[evlatests] All sensitivities are fine

[Rick Perley]

    I reported last week that sensitivities appeared to lowered by 
factors of a few, based on data taken in April with 0.417 seconds 
averaging, while data taken with 1.67 seconds provided correct 
sensitivities.   These observations were taken on a strong source. 

    To check on this surprising result, I ran an hour-long test 
yesterday evening, observing two sources (a strong calibrator and a 
nearby blank field), with three different averaging intervals:  0.417, 
1.67, and 6.67 seconds.  Equal durations were spent on all targets (i.e. 
about 10 minutes each), so we should get the same sensitivity in all 
images. 

    Results:

    All noise histograms are perfect -- perfectly gaussian, and of the 
expected widths for the bandwidth and time durations.
    All images rms's are sqrt(N) smaller than the noise histograms. 
    Using the observed SEFD values (from calibration -- the data were 
filled as correlation coefficients) of 300 Jy, the predicted values for 
the sensitivites are exactly matched to the observed values. 

    Conclusion:

    There is nothing wrong with the system. 

    To explain the surprising April results, we must invoke one of:  (a) 
An error in our data calibration process, or (b) An effect related to 
the strong calibrator observed, or (c) Something was wrong with the 
system then.