[evlatests] 12.5 MHz BW spectral line mode noise

[Rick Perley]

    All is *not* well with these modes.  To test if turning off the 
self-test has 'fixed' the extra-noise problem, I observed a blank field 
in C-band, in continuum mode, and in mode 'PA' with BW = 25, 12.5, and 
6.25 MHz.  The data were calibrated using 3C147.   The channelwidths 
provided are 3.125, 0.781, and 0.195 MHz.
    Data quality was excellent -- even outstanding -- for all 
observations.  The only complaint (other than 5B, which is already 
fixed) is with the 'short shallow drop' problem, which affects most EVLA 
antennas (but with negligible effect on sensitivity). 

    Noise histograms were made of the imaginary part of the Stokes I, Q, 
U, and V correlations of the noise field for the EVLA antennas.   In all 
cases, the width of I equaled that of V, and Q equaled that in U.  All 
four should have equal width, but as the table below shows, they do 
not.   The values given are in mJy. 

 BW                  I, V                  Q, U                       
Comment
----------------------------------------------------------------------------------
50 MHz          17.0                     16.8                    Actual 
BW ~ 44 MHz
3.125              60.0                     60.0                   
Exactly as expected
0.781             200.0                   130.0                  Q, U 
are correct.  I, V are 50% too high.
0.195             250                      255                     
Exactly as expected. 
-----------------------------------------------------------------------------------------------

I can't confirm if the 'self-test' was actually off.  But if it was, we 
still have a problem with extra noise in Stokes I and V (actually, in 
the correlations RR and LL).