Granetz	The plasma current flattopped from 0.1 to 0.3 s, just as programmed. However, the value was 0.30 MA, rather than the 0.25 MA that was programmed. Examination of the hybrid feedback scopes showed that everything worked as expected. Either the gain is not enough, or the RCUR feedback is fighting the OH1 IP feedback. For the next shot, double the proportional gain to 1.0 on the IP feedback.
Jun 30 1993 10:45:56:960AM	930630008	Robert Granetz	Additional note: the prefill pressure has been 4.0e-5 torr today.
Jun 30 1993 11:00:51:210AM	930630009	Robert Granetz	Fizzle. Everything appeared to be okay. The gas prefill pressure was not reduced after the previous good shot...that's the only thing we can think of. Try again with no changes.
Jun 30 1993 11:24:26:510AM	930630010	Robert Granetz	Good shot. Ipmax=0.29 MA, duration=0.38 s. Feedback control of Ip still doesn't seem to be trying hard enough to keep the current at 0.25 MA.
Jun 30 1993 11:27:23:220AM	930630010	Robert Granetz	After looking at the hybrid traces, we think the problem is due to conflicting demands with the RCUR feedback (especially the derivative part of RCUR). We will change the IP and RCUR controller M values AFTER the next shot. For the next shot, the only changes will be: reduce the prefill pressure from 4.0 to 3.6e-5 (since this last shot was a good one) and increase the pulse gas programming.
Jun 30 1993 11:47:30:620AM	930630011	Robert Granetz	Another good shot. This had the funny behavior at 0.2 s where radiated power and x-rays take a jump, and the RF limiter probe drops. The plasma current droops starting at 0.2 s, probably not due to feedback control. Ipmax=0.27 MA, duration=0.34 s. Sawteeth start at 0.134 s.
Jun 30 1993 12:17:23:440PM	930630011	Robert Granetz	For the next shot, we are going to make a big change: input a new set of M matrix values for the RCUR feedback, and develop a completely new controller for IP which uses OH1, OH2's, and EF3's. We calculated the M values using Steve Wolfe's IDL routine "controllers.pro". They should be more diagonal than just using OH1 for IP feedback.
Jun 30 1993 12:18:15:010PM	930630011	Robert Granetz	Note: while putting in the new controllers, there were a few that were limited by the max permissibles in the hybrid.
Jun 30 1993 12:07:51:960PM	930630012	Robert Granetz	Runaway shot. Intense hard x-rays.
Jun 30 1993 12:11:29:830PM	930630012	Robert Granetz	This shot was in trouble from the beginning. Bz has been creeping up a tiny bit. For the next shot, drop EF4 by another 40 amps. Also, reload the old RCUR M matrix values (since the old and new sets were about the same except for a constant of proportionality). Also turn on argon puffing.
Jun 30 1993 12:25:48:510PM	930630012	Robert Granetz	Also for the next shot: raise the prefill from 3.6 to 4.5e-5 torr, because of the expected molydenum gettering that usually follows a runaway shot.
Jun 30 1993 12:20:48:910PM	930630013	Robert Granetz	Got a shot back, but it started disrupting at 0.20 s. It was sawtoothing until the first disruption. The radiated power was higher than normal, perhaps because the previous runaway shot splatted molybdenum all over.
Jun 30 1993 12:27:45:610PM	930630013	Robert Granetz	For the next shot, the only change will be to reduce the prefill to 4.0e-5 torr.
Jun 30 1993 12:35:10:300PM	930630014	Robert Granetz	Very similar to the previous shot. The first disruption happens slightly later... at t=0.22 s.
Jun 30 1993 12:54:52:230PM	930630014	Robert Granetz	We can't figure out the hybrid feedback signals. The error outs appear to be correct, but the final M out for the OH1 voltage doesn't seem to be properly summed up. In fact, the M out OH1 voltage didn't change between shot 004 (which didn't have feedback enabled) and shots 010 or 014 (which had IP feedback enabled, with two different controllers). For the next shot, try demanding 0.35 MA for Ip. Since the plasma current has been naturally reaching 0.30 MA, this new demand should change the error from -50 kA to +50 kA, resulting in an obvious change in the M out OH1 voltage.
Jun 30 1993 12:55:43:350PM	930630015	Robert Granetz	Fizzle.
Jun 30 1993 01:15:06:650PM	930630016	Robert Granetz	Got a little more plasma current this time. Ipmax=0.32 MA, duration=0.30 s. There were disruptions starting at 0.22 s.
Jun 30 1993 01:16:25:430PM	930630016	Robert Granetz	The hybrid reacted properly to the different IP programming. For the next shot, increase the PID proportional gain on IP to 2.5. Also decrease the prefill pressure from 4.0 to 3.7e-5 torr.
