| Miniproposals | ||||||||||
|
| Operators | |
| Session leader(s): | Steve Wolfe |
| Physics operator(s): | Steve Golovato |
| Engineering operator(s): | Steve Fairfax,Vinny Bertolino |
| Engineering Operator Run Comment |
| ece alignment |
| Session Leader Plans |
| Physics Operators Plans |
| Session Leader Summaries |
| Entered: Jul 7 2004 03:49:09:743PM |
| Author: To Be Determined |
| Session Leader's summary for run 940505 Miniproposal 41 - Startup Cmod/1994 SL: Wolfe PO: Golovato EO: Bertolino, Fairfax The run plan was to start from the last shot frm yesterday, see what the new recon_vac and other reconstruction routines think of it and if we can make sense of the flux loop problem. If the new coefficients seem to make sense, then load the predictors based on them and proceed. Thereafter it should be be a matter of field tuning and pressure scans to try to get the current rise. The main goal was to get a plasma, as opposed to fizzles and duds. This goal was achieved. The first shot, loaded from 940504023, gave a reasonable looking null using the average replacement for F25, with Bz slightly positive according to the recon_vac signal. There was no breakdown. Tweaks to EF4 and Br_0 got 150kA shot on shot#3. The null looked great! Having gotten a current rise with predictors looking at bad signals, we proceeded to break it by changing to good predictors. After changing to the BP24_DE and BP26_DE Bp coils (see under magnetics problems below), we tried using a recalculated RCUR predictor (FLUXIHH) on shot 7. After several fizzles we reloaded shot 6 for #11, and that went well. We then proceeded to make a small adjustment to BR_0 offset (still using the average F25 value in the predictor), and recalculated the ZCUR predictors. We got good results on shots 15,17,&20. We concluded that 10000. worked well for an offset. We turned off the chopper for shot 12 and didn't turn it back on until 21, which fizzled. The present state of the PCS model is: 1. RCUR Predictor is old, physics inputs refer to _BC coils. Should be recalculated, but works as long as you don't look at it. This predictor uses F25. 2. ZCUR Predictor is new and seems to do about the right thing. 3. ZCUR Predictor is new, not tried on a real plasma yet. 4. ZXL - Old predictor. needs recalculation. 5. RXL - Old predictor. needs recalculation. 6. CLEARIN - Old Predictor. needs recalculation. 7. Br_0 - Predictor uses F24A and F26 to mock up F25. Could and should be recalc'ed, but is working for now. 8. ZCUR - ancient predictor, not used. 9-16. Rogowski or custom predictors, should be up to date. Problems: --------- Magnetics Problems: 1) F25 is definitely bad. Bob has replaced \MAGNETICS::MAG_FLUXES:FLUX_24 by the average of F24A and F26 in the magnetics tree. This seems to "fix up" MFLUX, and ought to work for FFIT and EFIT as well. I need to turn up the sigmas there. Ian has fixed his tree to give a low weight to F25 in MFIL and the FLUXIHH predictors. 2) BP24_BC and BP26_BC, which were hybrid inputs, were bad, and have been replaced by the corresponding _DE coils. NOT ALL PREDICTORS HAVE BEEN UPDATED to use the new loops. PCS Problems: 1) Using EDIT on a wire with a bad match between the PHYSICS predictor inputs and the\HYB_INPUT_NAMEs causes a bomb out. This means you can't use EDIT to look at the coefficients of a predictor if you change the names of any of the inputs it uses, as for example, when we changed from the BC loops to the DE loops. This problem cropped up late in the run and caused a mild panic such that we decided not to take the last shot out of concern that the PCS tree might be corrupt. 