Miniproposals

Miniproposal: Date Filed: Title: First Author: Session Leader: Steve Horne (shots 1-21)

Engineering Operator Run Comment

startup

Session Leader Plans

Physics Operators Plans

Session Leader Summaries

Entered: Jul 7 2004 03:49:09:757PM

Author: To Be Determined

Session leader report, run 940510 Horne, Marmar. The plan for today was to continue on MP 41, startup. We set up for long shots (invert initially at 1.0 sec, later moved to 1.2 sec) In terms of good plasmas, the run was a success; however we didn't make much progress on MP 41 because, once again, a flux loop (f23) failed (shot 4) and we had to think our way around it. While Wolfe frantically recalculated predictors, we continued to run with a kludged-up predictor for BR, which didn't work. When Steve W was ready (shot 9) we reloaded the vacuum predictor B_r, which we need to get breakdown. After an offset tweak we were running again, and got 9 long shots in a row, IP around 550 kA, ramping down to typically 200 kA. We found ourselves in the peculiar position of having long fairly reproducible shots and being unable to properly control or diagnose them. These shots tended to ride on the outboard limiter; Steve G and Yuichi seemed only mildly concerned as the limiter temperature ratcheted up. (It never went much beyond 100 C; apparently 150 C is the discharge cleaning limit. The very last shot, however, may have damaged an RF probe. Starting shot 12 we began miniproposal 40 -- Amanda's set of constant condition shots. Meanwhile, Wolfe was trying to rebuild all the FLUXIHH predictors. On shot 16, F23 (which was still in use in the FLUXIHH predictors) failed in a different way (perhaps tried to work?) making an entirely different shot. So we bit the bullet, and recalculated all predictors which were switched in after we had substantial current -- all except rcur. Then recalculated rcur, yielding a dud due to unknown causes; 19 was also a dud. Shot 20 was reloaded from 17 and ran. 21 was an attempt at segment switching which revealed a bug in PCS. After the run Bob G & I discovered that f23 was failing due to an intermittent connection between the cable and the feedthrough. He and I (and later Joe Snipes) took turns wiggling, bending and cursing the connection. When we were finished the connection was no longer intermittent -- was consistently open instead. Fixing this is a job for a maintenance day. Shot list -- Move all inverts to 1.0 sec. fizzle detector on all but TMX supplies. initial fill at 1.2e-5 1 Worked.Plasma died at 0.340 sec. There's a nasty up/duwn wiggle at about 60 msec. This may be a symptom of mis-tuned slow Z control loop -- chopper not on here. Try reducing proportional gain from 2.5 to 2.0 for next shot. We seem to loose the plasma just after oh1 crossover, which is taking a long time because we have low demand to OH1 there -- the plasma current has hit the 400 ka flat-top. Try raising it to 500 ka next shot. 2 Fill to 1.0e-5. Plasma current flat-top at 500 ka. (Change didn't get in. Operator error or bug? Probably former.) "After dirty, disruptive shot, drop fill pressure" -- Earl. Result -- Very strange shot. The stack crowbarred 3 Raise fill to 1.1e-5. since shot was less disruptive. Result Best shot this year -- 800 msec. Flux loop 23 is sick, though. Since we're running ok, leave everything be for now. This will be a difficult loop to develop a software replacement for. Don't change fill. Not disruptive, so don't drop; no runaways, so don't raise. Really change IP to 500000 A. Try integral gain on slow Z on at .08 to try to damp some of the oscillation in z. 4 dud. Is it the bad flux loop? The weight of f23 and f5 are about -/+ 1.8 e-2. Typical signals at commutation are 1.3 weber. If f23 is producing only .7 weber (as measured for this shot) then an error in BR of about 8 millitesla. Of course, the post shot analysis is also messed up. 5. Try setting f23 and f5 to zero in phys predictor for b_r. This will reduce the sensitivity somewhat. Result -- dud. I threw together a routine to pull the new (modified) b_r predictor out of the hybrid and evaluate it with the digitized flux loop data. It says that b_r near t=0 on shot 5 was about -1 to -2 millitesla, and that on shot 3 (which ran) was -4.5 millitesla. Since the modified predictor is off by almost exactly the minus_one_volt value, I set it to zero for shot 6. 6. Dud. No good. 7 Adjust bR offset feedback. Dud. 8 further tweak. Dud. 9 Recalculated b_r to ignore bad loop. Dud. 10 dropped offset 1 negative to -5.e-3; worked! Lost vertical control at end of shot when OH2 hits current limit. 11 Drew ramp-down to start earlier; inncreased IP to -550000. Very nice long shot. Dies at 1.00 sec when power supplies invert. We can't diagnose these shots magnetically; the bad flux loop has very large weight in the gap control and xpoint control so we cant really control them either. 12 Committed to several shots with no substantial changes; begin miniproposal 40. Also: moved inverts to 1.2 sec; disable fizzle detector at .96 to try and ramp below 100 KA. Worked;plasma died at .988 sec -- repeat of last shot. 13 No changes; good shot. 14 Try ramping down faster to keep oh2L off it's current limit. Worked fine -- ho hum. Perhaps a bit dirtier. 15 Fine -- hard xrays though. Raise fill to 1.2e-5 next shot Still loosing vertical control late because oh2l kisses it's current limit. Try ramping IP down (more positive) even faster next shot. 16 Worked -- entirely different shot. F23 is different. Give up on MP 40. (Amanda claims she has enough for now.) Recompute clear_in,r_xl, z_xl, and both zcur predictors. Don't do rcur yet. Change tf invert to .7 sec to save some cooling time. 17 It all worked! Now recalculate rcur. 18 that broke it. No obvious field problems. Raise fill, try again. 19 Still no flash. Reload 17. (failed due to EF3 turning on early). 20 Plasma. 21 Have loaded segment 1 from shot 18 into segment 2, starting at 100 msec. Plasma, but subtle failures. On examination, bitwise pid gains loaded from second segment are trashed. This looks like a programming bug in the PCS code that assembles the segments.

