[evlatests] Switched Power Variation with Elevation or Temperature

[rperley at nrao.edu]

The gain variation tests done Friday afternoon/evening were taken under
ideal weather.  The source was 3C345, and observations began at transit at
5PM and ended when the source set around 11 PM.  The ambient temperature
at the beginning was in the high 60s, and was near freezing at the end.

It was noted that the PDif values -- which are primarily used to monitor
gain variations in the electronics, rose steadily at K-band throughout the
observations -- with a function that looks remarkably like the inverse of
the temperature.  Does this rise really reflect a gain change, or is it an
artifact of some temperature sensitive component (such as the noise
diodes) which do not reflect a gain variation?

I have now reviewed all the PDif values at all eight bands for this run. 
A very clear relation is found:

The Ku, K, Ka and Q band values for PDif all rise during the period, and
all by the same amount -- ~10%.  There is some evidence that the BD
channel has a greater rise than the AC channel.

The L, S, C, and X band values are all dead flat during this period,
except during the last hour or two, where for some antenna-IFs, the values
drop.  This is almost probably due to the 'PDif Compression' problem,
since at very low elevations, the system temperature rises dramatically at
all bands.  (Curiously, this drop is not seen at the high frequency bands
-- or if it is, it is masked by the more general rise noted above).

So we must suspect the UX converter for the origin of this rise in
switched power.

The question of whether the change in PDif at the high frequency bands
represents a *real* change in system gain is harder to answer than one may
think.  We have competing, elevation-dependent effects:
   a)  Atmospheric absorption, which mimics a gain loss with increasing
elevation.  I have used the latest opacity functions available in AIPS,
based on the weather information, to remove this effect.
   b) The antennas lose forward gain at low elevations, due to dish
deformation, and to any errors in our attempts to implement the
'subreflector rotation trick'.