Jun 30 1993 01:20:46:740PM	930630017	Robert Granetz	Fizzle
Jun 30 1993 01:29:42:690PM	930630017	Robert Granetz	Vertical field was -5 mT at breakdown, which is marginal. For the next shot, increase EF4 by 25 amps (about +2 mT), and increase prefill to 4.0e-5 (since this shot was a fizzle).
Jun 30 1993 01:48:16:250PM	930630018	Robert Granetz	Good shot. Ipmax=0.35 MA, which is exactly what we programmed. There is a disruption at 0.22 s, which is where OH1 runs out of current. The plasma current also starts to rampdown at 0.22 s. Radiated power, Zeff, and x-ray signals are all about twice there normal levels. Other signals are near normal. No hard x-rays.
Jun 30 1993 02:01:19:520PM	930630019	Robert Granetz	Good one! Ipmax=0.387 MA. A new record. Duration=0.29 s. Ip starts to ramp down at 0.18 s, which is exactly when OH1 runs out of current. The centroid position is not well controlled after 0.2 seconds, probably due to the supplies running out of poop. Radiated power, Zeff, and x-rays are all very high.
Jun 30 1993 02:10:36:170PM	930630019	Robert Granetz	For the next shot, we wanted to move the PID turn-on and the programmed turn-on earlier (from 0.100 to 0.080 s.) However, we forgot to change the PID, and only changed the S demand waveform. Therefore, the next shot should be a repeat of this one.
Jun 30 1993 02:20:03:300PM	930630020	Robert Granetz	Good shot again. As expected, it was nearly identical to the preceding shot. The discharge ended a little earlier this time, duration=0.267 s. Ipmax=0.387 again. As before, the radiated power, Zeff, and x-ray signals are very high. This may be due to the argon puffing.
Jun 30 1993 02:20:56:250PM	930630020	Robert Granetz	For the next shot, move the PID earlier, from 0.100 to 0.080 s. Also decrease the pulse gas programming for the argon puff.
Jun 30 1993 02:37:11:480PM	930630021	Robert Granetz	Not so great. Ipmax=0.35 MA, duration=0.215 s. The position starts to get bad starting at 0.080 s, which is where the IP feedback turns on. The OH2's are pushed to zero hard by the turn on of IP feedback on OH1.
Jun 30 1993 02:39:42:400PM	930630021	Robert Granetz	For the next shot, the only change will be to delay turn-on of IP feedback from 0.080 to 0.126 s. The PID value is still 2.5 for the proportional component.
Jun 30 1993 02:50:03:520PM	930630022	Robert Granetz	Good shot. Ipmax=0.380 MA. Duration=0.345 s. X-ray emission saturated, but not until after 0.2 s. Zeff was less on this shot.
Jun 30 1993 02:51:11:170PM	930630022	Robert Granetz	Next shot: do a repeat for the ECE diagnostic. Amanda just fixed a problem that's invalidated their data for all previous shots.
Jun 30 1993 03:17:25:980PM	930630023	Robert Granetz	Not as nice a shot. It should have been a repeat of the preceding shot, but it had a major disruption at 0.22 s. We'll blame it on the Michelson.
Jun 30 1993 03:20:16:580PM	930630023	Robert Granetz	Big changes for the next shot: change the IP controller in order to reduce its use of the OH2's (so that they can be saved for ZCUR control). The new controller has been calculated using OH1, OH2's, EF1's, and EF3's. Due to hybrid limits, the new M values had to all be reduced by 1.5, so the PID gain was also increased from 2.5 to 4.0. Here goes nothing...
Jun 30 1993 03:30:53:000PM	930630024	Robert Granetz	Nice shot, but we still didn't make 0.400 MA. Ipmax=0.37 MA, duration=0.34 s. Position is still going bad, this time at 0.15 s.
Jun 30 1993 03:31:42:710PM	930630024	Robert Granetz	For the next shot, change IP feedback turn-on back to 0.100 s.
Jun 30 1993 03:43:29:980PM	930630025	Robert Granetz	Good shot. Ipmax=0.37 MA, duration=0.347 s. Similar to previous shots. The position control goes to hell starting at 0.12 s.
Jun 30 1993 03:44:38:440PM	930630025	Robert Granetz	For the next shot, back off on the Ip demand from 0.400 to 0.340 MA. We want to see if the controller really works, and if we get a flattop.
Jun 30 1993 04:02:29:530PM	930630026	Robert Granetz	Well we got Ipmax=0.31 MA, rather than 0.34 MA. And it flattopped. There was a major disruption at 0.225 s. Position control was a little better that time, maybe because the lower Ip demand left more current in the OH1 and OH2's.
Jun 30 1993 04:03:12:330PM	930630026	Robert Granetz	For the next shot, just increase the proportional gain on the Ip feedback from 4.0 to 10.0.
Jun 30 1993 04:14:04:850PM	930630027	Robert Granetz	Good shot. Ipmax=0.36 MA, which is higher than we wanted. Position control is also not too good.
