RFBG Commissioning with Adaptive Secondary - 25 March 2010 UT
Observer: DMiller, JHeidt, JHill, JStorm (LBTO)
Telescope Operator: DHuerta
Instrument Support:: TShih
Telescope Support: JLittle
Adaptive Support:: MXompero, GBrusa (LBTO)
The night started out with bad seeing due to the cold front that went through yesterday. Although by the end of the night the seeing was quite good at 0.5 arcsec (see sample images below)
. Scintillation was almost non-existent to your eye. Cirrus at the beginning, but clear at the end.
We learned about high order mode filtering on the shell. You must do it, but you must not set the threshold too low. That makes the difference between a mirror that is a mess of intermediate Zernikes, and a mirror that is indistinguishable from a rigid one.
We applied Zernike terms Z4-Z11 to M2 and measured them with GCS.
We found that the trefoil terms on M2 were rotated by 10 degrees compared to what you expect.
Other Zernike terms have inconsistencies (flips and/or rotations )between M1 and M2 that we haven't figured out.
(See additional notes below.)
It took a long time to sort out that the last patch (related to pupil shift) that fixed IDL broke GCS. We are still sorting out that paradox, but GCS works fine after we rolled back to the previous version.
After lots of messing around with various active optics programs and issues, we got some nice transformation data in good seeing during the last hour of the night. The guider images were as good as 0.43 arcsec FWHM in r'. Best guider image was 0.38 arcsec FWHM measured with IRAF.
JMH estimates that we were 30% complete with the AGw commissioning phase at the beginning of the night. We dropped back to about 18% during the night, and finished at about 33%.
Building has been rotated to melt snow off all 4 sides.
23:05 switch snow melting to ventilation mode
DThompson updated the keyword defaults to be aware of left and side side DD parameters.
Single frame names refer to: irtc.20100325.xxxxxx.fits
Cube names refer to: irtc.20100325_00xxx_cube.fit
01:40 Open, some clouds in west + some Cirrus, T_out = -3.0C, T_Mirror = 3.0C, T_Shell = 0.7C, Hum = 40%
Initial Pointing and Collimation Adjustment
Each night in this RFBG commissioning run, we go to a bright star (Pollux) at Elevation 80 degrees. Then we adjust the vertex tip-tilt (RX/RY) of the secondary to get rid of the coma
that we see in the extrafocal pupil on the IRTC.
01:49 pointto Pollux gs=0 TelM = "Track" , OK, already close to Pollux, adjust: IE = -69, CA = -60 at EL = 76, AZ = 105. Some Coma, try to adjust: IE = -42, CA = -41
DX M2 Global X = 2.5, Y = -3.63, Z = 0.5, RX = -520, RY = 0
, not too bad
Apparently some temperature effects visible on images with plumes of warm air streaming off the center of the pupil, 2 takeseq's. with 0.02sec's cubes 79 and 80
02:22 pointto M5_1210 gs=0 TelM = "Track" , EL = 26, AZ = 170, at very-low elevation to adjust AZ-zeropoint, pointing OK, star close by, Adjust IA = -290
, keep IE=-42 CA = -41
This follows the trend we saw on the left side of the telescope. IA is near -390 arcsec if the temperature has warmed by 5 degC, and IA is near -270 if the temperature has cooled.
Comparison of M2 Zernikes
02:34 pointto StoneEagw
gs=0 TelM="Acquire", OK, centering,
02:45 pointto StoneEagw
gs=0 TelM="Active", OK, centering, let collimate, seeing not very good,
guider reports ~ 1.2"
Between 03:20 and 03:56 we successively applied (and removed) +1.0 microns of Z4 through Z11
to adaptive M2. GCS measured the corrections, but did not send them to PSF except when we paused to refocus. The results from the GCSR syslog are attached below, and summarized in this table. The expected correction if everything were perfect would be -800 nm.
|| GCS correction
|| GCS gain
|| AOS Gain
|| 0 Z6
|| +250 Z5
|| +100 Z8 ?
|| 0 Z7
|| -300 Z10
|| -320 Z9
Legend: The first column is the Zernike term; the second column is the correction that GCS wants to apply; the third column is any noticeable crosstalk term in the correction, the 4th column is what GCS measured (correction/-0.8); the fifth column is the gain factor used by GCS and IDL.
