[evlatests] 2GHz-wide passbands

[Rick Perley]

    Time for some good news. 

    I've looked at one of the 3-bit 'wideband' tests taken yesterday 
afternoon. 

    I removed a multi-subband delay from each, and then solved for the 
absolute bandpass.  The results show a smoothly varying (across 
frequency) phase and amplitude across the entire 2 GHz subband width. 
    There is negligible overall slope in the amplitude -- the adjustment 
by the 'slope filters' has worked well. 
    The maximum variance in spectral power is quite small for all 
antennas except 22A.  For the others, the range is about 5 dB.  For 22A 
it is 7 dB. 
    The 'mean' spectral power, as judged by these cross-power bandpass 
solutions looks like this (in dB)

    Mean Sp. Power

12A      -24
12C      -21
15A      -21
15C      -20
22A      -28
22C      -21
28A      -23
28C      -22

    Note that, once again, 22A stands out.  It has (by far) the lowest 
correlated signal, and it has the largest variation in its bandpass.  It 
also has -- by far -- the largest rms noise in the correlation 
products.    Presuming that the analog power meter being used to set the 
attenuators does not lie, there is within this particular antenna, prior 
to the sampler, a large additional source of uncorrelated noise. 

    I tried to look at the autocorrelation spectra -- curiously, these 
were only available for antennas 12 and 22.  I thought we were to get 
all of them now, with the change of antenna 27 to 12? 
    Amongst the total power spectra available, there are some bizarre 
problems ...

    Rick