Subject: Re: [sloan-25mlog 3452] 2.5m obslog 20170410 (SJD 57854)
From: Paul Harding
Date: Tue, 11 Apr 2017 16:01:06 -0400
The gcam data are not at=20
https://data.sdss.org/sas/sdsswork/apo/gcam/
so I can=E2=80=99t help with the analysis.
The problem is that the guider cannot keep up with the tracking error.
Using summed exposures it the worst thing to ~5 sec dead time per read.
Using single 5 sec exposures should be better.
The reason for the variation in behaviour has to do with with other =
things that cause dead time.
Scale and focusing corrections. These take time and positional updates =
do not happen until they are complete.
Paul
> On Apr 11, 2017, at 8:31 AM, asimmons@apo.nmsu.edu wrote:
>=20
> 08:42Z - Refer to the replied 'GuiderErrorAxis9773' plot to this =
nightlog. We
> noticed that at an altitude of 85 degrees, the average errors (bold =
line) in
> guiding were outside of the tolerated range (THANKS TO ELENA FOR NEW =
AXISERROR
> SCRIPT!). Luckily the exposures for APOGEE and MaNGA were at the tail =
ends of
> their exposures, so we waited 1-2 more minute until both projects were =
reading
> out / transitioning before proceeding with testing...
>=20
> As a test to see if the guiding would improve, we decreased the guider =
exposure
> time from 5 seconds to 4 seconds with stack=3D3 at point (1). There =
was hardly any
> change to the error trend (but maybe we should have waited a bit =
longer before
> proceeding with point (2)). At point (2), we stopped guiding within =
the guider
> GUI and promptly restarted guiding (in other words, we did NOT restart =
the
> actor, just the guiding within the GUI within 5 seconds). At point =
(3), we
> increased the guider stack from 3 to 4 (hardly any change). At the end =
of the
> plot, we were at an altitude of 86 degrees. We remained exposing for =
expTime=3D4
> seconds with stack=3D4 for the remaining duration of this plate. The =
combination
> of all of these resulted in the guiding slowly improving before =
switching out
> the next cartridge. After completing these tests, most likely the =
stopping and
> restarting of the guider caused the guiding errors to improve. We are =
puzzled at
> the reason why.=20