Date: Tue, 13 Mar 2001 09:09:33 -0800
From: Russell E Owen
Subject: Re: 2.5-m Obslog 20010312 (MJD 51981)
>PROBLEMS IN DETAIL:
>===================
>
>TCC/Az-Alt-Rot:
>AxisDTIme:alt (-0.5: Mar 12 21:09:56 MST)
>AxisDTIme:alt (-0.6: Mar 12 21:09:56 MST)
>AxisDTime:alt (-0.7: Mar 12 21:13:12 MST)
>AxisDTime:az (-0.5: Mar 12 21:14:07 MST)
>AxisDTime:rot (-0.7: Mar 12 21:14:07 MST)
Ouch! The axis controller clock should always be accurate in the
fractional part due to its synchronization with the 1 Hz pulse. Hence
I expect to see errors of a few 1/10ths of a second if things are
working well, and some integer number of seconds +/- a few 1/10ths of
a second if SET TIME is broken in the MCP (as it was for awhile).
Having the fractional part of time off by nearly 1/2 second is a very
different thing, and quite disturbing. This suggests two different
scenarios, neither very comforting:
- The TCC clock has drifted. The TCC sets itself from the GPS every
hour, so this seems unlikely.
- the MCP is not correctly synchronizing to the 1 Hz pulse. Again,
this does not seem very likely, though I have no idea what that code
looks like.
Here is a test I would like you to perform next time you see an
AxisDTime with a fractional part near 1/2 second (e.g. 0.4, 1.6,
-1.5...). Please execute the following TCC commands:
AXIS STATUS
SET TIME
AXIS STATUS
AXIS INIT
AXIS STATUS
This synchronizes the TCC's internal clock to GPS and determines what
affect that has on AxisDTime, then re-initializes the axes (which
sets the MCP's clock) to see if that helps.
Thank you very much for the clear report. Is there a PR I should
reference or update?
-- Russell