[evlatests] K-band sensitivity ...

[Rick Perley]

    One thing is not as good as expected -- the sensitivity. 

    The data form perfectly gaussian histograms with width 2.97 Jy (1 
second averaging, 128 kHz channelwidth).  This is about three times 
higher than it should be.  (Rule:  sigma = 
5.62*Tsys/epsilon/sqrt(2*dt*dnu)).  I've assume 100% correlator 
efficiency. 
    I checked this against X and C band data, taken a few days ago, with 
the same averaging and channelwdith -- both give widths of 0.65 Jy.  

    The map rms is exactly what it should be based on noise statistics:  
1sigma = histogramwidth/sqrt(Nsub*Nchan*Nvis).  For my map, Nsub = 4, 
Nchan = 951, Nvis = 168300.  The observed sigma is 0.12, which is the 
predicted value.