[evlatests] No gain change?

[Rick Perley]

    I reported earlier on the remarkable drop in rms noise when the 
fringe rotators are turned off (in 3-bit mode).  I also noted that we 
couldn't rule out the possibility that the gain changed, since turning 
off the rotators destroys the coherence of the source flux. 

    This coherence loss won't affect the autocorrelations, so I looked 
at these, from times when the rotators were on and when they were off. 

    Autocorrelation amplitudes remain the same for both.  Confidence 
that these are actually meaningful numbers is provided by noting that 
they change between the 'on-3C286' and 'off-3C286' observations by about 
the right value -- 5 Jy (and remains this for fringe rotation on and 
fringe rotation off). 

    I thus claim -- much more strongly now -- that the noise reduction 
observed when the rotators are turned off is real, and is at about 20%. 

    Ken is now repeating this experiment using the 8-bit path. 

    Results should be ready shortly...