[evlatests] More on 12.5 MHz spectral line noise ...

[Rick Perley]

    Ken returned the system to January 2007 state, and ran the same file 
that we have used in the past to test the noise distributions in various 
spectral line modes.  The date is potentially significant, as the last 
time we are sure the noise distributions were correct was in late Dec 
2006. 
    The various modes and BW were observed twice each, once with 
self-test on, once with it off. 

    The mode 'PA' data in this latest test are more bizarre than ever. 

    I calibrated and plotted three of the tests:

    A) Mode 1A @ 50 MHz BW.  The channel width is 3.125 MHz.  For both 
selftest on and selftest off, the sigmas of the noise distributions are 
the same:  0.178 Jy.  This establishes the standard -- all others should 
scale inversely with the sqrt of the change in channelwidth.

    B) Mode PA @ 25 MHz BW.  Here the channelwidth is also 3.125 MHz.  
The noise distributions should be the same.  And they are, for both 
selftest on, and off. 

    C)  Mode PA @ 12.5 MHz BW.  Now the channelwidth is 0.781 MHz.  The 
noise should be two times greater -- i.e., about 0.34 Jy.  The table 
below gives the remarkable results:

   Self-Test     RR    LL    RL    LR
------------------------------------------------
        On        1.77    1.70  0.57  0.58 
       Off          .67     .66     .34    .35  
-------------------------------------------------

    Only for the cross-hands, and only with self-test off, is the noise 
correct.  All others are wrong, by varying factors.