[evlatests] Feed alignment, and subreflector rotation

[Rick Perley]

    Holography was run last evening at K and Q bands to determine if our 
troubles with implementing the subreflector rotation to offset 
gravitational sag are due to misalignments of the feed. 

    The short answer is:: Clearly Yes. 

    Details:

    Holographic rasters were run on 3C286 near transit (elevation 80 
degrees).  The raster was 13 x 13, with oversampling of three -- this 
gets us out to the 2nd null, and allows careful examination of the 
data.   Referenced pointing was done to align the main beam.  We are 
looking for the asymmetry in the first sidelobe, indicative of a 
misalignment in the optics. 

    Beam maps were made of the non-referenced antennas (7 VLA antennas 
were utilized as non-moving references, to safeguard against spatial and 
temporal phase gradients).  The holography program solved for the 
equivalent misalignment of the subreflector. 

    The results are very clear.  There are misalignments of up to 11 
millimeters in the subreflector position, largely in the 'Y' (vertical 
direction with the antenna pointed at the horizon) coordinate, with 
almost all antennas having the same sign of the offset.  The measured 
phase error across the antenna aperture shows a sinusoidal variation of 
peak-peak amplitude of ~90 degrees for the ~6 worst antennas -- this 
degrades forward gain by about 1 dB (25%).    For these antennas, the 
beam maps show a large coma lobe -- the first sidelobe on one side of 
the main beam is essentially non-existant, while the opposite side is 
about double its normal amplitude. 

    Most encouraging -- the antennas for which the subreflector rotation 
'trick' appeared to work in the data taken from the previous night are 
exactly those for which the misalignments as measured last night are 
small.  The converse is also true in all cases -- the worst failures in 
the subreflector rotation are exactly those for which holography shows 
the largest 'Y' offsets.  This strongly indicates that the 'trick' is 
coded correctly, and our difficulties with its application are entirely 
due to feed subreflector/feed alignment issues. 

    Ken and I are discussing methods to efficiently measure the 
misalignment through rotation of the subreflector in small increments, 
followed by vertical beam cuts.