Alcator C-Mod Run 1050311 Information

Back to runs

Miniproposals
Miniproposal:405
Date Filed: 1/18/2005
Title:L-mode Momentum Confinement Time De-termined from Fast SSEP Change
First Author:John Rice
Session Leader:Ed Synakowski (shots 1-27)

Operators
Session leader(s):Ed Synakowski
Physics operator(s):Steve Wolfe
Engineering operator(s):Bill Parkin,Bill Byford

Engineering Operator Run Comment
MP#405 L-mode momentum confinement time determined from fast SSEP change

Session Leader Plans
Entered: Mar 10 2005 05:54:20:823PM
Author: John Rice
MP 405

L-mode Momentum Confinement Time Determined from Fast SSEP Change

set up vanilla shot, 0.8 MA, 5.4 T, nl04 1.0 in LSN then switch to USN
with a step in SSEP going from -0.5 to +0.5. determine toroidal velocity
profile evolution from HIREX array. if successful, map out momentum confinement
as a function of plasma current.
Entered: Mar 10 2005 06:33:54:400PM
Author: Ed Synakowski
Perturbative momentum transport - technique development and initial documentation

This experiment is aimed at developing a new technique for measuring momentum transport. The highest priority is to develop the control techniques for modifying ssep rapidly, where "rapid" is defined to be on the order of or faster than an energy confinement time.

The basic idea picks up on the observations of LaBombard, Rice et al. over the last year or so regarding the dependence of edge flow velocity on configuration (LSN, USN, balanced), and how this flow velocity defines a boundary condition for core flows.

The experiment will be done in L mode. The first step will be to make a rapid swing from LSN to USN, with Wolfe at the controls to minimize the time required to make this change. In the meantime, the core rotation measurements will be made to assess the size change in the rotation that is measured during this swing. It is likely that a density of 1.5e20 or more will be best for this measurement, as the largest difference in the rotation velocities observed in steady-state revealed a moderately strong density dependence.

An important consideration in the experiment is the quality of the Ar spectra, so adjustments will likely be made in Ar puffing. Presuming that a momentum pulse is observed in the HIREX measurements, optimization will take place in the data acquisition time, Ar levels, etc..

This optimization will likely take place as the ssep changes are modified as well. In particular, one goal is to have a return configuration change, i.e. a square hat in ssep from LSN to USN and back. Having a flattop of at least a confinement time in each configuration may allow boxcar averaging of spectral data.

Another goal is making the configuration change as brief as possible while still consistent with a good measurement. This may prove to be a valuable technique in the H mode, where the goal will be to be as minimally perturbing to the H mode state as possible.

If this proves successful, then an optimized setup in ssep changes, Ar levels, and HIREX data acquisition will be used to scan a small set of plasma parameters. Plasma current will be the first to be scanned. Second priority is the plasma density.

In short, the plan is as follows:

- minimize the time for configuration change from LSN to USN, aiming for a ssep change from about -5 mm to +5 mm. Starting condition is 800 kA, 5.4 T, nl04.

- increase and reduce the plasma density to identify the condition for maximum change in the density.

- optimize measurement of rotation change through Ar gas puff, data acquisition.

- docuement optimized condition well, e.g. multiple shots for statistics

- develop square wave change in ssep (-5 mm to +5 mm to -5 mm) with flattops and dwell times of order or greater than taue.

- shorten flattop periods to identify limiting factors in rotation pulse measurements


- with optimal setup, perform variations in Ip. Likely at least two densities at the new currents. 1.2 MA is a likely starting point, then down to 0.5 MA.

- higher field and lower field, assuming a wildly, amazingly successful day.

Success will be declared if the basic elements of the technique are developed, and one condition is well documented.





Physics Operators Plans
Entered: Mar 10 2005 04:40:07:887PM
Author: Steve Wolfe
Engineering setup and Physics Operator plan
for Friday, 11-Mar-2005

MP#405 L-mode momentum confinement time determined from
fast SSEP change

SL: Rice, Synakowski
PO: wolfe

Engineering setup:
-------------------

Overnight ECDC in D2, bake at 60C

Power system as on 1050303004
Note: A-coils must be put back in standard configuration
+Dtop -Dbot -Jtop +Jbot
before run.