It looks like there could be a factor of ~2 (-2) between the two systems, although both are supposed to be rms wavefront terms.
The crosstalk between Z9/Z10 seems to indicate that the M2 trefoil terms are rotated by 10 degrees. Marco corrected a 14 deg rotation of Z9/Z10 in April.
We have not yet understood what combination of flips and rotations can account for the other sign flips. Note that IDL/GCS have a similar set of sign flips in their conversion from measurement to correction.
JMH (updated interpretation 26-April-2010): I've added a column to show the gains being used by GCS to match the S-H measurements to the M1 Zernikes.
It appears that these gains account for all the flips seen in the table other than a 180 degree rotation (needed to make the gains of Z7-Z10 work).
I've added another column to shown the gains that would presently (March 2010) be required for active optics conversion between M1 (PSF) and M2 (AOS). These gains are equivalent to a 180-degree rotation (e.g. Z6 is invariant, Z7 changes sign, etc).
That means that both M1 and M2 are using identical (or very close to identical) versions of Noll Zernikes, except that we have a 180-degrees rotation between the two. No special gains or conversions should be required. Marco is going to investigate whether he can recompute the M2 Zernikes to match M1.
We should remeasure Z5-Z8 to verify if there any real cross-terms there.
4:32 Pointto M5_1246, GS=2, PA=0, TelM
check for orientation on IRTC-Panel vs. DS9-panel
image at central position, 2sec, 043335
image with offset RA +5", 2sec, 043521
image with offset Dec + 5", 2sec, 043943
=> IRTC-Panel: East is up, North to the left
=> DS9-Panel: East is down, North to the left
X-Y motion as a function of focus AND assorted other problems
04:51 pointto BS9143 GS=0, TelM
=ACTIVE, check if focus is moving along the
optical axis at the AGw.
Doug's having problems, redo preset
04:49 pointto BS9143 GS=0, TelM
05:05 pointto BS9143 TelM
Jesper moves probe back to center
5:11 Pointto M5_1246 TelM
=ACQUIRE, GS=0, better than BS-star since brighter (especially once AGw is defocussed
Check if focus is moving along the optical axis at the AGw. Tests done by Doug & Jesper using IDL-scripts.
No collimation (actually its the opposite!)
05:27 pointto BS9143, GS=0, TelM
=active, collimation is very slow, seeing degrades (~1.5")
05:42 pointto BS9143, GS=0, TelM
=track, Doug & Jesper test, sighhhh...not a good choice
05:49 pointto M5_0444, GS=2, TelM
=active, GS too faint
05:56 pointto MS_1311, GS=2. TelM
=active, hexapod gets stuck, reset OSS
06:03 redo MS_1311, gnagna...M1 hits limits in x, cleared active optics by David,
global offset of z abs 2.5 at M2
06:18 redo MS_1311, GS=2, TelM
=guide, still something wrong, new IE = -30, CA = -32
Collimation via IDL by Doug, apparently some higher order Zernikes who makes collimation difficult
07:00 JMH: Things have been very confusing for the past hour or so with GCS seeming to diverge in active optics. Then IDL struggling with big mid-scale residuals in the S-H spots. We think we have figured out what happened. When Marco and Guido were applying Zernike modes to the primary, they turned off "high order rejection". Apparently this allowed these mid-scale errors
to creep into the shell. Then we think that 5th-order coma from the shell was cross-talking into 3rd-order coma in GCS and causing runaway of the coma. When Marco re-flattened the shell with "high order rejection" on, we instantly got the best looking set of S-H spots that we have seen. Now we are setting out to see if we can repeat the experiment.
