[evlatests] Referenced Pointing Failures and Holography

[Rick Perley]

     We are continuing our efforts, begun last summer, to align the 
VLA's optics for optimal high frequency observing.  Progress is excellent.

     This is high precision work, and it is critical that the antennas 
point as precisely as possible during the raster beam cuts needed to 
determine subreflector positioning.  For this, we utilize the holography 
mode.

     Last summer, we made the happy discovery that the holography mode 
utilizes 'secondary referenced pointing' by default.  Within the 
holography function, referenced pointing is always implemented (with a 
cadence determined by the user).  By default, the function utilizes 
offset pointing values previously determined, and determines an 
'updated' value which is then applied.  What makes this especially 
useful for our studies is that these coefficients are remembered for a 
subsequent holography raster call -- so long as a 'normal' observation 
is not made in between.

     However, another 'feature' of the holography function has been 
discovered.  And this is not a happy one.

     For reasons unclear, the referenced pointing determination within 
the holography function occasionally (about 5% of the time) fails to 
find a solution.  When this happens, it is very clear that the result is 
then to 'forget' the previously determined offsets, apparently setting 
them to zero.

     In some cases, this is not harmful, but in most cases it is, as we 
are making use of the 'memory' to determine small changes in large 
offsets (compared to the primary beamsize) at high frequencies.

     It would be useful if this 'reset to zero' feature were replaced 
with a 'stay where you are' feature when the referenced pointing fails.

     Besides this, we need to find why referenced pointing fails. In 
every case in these recent studies, the quality of the data, the 
strength of the source, the stability of the weather, and the size of 
the offset correction are such that the solution should be easily 
determined.  To emphasize this point -- there are more failures at Ku 
band than at any higher band (including Q).  This cannot be explained by 
any of the normal reasons listed for pointing mode failures.

     Rick