[evlatests] A most curious end-of-scan problem ...

[Rick Perley]

    I continue to experiment with our new switched power capability.  
The most recent test (run Saturday) combines 3C286, Cygnus A,and a phase 
calibrator (the plan is to see if we can finally do *real* gain 
transfer, using an absolute source (Cygnus A) to determine the flux of 
3C286).  The script cycled around these objects,, observing each source 
at the four lowest frequency bands in OSRO mode. 

    For C and X bands *only*, the final 17 integrations (17 seconds) for 
each source were completely noise on all antennas and both frequencies 
within each of these bands.  No such effect was seen at L or S bands.  
The duration of the 'dead' data was exactly, and only, 17 seconds. 

    A single exception was found -- the effect was not seen on the last 
observation at X-band -- but this scan was unique since it was the last 
one in the script. 

    In case it is important, I note that the C-band observation was 
followed by X-band, and X-band followed by L-band.