Miniproposals

Miniproposal: 147 Date Filed: 7/15/1996 Title: Enhanced Startup Techniques (Higher Pre-charge and Parallel EF4) First Author: Ian Hutchinson Session Leader: Steve Wolfe (shots 1-25)

Miniproposal: 159a Date Filed: 7/29/1996 Title: Restart Operation: August 1996 First Author: Steve Wolfe Session Leader: Steve Wolfe (shots 1-25)

Engineering Operator Run Comment

Enhanced startup/hutch/wolfe/silva

Session Leader Plans

Physics Operators Plans

Session Leader Summaries

Entered: Jul 7 2004 03:49:15:883PM

Author: To Be Determined

Physop/Run Summary for 960809 SL: Wolfe/ PO: Hutch/WOLFE EO: Silva/Daigle MP#159A,147 Restart cmod, enhanced startup The idea of this run, as of all the runs this week, was to get good plasmas, preferably with high pre-charge and parallel EF4s. We have not yet succeeded and today's run was the least successful all week, despite the fact that we did get one reaonably good 780kA discharge to run past 300msec. ECDC in D2 was carried out since after the run on Wednesday, with the exception of about 5 hours on Thursday, until about 7AM on Friday. The TF crowbar circuit was put in an EMI-shielded box to try to get rid of the mysterious crowbars with EF4 in parallel. The hybrid gain on the EF1 supplies was changed from 1/50 to 1/75 to correspond to the actual state of the hardware (since 1/30/96!). The run got off to a good, if somewhat late, start, with Hutchinson at Physop and Wolfe at Session Leader, but it went downhill fast; fell off a cliff is more like it. Shot 1 was a TF only test, to make sure the crowbar board wasn't going to trip out or otherwise screw up. It had been tested yesterday to demonstrate that it would function as expected on a real overvoltage. Shot 3 was the first shot with all supplies firing, and we got a plasma which made it essentially to flattop and looked good until the EF1L supply tripped out. On shot 4 the TF tripped out again, so the shielded box trick didn't work. We punted the parallel EF4 until the crowbar problem is dealt with, and put them back in series. Shot 5 was a dud, then #6 got a fizzle. That was the last sign of light for the rest of the day! After considerable tweaking, Ian gave up after #10 and turned over Physop to Wolfe, who continued tweaking all day with no success. BR, BZ, and pre-puff were scanned, although obviously not far enough. At shot 18, I raised the TF to 6.2T, which didn't help. At shot 22 I reloaded a shot from Tuesday which had worked with the coils in series at 5.3T; it didn't work at 6.2 or (shot 23) at 5.2T. More tweaks to fields and pressure on #24,25 did not produce a flash. I think 19 duds in a row may be a record. ECDC was started after the run with no difficulty, and Terry removed the filter from the A-side camera so we could look for dangling objects in the vessel. There are none in the field of view of either camera, and it seems unlikely there is any such problem. The reason for the long string of duds is completely unknown. All week we have had a pattern of successful startups being followed by duds rather than fizzles, which is somewhat unprecedented in our experience. It may be that today was just a more severe manifestation of this phenomenon, rather than a new problem, but even that is unknown. Speculation includes the possibility that we have a bad flux loop signal which is corrupting the reconstructions, leading us to tune up to a poor null and perhaps also mis-driving the hybrid feedback; a permanent magnetization of some new internal, which would not show up on the loops but could screw up the fields; a gas problem, perhaps leading us to use too high a pressure at t=0+; and some unknown difficulty with high pre-charge