07:15 weather now very good, almost completely clear, seeing seems to improve, guider reports ~0.7"
07:25 pointto MS_0545 GS=4, TelM
=active, collimation OK now
Since seing sems to be good now, go to Stone-I
07:58 pointto StoneIagw
=active, and let collimate
08:19 cl < tran_StoneI_mix1.cl
PA=0: transtar(3) 082058 cube 82
transtar(24) preset failed, GCS apparently hangs up, stopped
08:29 cl < tran_StoneI_mix1.cl,
PA=0: transtar(3) 083017 cube 84, collimation diverges again
transtar(24) preset failed, stopped,
John changed WFS_proberotationfactor from +1to -1
08:36 pointto StoneIagw
=active, and let collimate again
08:39 cl < tran_StoneI_mix1.cl, practicing a bit
PA=0: transtar(3) 084038 cube 86
transtar(24) 084333 cube 87
transtar(27) 084615 cube 88
PA=180: transtar(19) 084930 cube, stopped, collimation continues to be poor
John changed WFS_proberotationfactor back from -1 to +1, restart GCS
08:56 pointto StoneIagw
=Active, PA=INDEF, and let collimate, OK, guider reports ~0.7" now
done along Y-axis (PA was 118)
09:06 pointto StoneIagw
=Active, PA=80, and let collimate again, but with 30deg off Y-axis
now lots of Z11 were applied, still not clear why collimation is not good
09:38 pointto StoneIagw
=Active, PA=INDEF, no star
09:44 pointto MS_0460, GS=0, TelM
=acquire, check pointing, OK
09:48 pointto StoneIagw
=Active, PA=INDEF, GCS hangs up after a while
10:05 rollback to previous version of GCS
(before the 20100321 patch). We did this because that version seemed to be working correctly during the transformation data of 20100321.
10:05 pointto StoneIagw
=Active, PA=INDEF, collimation OK
10:09 pointto StoneIagw
=Active, PA=80, collimation OK
10:18 pointto StoneIagw
=Active, PA=50, collimation OK
10:18 pointto StoneIagw
=Active, PA=30, collimation OK
Transformation Data with IDL (not yet suitable for final reduction)
10:25 collect transformation data at StoneI
-field using Dougs IDL-script
script has some weak features, several updates/modifications,
several images/cubes taken which are crap (but partly not)
Images 102607 - 110625 + cubes 100 - 117 taken by Doug for test purposes
and optimizing his IDL-script
Probe Transformation Data on StoneL
11:12 pointto StoneLagw
=active, lots of coma, hit limits while trying to collimate,
reset, lost guiding
11:28 pointto M5_1569, GS=0, TelM
=track, check positioning, new CA = -23, IE = -26, no coma
11:33 pointto StoneLagw
=active, coma, no coma ?????
looks at least OK by now,
11:41 cl StoneL
_mix1.cl seeing on guider mostly 0.5-0.7"
PA=0 transtar(35) 114232 cube 119
transtar(31) 114510 cube 120
transtar(21) 114751 cube 121
transtar(19) 115026 cube 122
transtar(15) 115302 cube 123
PA=180 transtar(23) 115621 cube 124
transtar(17) cube 125 Binary, not useful
transtar(20) 120129 cube 126
transtar(18) 120406 cube 127
PA=45 transtar(35) 120706 cube 128
transtar(31) 120944 cube 129
transtar(13) 121224 cube 130
transtar(24) 121504 cube 131
transtar(30) 121743 cube 132
PA=210 transtar(12) 122048 cube 133
transtar(22) GCS hung data bad cube 134 + rotator error
stopped and started modified script
12:30 cl StoneL
_mix1_mod.cl seeing on guider mostly 0.5-0.7"
PA=210 transtar(22) 123032 cube 136 0.43" on guider
transtar(26) 123306 cube 137
transtar(34) 123542 cube 138
PA=315 transtar(31) 123908 cube 139
transtar(21) 124145 cube 140
transtar(19) 124430 cube 141, guiding failed, not useful
transtar(15) 124726 cube 142, guiding failed, not useful
transtar(13) 125016 cube 143, guiding failed, not useful, too bright outside now
Search for the "typical" (best I could find) image from the guider
See JPEG of guiderimage006589.fits below. 2.1 sec r' image taken at 11:41 UT. 7.5 pix FWHM = 0.41 arcsec FWHM.
See JPEG of guiderimage007241.fits below. 0.5 sec r' image taken at 12:37 UT. 7.1 pix FWHM = 0.38 arcsec FWHM
| 2.1 sec
|| 7.5 pix
|| 7.7 pix
|| 7.6 pix
| 0.5 sec
|| 7.07 pix
|| 7.18 pix
12:50 Close for impending dawn, T_out = 0C, T_Mirror = -0.3C, Hum=6%, sky is clear
Doug and Jesper inspected the transformation data afterwards, a new transformation out of them was not better than the old one. Main reason is the low number of observations, not sufficient coverage of the plane and also the fact that some of the images are elongated.
- 25 Mar 2010