Miniproposals

Miniproposal: Date Filed: Title: First Author: Session Leader: Steve Golovato (shots 1-19)

Engineering Operator Run Comment

Plasma run.

Session Leader Plans

Physics Operators Plans

Session Leader Summaries

Entered: Jul 7 2004 03:49:07:680PM

Author: To Be Determined

Run Summary for Friday, 6/4/93 (revised) Physics Operator/Session Leader:wolfe Assistant Physics Operator: golovato Engineering Operator: Daigle/Gwinn The goals for this run were to continue the effort to improve startup reliability, following up on Ian's run yesterday, and to do some more work on position control. The starting point was 930603028. With respect to startup improvement, we planned to carry out two studies: a scan of pre-fill pressure, and tweaking of OH1 and EF3 pre-programmed voltages to modify the current evolution between 10 and 25 msec (the Hutchinson Hesitation). With respect to position control and evolution of the discharge at later times (i.e. after 50 msec), one study was planned: varying the pre-programmed voltages on the EF3, OH1, OH2, and perhaps EF2 supplies to make the feedback demands smaller and directly observe the results of voltage changes on the plasma evolution rather than relying on the feedback loop. The feedback gains were left at the values set previously, namely: IP_RCUR: PID = [0.2,0.,0.] from 0.05 to 0.3 sec M =[-9.958589e-03, -1.985415e-03,-1.985415e-03,5.042325e-02] on [ OH1, OH2U, OH2L, EF3 ] IP_ZCUR: PID = [0.1,0.,0.] from 0.05 sec to 2 sec M = [ 2.904189e-03,-2.875147e-03] on OH2U/L Note that the small (1%) assymetry in the ZCUR controller was not noticed during the run, and was inherited from the starting point (as were all these values). In addition to the position control, which was intended to be set to relatively weak level, BR_0 feedback was left on with the following parameters: BR_0 PID = staircase to 1.0 between -1.5 and -1.0, drop to 0.48 at 0.05sec M= [-1536.71,1563.67] on OH2U/L Again, I don't know the origin of the slight assymetry in this nominally anti-symmetric controller. Power supply problems (see below) appeared on the first shot of the day, and were not resolved until shot 7, which did not occur until after 14:00. We were therfore only able to complete the first startup study, the fill pressure scan, and make one variation in supply voltage pre-programming. Some initial field tweaking was carried out to optimize (?) the field null after the power supply problems were resolved. In particular, for shot 11 the EF4 current was increased 50a to raise Bz, and the PID on the IC_OH2U&L was reduced by about 1/3 from -1 sec on to give greater weight to the BR_0 feedback. Also, on shot #7 we decreased the pre-programmed inversion voltage on the EF2's from -870 to -600 to accomodate the change from 3 series TMX's to two; the response of the upper and lower supplies to demands outside the available range was observed to be different, with the upper supply staying in inversion too long. Pressure was scanned up to 7e-5 and down to 4e-5 Torr between shots 12 and 19. The two shots at 7e-5 both fizzled, i.e. the current never rose above 20ka. Breakdowns were at 5.5msec, slightly later than at the lower pressures. The lower pressure shots all broke down and took off, except for one of the three shots at 4e-5 (17) on which the fields may have varied a bit. However, all the "successful" shots at 4e-5 were runaway-dominated discharges with horrendous hard xray emission. One of these resulted in the demise of one of Yuichi's probes on the outboard limiter. Increasing the gas puff (shot 17-19) did not significantly modify the runaway problem. It became apparent that the gas-fill and inner-wall recycling is completely controlling the early (and most of the late) density. Furthermore, the maximum current level obtained was also strongly correlated with the fill pressure, though not monotonically! The radial field during the 20-50msec time interval exhibited some variation, which may partially account for the differences in the current rise rates. Only one variation in voltage programming in the late phase of the discharge (>75msec) was made. Comparable shots are #11 and #19, both with a fill pressure of 6e-5. Unfortunately, the current rise in these two shots is completely different from the beginning, while the voltage change was made only after 75msec, so no conclusions can be drawn. Essentially all the positional variation in the non-fizzle shots is attributable to changes in plasma current. A simplistic post-run analysis of the controller used for RCUR indicates that it will apply a negative loop voltage while trying to move the plasma out, which would also result in decreasing the current, which would tend to move the plasma back in! The controller also seems to have the property of oblating the plasma as it tries to move it out (i.e. it generates a more positive decay index). Note: I had the sign wrong in the original version of this report. This effect may actually be observed in the flux reconstructions of shot#18. It was also noticed that the RCUR predictor has a substantial positive offset (~5000 A-m) which appears to be constant in time and is likely due to offset voltages on the hybrid inputs or in the DACs. Statistics/Score-card: ---------------------- 7 with current >150kA and longer than 100msec 11 fizzles (5 while chasing power system problems) 1 no breakdown (EF2L failed to pulse) -- 19 total shots Conclusions: ------------ Breakdown and subsequent current rise is feasible at fill pressures down to 4e-5 and probably lower, but this mode of operation leads to bad runaway discharges. Careful adjustment of Bz at breakdown and control of Br seem to improve reliability of the current rise somewhat. The B-top valve arrangement is apparently much too slow to affect the plasma density with on a 10msec time scale. The controller used for RCUR should be analyzed and if necessary adjusted to produce more of a pure vertical field. A quick and dirty analysis suggests having the coefficient on EF3 and doubling that on the OH2's. Problems: --------- OH2U supply had a problem which appeared to be a false current limit coming in at about 14kA, with the PLC limit set to 18 or 18.5kA. This problem vanished on shot 6, after Daigle and company had set up to look at the PLC line and the IOC had been lowered to 50% of rated current. This problem is probably an intermittant fault on the control board that is still there, and it should be found and fixed ASAP. EF2L Supply A blew three SCRs and "rearranged" its transformer winding; it had to be removed from the circuit. One supply was also removed from the EF2U set to avoid problems with assymetries. The hybrid tree was modified to correspond to the new configuration. Since we seem to be able to operate with only two TMX supplies in series on each of the EF2s at these low voltages, I recommend that the D&E supplies be put back on EF4 after the EF2L-A is repaired or replaced. The EF2U supply was noted to have a different response to its demand signal than EF2L, particularly after a demand that was too high to achieve. Apparently this behavior had been observed previously, but not fixed; it sounds like an accumulating integral error problem, as had been found on other supplies. It was avoided in this run by restricting the demands more carefully, but this would seem to be a fixable hardware problem. The EF2L regulation seems to behave more like it should.

