[evlatests] One more (confusing) results

[Rick Perley]

    Ken ran an 8-bit path test, turning off the fringe rotators, to 
compare the noise levels with those with the rotators on.  The goal was 
to see if there was a similar, mysterious reduction in the apparently 
noise when the rotators are off. 

    There is!  In fact, the rms noise levels with the fringe rotators 
turned off, are definitely below theoretical, given what we think the 
antenna sensitivities are.  With the fringe rotators on, we get, for all 
baselines except those with 12R, a noise level of 0.140 Jy.  With the 
rotators off, these values drop to 0.115 -- a ratio of 0.82, the same 
ratio we get with the 3-bit path. 
    For the record, the rms noise expected is given by, for a single 
polarization product:

    SEFD/(CorrEff.sqrt(2.B.t)), where

    SEFD = antenna system equivalent flux density:  about 250 Jy at best 
for Cband.  CorrEff is the correlator efficiency -- perhaps 0.92?  B is 
the BW -- 2MHz for these tests, and t = integration time -- 1 second.  
Plugging these numbers in, we expect  0.135 or so -- as we see with 
rotators on.