2) See Hybrid Problem #1 below. There is a kludgy workaround if Josh hasn't fixed the code yet. The problem is that the code to write the expression to the \PCS::P_nn:OUTPUT nodes is not getting called when it should. To force it to be called, you can go to the wire screen, change the name of the thing on the wire, hit the apply button, change it back, and hit the apply button again. You should see something like Writing IF_ERROR(10.*(\PCS::TOP.SEG_01:P_01 -.665)*\PCS::TOP.SEG_01:P_01:PREDICTOR:FACTO R * \HYBRID::TOP.CONFIG.WAVEGEN:REF_IP:P_TO_V_EXPR / \HYBRID::TOP.CONFIG.WAVEGEN :REF_IP:GAIN,0) to \PCS::TOP.SEG_01:P_01:OUTPUT in the IDL screen. Hybrid Problems: 1) The Wavegen IP reference wires are not putting out any output, even when we had current; nothing but bit noise on the first 8 wires, even with 300kA. Therefore, we are feeding back on ZERO on all the Ip referenced wires. Horne is working on the problem. The problem seems to be that the \PCS::P_01:OUTPUT expression in the model file seems to be left over from Gerry's run, and refers to REF_04 instead of REF_IP; furthermore the \PCS::P_05 points to REF_10 instead of REF_IP. Apparently Josh's code that deals with this is severely broken or is not being called at the right time. This is in fact a PCS problem. (see above). 2) On one shot, #18, the wavegen output on wire 16 (Ip) did not make it through the A-matrix summer, resulting in an error signal that was identical to the Ip signal (a fizzle). The wavegen output was digitized on the Pout CAMAC, and looked right. The problem appeared on the predictor scope as a predicted current that looked like the drawn Ip. As far as I know this was a one-time glitch. It needs to be watched. Power Supply Problems: There were no power system problems that impacted the run today. Conclusion - best run so far this year! |
| Physics Operator Summaries |
| Entered: Jul 7 2004 04:36:51:020PM |
| Author: To Be Determined |
| Physics Operator Summary - Run 940505 ------------------------------------------ Since the goals of this run were the same for the physics operator and session leader, read steve wolfe's (session leader) summary for this run. an additional summary from me would be redundant, and repetitious too. However, below is my shot-by-shot comments from the run. they also appear in the logbook under the topic physics operator. ------------------------------------------------------------------------- 01 ECDC in D from 6:30-7:40am pre-fill 1.2e-5 loaded last shot from yesterday (940504023) which broke down at 9ms and had max Ip to date (34ka). prefill was only 1.e-5 yesterday. we thought bz ramp might be too slow. the powers that be are still fighting the broken flux loop problems, trying to decide how we can correctly use the remaining data to reconstruct what is happening. no flash. all power supplies worked. 02 try making bz more negative, make zero crossong later, 8 instead of 2 ms. lowered ef4 by 22.5 volts, -997.5 to -1020. no flash. all power supplies worked. bz looks reasonable, br a little positive. 03 try lowering br, change br_0 offset from -2e-3 to -3e-3 mT. flashed! breakdown at 7 ms, 150 ka, 55ms. h_alpha says plasma started well centered up-down, then went up. 04 repeat. no flash. everything looks like previous shot. 05 try raising fill at little, 1.3e-5 from 1.2e-5. flashed, no Ip rise. beter reduce gas. ----------------------------------------------------------------------------------------------- two b-c Bp loops are not working, run delayed for cell access to change connections to use equivalent Bp loops from d-e ----------------------------------------------------------------------------------------------- 06 lower fill to 1.1e-5. another plasma, breakdown at 9ms, 80ms, 280ka. 07 repeat. recalculated predictors for rcur correcting for bad flux loops, swapped Bp coils. there had been cluges to account for bad flux loops up to now. will do zcur next shot if this works. no flash. 