MODS1 Commissioning Run 7: UTC 2011 April 17
Observer(s): R. Pogge, R. Stoll (OSU), M. Fumana (INAF)
Telescope Operator: G. Bechetti (LBTO)
AO Support: Remote from Tucson as needed
Telescope Support: M. Midkiff (LBTO)
Instrument Support: D. Officer (LBTO)
Observed all night. Winds were light. We had a few bands of cirrus blow by early and late, otherwise it was fairly clear. The bright moon
was the biggest obstacle, as the notes show at various times early when cirrus was a factor. Seeing varied from 0.6 to 1.5.
We have about 5 episodes of rapidly growing Z22 on the primary that required a reset.
We accomplished our main goals for tonight of getting the kinks out of the target acquisition process, and trying a couple of more
involved acquisition procedures (blind offset at various PA, A-minus-B differential acquisition with offset), as well as finishing up the
remaining stars in our growing standard star library. We also got the last of the twilight flat so we now have all the calibration comparison
data we could not acquire before the baffles were installed.
Overall a highly successful run. Too bad they could not all be like this. We got through all of the major items in our commissioning
plan, and next up is a whole bunch of data analysis and reportage.
Our sincere thanks to everyone who made this run great!
Minor modification of acqMODS based on last night's experiences.
All times are UTC
unless otherwise noted.
- Nice at twilight, warmer than last night (+10°C) and light winds. Some clouds off to the NW of us that might encroach
Took twilight sky spectra with the LX5x60x0.6 long slit and the Prisms (dual, red-only, and blue-only)
- Setting the pointing coeffs on a bright star
Rotator fault during the first preset. Master Control Relay on the LDG rotator is off. Might be a left-over from an MCR fault this afternoon.
Locking elevation at zenith and going into the upper-right tree house to reset the MCR on the rotator controller.
- Running again after the reset.
- Collimation Star - BS9134
Relative slow convergence, with a bit of dome seeing. Probably 1-arcsec at the start with some settling time ahead of us for the system
to calm down and the enclosure to vent and stabilize.
Spectrophotometric Standard Stars
library target - PG0823+546
Some cirrus going over, but we have this one from a past photometric night in grating mode, so this will give us a good shot in all modes,
and give us something to exercise the system while the telescope and mirrors equilibrate and settle. Running in the LS60x5 spectrophotometric
Combination of bright moonlight and passing cirrus led to the GCS grabbing the bright bar and spoiling the guiding.
- Tried Feige 34 - GCS acquire fail due to bright background. Cirrus now covering a good fraction of the sky with bright moon.
- Tried GD140. GCS acquire fail due to bright background. Also the TCS left us at Az=447° after the preset, so we are unwrapping.
We seem to be getting very close to azimuth wrap limits a lot on this run. Geno comments that he's not experience as many instances of
ending the presets so near wrap limits before. Something about BP10???
- Trying Hiltner 600. This one is down at 1.6 airmasses but well away from the Moon and in a relatively clear patch of the sky.
Before committing to the spectroscopic observations, we took u and z images through the LS60x5 slit in dual-imaging mode to check on
atmospheric dispersion at this airmass. With the 5-arcsec slit we are still well inside the slit at the extreme wavelength limits, so no need
to modify our targeting for a different PA to match the parallactic angle.
Took the opportunity of being at low elevation to check the IMCS tuning. Will consider revising the zero points for the stock IMCS Collimator TTFs
from Zenith where we measure now to a median set based on the elevation/rotation IMCS map from a couple of months ago. Note that these
will need to be measured and changed anyway when we do the final realignment of the blue channel before the May run, and this is a routine
instrument "maintenance" measurement/calibration that would be repeated at least annually.
- PG0823 again, this time in a hole in the cirrus so the background is lower.
On the initial preset the RMS WFE was >1000nm. Geno noticed that Z22 was getting into the bad zone. Applied the Z22 correction
at 0404UTC, and it behaved much much better after that.
