Alcator C-Mod Run 951205 Information

Back to runs

Miniproposals
Miniproposal:067
Date Filed: 1/9/1995
Title:Plasma Radial Position Feedback to Maintain Constant Antenna Loading
First Author:Yuichi Takase
Session Leader:Steve Golovato (shots 1-43)

Operators
Session leader(s):Steve Golovato
Physics operator(s):Earl Marmar
Engineering operator(s):Frank Silva,Joe Daigle

Engineering Operator Run Comment
RF Feedback Control

Session Leader Plans

Physics Operators Plans

Session Leader Summaries
Entered: Jul 7 2004 03:49:14:413PM
Author: To Be Determined

Run summary - 951205

SL: Golovato
PO: Marmar
EO: Silva/Daigle

Run 951205 - Plasma Position Feedback

Goals - (1) check out and learn to use plasma position feedback to
control rf loading, keeping it constant to avoid trips at
h-mode transitions

(2) also, nitrogen puffing at a level that does not affect shot,
from inner wall using ninja system.


The run was quite successful. We got the position feedback going early and
moved on to H-mode studies, getting some of the best H-modes we've seen.
The initial problem was to determine the approriate gain in the A-matrix to
influence rcur. We had some guidance from last year, when -2500 was too much
and -400 too little. Starting on shot 7 we saw evidence of the feedback in
the rcur error signal. This allowed us to get the sign right. The sign
convention is as follows: when loading decreases, the feedback signal goes
more positive and this moves the plasma out (toward antenna) when the
hybrid gain is negative. It is important to remember that there is another
gain at the hybrid input which is set to -10. Finally, the zeroing of the
feedback signal to set the desired loading level is done by a bira arbitrary
function generator camac module. This signal moves the level higher the more
negative it is and the actual voltage out is one half of what you draw. Got
all that, there will be a quiz on monday. For shot 10 the gain was -3000
and the rcur followed the feedback signal, moving the plasma in and out.
Shot 10 was also the first N2 puffing shot. On shot 11 we went too far with
a gain of -5000. The plasma went vertically unstable when the feedback pushed
it away from the antenna too far (remembering that clearin is also maintained)
elongating it too mcuh. With this gain the feedback was unstable overshooting
and oscillating. Ian was not pleased. We decided that it is not really
necessary to push the plasma away from the antenna, we can deal with too much
loading. We will alter the feedback circuit so the feedback signal will have
only one sign (positive). The plasma can still be moved in and out but not
out beyond zero. We accepted our punishment of 4 fizzles, reduced the gain to
-2000, and started raising the power to get h-modes. Thus far we were not
overlapping the rf powers and kept e-port (which controls the feedback) at
800kw. We got h-modes starting on shot 18. We found that the feedback did not
keep the loading constant during h-modes but reduced the decrease sufficiently
to avoid rf trips during h-mode. This allowed us to get arguably the best
h-modes we've seen to date. D-port was tripping frequently but this is a
separate problem not related to loading. Contrary to conventional wisdom, our
h-modes get better at higher density and with more gas during the h-mode.
However they do begin to degrade with too much gas. Shot 29 was the best, a
steady-state h-mode through the entire rf pulse with very low level (grassy?)
elms. Reducing the gas during the h-mode shortened the h-mode (shot 28) and
more gas degraded them somewhat (shot 39). It seems that moly accumulation and
high radiated power are key elements in the quality of our h-modes. Therefore
we get better h-modes at higher density where moly is better shielded. We got
record stored energy for 800ka but h-factors were still about 1.4. We went to
900ka on shot 33 and more feedback gain (-2750) on shot 37. Finally on shot 40
we went to 1.05ma. The plasma disrupted when the rf turned off, too much
density and moly with not enough input power. Earl predicted this in advance
but still took the shot. We were again punished by 3 fizzles to end the day.
It seems recovery from disruptions, especially full current disruptions, is
more difficult now.

For information on engineering problems and shot statistics, see the physics
operator summary. For shot-to-shot comments, see the logbook under the
session_leader topic.

Physics Operator Summaries
Entered: Jul 7 2004 04:36:55:490PM
Author: To Be Determined

From: PHYSOP::MARMAR
Date: 5-DEC-1995 18:01:56
Subj: Physop run summary 951205

951205

EO: Silva/Daigle
PO: Marmar
SL: Golovato
PB: McCracken

MP67a RF loading position feedback with MP127 N_2 puffing on the back of the pig.

