[evlatests] Subreflector Rotation off since Dec 15

[rperley at nrao.edu]

I had noted in earlier reports that the antenna gain corrections derived
in the 'Flux Densities' data, taken in Jan 2012, seemed notably different
for some antennas than the corrections determined in the previous run (Dec
2010).  Until today, it seemed the problem was isolated to a few antennas.

As part of the effort of getting the corrections into AIPS and CASA, it
was noted that most antennas were notably different in Jan 2012 than
before, with the differences always in the same direction -- the elevation
of maximum gain was much higher now than before:  typically 90 degrees (or
higher) compared to typically 55 degrees.

This encouraged close examination of a special run I took in October, of a
single source from 12 midnight to dawn.  These data very clearly showed
that a significant change has occurred in the antenna gain function
between October of last year, and January of this year.  These results
even suggested that the 'subreflector rotation' trick, which counteracts
the bending of the quadrupod supports, had dramatically changed.  I asked
Ken this morning to see if he could determine whether this important
function was still working.

Ken soon found that all the subreflector rotation coefficients which
describe the necessary rotation as a function of zenith distance had all
been set to zero in the evening of Dec 15.  We have not found out why this
was done.  So, since that day, *no data* have had this real-time
correction applied.

Ken restored the correct correction coefficients at 5PM this afternoon, at
the start of the next science run.

I have carefully compared the gain curves from the October run to that of
January, in order to roughly estimate the effect.

1) At Ku band, and below, the effect is negligible.  The maximum change in
gain at low elevations is only 1 to 2 %.

2) At K-band, the effect is small.  For a corrected 'tuned' antenna, the
change of gain at 20 degrees, compared to the desired maximum at 55
degrees (which is what it is for most antennas) is 10% in amplitude (20%
in power), or less.  For a few antennas (notably ea15 and ea18, for which
the optimum angle was too high to start with), the loss is a bit larger.

3) At Ka-band, the effect is larger, ,typically 10 to 15% in amplitude (or
20 to 30% in power) at an elevation of 20 degrees.  At middling
elevations, the effect is small by about a factor of four.

4) At Q-band, the difference is quite large, and also quite uniform across
the antennas.  For properly 'tuned' antennas, the loss of gain at 20
degrees (and at 80 degrees) w.r.t. the optimum at 55 degrees, is 20 to 30%
in power.  For data taken after Dec 15, the loss of gain at 20 degrees
compared to 90 degrees (the new maximum) is typically 40%, and for some
antennas, over a factor of 2!  (To be more fair, this extreme effect is
only on the clearly discrepant antennas ea15 and ea18). The more typical
maximum loss is about 70%, compared to 30% in the past.
         Put another way:  The loss of power gain (hence sensitivity) for
the typical Q-band antenna, after Dec 15, is about:
         < 10% for elevations above 60 degrees
         ~ 20% at an elevation of 30 to 40 degrees
         > 40% at elevations less than 20 degrees.

There is likely an effect on pointing, but this will largely be corrected
by the routine pointing runs.  The gurus are still discussing the
magnitude of the residual effect.