startups. There is no obvious problem with any flux loops, but a subtle error could conceivably be sufficient. The pressure scans did not go below 18msec pre-puff, which is quite long by historical standards; on the other hand higher pressure are usually seen to produce fizzles not duds. We have no way of identifying a permanent magnetization effect, and have not been able to come up with a reason why high pre-charge should be intrinsically more subject to duds. In the absence of further insights into the cause of the problem, I recommend that the next run begin with a standard pre-charge startup and seriesed EF4 coils, i.e. go back to square one. A possible starting point is 960105009, a pre-boronization fiducial. Note that the null looked horrible compared to many today, but it broke down quite early, and had a good current rise. Note also that the engineers have to put back the old commutation resistances and change the cap bank voltage for the lower (standard) pre-charge currents. I also recommend that we change physops, as this week's victims are pretty much used up! Below are Ian's notes from during the run. Mine are in the logbook: ********************************************************************** Physop 960809 MP 159A/147 Enhanced Startup. Plan was ________ Run Plan for Friday, 960809 ---------------------------- MP# 159A Startup machine for August 1996 MP# 147 Enhanced Startup - high pre-charge and parallel EF4 SL: Hutch >Or maybe PO: wolfe >the other way around EO: Silva/Daigle ECDC overnight in D2. Since the tests of the new TF crowbar circuit were not done on Thursday, we will have to begin with those. If it appears that the parallel EF4 configuration is not causing TF crowbars, we will leave them that way. Otherwise, put them back in series. Assuming we have the EF4's in parallel, we will start with the same engineering setup as at the end of Wednesday (960807033). If we end up in series, we will want the setup from 960807012, (Physop:use PCS from shot 8?). Note that we want the OH crowbars at .12sec. In either case, gas setup as on Wednesday: A-side 20psi D2 B-top 5psi D2 B-sideL 4psi Ar Turn on John Rice's gas bottle. Spend the day trying to get good shots with the high pre-charge, and if possible, parallel EF4. We want to get enough full plasma shots to allow diagnostic alignment and get a pre-boronization fiducial baseline. If we succeed we will have completed all of the goals for this week except RF conditioning. Load from 960807025. Fill puff 21ms. Lower the early gas puff. set def cmod$models:[hybrid.tcl] @hybinit_dispatch type /tail=100 mds$root:[syserr]shape_bitbus_server.log 1 Pstest At 10 am. TF test looks OK. 2 Psdud No EF2, EF3, EFC, OHs. EF4s rather imbalanced U=1375, L=1458A. Consistent with the upper being rather warm. 3 Pla Would have been a good plasma. Power supplies died. Got 5.7T at breakdown, 5.9T at flat-top. Bdown at 6.3ms. Fast current rise: 560kA at .1s. CII .01. Problem is EF1 tripping out when it is being asked for a big negative current. EF4 gets through cross-over, but takes a long time (100ms). TF back to 5.3T A-side (K-side, PG3) to 30 psi. 4 Psdud TF trips as usual. The problem is not fixed. Series the EF4. 5 Dud EF4 was too low 1394/1370A because I did not tweak the voltage demand to compensate. EF2s were 2127 2186. EF2 P to -2140 from -2100. 6 Fiz Got EF2=2137,2209 bdwn at 6.3ms. CII .024 EF2 P to -2180A. 7 Dud Got 2183,2248. Gradient looks better. p_o=.06mtorr. prepuff to 23ms. 8 Dud Reload shot 3. Increase EF4 voltage to 140V to compensate for the seriesed coil need. 9 Dud Reload shot 8. prepuff to 26ms. EF4 to 1390 from 1405. 10 Dud Out of patience. Hand over to Wolfe.