A successful day.

The RF loading feedback appears to work well, and we had many shots with
"steady-state" H-modes. Density scans indicated that the generally better
H-modes happened at higher densities, where the moly and radiated power were
lower. We finally settled on a value of -2750 for the contribution of the
RF_load signal to the RCUR estimator. Shot 11 demonstrated that -5000 was too
large, causing a large inward excursion, increasing kappa and leading to a
disruption. It has been suggested that the circuit be modified so that it calls
for excursions toward the antennas when the loading is too low, but ignores
loading that is higher than the nominal setting. Some adjustment of the maximum
allowed excursion may also be necessary.

Some effort was spent adjusting the early gas puffing to speed the density rise
and reduce early moly influx. From shot 26 on, this appeared to be quite
successful.

The nitrogen piggy back also appeared to get good data.

The TF cooling was working much better today compared to last week, with cooling times below 10 minutes on most shots.

1 PF test
2 load 951130019, increase prefill puff by 3 msec, add .5 mT B_R offset, add early gas puff, TF invert at 1.0, TF on PLC;
plasma, H-mode during RF (2MW); this was a fiducial-style shot.
3 repeat; plasma; lower RF power (1MW)
4 raise nl_04 to 1.2e20; plasma; finning and slightly disruptive on rampdown
5 -100. into RF loading RCUR feedback; fizzle
6 2 msec less pre-fill puff; plasma; H-mode on second RF pulse (1.2 MW)
7 -1000 on RF loading RCUR feedback; plasma; 2.5 mm jog on X-ray centroid at time of programmed jog on input 81
H-mode, 1.1MW RF
8 repeat; plasma; higher RF power; H-modes
9 plasma; H-Mode; 1.2 MW
10 N2 puff at ~.75 s; -3000 on RF loading RCUR feedback; plasma; disrupt on rampdown (500 kA)
11 remove 1 msec from pre-fill puff; -5000 on RF loading RCUR feedback; plasma; disrupt at .68 s (.8 MA); RCUR decreased, kappa
too high
12 remove 1 msec from pre-fill puff; -2000 on RF loading RCUR feedback; fizzle
13 repeat; fizzle
14 add 2 msec to pre-fill puff; N2 puff off; fizzle
15 add .5 mT to B_R; disrupt on rampdown (400 kA)
16 draw nl_04 down after 0.9 s; plasma; still ratty on rampdown, but made it to the end.
17 N2 back on; power supply dud; no PF's; breaker pulled
18 plasma; H-mode; looks like the feedback is doing the right thing
19 repeat; plasma
20 no N2; plasma; in and out of H-mode
21 N2 back on, start ~.8 s; H-Mode(s)
22 plasma
23 RF loading RCUR feedback turned off; plasma, multiple RF trips during H-modes
24 reduce nl_04 just before RF turn on; RF loading RCUR feedback on; N2 off;
25 fiducial; plasma; magnetics screwed up, so no efit.
26 reload 23; add early gas; raise nl_04 to 1.4e20; raise preprog gas voltage;
plasma
27 more early gas; more preprog gas voltage; refill A side plenum to 21 psi;
plasma; disrupt on rampdown (450 kA); long ELM free H-mode
28 turn gas off at 0.6 (program nl_04 to 7e19 at 0.601); plasma; in and out of
ELM-free; 2.9 MW
29 program nl_04 up to 1.9e20 after RF on; N2 on; nice long H-mode; 120 kJ
stored energy; disrupt on rampdown (530 kA)
30 repeat; RF tripped; plasma
31 increase nl_04 to 2.2e20 during RF; plasma; early funniness on density rise
(fast) and current rise (slow)
32 raise EF3 PLC limit to 12 kA, tweak early segment 2 gas (lower); plasma;
lower density during RF
33 increase Ip to .9 MA; plasma
34 turn off gas during RF (nl_04 to 7e19); fizzle
35 reduce prefill puff by 1 msec; plasma; RF tripped
36 retry; plasma; similar result; RF tripped
37 -2750 on RF loading RCUR feedback; fizzle
38 B_R offset .5 mT smaller (now .0065); plasma
39 raise density pre and during RF; N2 off; plasma
40 Ip 1.1 MA; disrupt at .9s
41 reload 29, raise Ip tp .9 MA; fizzle
42 decrease pre-fill puff by 2 msec; fizzle
43 decrease pre-fill puff by 2 msec; fizzle

