Date: Sat, 2 Oct 1999 13:59:01 -0600
From: Russell E Owen
Subject: Re: 2.5m obslog 19991001 (MJD 51453)
>At about 23:00 we had trouble relaxing the primary. We had moved to
>87 degrees in altitude using the MCP menu. Russell told us to move
>using the track command in the TCC instead. We were then able to
>relax the mirror normally.
The underlying problem was that the primary wasn't at the collimated
position. We're not sure how or why that happened. The procedure
mentioned above was one way of getting the primary to its collimated
position before trying to relax it.
>At about 1:00 we acquired our first star for the pointing model with
>the engineering camera. It appeared as a streak across the CCD, with
>an oblong image (of low counts) near the center of the streak.
>Russell soon diagnosed this as due to an oscillation of the primary
>mirror. This was dealt with by:
>    issuing STOP T_COLLIMATE
>    moving the mirror to the reverse axial limits
>    issuing a RESTART COLLIMATE
>    issuing a SET FOC/INC
>The star images then looked normal and we proceeded with a 12 star
>pointing model, followed with the 48 point model mentioned in the
>first paragraph.
A comment: debugging this problem was difficult. It's possible that
the linear gauges would show it, but they're not meant for rapid data
collection and I confess we didn't try it. Actually looking directly
at the primary appears to be nearly impossible with the baffling in
place. I finally shone a flashlight at the secondary and examined
reflections of the baffling and trusses and saw some of them moving
side to side.
It would be nice to have some feedback when the mirror support system
goes unstable. Perhaps regular output of the linear gauges would be
sufficient.
>Twice during the pointing model, a track command failed, with an
>error message that it failed to converge. issuing the "track"
>command a second time worked.
This is a known problem with the TCC slew computation. It usually
occurs only if the end point is right near a wrap or if the initial
velocity is too high (as would happen if you issue one track command
and then another one when the telescope is at full speed, but NOT
during pointing models). Issuing another TRACK command (just plain
"TRACK") usually takes care of it. (Did this really happen twice? I
only remember once.)
Finally, regarding the model fit...we were taking only 1 second
integrations to avoid saturating the chip. Hence it is not surprising
the model fit more poorly than usual. Presumably a darker neutral
density filter should have been installed for viewing FK5 stars. I
wonder how many NDs we have for the engineering camera, and how easy
it is to tell which one is installed.
Also, the star images were not symmetrical, quite possibly due to the
short integrations, but this is something to keep an eye on. I would
not be surprised if the primary never fully stabilized.
-- Russell