Gas Setup:

fill B-top with 6psi D2, Hybrid ENabled
fill B-main (ex C-side) 30 psi D2, ENabled
fill B-side lower 2 psi Argon, ENabled
leave B-side upper as is (He3), DISabled
fill lower capillaries (NINJA plenum) 10 psi D2

Enable ECE, Z-bolo shutters (assuming no vacuum problems)

GV1 closed; bypass to RGA open
Pump valves closed in pulse and 5minutes after shot

Run Plan:
----------

The main problem for the physop in this run is
to get a fast SSEP transition from lower null to
upper null, and if possible back, perhaps for multiple
times in a single shot. The goal is to change the dominant
xpoint faster than a momentum diffusion time, which
is approximately 20-25msec. A 10ms transition would be
considered good.

Start from seg4 of 1050303004 (SSEP control)
reduce current to 0.8MA
increase nl04 to 1.0e20
flatten rcur to produce outer gap constant ~1.5cm
(check with Brian, in case he wanted this change)
Program SSEP change from -.005 to .005 between 0.9 and 1.0sec
Note that zcur is programmed to be -1cm. Probably want
to either set this to zero or move it up simultaneously.

Once I get the before and after conditions satisfactory,
begin trying to shorten the transition time. We should have
fast efit running on these shots, programmed to run at 1ms
intervals during the time of the transition.

Determine how fast we can actually make the switch from LSN
to USN. If this is short enough, proceed to making a reverse
transition. Then try multiple transitions, as called for
by the SL's.

If we get that far, proceed to different plasma conditions,
as called for in the MP: Higher and lower density, then
higher and lower current (1.2MA and 0.5MA). Take 2 or 3 shots
at each condition.

Session Leader Summaries
Entered: Mar 12 2005 12:43:43:043PM
Author: Ed Synakowski
Perturbative momentum transport summary - Synakowski, Rice

â?¢ With a primary goal being the assessment of the viability of using a rapid LSN -> USN transition as a means of inducing a velocity pulse originating at the edge, it looks as though the response time limit is about 50 - 60 ms for a requested control system change. This is slower than expected, and if true may limit the utility of this for L mode. HOWEVER, it is likely that there is still value of such an edge perturbation in H modes or ITB plasmas, where expected momentum confinement times (assuming it is similar to tau-e) are considerably longer than this.

â?¢ Operations were tough all day, with influxes in the USN configuration compounding the control challenges.

==============================

The effect of configuration variation from LSN to USN on the core rotation was in fact observed in a single discharge. The primary condition was ohmic, 5.4 T, 800 kA, average target density of 1.4e14 cm-3.

The battles of the day were configuration control (managing the rapid changes) and impurity influx in the USN configuration. It looks like shot 004 may have marked the start of the influx problem. Rapid EFIT analysis (EFIT01) revealed that fluxes may have struck an upper chamber structure after the switch to USN. It is not certain that this is the cause, but nevertheless increases in Prad were routines seen upon completion of the switch, and generally resulted in early termination of the discharge. Late in the day, plasmas in USN throughout the pulse were run, and some resulted in impressive displays of material influx as seen in diagnostics such as H-alpha as well as some of the RF/LH camera views.

Adequate Ar signal levels were obtained that suggest that multiple configuration changes in a shot, or shot averaging, might yield a good measure of momentum confinement, especially in plasmas with longer confinement times such as H modes, EDA H modes, or ITB plasmas. So to this extent the experiment was a success, and it is worth considering development of an H mode/ITB momentum confinement proposal if the USN impurity influx problem can be managed (hopefully a good weekend of GDC will go in the direction of helping this problem).

Due to the influx and control problems, no effort was begun regarding performing a parameter scan of any sort.


Physics Operator Summaries
Entered: Mar 11 2005 07:09:53:980PM
Author: Steve Wolfe
Physop Summary for Friday, Mar 11
MP#405 Fast SSEP change for momentum diffusion measurements

SL: Synakowski
PO: wolfe
EO: Pfeiffer, Dekow, Parkin, Byford

This was a difficult day, partly because the experiment was
intrinsically hard, and partly because after shot 4 we had problems
that seem to be indicative of dirt or something out of place in the
upper divertor. This manifested as increased radiation and injection
events when we went to positive ssep. Hopefully a weekend of ecdc
will help clean up.