Tally: 43 shots, 32 plasmas, 9 fizzles, 1 PS dud, 1 test

Session Leader Comments
Dec 5 1995 09:11:55:010AMSteve Golovatorun 951205

goals - (1) check outand learn to use plasma position feedback to
control rf loading, keeping it constant to avoid trips at
h-mode transitions

(2) also, nitrogen puffing at a level that does not affect shot,
from inner wall using ninja system.

Dec 5 1995 09:12:36:800AM951205001Steve Golovatoshot 951205001 - magnet test shot.

Dec 5 1995 09:12:56:660AM951205002Steve Golovatoshot 951205002 - fiducial shot,

Dec 5 1995 09:14:10:310AM951205003Steve Golovatoshot 951205003 - setting up rf for l-mode with around 7 ohms loading,
outer gap of 1.5-2cm.

plasma. still getting dithering, raise nl form 1.0 to 1.2

Dec 5 1995 09:16:27:330AM951205004Steve Golovatoshot 951205004 -

plasma. getting nl=1.1, loading about 6 ohms, this is ok.
set bira to down .05 volts to 2.95 to zero feedback.

Dec 5 1995 09:23:25:530AM951205004Steve Golovatoshot 951205004 -

plasma. getting nl=1.1, loading about 6 ohms, this is ok.
set bira to down .05 volts to 2.95 to zero feedback.
put +/- 0.05 volt wiggle in bira signal and -100 in A matrix.
lokk for level of error caused by wiggle to calibrate A matrix
level. this should not actually move plasma enough that we can see it.

Dec 5 1995 09:44:49:850AM951205005Steve Golovatoshot 951205005 -

fizzle.

Dec 5 1995 09:47:15:200AM951205006Steve Golovatoshot 951205006 -

plasma. bira comews in about 20ms earlier than programming says.
turning off bira in 1ms to avoid large offset at end of rf pulse.
see jog in feedbeack signal but not on rcur phys a-out. raise
a-matrix to -200 from -100 for next shot.

Dec 5 1995 10:19:12:450AM951205007Steve Golovatoshot 951205007 -

changed a-matrix to -1000 for this shot.
plasma. 0.2 volt offset at end of rf pulse (which shouldn't be there)
can be seen on rcur error signal, plasma moves about 2.5mm. we now have
a calibration. increased programmed wiggle to +/- 0.2v, remembering
now that the actual wiggle will be half of this or +/- 0.1v.

Dec 5 1995 10:20:44:360AM951205008Steve Golovatoshot 951205008 -

plasma. rf timing wrong by mistake.

Dec 5 1995 10:35:10:310AM951205009Steve Golovatoshot 951205009 -

plasma. can see plasma move at transition from - to + .1v (.2v change).
moves out (toward antenna) about 2mm when signal goes more positive.
this is correct direction since the siganl goes more positive when
loading decreases. increase a-matrix to -3000. form -1000. for next
shot.

Dec 5 1995 10:45:06:180AM951205010Steve Golovatoshot 951205010 -

plasma. can now see all the changes in the feedback signal in the
phys a-out rcur signal. gain looking good.

first shot with nitrogen puffing, got what was expected but see density
increase.

Dec 5 1995 10:56:48:540AM951205011Steve Golovatoshot 951205011 -

increase a-matrix to -5000 from -3000.

plasma. disrupted, too much feedback gain. go back to -3000 and
eliminate programmed wiggle, see if the loading is kept constant.

Dec 5 1995 11:00:32:180AM951205011Steve Golovatoshot 951205011 -

increase a-matrix to -5000 from -3000.

plasma. disrupted, pushed too hard, with clearin set, plasma
elongated too much.
too much feedback gain. go back to -2500 and
eliminate programmed wiggle, see if the loading is kept constant.

Dec 5 1995 11:09:23:120AM951205012Steve Golovatoshot 951205012 -

fizzle.

Dec 5 1995 11:14:09:070AM951205013Steve Golovatoshot 951205013 -

fizzle. a-matrix at -2000, not -2500.