Physics Operator Summaries

Entered: Jul 7 2004 04:36:51:193PM

Author: To Be Determined

Steve Horne's session leader summary pretty well covers it. I add the following thoughts on the run. It appears that we are over the worst of our dirty plasma problems, and every shot that had a null also got plasma current. The broken flux loops are a real pain, but the good news is that at least one appears to be broken outside the machine and so should be fixable. We should concentrate on fixing the loops that we can, and/or modifying the feedback algorithms to work with those we have, before we proceed with running a lot of shots. I believe the machine and power supplies are ready for real physics runs, once the magnetics get sorted out. Totals: 14 plasmas, 7 duds, 0 fizzles; all shots with deuterium fill and puff. From: PHYSOP::MARMAR Date: 10-MAY-1994 21:18:45 Description: Physics Operator's Run Summary: 940510 Steve Horne's session leader summary pretty well covers it. I add the following thoughts on the run. It appears that we are over the worst of our dirty plasma problems, and every shot that had a null also got plasma current. The broken flux loops are a real pain, but the good news is that at least one appears to be broken outside the machine and so should be fixable. We should concentrate on fixing the loops that we can, and/or modifying the feedback algorithms to work with those we have, before we proceed with running a lot of shots. I believe the machine and power supplies are ready for real physics runs, once the magnetics get sorted out. We may still have some divertor conditioning to do. Totals: 14 plasmas, 7 duds, 0 fizzles; all shots with deuterium fill and puff.