In the end, I think we may have demonstrated that the maximum rate
at which we can swing ssep from negative to positive is about 50-60ms.
This is longer than the momentum confinement time in L-mode
plasmas, and so implies that the technique is not feasible for
the measurement as proposed. The best example is the final shot, #27,
in which the ssep demand was a 10ms ramp from -7mm to -3mm (which
actually corresponded to -5mm to -1mm) did not reach target
until 55msec. This was about half the slew we actually wanted for
the experiment. While there is some headroom on the gain, we seem
to be at power supply limits for the controller used for SSEP.
One way out might be to try to do the experiment using fast zcur
control to move a nearly balanced DN up and down, rather than
trying to do the transition by driving the X-points up and down
with the plasma more or less stationary. This technique would
definitely require additional development.

Note that all ssep control shots today used seg4. We also ran a few
upper null plasmas using seg3.

We still have not demonstrated acceptable control of an upper null
using SSEP control. Using a clearin predictor referenced to the
upper null location was somewhat successful (see shot 4), but still
allowed the gaps to change from the previous values. The problem
may be that the ssep control moves the upper xpoint below the nominal
location. Shot 4 was the best (only) example of a lower to upper
transition more or less controlled into rampdown. The L->U ramp
in this case was 100msec. This shot may have damaged something in
the upper chamber, because all subsequent shots that were run upper
null had lots of injections and very poor control. The strikepoints
during shot 4 were on the ceiling during flattop, but may have
interacted badly with the gussets during rampdown.

After shot 4, we tried to speed up the ramp, first to 50msec
and then to 25. These shots had poor gap control during and following
the transition and disrupted by 1.2sec. At shot#8 I tried changing
the SSEP controller to be orthogonal to ZXU and not to ZXL, and
reducing the gain on ZXL. However, I later discovered that I had
somehow expunged the 12700 A-m offset in the second clearin predictor
in setting up this shot. This introduced a large thrash in the
clearin error signal at 0.9 sec when the predictor switched in. That,
rather than the change of controller was probably responsible for
the nearly instantaneous disruption on this shot, and on shot 10
where I removed the new ssep controller.

For shot 11 I reloaded the seg from shot 7, which reproduced the
disruption on that shot, after a couple hundred msec on the upper
null. After a few more unsuccessful attempts to get decent operation
on the upper null, changing gains and such, we decided to try a
straight upper null program, using seg 3 which had also been
loaded from 1050303. These shots, #20-23, were also dirty and full
of injections and early disruptions, demonstrating that at least
part of the problem was with the upper divertor, rather than with
SSEP control or transitioning from lower to upper.

Since the upper divertor shots did not seem to be improving, we
finished the run by going back to lower null and doing the
SSEP sweep only as far as DN (program to -.002 and -.003) on shots
24-27. Of these, shot 27 is the best example, as described above.

Startup continued to be problematic today, although fizzles were less
of a problem than the last couple of days. There were a number of
high runaway discharges, all arising out of bouncy (outside strikes)
startups. I raised the rcur offset in seg1 to 2000 at shot 3 and that
seemed to help but did not eliminate the problem. The pre-puff still
seemed to need to be high, relative to previous weeks.

Some of the early shots had unexplained excess density, reminiscent of
the Wednesday experience. Since NINJA had been puffing at about
the time we saw the rise and the pg3 shut off, I had them turn off,
but that was almost certainly not the cause of the problem.

Scorecard:
Tests 0
Duds 0
Fizzles 3
Plasmas 24
------- --
Shots 27




Session Leader Comments
Mar 11 2005 04:33:33:027PMEd SynakowskiUSN only. Short and bright (just over > 0.5 s). SO the problem is not only control...
Mar 11 2005 04:48:55:747PMEd SynakowskiLonger! Under considerable distress. Went out to 1.25 s or so, the longest in a quite a while.
Mar 11 2005 11:12:54:367AM1050311001Ed SynakowskiThis shot will be an attempt to go from LSN to USN from 0.9 to 1.0 s. ICRF will condition starting at 1.4 s.

Died around 1.2 s, after the move to USN. Lots of hard X-rays as seen on the speckles of the plasma TV monitor. Undershot on the requested current - looked like it was in distress. Overshot on density.

EFIT says that the requested quilibrium change was successful, however.
EFIT01 being run as well as usual analysis to track fast changes in the equilibria.

ninja puff request will be cut way back.

Steve has to make a large number of changes...