Dec 5 1995 11:21:51:140AM951205014Steve Golovatoshot 951205014 -

fizzle..

Dec 5 1995 11:26:43:560AM951205014Steve Golovatoshot 951205014 -

fizzle..

Dec 5 1995 11:44:51:770AM951205015Steve Golovatoshot 951205015 -

lowered bira to 2.85 from 2.95 to better zero feedback for current
conditions.

plasma. feedback zeroed.

Dec 5 1995 11:46:17:270AM951205016Steve Golovatoshot 951205016 -

raised power to 1mw from .8mw to get h-mode.

plasma, no h-mode during e-port, raise power some more.
h-mode during d-port at 1,.2mw.

Dec 5 1995 11:59:09:820AM951205017Steve Golovatoshot 951205017 -

raised power to 1.5mw from 1.mw

dud. oh's pulled out.

Dec 5 1995 12:12:20:260PM951205018Steve Golovatoshot 951205018 -

plasma. 1.4mw, elm-free h-mode for 60ms. outer gap reduced from
16mm to 9mm, to maintain loading. loading dropped to 4 ohms from
7.5, which is less than usual for elm-free h-modes, where loading
can go below 3.

Dec 5 1995 12:34:48:920PM951205019Steve Golovatoshot 951205019 -

plasma. repeat of shot 18.

Dec 5 1995 12:37:29:830PM951205020Steve Golovatoshot 951205020 -

plasma. extended e-port to .9 sec, got train of elm-free, elmy,
elm-free h-modes. no interrupts. now add d-port power for next shot.
outer gap shows plasma motion toward antenna but xray centroid seems
to move out of phase with h-modes, away from antenna during elm-free,
toward during elmy.

Dec 5 1995 12:45:52:770PM951205021Steve Golovatoshot 951205021 -

plasma. 65ms elmfree h-mode, about the longest we've ever seen.
raising e-port power. d-port tripped early.

Dec 5 1995 12:56:45:830PM951205022Steve Golovatoshot 951205022 -

plasma. 1.8mw from e-port. d-port 1.1mw in lmode, .9 in h-mode.

Dec 5 1995 01:07:36:080PM951205023Steve Golovatoshot 951205023 -

feedback turned off using gate enable.

plasma. hmodes are shorter, lots of rf interrupts. outer gap
flat at 16mm. clearly the feedback helps, though it doesn't make
the loading constant. drops to 3 ohms, 40kv at

Dec 5 1995 01:16:29:580PM951205024Steve Golovatoshot 951205024 -

feedback back on. try reducing gas before rf pulse to reduce
density rise.

plasma. looks much like last shot, but with much fewer the rf trips.
h-modes not as good.

Dec 5 1995 01:29:02:160PM951205025Steve Golovatoshot 951205025 -

redo fiducial with more diagnostics.

plasma. go back to feedback studies.

Dec 5 1995 01:50:12:630PM951205026Steve Golovatoshot 951205026 -

raise density to nl=1.4e20.

plasma. 75ms elmfree h-mode. plasma movee about 5mm according to xray
centroid. variation of loading with outer gap seems less sensitive
in h-mode than in l-mode we still see a large decrease but not as much
as without feedback and we don;t get as many trips, usually we get none
on e-port. d-port has other problems which lead to crowbars.

density still not what was requested, only got 1.1e20 before rf.

Dec 5 1995 02:06:11:140PM951205027Steve Golovatoshot 951205027 -

plasma. long elmfree h-mode, 130ms, lost when d-port trips.
density stops steep rise when gas valve shuts off, stays
in elmfree h-mode with slower density rise. wp=114kj, highest
we've seen at 800ka, non-pep.

Dec 5 1995 02:12:46:710PM951205028Steve Golovatoshot 951205028 -

turned off gas at rf turnon, see effect on density rise.

plasma. 2.9mw. no trips. h-modes shorter, not as good as previous
shot. try keeping gas on instead of turning it off.

Dec 5 1995 02:25:25:670PM951205029Steve Golovatoshot 951205029 -

plasma. elmfree? or grassy elms for the whole rf pulse.
it wants more gas, not less. wp=120kj. 3mw, no trips.

