[evlatests] Time-variable switched power at C-band

[Rick Perley]

    The flux densities run is providing strong evidence that we have a 
time-variation, with a 24 hour period, in the antenna gains! 

    The data are of excellent quality.  Following the same procedure as 
summarized yesterday (for X-band), the PDif values were applied to the 
visibilities to correct for any variations in system gain.  At this 
band, elevation gain effects are known to be very small, so a plot of 
system gain as a function of time should provide a clear measure of 
basic stability.  The primary calibrators used (3C48, 3C138, 3C147, 
3C286) have well known flux densities. 
    Plots of the values of the switched power show a strong (7% 
peak-peak) variation in the values of PDif.  (It's not possible to see 
this in the PSum values, since variations in the ground spillover and 
atmospheric emission are larger than the expected variations). 
    Plots of the antenna gains, after having the PDif values applied, 
show a similar variation -- *** but of much smaller amplitude*** -- only 
about 1% (pk -- pk). 
    Plots of the values of antenna gains, without applying the PDif, 
show the same variations, but with opposite sense, as seen in the PDif.  
The visibility amplitudes are lowest in the afternoon.  The amplitude of 
the variation is larger on the first afternoon than on the second -- the 
first day was sunny, the second was cloudy. 
    The effect is not due to elevation -- the AIPS program ELINT could 
find no significant solutions. 

    From this, it appears we have a temporal dependence of gain, likely 
caused by temperature effects.  Most of the effect is removed by the 
application of switched power -- but there is a small residual, which 
presumably is due to the switched power itself reacting to temperature 
variations. 

    Note that no such effects were seen the the X-band data.  The two 
receivers are adjacent, separated perhaps by 40 cm.