LH will go in at 400 ms on the next shot.
Mar 11 2005 11:36:56:663AM1050311002Ed SynakowskiThis plasma died even earlier, around 0.5 s. Density excursion at around 0.3 s. Will ask for no NINJA gas on the next shot.
Mar 11 2005 11:40:25:880AM1050311003Ed SynakowskiBetter in terms of pulse length. Died just after 1.5 s. Density still too high, but better (line average peaks at around 2 near 850 ms). Made a decent configuration change from 0.9 - 1.0 s.

HIREX was hit hard by hard x-rays.

Ip was closer to 800 kA level.

In LSN, MARFE in upper part of vessel. In LSN, MARFE moves to the lower chamber.
Mar 11 2005 11:49:30:100AM1050311003Ed SynakowskiBetter in terms of pulse length. Died just after 1.5 s. Density still too high, but better (line average peaks at around 2 near 850 ms). Made a decent configuration change from 0.9 - 1.0 s.

HIREX was hit hard by hard x-rays.

Ip was closer to 800 kA level.

In LSN, MARFE in upper part of vessel. In LSN, MARFE moves to the lower chamber.

Steve has worked on startup in an effort to beat down the density, and perhaps help the hard x-ray situation.
Mar 11 2005 12:11:25:553PM1050311004Ed SynakowskiHard x-rays better. A burst of D-alpha at the start of the USN phase - hitting something? Nel is lower, close to the target value.

First sign of success in the HIREX signal! A change in the central rotation that takes place on the same time scale as the SSEP change.

This shot actually made out to nearly 1.7 s, well into the Ip rampdown.

John will ask for more Ar for the next shot (20 ms more).
Next shot will try to make the ssep change 50 ms faster, from 900 to 950 ms.


Mar 11 2005 12:25:38:443PM1050311005Ed SynakowskiShot terminated after 1.1 s. Vertical instability? Small glitch on D-alpha at the top of the SSEP change. MARFE towards the end. Density a bit high.

Ohmic heating takes off right after the SSEP change. Bolometer nearly triples in signal.

Fast EFIT01 did not go through.
Mar 11 2005 12:41:24:473PM1050311006Ed SynakowskiShot lost right after the SSEP transition, near 1 s. Vertically unstable.

Will raise the density a skootch, from .8 to .9 nl04
Mar 11 2005 01:06:34:600PM1050311007Ed SynakowskiAgain, terminated early. Distress again begins at 950 ms time, start of USN period.
Mar 11 2005 01:10:31:600PM1050311008Ed SynakowskiUnstable again. Steve is scratching his chin...
Mar 11 2005 01:54:17:507PM1050311010Ed SynakowskiQuick demise again at 950 ms at top of LSN -> USN transition. Read Steve's notes.


Mar 11 2005 03:12:25:960PM1050311011Ed Synakowskivirtually identical to 007, which has its own difficulties but at least they are reproducible.

John has some HIREX data that shows the general shift to more counter-directed rotation after the switch. Since this is so similar to 007, I am wondering if these data can be averaged.


Mar 11 2005 02:31:15:803PM1050311012Ed SynakowskiDied near 830 ms, BEFORE planned SSEP step. Something fell in, perhaps. Mo? Something radiates 10 MW, albeit briefly.

Will redo, since all of programmed changes were for after this time.

Mar 11 2005 02:47:04:960PM1050311013Ed SynakowskiAnother early demise. Unclear what is going on.
Mar 11 2005 03:04:27:367PM1050311014Ed SynakowskiTerminated just before 1 s.
Mar 11 2005 03:10:57:113PM1050311015Ed SynakowskiLost density feedback, and thereful control. Loads of runaways. Still died at the top of the LSN -> USN switch.
Mar 11 2005 03:48:57:937PM1050311017Ed SynakowskiDistress at .73, early termination.
Mar 11 2005 04:10:19:213PM1050311018Ed SynakowskiMade it farther (almost 1.2 s).
But ohmic input power skyrockets after the switch.

Is it beginning to limit on the outboard side after the switch? Looks like it, according to EFIT01 analysis. Maybe hitting something out there that is then introducing gunk.
Mar 11 2005 04:27:32:073PM1050311019Ed SynakowskiSimilar pathology. Rained on with sparks near 1 s, according to J port antenna view.

Steve will try UN all the way through on the next shot.
Mar 11 2005 04:36:08:760PM1050311020Ed SynakowskiUSN only. Short and bright (just over > 0.5 s). SO the problem is not only control...
Mar 11 2005 04:49:19:200PM1050311021Ed SynakowskiLonger! Under considerable distress. Went out to 1.25 s or so, the longest in a quite a while.
Mar 11 2005 05:36:58:870PM1050311024Ed SynakowskiThis was an attempt to stop just shy of diverted USN. This plasma lasted longer than any other, except shot 004.

