K. Sowinski
Several subtle, and not so subtle, errors were corrected in this
update. I want to describe them in a broad forum because their
effects have at various times been described as not reproducible and a
sizable body of folklore has grown which ought to be dispelled. Most
of the problems to be described are interactions between multiple
subarrays and have become more obvious with the recent increase of Q
band observing. That all of these errors have been in the system
since the dawn of time makes it clear that there is only a limited
combination of circumstances which I can test in a finite amount of
time and it is always possible that there are untested situations for
which ANTSOL, or pointing or, indeed, almost anything else may fail.
It has long been noticed that when two or more subarrays experience a
source change at the same time that some antennas will not receive the
necessary commands to configure the subreflector. The cause of this
was understood, but repairing it took a long time to percolate high
enough on the list of things to do because it is a situation which is
easily avoided. I have now made an attempt to correct this problem
and have not seen it occur over a period of strenuous testing.
Testing with the software in use prior to the update produced many
errors over a similar interval of time. This means that there no
longer need be any artificial restrictions on stop times for scans in
multiple subarrays.
Despite the fact that determining pointing offsets always works
reliably for pointing runs, the process has seemed to be less reliable
when determining pointing offsets for use in referenced pointing. A
serious problem was found in ANTSOL which probably explains most of
the pointing failures when two or more subarrays are in use. If a
calibrator scan is in progress in one subarray and another subarray
begins a calibrator scan, ANTSOL will produce one corrupt solution for
the original subarray when the second subarray produces its first
record. This is not of concern unless the ANTSOL solutions are
actually being used for something. There are two cases where this is
true; when doing pointing, and when doing an active phased array
observation (not VX). This interaction between subarrays has now been
removed and ANTSOL has been shown to behave with true independence
when two or more subarrays are observing calibrators. I examined the
results of a recent two subarray Q band experiment and found that all
but two of the failed pointing scans can be attributed to this
problem. The two remaining failures were due to operator error and
another problem in ANTSOL to be described in the next paragraph.
The remaining problem in ANTSOL was evident even in the one subarray
case. At IAT midnight, if a calibrator scan is in progress, ANTSOL
stands a good chance of getting confused and doing nothing for one or
two minutes. If this should be a pointing scan, then it is likely
that the scan will expire before ANTSOL recovers itself and no
pointing results would be obtained. This, too, is now fixed.
In summary, I expect the system in general to be more robust while
doing multiple subarray observations and, in particular, to do a much
more reliable job of referenced pointing for Q band observations. I
encourage you to let me know, in as much detail as is possible, if
this should not turn out to be true. Complaints may not always get
things fixed, but not complaining is even less likely to do so.
======================================================================
|