08 added -5000. offset to rcur. try again. breakdown but no current ramp. bz ramp not right. 09 went wrong way, try +5000. no flash. bz, br look fine, try more gas, 10 fill to 1.2e-5. no breakdown. 11 try reloading shot 006, which worked. this worked again, 200ka, 70 ms. 12 repeat without chopper, which according to horne may be screwing us up when it comes on at about 60ms. we are also gettering so we may not have enough gas. no breakdown, raise gas. 13 fill to 1.3e-5. breakdown, no current rise. perhaps a classic fizzle with too much gas. 14 fill to 1.2e-5. no breakdown. maybe br is a little high? 15 drop br offset to -4.e-3 from -3.e-3. also recalculated algorithms for slow zcur (wire 2). zcur feedback coes on at 37.5ms, doesn't affect breakdown. good shot. 330ka, 170ms. looks like an injection at 135ms killed it. 16 added offset of 6000 to zcur (wire 2). when feedback comes on, thinks plasma is low when it is actually high, plasma almost lost up at 60ms on shot 16 when it finally kicks in. breakdown, no current. 17 try again. plasma. 220ka, 160ms. 18 fill to 1.3e-5. increase offset on zcur2 to 10000. late breakdown 12 ms. seem to have every other shot symdrome. 19 recalculated algorithms for fast zcur (wire 3). this will do nothing this shot since chopper is off. also set offset to 10000. breakdown, no plasma. 20 lower fill to 1.2e-5, try again. plasma, 308ka, 165ms. 21 what do we do to get two in a row? turn chopper back on. eliminated bump at 50ms in slow zcur prop gain. breakdown, no plasma. |
| Session Leader Comments | |||
| May 5 1994 08:41:53:480AM | Steve Wolfe | Run Plan for 940505:
SL:wolfe PO:Golovato EO:Bertolino/Fairfax Start from the last shot, see what the new recon_vac and other reconstruction routines think of it and if we can make sense of the flux loop problem. We also need to keep an eye on F23. If the new coefficients seem to make sense, then we will load the predictors based on them and proceed. Thereafter it will be a matter of field tuning and pressure scans to try to get the current rise. I have the feeling we're really flying blind! | |
| May 5 1994 09:26:55:210AM | 940505001 | Steve Wolfe | Shot 940505001
Start from 940504023. P=1.2e-5 Result: No flash. Null looks fair, with F25 replaced. Slightly positive Br, Bz hangs around near 0. |
| May 5 1994 09:41:37:070AM | 940505002 | Steve Wolfe | Shot 940505002
Make EF4 more negative by 22.5 A (to -1020) Result: No flash. Null with MFLUX looks real good. |
| May 5 1994 10:10:06:550AM | 940505003 | Steve Wolfe | Shot 940505003
Change BR_0 offset from -2mT to -3mT. Result: Got a flash! Got current - 150 kA , lasts 60msec. Null looks beautiful. |
| May 5 1994 10:16:45:600AM | 940505004 | Steve Wolfe | Shot 940505004
Repeat Result: Nothing. The null looks "the same" . Maybe the alternate shot effect? Granetz/Horne report there are problems on the BP coils, including some in the BC array that go to the hybrid. |
| May 5 1994 06:05:08:310PM | 940505005 | Steve Wolfe | Shot 940505005
Repeat at 1.3e-5 pressure. Granetz is changing the \MAg_processed:flux_24 to use the average of the adjacent loops. Result: GOt the flash back, looked short on the TV. It was a fizzle? Well we know we can do it, anyway. We need to deal with the dead Bp coils, so we'll open the cell and take care of that. The bad coils are BP24_BC and BP26_BC. replace from the DE arrays. What the hell happened to all these magnetics???!!! |
| May 5 1994 12:54:27:920PM | 940505006 | Steve Wolfe | Shot 940505006:
