[evlatests] Still more on spectral line noise

[Rick Perley]

    More tests were run this afternoon -- same setup as before 
(calibrator, plus blank sky).  This time, I observed in modes:  
Continuum, PA, PB, 4, 2AC, 2BD, 1A, 1B, 1C, and 1D, all at each of three 
bandwidths:  25, 12.5 and 6.25 MHz.  That's 27 different setups. 

    In short, only the observations taken with 12.5 MHz BW, in spectral 
line, give discrepant results.  But the higher rms seen at 12.5 MHz BW 
is not the same for all modes.  I present a table, below, showing the 
SEFD for each setup. 

Mode    BW=25      BW=12      BW = 6
-----------------------------------------------------
Cont.          340          340            325
PA             380      630/370         350
PB             380           580           380
4                370           580           330
2AC          340           430           320
2BD           340          470            350
1A              360          520            360
1B             340           405            380
1C             350           420            380
1D             350           660            340
----------------------------------------------------

    Overall, the values are higher t(by ~10%) than the morning 
observations -- but the cause is very likely the winds, which were near 
stow limits throughout. 

    I can not think of any more useful tests.  Perhaps somebody else 
can.  Better -- perhaps somebody can explain why the use of spectral 
line modes with 12.5 MHz BW causes an apparent loss of sensitivity by up 
to a factor of two.  Note that for all observations, the data were 
loaded as correlation coefficients, and for all datasets, the correction 
required to convert these to flux density were the same.  In other 
words, the apparent loss of sensitivity seen in the noisier maps and 
wider histograms for the 12.5 MHz data were *not* seen in the 
correlation coefficients.  How can this be?