Physics Operator Summaries

Entered: Jul 7 2004 04:36:56:427PM

Author: To Be Determined

Physop/Run Summary for 960809 SL: Wolfe/ PO: Hutch/WOLFE EO: Silva/Daigle MP#159A,147 Restart cmod, enhanced startup The idea of this run, as of all the runs this week, was to get good plasmas, preferably with high pre-charge and parallel EF4s. We have not yet succeeded and today's run was the least successful all week, despite the fact that we did get one reaonably good 780kA discharge to run past 300msec. ECDC in D2 was carried out since after the run on Wednesday, with the exception of about 5 hours on Thursday, until about 7AM on Friday. The TF crowbar circuit was put in an EMI-shielded box to try to get rid of the mysterious crowbars with EF4 in parallel. The hybrid gain on the EF1 supplies was changed from 1/50 to 1/75 to correspond to the actual state of the hardware (since 1/30/96!). The run got off to a good, if somewhat late, start, with Hutchinson at Physop and Wolfe at Session Leader, but it went downhill fast; fell off a cliff is more like it. Shot 1 was a TF only test, to make sure the crowbar board wasn't going to trip out or otherwise screw up. It had been tested yesterday to demonstrate that it would function as expected on a real overvoltage. Shot 3 was the first shot with all supplies firing, and we got a plasma which made it essentially to flattop and looked good until the EF1L supply tripped out. On shot 4 the TF tripped out again, so the shielded box trick didn't work. We punted the parallel EF4 until the crowbar problem is dealt with, and put them back in series. Shot 5 was a dud, then #6 got a fizzle. That was the last sign of light for the rest of the day! After considerable tweaking, Ian gave up after #10 and turned over Physop to Wolfe, who continued tweaking all day with no success. BR, BZ, and pre-puff were scanned, although obviously not far enough. At shot 18, I raised the TF to 6.2T, which didn't help. At shot 22 I reloaded a shot from Tuesday which had worked with the coils in series at 5.3T; it didn't work at 6.2 or (shot 23) at 5.2T. More tweaks to fields and pressure on #24,25 did not produce a flash. I think 19 duds in a row may be a record. ECDC was started after the run with no difficulty, and Terry removed the filter from the A-side camera so we could look for dangling objects in the vessel. There are none in the field of view of either camera, and it seems unlikely there is any such problem. The reason for the long string of duds is completely unknown. All week we have had a pattern of successful startups being followed by duds rather than fizzles, which is somewhat unprecedented in our experience. It may be that today was just a more severe manifestation of this phenomenon, rather than a new problem, but even that is unknown. Speculation includes the possibility that we have a bad flux loop signal which is corrupting the reconstructions, leading us to tune up to a poor null and perhaps also mis-driving the hybrid feedback; a permanent magnetization of some new internal, which would not show up on the loops but could screw up the fields; a gas problem, perhaps leading us to use too high a pressure at t=0+; and some unknown difficulty with high pre-charge startups. There is no obvious problem with any flux loops, but a subtle error could conceivably be sufficient. The pressure scans did not go below 18msec pre-puff, which is quite long by historical standards; on the other hand higher pressure are usually seen to produce fizzles not duds. We have no way of identifying a permanent magnetization effect, and have not been able to come up with a reason why high pre-charge should be intrinsically more subject to duds. In the absence of further insights into the cause of the problem, I recommend that the next run begin with a standard pre-charge startup and seriesed EF4 coils, i.e. go back to square one. A possible starting point is 960105009, a pre-boronization fiducial. Note that the null looked horrible compared to many today, but it broke down quite early, and had a good current rise. Note also that the engineers have to put back the old commutation resistances and change the cap bank voltage for the lower (standard) pre-charge currents. I also recommend that we change physops, as this week's victims are pretty much used up! Below are Ian's notes from during the run. Mine are in the logbook: ********************************************************************** Physop 960809 MP 159A/147 Enhanced Startup. Plan was ________ Run Plan for Friday, 960809 ---------------------------- MP# 159A Startup machine for August 1996 MP# 147 Enhanced Startup - high pre-charge and parallel EF4 SL: Hutch >Or maybe PO: wolfe >the other way around EO: Silva/Daigle ECDC overnight in D2. Since the tests of the new TF crowbar circuit were not done on Thursday, we will have to begin with those. If it appears that the parallel EF4 configuration is not causing TF crowbars, we will leave them that way. Otherwise, put them back in series. Assuming we have the EF4's in parallel, we will start with the same engineering setup as at the end of Wednesday (960807033). If we end up in series, we will want the setup from 960807012, (Physop:use PCS from shot 8?). Note that we want the OH crowbars at .12sec. In either case, gas setup as on Wednesday: A-side 20psi D2 B-top 5psi D2 B-sideL 4psi Ar Turn on John Rice's gas bottle. Spend the day trying to get good shots with the high pre-charge, and if possible, parallel EF4. We want to get enough full plasma shots to allow diagnostic alignment and get a pre-boronization fiducial baseline. If we succeed we will have completed all of the goals for this week except RF conditioning. Load from 960807025. Fill puff 21ms. Lower the early gas puff. set def cmod$models:[hybrid.tcl] @hybinit_dispatch type /tail=100 mds$root:[syserr]shape_bitbus_server.log 1 Pstest At 10 am. TF test looks OK. 2 Psdud No EF2, EF3, EFC, OHs. EF4s rather imbalanced U=1375, L=1458A. Consistent with the upper being rather warm. 3 Pla Would have been a good plasma. Power supplies died. Got 5.7T at breakdown, 5.9T at flat-top. Bdown at 6.3ms. Fast current rise: 560kA at .1s. CII .01. Problem is EF1 tripping out when it is being asked for a big negative current. EF4 gets through cross-over, but takes a long time (100ms). TF back to 5.3T A-side (K-side, PG3) to 30 psi. 4 Psdud TF trips as usual. The problem is not fixed. Series the EF4. 5 Dud EF4 was too low 1394/1370A because I did not tweak the voltage demand to compensate. EF2s were 2127 2186. EF2 P to -2140 from -2100. 6 Fiz Got EF2=2137,2209 bdwn at 6.3ms. CII .024 EF2 P to -2180A. 7 Dud Got 2183,2248. Gradient looks better. p_o=.06mtorr. prepuff to 23ms. 8 Dud Reload shot 3. Increase EF4 voltage to 140V to compensate for the seriesed coil need. 9 Dud Reload shot 8. prepuff to 26ms. EF4 to 1390 from 1405. 10 Dud Out of patience. Hand over to Wolfe.

