[evlatests] Applying Switched Power -- another caveat

[Rick Perley]

     At the test-engineering meeting  yesterday, I gave a ringing 
endorsement of automatic application of switched power.

     My single significant caveat was that the PDif values are greatly 
sensitive to RFI, so that we will certainly need good filtering software 
before blindly applying these value.

     I should have mentioned one more -- that the switched power 
mechanism be working properly!  In the October data, there are two good 
examples of what is *not* working well:

     At Q-band, ea26 had zero-mean PDif -- either the diode is 
non-functional, or the 10 Hz switching was off.  Application of the 
(zero mean gaussian noise) would really mess up the visibility data!
     More subtle was ea23 in the RCP at C-band.  The plot of PDif looks 
normal -- but instead of ~ 100 millicounts, the mean level was a measly 
4 millicounts --- and the PDif values for Cygnus A and Taurus A were 
actually negative (this offset due to the PDif Compression).  I have no 
idea how this could have happened.

     This note is not a call for action on these antennas -- these data 
are 11 months old, so it's likely these problems have been found and 
corrected.  It's just a reminder that constant vigilance will be needed 
to ensure good results from the on-board calibration.