|
NIRSPEC Reliability Improvement
Progress report: 5
July 2004
|
|
Overview:
The last report suggested three options with regard to how to proceed given
the feeling that our primary goal (greatly reducing the frequency of server
crashes) was within reach. It was decided to continue
the project at a reduced effort level, and to simplify the project plan
to just two remaining tasks. Since events
and schedules have conspired to make an early June full meeting and report
impractical, a short status report is presented here in the interim.
Early in June, NIRSPEC suffered a failure of one of its filter wheels.
In the short term this actually benefitted the project because it allowed
us to have most of the night of June 11 for code testing.
Subsequently though, warming of the instrument was begun in preparation
for repairs. The project is largely on hiatus now as result.
Upgrading the host.
Since no spares exist for the host computer this represents a critical
vulnerability and the project was granted permission to continue with our
efforts to upgrade the host. Unfortunately though, in this
reporting period, other more urgent tasks have diverted resources and there
is no progress to report.
Test and release new server codes.
-
At the time of the last report (May 6) two issues remained prior to release
of the new codes designed to reduce the frequency of server crashes.
-
The rotator GUI reported quantities differently than the current
codes. At the very least this would confuse observers
and perhaps indicated the code was not working correctly.
-
Testing of the rotator performance had been somewhat limited, leaving
some doubt as to whether the requested and obtained PA's matched regardless
of pointing and instrument configuration.
-
Testing on the night of June 11 revealed we are very close to releasing
the new code.
-
Issues with regard to the rotator control GUI appear resolved.
-
About a dozen double stars were observed all over the sky, including
several that passed close to the zenith. Once the PA was set,
the rotator held both stars on slit and no server crashes were experienced.
-
Unfortunately, small but possibly significant, offsets in PA were
necessary in many cases to get both stars on slit. The offsets ranged
from 0 to 2 degrees and averaged about 0.5 degrees.
Proceeding From Here:
As mentioned above, it is possible that an offset exists between
requested and obtained PA. A number of possible explanations
exist and further testing is necessary prior to release:
-
There is no problem with the code because the offsets are just the result
of differences between present and tabulated double star PA's.
-
There is a problem but the old code would yield the same results because
some parameter simply needs revision.
-
There is a problem with the new code.