| Miniproposals | ||||||||||
|
| Operators | |
| Session leader(s): | Ian Hutchinson |
| Physics operator(s): | Steve Horne |
| Engineering operator(s): | Vinny Bertolino,Joe Daigle |
| Engineering Operator Run Comment |
| orthog controllers |
| Session Leader Plans |
| Physics Operators Plans |
| Session Leader Summaries |
| Entered: Jul 7 2004 03:49:11:337PM |
| Author: To Be Determined |
| Session Leader's Summary 941129 SL Hutchinson, PO Horne, EO Bertolino Miniproposal 65 continued: Orthogonal Controllers. Plan: Check out Tweaked Ip controller. Turn on RXU,ZXU at lower gain. Turn off EF1/2U gain. Exercise these controllers. Results: Most of the day was spent struggling to make the RXU,ZXU controllers work. Finally it was found that they were overflowing in the wavegen. Shots 24 and 25 fixed this problem. This made the feedbacks do the right thing at last. Shots 22 and 24 had 0.4s H-modes with ELMS but quite good confinement. Their upper shaping was quite different. This might be an interesting comparison. We have not yet tuned the controllers. But this should be relatively easy now we have them under proper control. A problem is that the place the controllers want to go maxes out the EF2 current at -4kA. Our running into this limit has not yet caused loss of control but it might. It was found that the analysis predictors are not the same as the hybrid is using. This needs to be fixed. It is mostly a calibration difference. Details: Start from 941123018 which has no RXU,ZXU control yet. Engineers say the OH1 cross-over ought to be shorter. Shot Set tauopt=0.3 in Ip controller which lowers the OH2 weight. Fill 1.2e-4 1 Pla Long but lots of nasty injections. No obvious changes from 23018. Fill to 1.4e-4 and loaded to verify that we got the changes. 2 Pla Looked nicer. Density dropped to .5 from 1.0e20 nL. Turn on nL4 feedback. Asking for .6. Lowered gain on ZXL to 2 from 9 at 1 sec to try to help the EFC step at end. 3 Pla Disrupt on ramp down at 550kA. OH2 behaviour different in down ramp. ZXL does move up rapidly during the down ramp. Actually there was no density feedback because the controller was zero. Fixed. Drop EF1/2U gain to zero at .35s. Turn on ZXU RXU then, @ 2 s^-1 Drop RXL gain to .5 from 3. at 1s. 4 Pla Ramp down seems fixed but early we have a position excursion. The switch at .35 causes EF2U to have a big negative voltage on it. This rails the supply. Although the controllers move the plasma in the right direction, it overshoots for no obvious reason. The predictor scope does not agree with the Aout. There is an offset difference. Put back EF2U gain at 0.5 after .35s. 5 Pla Still a big jerk on the EF2 voltagespso. Programmed the plasma further in early to fix the hot spot. ZXU gain to .05 (nearly off). 6 Pla Almost no difference. Changed the V on EF2U/L to be +100V from -100V during the relevant period. 7 Pla Big improvement in the lurch. The hot-spot is still there. Raised V EF2U/L to 200 V, programmed R to .65 early 8 Pla Seemed to get rid of the hot spot. Injections gone. Lurch not changed much. Programmed R to move in to .64 during ramp down. Prog RXU to -.03 because there is a discrepancy between the Hybrid and the analysis values. 9 Pla Plasma did move in. RXU control changes minimal Raised gain to 10s^-1 from 2s^-1 Programmed the current end to be a linear ramp down. 10 Pla 10-17 See Horne's Log. Observed disruption problems on ramp down. 17 Pla Disrupt at 1.04s 18 Pla Runaway Shot. Change to ZXU gain to step up slowly. Fill to 1.6 19 Pla Long. Not all the steps visible Moved the .55 switch to .5s. Drew -.04 RXU and doubled gain to 4/s. 20 Pla Disrupt at 1.06s. Seemed to do something to RXU. Drew -.07 RXU, gain to 8/s. 21 Fiz EF1/2L gains were previously left on till .5s. Turned off at .35s. Fill 1.4. 