Date: Fri, 26 Mar 2004 08:03:10 -0500 (EST)
From: John Monnier <monnier@umich.edu>
Subject: MONNIER: End Run Report


End of Run Report:
JDMonnier
2004Mar25


Current Configuration: N35 S15 C10 AFIX
Software:
 ~/iotamc_g3.2/vxworks-start-LL        #for new lincoln lab startracker
 ~/iotamc_g3.2/lmtmc_start
 ~/iotamc_g3.2/daq_start
 ~/observe_g3.2   # start IDL software (do BEFORE lmtmc_start)


KUDOS
-----
The new YAW on YAW 1 is GREAT! Did not Lose ANY time to this
recently-common problem of running out yaw-range. Thanks nat!

New Lincoln lab CCD replaces Geary CCD. Everythings seems to work the same
except now we can readout at tint=2ms => 500Hz (nice!)

Thanks to Kamal and Pete one late night for figuring out a solution to the
corrupted shared memory problem (now, if we could only figure out the
cause too!).  Kamal instituted an automatic shmem purge so this problem
should not ever recur.....(ha!)


WARNINGS
--------
{ettore:}CPU2 often (~50% or more of the time) hangs when going from
quadrant mode to fringe taking mode.  Is there some way to more kindly
'stop' the cpld from doing quadrant mode before loading the new program?
Maybe this will help with the problem.  [This kind of crash --PMC not
found-- requires a complete restart of the whole vme bus and is Very
disruptive!]

Marc has put some new alignment wands that move in/out of beam.  These are
proving useful but operation can be a little trickey (involving various
3-position switches). IMPORTANT: DO NOT LEAVE WAND IN BEAM AFTER DOING
YAW (easy screwup to make).

The limiting magnitude of the Star Tracker is actually worse now due to
excessive dark current and higher-than-expected readnoise -- see previous
JDM memo.  Not sure the limit, but V=9 mag at tint=30ms is ok as long as
there is no Laser light.

LD2 laser seems to appear in FIXED beam often (for LD2~1900cm).  In the
end, a good solution is to note the exact LD2 position (from Delay
server), go out and block LD2 (left) laser in hut, Observe. When ready to
move LD2 again, you must unblock laser and re-index at last known
position.  This procedure is foolproof at removing the laser light and
gives one the opportunity for midnight exercise (thanks marc for the
idea).

Telescope B has a temperature-dependent focus due to misalignment of
'glass' spacing rod (bottom).  Hopefully this will be re-alignment from
scratch before next observers.  If not, then beware of strongly variable
focus positions from dusk to dawn (particularly atrocious at dusk
sometimes -- ~10-15" image quality!!!).

IDL occcasional freezes (I'm pretty convinced that this is an IDL problem
and nothing I can fix).  In nearly ALL cases, this can recovered by ^C. If
you get the IDL prompt, then simply type '.continue'.  If you don't, then
type ^Z, kill IDL process, restart 'observe_g3.2'. This is all outlined in
the DAQ Manual Appendix (accessible by button on netscape). Be Sure to
reset observing parameters (reads/loops/lambda) since this is set to
defaults when observe_g3.2 starts!

HINT: To get good IDL color table while observing, start observe_g3.2
BEFORE lmtmc-start

Tel A/B pointing models in scripts/jdm_march04  work well. Tel C is not so
bad, but not great (not sure why... just high residuals in ModelFit..)
Tel A Tilt axis home was replaced and there is an offset (see Model A in
scripts/JDM_march04)

At one point, we had a low pixel 2 on the quadrant. After blowing
canned-air at the Integrated Optic aggressively, the flux suddenly
increase by x3. Perhaps we should clean this regularly? What is safe?


NEW FEATURES
------------
IDL Fiber Explorer (#5): Due to popular demand, added new option (b) to
make square grid of specified size for mapping out fiber image.  Hit (b)
again to abort the grid motion one peak is found.

IDL Find Fringes Display (#4): Added nifty closure phase plot to main (4)
find fringes display

IDL Align Fiber (#6): Updates for use with new star tracker: Axes were
rotated by 45 degrees (which caused the feedback loop to be slightly
confused, to say the least).

"Align IO on camera" idl display that shows the percentage of flux on
the central pixel (light in the central pixel divided by sum flux on a
3x3 square pixel around it).

I added new pointing models for B/C to the point model spreadsheet.

Some IDL startracker debug utilities -- for instance, try running
'~/Idlstuff_g3.2/Startracker/st_stats.script

Attached is a ps (and xfig file) 
of a useful LOG sheet for observing with IONIC3, 
developed by JDM, JPB, RMG (feel free to use!!) 


MISC COMMENTS
-------------
Earlier observers reported common and annoying problems with startracker
glitches, where the image would be striped or in someway screwed. These
glitches appear to have completely disappeared after removing the extender
board which had been in place for debugging (good!)

Telescope B coupling might also be hurt by vignetting in beam train which
depends on position on LD1.  This can theoretically be adjusted using
Short Delay Yaw, however Short Delay Yaws are not often adjusted because
of a flakey electronics connection to one of Yaws (don't remember which
one -- I had marked it on the alignment pages but apparently these have
been replaced and so ...?) My advice: DONT RISK IT (DO NOT ADJUST SD
YAW!) [note: both rafael and I have replaced the offending picomotor and
check spring tension -- NOT IT!  the most likely explanation is bad
cabling.]

Home switch for North/A Telescope Tilt Axis was replaced. If you use an
old Pointing model, then be aware of an offset.

Someday we should just buy some laser filters to remove this long delay
laser light problem!

The PICNIC dewar seems to get colder (outside metal) than I would think.
Maybe the vaccuum is getting soft -- somebody might think to pump it down
again soon.

Using broad H, the 6-pixel image of the IO combiner seems a little off --
the scale is slightly wrong and so one can not optimize all pixels
(currently, about 70% of flux is going into each central pixel).


"Easy"* things to do that would greatly improve efficiency!
----------------------------------------------------------
       *something is always easier if someone else does it

We keep talking about putting the Star tracker mirrors on sensible x-y
mounts (not these horrible 3-screw 120 degree axes mounst) with
picomotors. This is critical for future automation and would allow one
person to do the alignment by looking at pupil through alignment telescope
and optimize pupil! or by using an IDL program to optimize flux on star
tracker -- the possibilities are endless!

PETE: it would be a great time save if the PM Search started from the
center and moved out rather than the left-bottom corner -- I know you are
swamped, but this improvement would really speed upa cquisition of faint
targets! [currently the sweep is too fast to 'catch' the star for slow
integration times, even lowering the Search Step]


Good luck to next observers, and it would great to see similar end-run
reports for other observers!  Send to iota_all [mailing list]. 

Send corrections/comments to me and I will update the report for Web.