[evlatests] 3-bit sensitivity, and power levels

[Rick Perley]

    Following my note this morning about an apparent relation between 
power level and correlator sensitivity, Ken adjusted the input power 
levels to the 3-bit sampler to be more uniform, and re-observed the 
standard source. 
    The results show that these corrections made almost no difference.  
The very poor sensitivities with all RCP baselines to 22 were notably 
improved, but not anywhere near the correct levels (as established by 
the 8-bit path).  All other cross-correlations remained the same. 

    This relative insensitivity to input power level (over the ranges 
supplied) is as expected theoretically. The problem of lost sensitivity 
with the 3-bit system, especially on the RCP side, remains.