Loads of hard x-rays.

Evidence for junk falling in after 1 s on D-alpha. Camera shows lots of gub coming in early, too (early in the ramp, 100 ms neighborhood).

Physics Operator Comments
Mar 11 2005 08:38:43:647AMSteve WolfeSetting up for the run, more or less as indicated in the plan:
Load from 1050303004. Note that while the titles of segs 2-4
say these are 0.8MA, they are not. Seg 2 is 1.2MA, and seg4, which
is the one I'm after, is 1.05MA.

In seg4: Change ip to 800ka, nl04 to 1.0e20,
SSEP program from -.005 to +.005 between 0.9sec and 1.0sec
copy trace from wire 11 to 12, just in case.
I don't need the second ssep wire, but I may want to look at
an alternative clearin that maps to the upper xpoint. Put a second
clearin on wire 9, with gains off and referencing zxa=+.394,offset=12700

On all_v, shorten acoil_cur trace from 10sec to 2.5
put oh2 voltages to ~20V through eof, as in seg2
I should probably put the zcur to 0, but I'll wait a shot on that. Also,
I'll leave the rcur and rxl programming as is until I get a targeting
evaluation from Blab.

Now for the startup. Joe was making a lot of changes yesterday, and seemed
to be getting improved startups toward the end of the day. However, the
breakdown was rather late on the last shot.
Import seg1 from 1050310026, change IC_ef4 from -1345 to -1355
leave pg4 at 21ms and Br0 at -.002

Note that in both seg1 and seg4 the chopper voltage is non-zero in flattop
(150 in seg 1 and 100 in seg4).This affects the bias (average) current and should no longer be set this way. Checking, the bias is low on 1050303004. Set the chopper voltage to zero on ALL_V's in both seg 1 and 4

Load into model
Mar 11 2005 09:22:42:287AMSteve WolfeNote that the gain switching in seg4 has SSEP gain at 6
from the start of the segment, and ZXU gain dropping from
2.0 to 0.2 at about 200ms.

I may want to consider a change in the SSEP controller to
make it ortho to ZXL instead of ZXU at the time I go upper
null, and also bring ZXU gain back on and drop ZXL gain (from 8)
at that time. This is an additional complication, on top of
the change in clearin reference, for the UN part of the discharge.

If I do this, I'll want the ZXU position program to be close
to nominal. The waveform is presently drawn to +.03, which is
probably appropriate for the lower null, and ZXL should probably
be more like -.03 in the upper null phase.
Mar 11 2005 09:42:01:680AMSteve WolfeAlso note that the upper null program in seg3
has ZXU=.005, RXU=-.02 straight across, while
the seg 4 program is ZXU=.03 and RXU moving out
from -.02 at 0.8sec to -.006 at 1.5sec. EFIT for
1050303004 does show a small, ~1cm, outward drift
in RXU. This may or may not be a good thing. To keep
the strike off the gusset I may want to leave the
program more negative, but I have to watch the ef2u
current to see it doesn't go to zero; it's ok on the
old shot, but that's at different current and density.

Go ahead and make the change to RXU=-.02 in flattop,
then back in in rampdown, and see what happens. Still
leaving the funny sweep in RXL.

Load into model
Load at 11-Mar-2005 09:41:50.00
Open tree /home/wolfe/pcs_scratch -1
Open tree done

Mar 11 2005 11:15:48:430AM1050311001Steve WolfeShot#1: First try at lower to upper.
Plasma to 1.2sec, didn't like something anbout the upper
null.
Startup is ratty, several bounces;lots of hard xrays.
Current is low.
SSEP looks like it is starting -.3 going to +.5
Clearin also takes a big jump.
Strike point is going close to the gusset by the time it disrupts.

Clearin on wire 4 thinks it's happy, but the one referencing the
upper null on wire 9 is at zero in the first half and shows
the increase to 10000A-m at the switch, i.e. it is right all the
way through.

The current is off because the seg4 still is using the old rogowski
values! It never got re-called, which means Ian or whoever ran the
whole day with the wrong value!

Next shot:
Call the alg of Ip, offset to -70000.

Add a second alg line to clearin referencing the upper
null starting 0.9sec. If I need to do this several times,
it isn't going to be possible to program multiple chops
with the present user interface.