22 Pla Long H-Mode. RXU Aout did not change. This proves that the hybrid is not working correctly. Seems to be a problem with the wavegen and factor. The vertical position lurch is improved. Dropped the factor on RXU and ZXU by x4. This overflows the controller. Dropped the timeconstants by x4. Raised the PID gain back to where we were before. 23 Pla Disrupt at .28s. Ugly start up. repeat. 24 Pla Ramp down went to 1.3s. Dramatic changes on RXU ZXU. They did the right thing. Raised ZXU gain to 8/s from 4/s. Raised drawn RXU to -.04 from -.07. 25 Pla Went to 1.33s. Another long H-mode Shapes are really doing the right thing. Nicely single null. EF2U saturates Scorecard. Shot,Duration(s),Ipmax(MA) Shot,Duration(s),Ipmax 1 1.271 0.817 2 1.289 0.817 3 1.132 0.815 4 1.300 0.833 5 1.328 0.830 6 1.345 0.833 7 1.306 0.835 8 1.337 0.838 9 1.330 0.838 10 1.276 0.838 11 1.069 0.835 12 1.071 0.830 13 0.000 0.024 14 1.050 0.843 15 1.290 0.838 16 1.053 0.848 17 1.054 0.851 18 0.389 0.559 19 1.321 0.840 20 1.057 0.843 21 0.000 0.024 22 1.055 0.840 23 0.385 0.592 24 1.311 0.833 25 1.344 0.843 Finished summary for 941129 Total of 25 shots. 25 shots. 23 plasmas. 2 fizzles. 0 duds. Fizzle = gt 17 lt 50kA. |
| Physics Operator Summaries |
| Entered: Jul 7 2004 04:36:52:053PM |
| Author: To Be Determined |
Run 941129 continuing MP 65, orthogonal controllers. Hutch, Horne, Bertolino, Daigle. We fought various problems today but by the end of the run Ian's controllers were in and were doing the right thing. The shot log below has most of the details. After a late start, we struggled against a PCS bug for most of the run; also there were cooling problemsand a couple of camac init failures. We had lots of fun on the way, including our longest ohmic h-mode (22, see Ian's notes) This may also be our first with density feedback, which was turned on today successfully. PCS/Operator notes -- the code which used to check whether the wavegen programming was within hardware limits was inadvertently deleted from PCS. This lead to the problems involving r_xu and z_xu starting at shot4, which we didn't discover until after shot 22. We fixed this (you'll be warned by PCS) but be aware that shots from today up to 23 have bad programming in r_xu and z_xu. Raise the gain on nl_07 from 5 to 8 or so. ====== Run starts===== Shot 941123018 loaded. 1.2e-5 fill 1 plasma, dirty, but ramped down nicely. Density feedback signal is on nl_04. It looks ok, check once more. fill to 1.4e-5 for next shot, 2 Plasma. density signal is on nl_4 ; looks ok; says 1.e20 last shot, 5.e19 this shot. (according to stored p_to_v). 3 Turn on density feedback at .2 sec, P gain of 5. Drop zxl feedback at rampdown from 8 to 2. Plasma 4 ef1 u and ef2 u gain to zero at .35 rxu and zxu set at 2/sec from .35 to 1, then drop to .5/sec rxl set to 2 at 1 sec. Start density rampdown at .95 sec. Turns out there was no controller loaded for gas valve. Agreement was coincidental. pulse_gas_3 controller Loaded for next shot. Long delay. One TF arm claims to be warm. Result: Plasma. very fast vertical oscillation at about 700 msec; (150 hz or so.) 6. There's a large difference between the r_xu as calculated via hybrid, and via Ian's scope. 