Physics Operator Summaries

Session Leader Comments
Jun 4 1993 09:03:35:610AM		Steve Golovato	930604 run. wolfe session leader. start from shot 930603028 (last good shot yesterday). plan for day is : (1) positionn control late in shot: make preprogramming resemble more closely what feedback is trying to do so feedback doesn't have to work so hard. (2) address reproducibility by gas scan and tweaking voltages early in shot during current rise (mainly oh1 and ef3). try to get rid of hesitation in current rise between 10-20ms.
Jun 4 1993 09:15:35:920AM	930604001	Steve Golovato	shot 930604001. same as shot 930603028. comutation resistors: ef1u\l 129mohm oh1 66mohm oh2u\l 13 mohm ef2 ballast 45mohm fizzle. oscillations on oh2u.
Jun 4 1993 10:59:56:160AM	930604002	Steve Golovato	shot 930604002. repeat previous shot (same as 930603028) except raise current limit on OH2U by 500A. oh2u acted as if is had too low a current limit (when compared to oh2l). ef2l (which also oscillated on shot 001) had three bad scrs in EF2L A-supply. This failure apparently occurred early in shot 930603029 (last shot yesterday). Transformer winding in ef2l A supply shifted, will jumper out this tmx (and one on upper for balance) and continue. Output gains for ef2u\l changed from 1/150 to 1/100 to get same voltage from 2 supplies. Maximum voltage we asked for was -650, which is a little more than 2 supplies can provide, so we may not get to maximum. Delay of a few hours to do this.
Jun 4 1993 11:59:25:020AM	930604002	Steve Golovato	shot 930604002. delay much less than a few hours. no plasma. no ef2l. contactor to ef2l A-supply open, needs to be closed even if supply out of circuit. oh2u looks better, less ripple. h_alpha and z_meter data not acquired, camac problem being fixed.
Jun 4 1993 12:22:26:870PM	930604002	Steve Golovato	shot 930604003. slight flash. same as shot 001 on oh2u. ef2 currents different after comutation, seems to be 50 ms oscillation. oh2u current limit back to 18ka.
Jun 4 1993 12:28:49:380PM	930604002	Steve Golovato	shot 930604004. again slight flash. oh2u still a problem. limit back up to 18.5ka for this shot. checking plc net shotto see if the correct current limit is being sent to oh2u supply.
Jun 4 1993 01:23:11:800PM	930604005	Steve Golovato	shot 930604005. same as 004. plc current limit is ok, problem must be in supply. may have been looking at wrong plc signal. try again looking at "correct" one. on plc, oh2u is oh2l and vice versa.
Jun 4 1993 02:14:47:480PM	930604006	Steve Golovato	shot 930604006. lowering ioc (hardware overcurrent protection) on oh2u from 100% of rated current to 50% (25ka). if software current limit is not working correctly, it might decide to limit high rather than low. fizzled but oh2u worked. ef2's the problem, asking for more than the supplies (2 not 3 tmx's each) can give. changed from -870 volts at t=0.0 to -600 volts.
Jun 4 1993 02:23:14:580PM	930604006	Steve Golovato	shot 930604007. finally a good shot.
Jun 4 1993 02:42:16:540PM	930604008	Steve Golovato	shot 930604008. adding current feedback on ef1u/l at 60ms, gain=1. lowering pre-fill from 6.e-5 to 5.e-5. fizzled. retry.
Jun 4 1993 02:43:14:710PM	930604009	Steve Golovato	shot 930604009. fizzled again.
Jun 4 1993 02:44:38:090PM	930604009	Steve Golovato	shot 930604009. fizzled. vertical field may be a little too negative.