EF2's are getting pretty low; reduce pre-program on EF4 from
2000 to 1000

Now for the startup. IC_EF4 from -1355 back to -1345
pg4 from 21 to 24ms
Change SSEP program to start at -.007 and end at +.003 (there
may be a problem with the -2500 value in the offset, but I don't
want to think about which way it needs to go).

Gas is off after .64sec. Request NINJA to reduce puff by at
least a factor 2.






Mar 11 2005 11:29:35:380AM1050311002Steve WolfeShot#2: Early disruption
big density increase at 0.3sec.
NINJA claims it isn't them, but turn them off anyway.
Startup still ratty, not as bad on the hards.

next shot: rcur offset from 1500 to 2000 in seg1

Mar 11 2005 11:46:53:287AM1050311003Steve Wolfeshot#3: plasma to 1.6sec
Density is too high early. Where's it coming from?
Valve is off from 600msec to 1.1sec

Gaps are better, still some thrash at the transition.
Startup barely made it- one big bounce and recovers.
Lots of hards again.

Next shot: reduce .1sec Ip point from 700kA to 550kA
Slow down density rate of rise, reduce seg4 pg3 pulse
Increase pg4 to 26ms
Mar 11 2005 12:08:23:490PM1050311004Steve WolfeShot#4: plasma to 1.7sec
density is better controlled, still has a step
in current at the gain switch.
Startup is better, no bounce.

Outer gap is dropping some at L->U transition
Inner gap is too, still need better clearin control
John wants 20ms more on argon
EF2U is going close to zero at the transition,
and EF4 swings negative, and it's during crossover
that we get a dalpha burst.

next shot: Try ef2byef4 target at -3000
shorten the transition to 50ms
Since I'm not sure what to do about the clearin, leave it
Mar 11 2005 12:27:59:320PM1050311005Steve WolfeShot#5: L->U programmed from .9 to .95sec
Changed ef2_by_ef4 to -3000A

Disrupts at 1.2
Voltage goes up right after L->U; why?
Marfe at 1.2sec
Gap control looks worse on this one.
ZXU is actually down to .36 when it goes to
UN.

John wants lower density.
Change zcur to zero, move zxl down at transition.
Mar 11 2005 12:44:26:320PM1050311006Steve WolfeShot#6:nl04 to 0.8e20, zcur to 0, zxl to -.01 at .95sec
Disrupts at 0.98sec, VDE

May have been the zxl change at high gain?

next shot: nl04 to 0.9,
remove the zxl jump, reduce gain on zxl from 8 to 1 at .95


Mar 11 2005 01:05:09:427PM1050311007Steve WolfeShot#7: nl04 to 0.9,
remove the zxl jump, reduce gain on zxl from 8 to 1 at .95

Hard xrays again, disrupts at 1.2 again.
Startup bounced again.
EF2U goes to zero at 1sec,
I may need to change the controllers on both clearin
and ssep during the upper null.

Next shot:
Change clearin controller at .95, zxa=.394. I don't think
that actually will do anything
Change ssep controller to be ortho to zxu instead of zxl after .95
That seems to require factor 2 longer time constant, not sure why
Increase pg4 from 26 to 28


Mar 11 2005 01:34:24:427PM1050311008Steve WolfeShot#8: Plasma to .95, seems to have died right
at the time I changed the controllers.

The new ssep controller is very different, in ways I don't
really understand. Maybe the controller code is broken.
Call it again.

The constraints should be such that the coefficients should be
proportional to the normal zxl controller, but it is very
different.

Recall the original one and get rid of the switch.
Mar 11 2005 01:35:10:710PM1050311008Steve WolfeShot#8: Plasma to .95, seems to have died right
at the time I changed the controllers.

The new ssep controller is very different, in ways I don't
really understand. Maybe the controller code is broken.
Call it again.

The constraints should be such that the coefficients should be
proportional to the normal zxl controller, but it is very
different.

Recall the original one and get rid of the switch.
This ought to put it back the way it was.
Mar 11 2005 03:45:57:803PM1050311008Steve WolfeSomehow on this shot the 12700 offset in the clearin predictor
got lost. This may explain the disaster at .9 which I attributed
to the controller switch which I also implemented on this shot.
Mar 11 2005 01:39:52:693PM1050311009Steve WolfeShot#9: fizzle. Moves in.
Drop pg4 back to 25msec
Mar 11 2005 01:54:40:850PM1050311010Steve WolfeShot#10: Still trying with a single ssep controller
This ought to be like shot 7, which isn't a great target
to shoot for, but I need to see if the ssep controller is
really the reason for the huge slam on shot 8

Disrupts hard at .952, just like 8. Apparently that was not
the problem.