Jun 30 1993 04:15:37:240PM	930630027	Robert Granetz	Big change for the next shot: use new controller that does not use the OH2's, only the OH1, EF1's, and EF3's. This is because the EF1's still have some current available. Also, the pulsed gas was increased a little.
Jun 30 1993 04:31:40:830PM	930630028	Robert Granetz	This was a disruptive shot...perhaps due to the increased pulse gas programming. Try repeating the shot with the gas programming copied in from several shots ago (from 026).
Jun 30 1993 04:41:51:750PM	930630028	Robert Granetz	Also for the next shot: turn off argon puffing.
Jun 30 1993 04:40:26:760PM	930630029	Robert Granetz	Beauty. Ipmax=0.33 MA, not quite flattopped though. EF1's had plenty of current left over. Centroid control was excellent. Radiated power, Zeff, and x-rays are all quite low. No disruptions until 0.335 s. Duration=0.38 s.
Jun 30 1993 04:41:18:630PM	930630029	Robert Granetz	For the next shot, change the S-waveform demand for Ip from 0.34 MA to 0.40 MA.
Jun 30 1993 04:42:12:580PM	930630029	Robert Granetz	Note: there was no argon puff on this shot.
Jun 30 1993 04:53:22:610PM	930630030	Robert Granetz	Another good shot. Ipmax=0.35 MA, with no flattop. The demand was for 0.40 MA. Centroid control was not bad. Radiated power, Zeff, and x-rays were low again, thanks to no argon puffing.
Jun 30 1993 04:55:13:900PM	930630030	Robert Granetz	For the next (and final) shot of the run: Raise Ip demand to 0.45 MA, and turn on PID earlier (from 0.100 to 0.076 s), and raise PLC current limit on EF3 to 18 kA.
Jun 30 1993 05:06:13:440PM	930630031	Robert Granetz	Good shot, but Ipmax=0.37 MA only. Duration=0.377 s. The supplies tried to get the current, but they ran out of current too soon. Radiated power, Zeff, and x-rays are all low again.

Physics Operator Comments

Engineering Operator Comments
Shot	Time	Type	Status	Comment
1	08:30:16:820AM	Test	Ok	EF3 test looks good
2	08:43:22:070AM	Test	Ok	EF3 test plc control; fast 10 khz digitizer
3	08:52:48:830AM	Test	Bad	EF3 did not fire; not the forced invert problem as yesterday
4	09:14:39:580AM	Plasma	Ok	First Plasma OK
5	09:35:17:970AM	Plasma	Ok	Fizzle. setting up OH1 for plasma current feedback
6	09:53:51:600AM	Plasma	Bad	No flash. EF3 supply misbehaving but did not cause fizzle;
7	10:12:35:970AM	Plasma	Ok	Fizzle. Using EF4 to compensate for EF3 bias. EF3 B supply still glitc
8	10:26:49:260AM	Plasma	Ok	Excellent shot with plasma current very flat at 300 kA for 200 msec
9	10:44:11:390AM	Plasma	Ok	Fizzle.
10	11:00:00:800AM	Plasma	Ok	good shot
11	11:27:32:020AM	Plasma	Ok	All OK
12	11:50:41:420AM	Plasma	Ok	Nasty runaway shot
13	12:11:40:590PM	Plasma	Ok	silva's back
14	12:27:59:080PM	Plasma	Ok	shot ok
15	12:53:15:120PM	Plasma	Ok	this ones for john
16	01:04:59:480PM	Plasma	Ok	good shot
17	01:17:11:690PM	Plasma	Ok	fizzle
18	01:31:19:780PM	Plasma	Ok	good shot; 350 kA with 100 msec flat-top; minor disruption at 220 msec
19	01:49:37:630PM	Plasma	Ok	0.38 MA plasma current; new record
20	02:09:07:820PM	Plasma	Ok	same as last shot; 385 kA
21	02:23:42:050PM	Plasma	Ok	330 kA disruption at 211 msec
22	02:40:19:780PM	Plasma	Ok	good plasma but no flat-top
23	02:53:12:630PM	Plasma	Ok	good plasma ; 150 kA/msec disruption from 300 ka
24	03:20:11:460PM	Plasma	Ok	New Ip controller but no flat top; OH2s running out of current
25	03:33:58:250PM	Plasma	Ok	no flat top
26	03:46:43:380PM	Plasma	Ok	no flat top; disruption at .23 sec
27	04:01:51:370PM	Plasma	Ok	Max PID gain but marginal Ip flat-top
28	04:14:58:330PM	Plasma	Ok	Very disruptive shot
29	04:32:45:760PM	Plasma	Ok	very nice shot, 325 +/1- 6 ka for over 100 msec
30	04:44:53:450PM	Plasma	Ok	Tried for 400 kA; got 350
31	04:57:26:050PM	Plasma	Ok	nice shot but no 400 kA