Jun 4 1993 03:07:57:480PM	930604009	Steve Golovato	shot 930604010. raised gas back up to 6.e-5. lowered ef1u/l programming to 1200a at 60ms and 1000a at 100ms, closer to what we're actually getting. fizzled again, even though gas back up. vertical field too negative. ian and bob claim plasma is vertically unstable, which wuld make sense if the plasma is toward the inside.
Jun 4 1993 03:29:50:630PM	930604011	Steve Golovato	shot 930604011. raised ef4 current at t=0.0 from 650 to 700. reduced pid gain on oh2 currents from 3 to about 1.8 so it works less hard on current feedback relative to the radial field feedback. worked.
Jun 4 1993 03:49:18:450PM	930604011	Steve Golovato	shot 930604012. changing pre-programming on oh1 and ef3 to make it better reflect what the feedback is asking for, i.e. -20 volts on oh1 from 75ms on and +100 volts on ef3 from 75ms (ef3 pre-programming extended out from 200ms to 350ms. gas fill lowered to 5.e-5 curent lower - 170ka compared to 220ka on previous shot, lower from 50ms on. first disruption at 90ms, followed by 50ms of oscillatory motion, followed by repeated disruptions through 250ms. position further in presumably due to lower current.
Jun 4 1993 03:57:36:960PM	930604013	Steve Golovato	shot 930604013. repeat of last shot to see if current remains lower. good shot. very similar to last shot. first disrupt a little later, fewer oscillations.
Jun 4 1993 04:10:19:390PM	930604013	Steve Golovato	shot 930604014. raise pre-fill to 7.e-5. fizzled. try it again. breakdown later. was earlier on shots at 5.e-5 compared to 6.e-5.
Jun 4 1993 04:39:57:360PM	930604016	Steve Golovato	shot 930604016. broke down earlier, lots of disruptions even during ramp, hard x-rays saturated, more current (280ka).
Jun 4 1993 04:56:12:240PM	930604017	Steve Golovato	shot 930604017. pulsed gas was 12-50ms, extended to 12-100ms at same level (60 volts), to supress hard x-rays which began at about 50ms. fizzled. Bz is tiny bit more negative at 20msec, and Bz' a little more Positive; this would be consistent with EF3 coming on a little harder?
Jun 4 1993 05:10:34:540PM	930604017	Steve Golovato	shot 930604018. runaway city. wild oscillations. hot spot on rf limiter.
Jun 4 1993 05:26:32:940PM	930604019	Steve Golovato	shot 930604019. pre-fill to 6.e-5. good shot, no runaways. first disruption 105ms, 215ka. jumps in vertical position following each disruption. no sawteeth. max nl 6.e19. moving in for whole shot.

Physics Operator Comments

Engineering Operator Comments
Shot	Time	Type	Status	Comment
1	09:02:17:270AM	Test	Bad
2	11:48:07:710AM	Plasma	Ok	Plasma shot no faults
3	11:58:58:990AM	Plasma	Ok	plasma shot oh2 current program not correct.
4	12:19:16:220PM	Plasma	Ok	Plasma shot OH2l problem
5	12:30:40:350PM	Plasma	Bad	oh2u limited at 14ka
6	02:02:23:800PM	Plasma	Ok	plasma shot no faults
7	02:13:44:050PM	Plasma	Ok	Plasma shot No faults.
8	02:24:31:120PM	Plasma	Ok	Plasma shot No faults
9	02:36:35:960PM	Plasma	Ok	Plasma shot No Faults.
10	02:55:06:980PM	Plasma	Ok	Plasma shot no faults.
11	03:10:51:060PM	Plasma	Ok	210 ka
12	03:33:57:170PM	Plasma	Ok	170 KA
13	03:50:55:000PM	Plasma	Ok	170 KA
14	04:01:21:320PM	Plasma	Ok	PLASMA SHOT NO FAULTS
15	04:13:17:140PM	Plasma	Ok	PLASMA SHOT NO FAULTS
16	04:29:50:280PM	Plasma	Ok	PLASMA SHOT NO FAULTS
17	04:42:48:410PM	Plasma	Ok	PLASMA SHOT NO FAULTS
18	04:57:26:150PM	Plasma	Ok	PLASMA SHOT NO FAULTS
19	05:11:00:510PM	Plasma	Ok	GOOD SHOT