Session Leader Comments
Aug 9 1996 08:10:58:390AM		Steve Wolfe	MP#159A,147 restart operation/enhanced startup SL: Wolfe PO: Hutch EO: Silva New crowbar circuit has been installed on TF, with hopefully more noise immunity. First we'll test that, then resume attempts at plasma where we left off on Wednesday. Since we've now gotten about 30 additional hours of ECDC, we may get better results.
Aug 9 1996 09:59:43:330AM	960809001	Steve Wolfe	960809001: TF-only test TF looks good.
Aug 9 1996 10:12:04:110AM	960809002	Steve Wolfe	960809002: Everything No OH's,EF2's,EF3,EFc. Did get EF1's,EF4, TF. EF4 imbalance is around 100A.
Aug 9 1996 10:48:51:760AM	960809003	Steve Wolfe	960809003: Try for Everything again. PLasma - to .34sec EF1L started to go bad at .1 sec, dies at .2sec Cii is at .01
Aug 9 1996 11:04:14:350AM	960809003	Steve Wolfe	960809003: Try for Everything again. PLasma - to .34sec EF1L started to go bad at .1 sec, dies at .2sec Cii is at .01 EF1L is getting slammed on hard at segment switch.
Aug 9 1996 11:30:07:650AM	960809004	Steve Wolfe	960809004: Raised plenum pressure on aside to 30psi. Go back to 5.3T. Engineers trying to fix ef1s We seem to have a N2 leak. TF crowbarred. It's not fixed yet. OH2U has coax resistance over 2uohm.
Aug 9 1996 11:54:37:790AM	960809005	Steve Wolfe	960809005: EF4's back in series. No program changes. dud EF4 needs adjusting, and the index looks too positive
Aug 9 1996 12:04:53:080PM	960809006	Steve Wolfe	960809006: Make ef2's 40A more negative. fizzle.
Aug 9 1996 12:17:55:590PM	960809007	Steve Wolfe	960809007: More ef2 tweak, same direction. dud. Null looks great. I say raise the fill.
Aug 9 1996 12:26:27:520PM	960809008	Steve Wolfe	960809008: raise the pre-puff 23msec. another dud! We've been here before
Aug 9 1996 12:47:46:940PM	960809009	Steve Wolfe	960809009: reload #3 and change ef4 voltage. dud
Aug 9 1996 01:11:18:010PM	960809010	Steve Wolfe	960809010: reload#8 again. More gas. ef4 15a less negative. dud. Ian gives up. Wolfe takes over.

