[evlatests] Autocorrelation spectral comparisons

[Rick Perley]

    It is very interesting to compare the autocorrelation spectra 
between the three parallel paths:  B0D0, A1C1 and A2C2.  I printed out 
these for the '012' 3-bit setup (it appears all 3-bit combinations give 
very similar spectra), and the 8-bit setup, plotting only the 8 subbands 
common to all three paths. 

    In short, there are more differences between these than 
similarities, even after discounting the 'shoulder slope' of subband 1 
in the B0D0 path.  In general, it's clear that the 8-bit spectra are 
'cleaner' -- in the sense there are fewer 'ups and downs' and 'curious 
features' in the spectral densities.  The previously mentioned 28 MHz 
standing wave is very clear on many, or most, of the 3-bit spectra.  
This is much weaker (but not absent) on the 8-bit spectra, where the 
dominant standing wave is more like 60 Mhz.  This latter periodicity is 
also seen in the 3-bit paths. 

    Probably none of this is particularly worrisome -- all indications 
are that these features are stable, and are easily calibrated.