Dec 5 1995 02:42:24:350PM951205030Steve Golovatoshot 951205030 -

plasma. lost d-port early, lost h-mode. in l-mode, loading is
higher than set point, feedback pushes plasma away from antenna
to 24mm.

Dec 5 1995 03:01:54:520PM951205031Steve Golovatoshot 951205031 -

raise density.

plasma. too much early gas. hmode not quite as good as shot 29.

Dec 5 1995 03:10:58:080PM951205032Steve Golovatoshot 951205032 -

tweaking early gas.

plasma. lose elmfree early, elmy the rest of time.

Dec 5 1995 03:29:16:200PM951205033Steve Golovatoshot 951205033 -

plasma. 900ka, 125kj. long elmy h-mode.

Dec 5 1995 03:36:40:070PM951205034Steve Golovatoshot 951205034 -

fizzle.

Dec 5 1995 03:46:06:700PM951205035Steve Golovatoshot 951205035 -

cutting gas during rf.

plasma. e-port port tripped early, screen overload. lost h-mode.

Dec 5 1995 03:53:44:030PM951205036Steve Golovatoshot 951205036 -

plasma. e-port tripped again.

Dec 5 1995 04:04:00:500PM951205037Steve Golovatoshot 951205037 -

raising feedback gain to -2750 from -2000, plasma not moving far
enough.

lowered e-port power a little.

fizzle.

Dec 5 1995 04:24:53:770PM951205038Steve Golovatoshot 951205038 -

plasma. hmode short, rf stayed on.

Dec 5 1995 04:32:44:290PM951205039Steve Golovatoshot 951205039 -

gas back on.

plasma. h-mode unitl rf trips out. almost elm free.

Dec 5 1995 04:46:49:880PM951205040Steve Golovatoshot 951205040 -

raise ip to 1ma.

plasma. 1.05ma. lost d-port early again.

Dec 5 1995 04:48:42:320PM951205041Steve Golovatoshot 951205041 -

last shot disrupted at e-port turnoff, lower ip to 900ka and
go back to shot 29.

fizzle after disruption.

Dec 5 1995 04:59:48:950PM951205042Steve Golovatoshot 951205042 -

fizzle.

Dec 6 1995 08:40:35:440AM951205043Steve Golovatoshot 951205043 -

fizzle. three fizzles for every full current disruption.

Physics Operator Comments

Engineering Operator Comments
ShotTimeTypeStatusComment
108:23:00:240AMTestOkoh test
208:40:11:840AMPlasmaOk
308:55:06:090AMPlasmaOk
409:07:17:100AMPlasmaOk
509:22:31:490AMPlasmaOk
609:34:43:080AMPlasmaOk
709:55:47:980AMPlasmaOk
810:06:34:180AMPlasmaOk
910:22:03:110AMPlasmaOk
1010:34:04:980AMPlasmaOk
1110:48:43:380AMPlasmaOk
1211:04:34:100AMPlasmaOkfizzel
1311:12:15:140AMPlasmaOk
1411:20:16:880AMPlasmaOk
1511:29:01:340AMPlasmaOk
1611:39:58:390AMPlasmaOk
1711:49:02:390AMPlasmaOk
1811:56:48:670AMPlasmaOk
1912:09:19:410PMPlasmaOk
2012:18:07:810PMPlasmaOk
2112:37:16:840PMPlasmaOk
2212:47:55:840PMPlasmaOk
2312:58:47:270PMPlasmaOk
2401:10:46:920PMPlasmaOk
2501:23:00:740PMPlasmaOk
2601:37:22:040PMPlasmaOk
2701:51:42:890PMPlasmaOk
2802:05:24:170PMPlasmaOk
2902:16:45:500PMPlasmaOk
3002:29:55:760PMPlasmaOk
3102:47:40:940PMPlasmaOk
3203:01:42:950PMPlasmaOk
3303:18:12:130PMPlasmaOk
3403:30:44:740PMPlasmaOk
3503:38:39:190PMPlasmaOk
3603:49:39:180PMPlasmaOk
3704:02:24:640PMPlasmaOk
3804:09:57:490PMPlasmaOk
3904:21:34:050PMPlasmaOk
4004:34:23:390PMPlasmaOk
4104:47:04:770PMPlasmaOk
4204:55:31:310PMPlasmaOk
4305:03:47:580PMPlasmaOk