[evlatests] sub-band to sub-band changes

[Rick Perley]

    ... one thing is clear from the autocorrelation spectra:  The 
'phenomenon' (the sampler resonance) is playing havoc at times with the 
level setting, since the resonance is strong enough on some antennas to 
influence the power levels -- to the detriment of the 'good' channels...

Rick Perley wrote:
>    ... and these power differences are visible in the BPASS 
> spectra...  Some are quite striking!
>
>    I'll also be looking at the autocorrelation spectra -- these are 
> more representative of what the requantizers are seeing ...
>
> Ken Sowinski wrote:
>> On Wed, 7 Dec 2011, Michael Rupen wrote:
>>
>>> We claim to be setting the requantizer gains such that the Psum for
>>> all basebands and all subbands should be ~14.2  This seems to be
>>> mostly working for the 8-bit, but not for the 3-bit side.
>>> This suggests we're not setting the requantizer gains at the right time
>>> in the script -- e.g., we may be adjusting the requantizer gains and 
>>> the
>>> attenuators/gain slope equalizers at the same time, which would 
>>> primarily
>>> affect the 3-bit paths.  This would not explain the other 3-bit Psys
>>> oddities of course.
>>
>> Right.  I plan to check for this later today.  Another
>> (unlikely) possibility is that the 3-bit system is not
>> as stable as the 8-bit.