Successful acquisition, nice and stable (background low), seeing ~1 arcsec.
- Z22 forces getting large again. Reset between observations. Took a little longer to recover this time. Note that backgrounds are not very high, certainly not high moonlight (we're in a clear spot, and 85° from the moon).
- Feige 34 Std library target
Seeing now around 0.7arcsec and the clouds have passed. Got rapid convergence on the WFS. Slit target acquisition is getting more efficient as we get more practice. A tool to take out a lot of the by-hand steps is up next - we now know what it needs to do and what order of measurements we need.
Imaging Acquisition Script Test
- SN2008es host galaxy - test of an imaging acquisition script, to see if we stomped some bugs in the acqMODS script engine (we've been
so focused on the spectroscopic acquisitions, we missed some tweaks needed in direct imaging mode).
Using the target to hammer imaging acquisition and exercise the instrument configurations, IMCS, etc. Works great.
Blind Offset Acquisition Script Test
- Blind Offset Acquisition
Using the field of one of the Persson standards, we setup a blind offset test. One star was designated as the reference,
another within 30-arcsec was designated as the proxy "faint target". A guide star is chosen that is valid with respect to
the guide patrol field limits for both. The acquisition used the LS5x60x1.2 long slit mask.
The goal was to setup acquisition scripts to do the following
ACQ Script 1: PA=0°
- Preset to the reference star with a guide star (active preset)
- Acquire the reference star and center it on the slit
- Execute and RA/Dec offset to the "faint" target star
- Take a thru-slit image to verify that the "faint" target was correctly landed on the slit
A second acq script was created to repeat the process, but at PA=40° and verify that it
works the same at two different PAs.
After doing this, we used the PA=40° blind offset to put the "faint" target on the slit,
and then issued absolute
DETXY offsets of +10 and -10 arcsec in Y (along the slit) to
confirm that the pointing reference update worked.
Bottom line is that everything worked as expected. This means we have the pieces necessary
- executing blind offsets in acquisition scripts
- properly resetting the DETXY coordinate origin for subsequent along-slit dithers of a faint target
This worked so well (we ran it a couple of times to make sure it wasn't a fluke) that we finished earlier
than the time we'd budgeted for this test.
That was the last really big target acquisition test. Some minor tweaks are needed in the acqMODS
script engine (the update pointing reference was done by-hand rather than sent by the script engine)
to fully implement this, but it looks like we're ready to rock.
Spectrophotometric Standards Part II
- Feige 66 - StdStar
Same as before, went fairly routine. We're on the same meridian as the bright moon, but there is currently no
cirrus and it is far enough away the background is reasonable (thank you, baffle!) wrt the GCS problem.
Seeing about 1 arcsec, but some puffing.
- Feige 67 - StdStar
This one is south of Feige 66, closer to the moon. A little higher background, but not enough to fuddle the GCS.
Seeing a touch worse than before for some reason.
Acquire/Observe Race Condition Test
- Acquire/Observe script race-condition test
This was a torture test of something an observer trying to shave off seconds on the observing might try. The
scenario is that they run acqMODS to start the preset and acquisition of the next target right when the last
readout from the last observation starts. This way the telescope preset fires during readout. We setup a really
short preset (basically hijacking an on-axis and off-axis WFS test preset pair from the reconstructor tests of
previous nights), and then firing the preset as soon as we see the image readout begin on the instrument.
This is admittedly tuned up to be somewhat pathological, but we were curious to see if there were any consequences
- crash the system
- cause failures that were confusing/hard to recover from
- something totally unexpected
The good news is that we had to try really hard to induce a race condition that resulted in weirdness. Weirdness is the
operative word, not badness. At worst we ended up with an odd set of warnings about trying to set the CCD binning and
region of interest readout when the acq script tried to set the
acquisition snapshot readout parameters while the CCD was still reading out (i.e., the acq script tries to alter the readout
parameters while a readout is in progress). Nothing broke, crashed, or caused serious problems, but we did
identify a couple of combinations that could be confusing to an observer not paying attention, but nothing that would
be either catastrophic or result in problems requiring difficult debugging and recovery. It would just act odd, but
later clear itself up. The worst that happened was the acquisition image was 8Kx3K instead of 3Kx3K. But the image
was still acquired...
