[evlatests] Strange System Temperature Changes

[Rick Perley]

    I secured a four-hour observation early this morning (3 to 7 AM) to 
further study EVLA polarization behavior.  As usual, I found other 
problems, not associated with what I was seeking ...

    The observations were of a single source, cycling through four bands 
-- L, C, X, and K.  Data quality in general is quite good.  (Details on 
what is not will be reported separately).   Observations were made at 
high elevation -- about 60 through 82 degrees, with meridian transit 
three hours into the run. 

    However,  the C-band calibration revealed a very odd, global, change 
in ampliutude over the 4-hour period, in that the antenna amplitudes on 
ALL antennas (both VLA and EVLA) dropped by typically 3 - 6 % over the 
four-hour period.    For some EVLA antennas (11, 16, 17, 23 and 24), the 
effect was considerably less.  All VLA antennas showed the decline at 
full amplitude.    The changes were smooth over the timerange.

    The decline is amplitude is clearly associated with an increase in 
system temperature.  All antennas recorded a dramatic rise in Tsys, with 
the same functional form as the loss in amplitude -- typically rising 
from 30 to 50 K through the time period!  By filling the data as 
correlation coefficients, it is clear that the SNR was declining 
throughout the time period, and that the Tsys correction mostly -- but 
not completely -- made the necessary adjustments. 

      I then checked the other frequencies:

       a)  No such Tsys variation is seen at L-band.  Everything was 
steady. 
       b) A similar -- and even larger -- effect was seen at X-band. 
       c)  At K-band, there are antennas whose Tsys rose, and others for 
which it fell, during the same time frame. 

    So what caused this? 

       - Weather can be eliminated.  The skies were reported to be 
mostly clear, and there was no wind.  Perusual of the 'wunderground' 
database shows the site temperature to be steady, and the winds light 
throughout.
       - It is not a purely elevation effect.  The antennas rose, then 
dropped.  Plotting Tsys, or gain, against elevation shows there is no 
direct correlation.  If elevation is involved, there has to be at least 
a two-hour lag.    This cannot be due to opacity. 
       - It's not due to pointing off the source.  Besides not being 
able to make the Tsys rise, the effects are much less noted at K-band.  
(See below). 

       I can only conclude that the antenna system temperature did 
change as reported, that this increase in noise power decreased the 
correlation coefficients, which were nearly, but not quite correctly, 
adjusted by the synchronous monitoring. 
    But as to *why* the system temperatures were changing so 
dramatically at X and C bands, on a clear, calm night -- I have no clue.