I still have the clearin changed controller in there, but those
looked identical. Also I recalled the alg for ssep.

Just reload the segment from shot 7 and try again.
Mar 11 2005 02:19:52:053PM1050311011Steve WolfeShot#11: reload the segment4 from shot 7 and try again.
plasma to 1.2, looks like shot 7.

next shot: Try going even faster, reduce clearin and ssep
gains after .9sec
Mar 11 2005 02:30:37:490PM1050311012Steve WolfeShot#12: ssep ramp in 25msec,
clearin gain from 9 to 5 at .9
ssep gain from 6 to 3 at .95
Increase argon another 20ms for JR

disrupts at .84, before anything is supposed to happen.
Possibly unrelated injection.
Try that one again

Mar 11 2005 02:46:46:380PM1050311013Steve WolfeShot#13: ssep ramp in 25msec,
clearin gain from 9 to 5 at .9
ssep gain from 6 to 3 at .95

Disrupts at .72sec. Something is coming in at 0.68
John says not Moly.
Startup is as little slow

one more try

Mar 11 2005 02:56:58:757PM1050311014Steve WolfeShot#14: ssep ramp in 25msec,
clearin gain from 9 to 5 at .9
ssep gain from 6 to 3 at .95
pg4 down to 23msec

Disrupts at .96 sec. Just past the switch.
Fast efit says the ssep is almost following
demand, gets to +2mm at .938, but then starts
being unstable. We're losing vertical stability at the same time.
That may be the limitation, the elongation
is going up, n/ncrit gets to 1.2 by .95sec
Both gaps are growing, which may be because
I reduced clearin gain.


Mar 11 2005 03:05:35:537PM1050311014Steve WolfeShot#14: ssep ramp in 25msec,
clearin gain from 9 to 5 at .9
ssep gain from 6 to 3 at .95
pg4 down to 23msec

Disrupts at .96 sec. Just past the switch.
Fast efit says the ssep is almost following
demand, gets to +2mm at .938, but then starts
being unstable. We're losing vertical stability at the same time.
That may be the limitation, the elongation
is going up, n/ncrit gets to 1.2 by .95sec
Both gaps are growing, which may be because
I reduced clearin gain.

Found something, maybe. There is a zero offset
in the second clearin gain. I could swear I set that, but
there is a definite jump in apparent clearin error.
Set the offset to 12700 and try again


Mar 11 2005 03:20:53:520PM1050311015Steve WolfeShot#15:Runaways, tci failed and turned off gas.
Disrupted at .96, so this is another ssep switch disruption
Still has a jump in the clearin value,
And it still seems to be vertically unstable.
The second clearin seems to have been there since shot2

It also still isn't following ssep at 25ms

Reduce the ramp back to 50msec


Mar 11 2005 03:37:24:130PM1050311016Steve WolfeShot#16:Reduce the L->U ramp back to 50msec
fizzle
Bounces? But it seems to be going in at the time.



Mar 11 2005 03:49:28:433PM1050311017Steve WolfeShot#17:Drop pg4 to 21msec
Still with the L->U ramp back to 50msec

Plasma to .75sec, probably another injection.
I'm going to drop the gain on nl04 before .6 sec
because the gas is flopping around, and try again.






Mar 11 2005 04:10:18:213PM1050311018Steve WolfeShot#18: drop the gain on nl04 before .6 sec
because the gas is flopping around, and try again.
Still trying to get to the switch with the reduced speed and
correct clearin predictor. While I'm at it, back off 20ms
on the argon puff

plasma to 1.14, past the L->U swing, but the control is
still bad. Outer gap is getting close to zero. Need to
recover clearin control. put clearin gain back up at 1sec




Mar 11 2005 04:29:49:480PM1050311019Steve WolfeShot#19: Notch in clearin gain between .9 and 1 to try to recover
some gap control.

plasma to 1.14sec
Almost the same. I can move the plasma in at the same time,
which may help.
However, Jim Irby points out that there is a rain of sparks
from above the JANT camera view.

