[evlatests] System Gain sensitivity on temperature

[Rick Perley]

    The date for the 30-hour flux densities run was partially selected 
to occur on a day with a large temperature change.  The change from 
night to day on May 02/03 was a bit more than 40 degrees F (from 20 to 
over 60F).  This enables a close check on our system gain sensitivity on 
temperature.  Two antennas were outfitted with temperature sensors (I 
think on the L and Ku band horns) to enable better determination of the 
gain sensitivity on temperature. 
   
    The switched power gives us an good indication of overall gain 
changes -- they are very significant at some bands.  Here is a short table:

    L-band:        ~2%
    S-band:          3 -- 5%
    C-band:         5 -- 9%
    X-band:         6 -- 10%
    Ku-band          25%
    K-band            25%
    Ka-band          12 -- 25% 
    Q-band:           20 -- 35% 

    Notes:

    1) The gain variations are almost certainly due to the 'heat pipe' 
effect (pointed out by Bob Hayward a couple years ago) -- the horns 
stick out into the night air, and are good heat conductors.  The 
temperatures of the post-amps are more affected by the horn temperature 
than the air temperature. 
    2) For all bands, the maximum gain is at the minimum temperature.  
This is as expected for a temperature origin.
    3) The gain change should be corrected for by application of the 
switched power -- *** provided that these devices outputs are either 
insensitive to temperature change, or are maintained at a constant 
temperature.  Application of the PDif values to the visibility data, and 
careful calibration, should enable us to determine by how much the 
switched power outputs are varying over temperature/time.    Analysis of 
the Jan 2012 data (the last flux densities run) indicated a ~2% residual 
at X and Ku bands -- the only bands whose pointing residuals are small 
enough to enable extraction of such a small effect.