Horne and Murray have switched the BP24_BC and BP26_BC at the hybrid inputs to the corresponding _DE signals. The Magnetics CAMAC should still read the right loops. Changing the HYB_INPUT_NAMES on 53 and 55 to reflect the change. They accidentally pulled the rogowski (56) while they were working, but think they got it back ok. On shot 3, the rogowski input to the hybrid was reading low (100kA vs 150kA). The only new thing is the fizzle detector, so we'll unplug it for shot 6. We'll take a shot with all set up the same, and then do the new predictor calculations. Lower fill to 1.1e-5 Result: Good flash - looked long like 3. JT reports the TV saturated on all frames; will put the filter in. 280 kA , lasted past 80msec. 3e19 nel_05. |
| May 5 1994 01:10:03:770PM | 940505007 | Steve Wolfe | Shot 940505007:
Recalculate RCUR predictor only. Doing one thing at a time. Leave offset=0. Result: Lost it. Null still looks good. Maybe gas after the last "good" one. |
| May 5 1994 01:18:00:240PM | 940505008 | Steve Wolfe | Shot 940505008:
From inspection of RCUR predictor scope, try RCUR offset=-5000. Note that it had been +10000 on the old predictor. Result: Fizzle. lasts till 55 msec but no current. The Bz turned around and went negative at 15-20msec. The RCUR predictor went from 5000 to 10000 at commutation. I think we went the wrong way. |
| May 5 1994 01:33:48:900PM | 940505009 | Steve Wolfe | Shot 940505009:
Change RCUR offset from -5000 to +5000. Result: Dud! Nothing obvious. |
| May 5 1994 01:47:51:690PM | 940505010 | Steve Wolfe | Shot 940505010:
Raise P to 1.2e-5 Result: No flash. Maybe a hair positive on Br, relative to earlier?? |
| May 5 1994 02:11:25:700PM | 940505011 | Steve Wolfe | Shot 940505011:
Reload shot 6 Result: Got a good flash again. 70msec, 200kA. We're gettering like crazy. Also got a PF crowbar. |
| May 5 1994 02:29:12:510PM | 940505012 | Steve Wolfe | Shot 940505012:
At Horne's suggestion, turn off the chopper for next shot. Result: No flash. Probably the gas needed changing. Br_0 still looks a bit positive? |
| May 5 1994 02:46:26:150PM | 940505013 | Steve Wolfe | Shot 940505013:
Raise the fill to 1.3e-5 Result: No flash on TV, which now has CIII filter, but there is a little on Halpha at around 10msec. |
| May 5 1994 03:02:08:880PM | 940505014 | Steve Wolfe | Shot 940505014:
Lower fill to 1.2e-5 and try again. Result: No sign of a breakdown on this one. |
| May 5 1994 03:26:38:390PM | 940505015 | Steve Wolfe | Shot 940505015:
Drop Br offset from -.003 to -.004 Recalculate the ZCUR predictor. It comes on around 38msec. Result: "A real shot!" - Steve Horne Lasted until 170msec, flattopped at 300kA Injections start at 130msec |
| May 5 1994 03:34:11:470PM | 940505016 | Steve Wolfe | Shot 940505016:
ZCUR offset from 0. to +6000. based on observed offset in predictor. Result: The usual post-plasma fizzle. Couldn't see it on the TV. |
| May 5 1994 03:51:09:400PM | 940505017 | Steve Wolfe | Shot 940505017:
If at first you don't succeed, try, try again... Result: 250ka , lots of injections. lasts till 160msec. nl5 says 6e19. Ugly, but better than nothing. |
| May 5 1994 04:02:23:680PM | 940505018 | Steve Wolfe | Shot 940505018:
Raise ZCUR2 offset to 10000. Raise P=1.3e-5 Result: No visible flash on TV. small breakdown. Horne and Paranay figured out that the reason the Ip hybird input was loaded by the fizzle detector is it was an inverting op amp fromt end. It has now been changed to a follower and will be reinstalled for shot 19. |
| May 5 1994 04:29:55:780PM | 940505019 | Steve Wolfe | Shot 940505019:
Recalculate ther ZCUR predictor on wire 3 and use offset=10000., but don't turn the chopper back on yet. Leave the gas alone at 1.3e-5 Result: No visible flash We've noted two problems with the hybrid. 1) On the good shots, eg 15, we have only bit noise on the Ip referenced P_out channels. This implies that the WAVEGEN Ip reference hasn't worked since Tinios' run. 2) On shot 18 , a fizzle, the Ip redictor on wire 16 looked like the drawn waveform, and the Aout waveform looks like the current. Furthermore the POUT digitzer sees the correct drawn waveform, so the summer in the A-matrix must not have worked on that shot?????? |