7 Plasma. The rxu as calculated by hybrid agrees with a calculation based on digitized magnetics. Therefore the hybrid calculation is correct -- no overflows, etc. The predictor as loaded seems to be qualitatively correct, except for a scale failure. We'll just fiddle it with the drawn waveform. Density feedback finally working -- controller wasn't right till this shot. 8 There's a hot spot on the antenna, and a kick in ef2u voltage which Ian's been trying to fix; see his log. 9 PLASMA. 10 INCREASE GAIN IN r_xu from 2/sec to 10/sec. Plasma. The gain change didn't happen. 11 zxu programming changed from +.1 to +.03 INCREASE GAIN IN r_xu from 2/sec to 10/sec. Plasma, bad injection of moly just after rampdown, some gettering. 12 Increase gain on zxu from .05 to .25 in hybrid units. Gain drops back to .05 at 1 sec. Plasma, Nasty disruption. 13 Start clearin negative going ramp a bit earilier (800 msec.) to try to fix the disruption. Camac delay, then alternator speed too low. FIZZLE 14 No change. Plasma, but still disruption when rampdown starts. 15 Reload shot 9, which ramped down nicely. This should test whether programming or machine conditioning. Plasma, Ramped down nicely. 16 Try to fix zxu first this time. Change zxu gain from .05 to .5 from .35 sec to 1sec; then zero. 17 Try leaving gain on, but drawing zxu back to zero before rampdown. Also turned argon puff off. No obvious change. 18 gain on zxl up to 5 /sec (was 3.5) Very dirty short shot. 19 1.6 on fill; raise zxu gain from 1 at .35 to 4 at .65 sec. 20 The switches at .35 and .45 show up; but not the later ones. Move switch at .55 back to .5. Drew rxu to -4 cm, and raised gain from 2/sec to 4/sec. 21 Fizzle. 22 ef1 L and ef2L current PID to zero at .35 sec. It was a mistake to have them on this late -- also drop fill to 1.4e-5 Plasma, disrupts on the way down. 23 It seems that the wavgen gain for rxu and zxu is overflowing. (Changing them didn't change the error.) So dropped both predictor:factors by factor of 4; rescaled controllers to fit by setting controller time constants down by factor of 4; set PID gains back to the same physics level. This was a PCS bug which crept in since June -- check_wavegen wasn't being called. 24 Much better. Got the rampdown. 25 zxu to 8/sec rxu drawn to -.04 instead of -.07 Best yet. It looks like the controllers actually have control. ====== End run ======= |
| Session Leader Comments |
| Physics Operator Comments |
| Engineering Operator Comments | ||||
| Shot | Time | Type | Status | Comment |
| 1 | 09:05:03:520AM | Plasma | Ok | no faults |
| 2 | 09:23:05:960AM | Plasma | Ok | no faults |
| 3 | 09:43:07:590AM | Plasma | Ok | no faults |
| 4 | 10:17:07:680AM | Plasma | Ok | no faults |
| 5 | 10:40:19:880AM | Plasma | Ok | no faults |
| 6 | 10:58:56:560AM | Plasma | Ok | no faults |
| 7 | 11:13:04:700AM | Plasma | Ok | no faults |
| 8 | 11:31:33:120AM | Plasma | Ok | no faults |
| 9 | 11:52:56:270AM | Plasma | Ok | no faults |
| 10 | 12:16:29:690PM | Plasma | Ok | no faults |
| 11 | 12:38:30:810PM | Plasma | Ok | No faults |
| 12 | 12:58:10:900PM | Plasma | Ok | No faults. |
| 13 | 01:13:57:570PM | Plasma | Ok | No faults |
| 14 | 01:31:21:760PM | Plasma | Ok | No faults |
| 15 | 01:48:00:780PM | Plasma | Ok | No faults. |
| 16 | 02:04:35:340PM | Plasma | Ok | No faults |
| 17 | 02:25:23:500PM | Plasma | Ok | No faults. |
| 18 | 02:45:13:640PM | Plasma | Ok | No faults. |
| 19 | 03:01:48:830PM | Plasma | Ok | No faults |
| 20 | 03:21:16:650PM | Plasma | Ok | No faults |
| 21 | 03:47:12:810PM | Plasma | Ok | No faults |
| 22 | 03:59:49:090PM | Plasma | Ok | No faults |
| 23 | 04:24:51:240PM | Plasma | Ok | No Faults. |
| 24 | 04:36:31:600PM | Plasma | Ok | No Faults |
| 25 | 04:54:21:920PM | Plasma | Ok | N0 faults |