The situation we setup is somewhat pathological, as we tuned both sets of scripts (acq and obs) to maximize the liklihood of
conflicts. We also did stupid things like fire the preset/acquisition while the last image was still in the last few
seconds of an exposure (read: tired observer jumping the gun), and nothing really bad happened. Possibly confusing,
but no badness.
We cannot protect a foolish observer from everything they might do (like Goethe said, "against stupidity the gods themselves
contend in vain"), but we think we established there are no bad gotchas in trying to do what would occur naturally to a
conscientious observer trying to be efficient.
It also gives us good data to use in planning future optimization of the observing process (our goal in commissioning
is to first establish basic function, then optimize). So far we've done a number of optimizations to boost
efficiency, but these mostly involve systems we have full control of (e.g., parallel vs. serial configuration of
instrument mechanisms, tight coordination of exposures and IMCS operation, etc.). What we were looking at here
involves closing up the choreography of pointing and data acquisition, which involves a system we don't have the
ability to live test often (the full live LBT TCS). Optimizations that involve a high degree of process concurrency
among independent systems in which operations have a component of contingency in their timing are much more difficult
to plan without hard empirical data to inform that planning. For example, filter wheel motions are highly deterministic - we
know how long a move takes. Presets are not deterministic by any model anybody has given us, how long it takes a
preset to finish depends on a number of contingent factors (e.g., how big the slew is, how fast Dave lets the bogies
run this week, what the guide star acquisition is like, etc.), so this choreography will be more challenging. We have
established function so far, and have data to help us consider what further optimization of acquisition and pointing we
Imaging Mode Acquisition Scripts
host imaging acquisition.
With cirrus gone but in bright moon, we're exercising the image mode acquisitions more. This particular run is on a faint target where
we are using the bright moon to simulate looking at a very faint target, including taking pairs of images with small dither offsets
to do an IR-style A-minus-B differential image to see if we can spot the fuzzball. This might be what one would do to acquire
an emission-line dominated but faint continuum source object for spectroscopy.
- during the first sequence of images for the obs script we were testing, we started getting errors on the WFS about
not being able to read images from the camera. The WFS camera image was frozen. Rick opened up VNC on the WFS camera
machine, and we saw the message
ERROR Readout counter stuck
over and over. We saw this 2 nights ago with the guide camera AzCamServer
We did the following to restart:
- stop the GCS
- stop the AzCamServer program
- reboot the AzCam machine (this took longer than we saw with the guide camera computer)
- when the AzCam machine rebooted, restart GCS
- when the GCS finished its startup, resend the preset
This fully recovered function. Note:
neither this time with the WFS camera nor 2 nights ago with the guide camera, did we
power-cycle the Leach controllers.
So far no problems.
- resumed imaging mode acquisition and observing script testing...
- Cirrus threatening from the WNW, will be here soon...
- PTF09bcl host acquisition field
Acquisition of something not stupid faint, this time to do a spectroscopic acquisition for a little change from imaging acks. Cirrus rolling around now, but luckily it is reasonably thin and the moon is far enough away it isn't a hindrance.
- Z22 reset required, the third one this evening thus far
Seeing 0.8 arcsec, but variable, cirrus is blowing by fast, so will be mostly out of our way soon (the thick parts, anyway)
- Z22 reset required again... forces built up quickly.
- Z22 reset required again, same deal as the last two. We're about 30 minutes into twilight at this point
- We're calling it a night, and a run.
- 17 Apr 2011