| May 5 1994 04:42:22:910PM | 940505020 | Steve Wolfe | Shot 940505020:
Lower gas to 1.2e-5 Fizzle detector back out. Result: Got a plasma. Lots of disruptions. INitial oscillation now goes down. |
| May 5 1994 04:48:29:410PM | 940505021 | Steve Wolfe | Shot 940505021:
Delete step in P gain on ZCUR 2 at 50msec (constant at 1.0) Turn chopper back on Result: NO visible flash - small breakdown. |
| Physics Operator Comments | |||
| May 5 1994 05:14:47:080PM | 940505001 | Steve Golovato | 01 ECDC in D from 6:30-7:40am
pre-fill 1.2e-5 loaded last shot from yesterday (940504023) which broke down at 9ms and had max Ip to date (34ka). prefill was only 1.e-5 yesterday. we thought bz ramp might be too slow. the powers that be are still fighting the broken flux loop problems, trying to decide how we can correctly use the remaining data to reconstruct what is happening. |
| May 5 1994 05:15:11:060PM | 940505001 | Steve Golovato | 01 ECDC in D from 6:30-7:40am
pre-fill 1.2e-5 loaded last shot from yesterday (940504023) which broke down at 9ms and had max Ip to date (34ka). prefill was only 1.e-5 yesterday. we thought bz ramp might be too slow. the powers that be are still fighting the broken flux loop problems, trying to decide how we can correctly use the remaining data to reconstruct what is happening. |
| May 5 1994 05:16:32:180PM | 940505001 | Steve Golovato | 01 ECDC in D from 6:30-7:40am
pre-fill 1.2e-5 loaded last shot from yesterday (940504023) which broke down at 9ms and had max Ip to date (34ka). prefill was only 1.e-5 yesterday. we thought bz ramp might be too slow. the powers that be are still fighting the broken flux loop problems, trying to decide how we can correctly use the remaining data to reconstruct what is happening. no flash. all power supplies worked. |
| May 5 1994 05:16:51:170PM | 940505002 | Steve Golovato | 02 try making bz more negative, make zero crossong later, 8 instead of
2 ms. lowered ef4 by 22.5 volts, -997.5 to -1020. no flash. all power supplies worked. bz looks reasonable, br a little positive. |
| May 5 1994 05:17:09:570PM | 940505003 | Steve Golovato | 003 try lowering br, change br_0 offset from -2e-3 to -3e-3 mT.
flashed! breakdown at 7 ms, 150 ka, 55ms. h_alpha says plasma started well centered up-down, then went up. |
| May 5 1994 05:17:27:130PM | 940505004 | Steve Golovato | 04 repeat. no flash.
everything looks like previous shot. |
| May 5 1994 05:17:42:050PM | 940505005 | Steve Golovato | 05 try raising fill at little, 1.3e-5 from 1.2e-5.
flashed, no Ip rise. beter reduce gas. two b-c Bp loops are not working, run delayed for cell access to change connections to use equivalent Bp loops from d-e. |
| May 5 1994 05:17:59:870PM | 940505006 | Steve Golovato | 06 lower fill to 1.1e-5.
another plasma, breakdown at 9ms, 80ms, 280ka. |
| May 5 1994 05:18:13:840PM | 940505007 | Steve Golovato | 007 repeat. recalculated predictors for rcur correcting for bad flux
loops, swapped Bp coils. there had been cluges to account for bad flux loops up to now. will do zcur next shot if this works. no flash. |
| May 5 1994 05:18:27:420PM | 940505008 | Steve Golovato | 08 added -5000. offset to rcur. try again.
breakdown but no current ramp. bz ramp not right. |
| May 5 1994 05:18:43:470PM | 940505009 | Steve Golovato | 009 went wrong way, try +5000.
no flash. bz, br look fine, try more gas, |
| May 5 1994 05:19:03:450PM | 940505010 | Steve Golovato | 10 fill to 1.2e-5.
no breakdown. |
| May 5 1994 05:19:14:350PM | 940505011 | Steve Golovato | 11 try reloading shot 006, which worked.