Physics Operator Comments
Aug 9 1996 01:29:37:130PM	960809011	Steve Wolfe	960809011: Wolfe in as PHYSOP. Raise pre-puff from 25 to 29 msec EF4 from -1390 to -1370 Another dud. So far, no bwtter than Ian. I ended up with less Bz than before?
Aug 9 1996 02:03:44:270PM	960809013	Steve Wolfe	960809013: EF4 from -1350 to -1360. I don't believe this small a change does squat. Another dud. It opened the null up again. I don't like that index. Still too positive.
Aug 9 1996 02:22:18:220PM	960809014	Steve Wolfe	960809014: Continue Ian's game: increase ef2's from -2180 to -2230a Fairfax is trying something on the EF4 cabinets to reduce spikes on the TF bus; he is removing caps from the ef4 primary? No flash. Hardly any change in the null.
Aug 9 1996 02:33:52:490PM	960809015	Steve Wolfe	960809015: Try lowering br by 1mT, despite the fact that it looks good. dud. Now try raising it.
Aug 9 1996 02:51:09:890PM	960809016	Steve Wolfe	960809016: Now try raising br by 2mT (1mT above shot 14) Br0 offset is now .007. DUD again!@ It looks like positive Br. Next, I'll try raising the gas once more, then dropping the gas. Then I'll reload shot 6
Aug 9 1996 03:05:36:040PM	960809017	Steve Wolfe	960809017: gas to 32msec! Br back to .006 Still nothing. I'm going to raise the TF to 6T. (180kA)
Aug 9 1996 03:21:36:170PM	960809018	Steve Wolfe	960809018: TF to 180kA(PLC request), start at -1.1sec Still didn't flash. ok, start dropping the pre-pulse
Aug 9 1996 03:36:39:200PM	960809019	Steve Wolfe	960809019: TF still at 180kA(PLC request), start at -1.1sec. Gave 6.2. pre-puff from 32 to 26msec Still no flash. What the H... os going on? We have gotten oh1 crowbars on each shot at 6.2T
Aug 9 1996 03:55:04:110PM	960809020	Steve Wolfe	960809020: TF still at 180kA(PLC request), start at -1.1sec. Gave 6.2. pre-puff from 26 to 22msec No flash. That null looks bad. negative Br and Bz. Try fixing that instead of dropping the pre-puff
Aug 9 1996 04:07:07:970PM	960809021	Steve Wolfe	960809021: Br to 6.5mt EF4 from -1360 to -1330 Leave pre-puff at 22msec Still no flash. Take Ian's suggestion and go back to 960806032 (seg 1 only).
Aug 9 1996 04:25:48:900PM	960809022	Steve Wolfe	960809022: Load seg 1 from 960806032 (Ian's suggestion). Strange behavior, after the ack widget came up the load seg widget didn't go away, so I ok'ed it a second time. It brought in the segment though (br is at .006 and EF4 at 1400, pre-puff at 18msec.) I'll try this once at the higher field and then drop to 5.3T Still no flash. Null looks so-so, maybe a little negative Br, Bz too high early. See what it looks like at 5.3T
Aug 9 1996 04:42:36:430PM	960809023	Steve Wolfe	960809023: Back to 151kA on the TF. No other changes. No flash. The null looks horrid, nothing like 960806032. Of course, that was before granetz fixed the flux calculation? One or two more shots: raise Bz and the fill.
Aug 9 1996 04:53:13:700PM	960809024	Steve Wolfe	960809024: One or two more shots: raise Bz and the fill. EF4 to -1380. pre-puff to 23msec no flash. Br looks negative
Aug 9 1996 06:20:50:150PM	960809025	Steve Wolfe	960809025: Br to .007 No flash. Put it to bed.

Engineering Operator Comments
Shot	Time	Type	Status	Comment
1	09:57:06:050AM	Test	Ok	tf only
2	10:05:52:490AM	Test	Ok	all power supp.did not come on
3	10:26:53:860AM	Test	Ok	ef1 lower problems
4	11:14:58:530AM	Test	Ok	tf xbar
5	11:46:12:820AM	Test	Ok	no faults
6	12:00:30:550PM	Test	Ok	no faults
7	12:10:08:800PM	Test	Ok	oh 1 comm fault
8	12:23:45:300PM	Test	Ok	oh1 comm fault
9	12:42:14:030PM	Test	Ok	comm fault oh1
10	01:04:13:820PM	Test	Ok	oh1 comm fault
11	01:22:23:180PM	Plasma	Ok	OH1 Comm Fault
12	01:39:08:630PM	Plasma	Ok	Oh1 Comm Fault
13	01:53:43:080PM	Plasma	Ok	no faults
14	02:17:03:850PM	Plasma	Ok
15	02:32:12:230PM	Plasma	Ok
16	02:47:05:210PM	Plasma	Ok
17	03:03:37:360PM	Plasma	Ok
18	03:18:17:610PM	Plasma	Ok	Comm Fault Bypass Cab. OH2
19	03:33:39:380PM	Plasma	Ok	Comm Fault Bypass OH1U
20	03:49:19:100PM	Plasma	Ok	Comm Fault Bypass OH2U
21	04:04:57:730PM	Plasma	Ok	Comm Fault OH2 U Bypass
22	04:21:21:350PM	Plasma	Ok
23	04:36:00:810PM	Plasma	Ok
24	04:50:09:570PM	Plasma	Ok
25	04:59:43:970PM	Plasma	Ok