Instead, let's run an upper null all the way through.
Mar 11 2005 04:39:07:340PM1050311020Steve WolfeShot#20: Try an upper null.
Turn on seg 3, change nl4 to 9e19
Give John 70msec of Ar

Disrupts at 0.55sec, looks ugly for awhile before.
This one is touching the outer limiter at 0.4sec,
which is not the cause of the problem but I probably should
move in a bit.


Mar 11 2005 04:49:31:590PM1050311021Steve WolfeShot#21: Adjust Rcur a little inward and try again for upper null.
plasma to 1.26 seconds but lots of junk coming in.

next shot: try again with increased clearin
Mar 11 2005 05:04:43:527PM1050311022Steve WolfeShot#22: clearin from 0 to .0025
Short again, .55sec
Startup is a little slow, and bd is a little late.

next shot: drop pg4 2ms and move clearin increase a
bit later
Mar 11 2005 05:22:25:963PM1050311023Steve WolfeShot#23: pg4 to 19ms, clearin .0025 point to .514sec from .414
plasma to .8sec, something bad happens at .67

Next shot: Go back to seg4 and swing up from -.007 to -.002
just shy of upper null
Mar 11 2005 05:36:16:950PM1050311024Steve WolfeShot#24:Go back to seg4 and swing up from -.007 to -.002 in 30msec
just shy of upper null
Remove clearin notch, second clearin predictor, ssep gain drop

Lots of runaways this time, but it lasted to 1.65
ssep is just about zero after 1sec.
Lots of injections after 1 second.
Startup bounced again, amaybe a couple times.
Looks like the ssep excursion takes about 60-70msec

next shot: increase pg4 by 2msec, decrease ssep final to -.003
Mar 11 2005 05:45:16:353PM1050311025Steve WolfeShot#25:increase pg4 by 2msec, decrease ssep final to -.003 in 10msec
fizzle, looks like one bounce and done.

Another 2ms on pg4 and hope it goes
Mar 11 2005 06:03:21:463PM1050311026Steve WolfeShot#26:increase pg4 by 3msec, to 24msec
ssep final to -.003 in 10msec

Early disruption .7sec. Something is falling in.


Mar 11 2005 07:11:12:073PM1050311027Steve WolfeShot#27: One last try.
Plasma into rampdown
Few or no injections
Hesitation in startup, but no runaways.
SSEP gets to about -.001 at .954 and oscillates about that level.
In some ways this is the best shot of the day. Thanks to Bill and
the alternator crew for allowing us one more shot after 18:00.

I think that 50-60ms is about as fast as we can swing it.
This probably means that this technique is not going to be
suitable for the momentum confinement measurement, even if
the control in the upper null state could be improved.
The limitation may be the EF1U and EF2U slew rates.
EF2U actually seems to have enough voltage, but the lag between
demand and supply voltage seems to be about 10msec.
EF1U demand is going from +200 to -200V, and the supply never gets
below -40V.

I missed removing one gain switch, the drop in zxl gain
at .95sec, which is visible as a drop in the ef1L voltage, but that's
not really at issue.

It may be possible to go a little faster by increasing the gain, but
this is already ringing and is pushing on power supply lags for the
thyristor supplies, especially ef2.

Engineering Operator Comments
ShotTimeTypeStatusComment
110:36:01:537AMPlasmaOk
211:14:16:507AMPlasmaOk
311:30:57:083AMPlasmaOk
411:48:30:570AMPlasmaOk
512:11:20:037PMPlasmaOk
612:29:27:477PMPlasmaOk
712:45:01:897PMPlasmaOk
801:05:09:350PMPlasmaOk
901:36:11:007PMPlasmaOk
1001:48:40:570PMPlasmaOk
1102:04:50:913PMPlasmaOk
1202:23:20:583PMPlasmaOk
1302:35:50:257PMPlasmaOk
1402:48:20:617PMPlasmaOk
1503:07:11:617PMPlasmaOk
1603:32:36:803PMPlasmaOk
1703:45:00:130PMPlasmaOk
1803:57:48:860PMPlasmaOk
1904:12:23:370PMPlasmaOk
2004:30:27:057PMPlasmaOk
2104:42:53:510PMPlasmaOk
2204:58:21:917PMPlasmaOk
2305:11:28:010PMPlasmaOk
2405:28:39:200PMPlasmaOk
2505:42:04:293PMPlasmaOk
2605:54:50:557PMPlasmaOk
2706:07:14:480PMPlasmaOk