[evlatests] Autocorrelation Spectra are useful!!!

[Rick Perley]

    According to theory, true autocorrelation spectra, when properly 
calibrated using the antenna gains from regular calibration, should give 
us the antenna SEFD, in Jy.  (Actually -- not quite, depending on your 
definition of SEFD.  If we consider the SEFD as a purely antenna-defined 
quantity (and I do), then the autocorrelation values will be bumped 
upwards by losses associated with the digital system).   An easily 
acquired value of SEFD is very useful in assessing antenna and array 
performance. 

    The correlator only can provide, at any one time, autocorrelation 
spectra for half the antennas (right?).  It has been arranged, 
fortunately, that the seven 3-bit antennas provide these 
autocorrelations, so I was able to check the assertion stated above. 

    It works!  The autocorrelation spectra (following gain and bandpass 
calibration) give nice plots with just about the right amplitude (250 to 
300 Jy for Ku-band).  And, the 3-bit autocorrelations are significantly 
higher than 8-bit, as expected due to the losses from the sampler. 

    Confidence that these numbers are indeed related to real sensitivity 
comes by comparing them to the histogram widths I've made for each 
correlator pair.  The autocorrelation spectra show ea06 to have a higher 
SEFD, especially on the LCP side.  This is easily seen in greater noise 
on all correlations involving ea06 in LCP.