Session Leader Comments

Physics Operator Comments
May 10 1994 10:13:45:690AM	940510001	Earl Marmar	Continuing MP 041 '94 Startup Load shot 940506026, I_P -400000 PLasma; disruptive, radiative, .445 MA, .34 seconds
May 10 1994 10:45:40:280AM	940510002	Earl Marmar	I_P to -500000, lower pulse gas, fill to 1.0e-5, Prop gain on slow Z_Cur lowered from 2.5 to 2.0 Plasma. OH2L problem at about .2 s. .474 MA, .56 s, fizzle detector finally killed it at 100 kA.
May 10 1994 11:04:48:740AM	940510003	Earl Marmar	fill to 1.1e-5, Good shot. .82 seconds, .47 MA
May 10 1994 11:30:22:660AM	940510004	Earl Marmar	Raise I_P programming to -500000 Changed slow Z_cur integral gain moved earlier Dud. F23 flux loop appears bad.
May 10 1994 11:53:50:000AM	940510005	Earl Marmar	Take F23 and F5 out of Br feedback. Dud.
May 10 1994 12:44:08:910PM	940510006	Earl Marmar	Change offset on Br feedback. Dud
May 10 1994 12:44:16:920PM	940510007	Earl Marmar	Change offset on Br feedback. Dud
May 10 1994 01:40:17:860PM	940510008	Earl Marmar	Wolfe has put new Br algorithm into recon_vac. Change offset again. Dud
May 10 1994 01:42:10:820PM	940510009	Earl Marmar	Recalculate B_r predictor. Dud
May 10 1994 01:45:03:570PM	940510010	Earl Marmar	Change B_r offset by 1 mT. Good plasma.
May 10 1994 02:03:57:820PM	940510011	Earl Marmar	I_P to -550000; I_p ramp starts a little earlier. Current limit on OH2's raised to 25 kA. Good plasma. .6 MA; lasted to inversion at 1 second.
May 10 1994 02:17:49:750PM	940510012	Earl Marmar	Repeat. Good.
May 10 1994 03:14:30:480PM	940510014	Earl Marmar	Changed I_p rampdown slightly. Good plasma. Starting to see impurity problem during current rise (also true on shot 13).
May 10 1994 04:14:15:100PM	940510017	Earl Marmar	940510017: Wolfe taking over as PHYS_OP Recalculate algorithms for: ZXL - change factor from 4e-5 to 3e-5 RXL - change factor from 7e-5 to 4e-5 CLEARIN - factor remains at 3e-5 ZCUR (2) - factor remains at 4e-5, offset=10000. ZCUR (3) - factor remains at 4.e-5, offset=10000. TF invert moved to 0.7 sec Result: Looks good on the TV Ran 850msec. Gaps look better inside and out.
May 10 1994 04:32:25:080PM	940510018	Earl Marmar	940510018: Recalculate RCUR - factor remain at 4.e-5,offset at 10000. result: No plasma. No breakdown. No obvious difference in the early fields.
May 10 1994 04:34:57:050PM	940510019	Earl Marmar	940510019: Raise pressure to 1.3e-5 and try again result: No plasma. No breakdown.
May 10 1994 04:46:21:600PM	940510020	Earl Marmar	940510020: reload shot 17 Before running this shot we noted that on 19 EF3 was on at commutation and Bz was bad. result: Got plasma, looked like lots of xrays on the TV.
May 10 1994 05:08:17:310PM	940510021	Earl Marmar	940510021: INcrease gas puff between 10 and 100msec. Copy segment 1 from #18 into segment #2, and use segment #2 after 100msec. result: Ran but died at 380msec. Looks like it was way outside. FFIT says it's on the outer limiter.

Engineering Operator Comments
Shot	Time	Type	Status	Comment
1	09:47:10:920AM	Plasma	Ok	no faults
2	10:16:33:350AM	Plasma	Bad	oh2l comm fault
3	10:47:58:120AM	Plasma	Ok	no faults
4	11:16:34:000AM	Plasma	Ok	Dud; all eng. systems ok
5	11:36:54:860AM	Plasma	Ok	Dud; all engineering systems worked well
6	12:19:44:740PM	Plasma	Ok	Dud. Extended delay to program around another bad flux loop.
7	12:30:04:890PM	Plasma	Ok	Dud.
8	12:51:08:510PM	Plasma	Ok	Dud. 5 minute delay for problem in alternator excitation control circui
9	01:03:03:140PM	Plasma	Ok	Dud. All else OK
10	01:20:43:580PM	Plasma	Ok	good 500 kA plasma. Operator panic due to extended TF pulse length; hit
11	01:49:44:200PM	Plasma	Ok	Nice 500 kA plasma; 0.5 sec flat top, 1 sec total length
12	02:11:26:420PM	Plasma	Ok	Another nice 600 kA plasma
13	02:32:48:050PM	Plasma	Ok	Another 600 kA shot; OH2U commutation fail induced by plasma disruption
14	02:56:45:830PM	Plasma	Ok	another good shot;
15	03:22:52:040PM	Plasma	Ok	Another nice shot; Amanda is getting ECE position calibration
16	03:40:25:510PM	Plasma	Ok	Another good shot with exponential plasma current ramp-down
17	04:05:31:260PM	Plasma	Ok	Another good shot
18	04:18:53:220PM	Plasma	Ok	dud! Snapped the streak.
19	04:31:10:480PM	Plasma	Ok	Another Dud!
20	04:42:11:980PM	Plasma	Ok	got 600 kA shot back
21	04:59:07:910PM	Plasma	Ok	Left a good shot on the drum