this worked again, 200ka, 70 ms. |
| May 5 1994 05:19:27:230PM | 940505012 | Steve Golovato | 12 repeat without chopper, which according to horne may be screwing
us up when it comes on at about 60ms. we are also gettering so we may not have enough gas. no breakdown, raise gas. |
| May 5 1994 05:19:40:350PM | 940505013 | Steve Golovato | 13 fill to 1.3e-5.
breakdown, no current rise. perhaps a classic fizzle with |
| May 5 1994 05:19:52:290PM | 940505014 | Steve Golovato | 14 fill to 1.2e-5. no breakdown. maybe br is a little high?
|
| May 5 1994 05:20:04:020PM | 940505015 | Steve Golovato | 15 drop br offset to -4.e-3 from -3.e-3.
also recalculated algorithms for slow zcur (wire 2). zcur feedback coes on at 37.5ms, doesn't affect breakdown. good shot. 330ka, 170ms. looks like an injection at 135ms killed it. |
| May 5 1994 05:20:16:990PM | 940505016 | Steve Golovato | 16 added offset of 6000 to zcur (wire 2).
when feedback comes on, thinks plasma is low when it is actually high, plasma almost lost up at 60ms on shot 16 when it finally kicks in. breakdown, no current. |
| May 5 1994 05:20:33:290PM | 940505017 | Steve Golovato | 17 try again. plasma. 220ka, 160ms.
|
| May 5 1994 05:21:05:120PM | 940505018 | Steve Golovato | 18 fill to 1.3e-5. increase offset on zcur2 to 10000.
late breakdown 12 ms. seem to have every other shot symdrome. |
| May 5 1994 05:21:37:680PM | 940505020 | Steve Golovato | 20 lower fill to 1.2e-5, try again.
plasma, 308ka, 165ms. |
| May 5 1994 05:21:49:820PM | 940505021 | Steve Golovato | 21 what do we do to get two in a row?
turn chopper back on. eliminated bump at 50ms in slow zcur prop gain. breakdown, no plasma. |
| Engineering Operator Comments | ||||
| Shot | Time | Type | Status | Comment |
| 1 | 09:10:07:830AM | Plasma | Ok | no faults |
| 2 | 09:31:45:080AM | Plasma | Ok | no faults |
| 3 | 09:44:51:940AM | Plasma | Ok | disruption induced comm faults |
| 4 | 10:07:01:950AM | Plasma | Ok | no faults |
| 5 | 10:26:08:680AM | Plasma | Ok | no faults |
| 6 | 12:32:02:860PM | Plasma | Ok | no faults |
| 7 | 12:56:42:240PM | Plasma | Ok | no faults |
| 8 | 01:11:30:530PM | Plasma | Ok | no faults |
| 9 | 01:22:17:060PM | Plasma | Ok | no faults |
| 10 | 01:41:23:770PM | Plasma | Ok | no faults |
| 11 | 02:02:53:800PM | Plasma | Ok | 200 kA, 70 msec plasma |
| 12 | 02:24:11:090PM | Plasma | Ok | Dud. |
| 13 | 02:38:03:550PM | Plasma | Ok | Dud. No faults |
| 14 | 02:58:04:100PM | Plasma | Ok | Dud. No faults. Testing new iso-amp in EF3 |
| 15 | 03:13:08:940PM | Plasma | Ok | 300 kA flat-top for 40 msec; 170 msec total |
| 16 | 03:28:38:610PM | Plasma | Ok | Fizzle |
| 17 | 03:41:11:260PM | Plasma | Ok | 250 kA for 170 msec |
| 18 | 03:57:38:450PM | Plasma | Ok | |
| 19 | 04:10:29:210PM | Plasma | Ok | dud. Engineering systems working well. |
| 20 | 04:32:27:710PM | Plasma | Ok | 300 kA ratty plasma; OH2L commutation fail at disruption; no damage |
| 21 | 04:45:01:830PM | Plasma | Ok | Fizzle |