[evlatests] Subreflector Flagging Issue

[Rick Perley]

    I have reviewed about two dozen 'QA2' reports from the calibration 
pipeline.  About twenty of these are at L-band. 
    Part of the QA2 output is a very nice plot of the on-line flags, 
separated by time, antenna and color-coded by cause. 
    Review of this shows an interesting problem, affecting about half of 
the L-band observations:
   
    *  ea28 is flagged for 'subreflector position' (i.e., rotation, not 
focus) error for typically half the observation duration.  Now, what 
makes this curious is that for all of these observations, there is only 
the single band requested -- so the subreflector can not have moved 
between scans.  Yet it's flagged for incorrect position on about half 
the scans.  Other antennas occasionally show this (but with much less 
than half the data being flagged on any one script).   

     I have not attempted to load any of these affected runs to see if 
the visibilities are actually materially affected when the on-line flags 
are set -- but I have a sneaky suspicion (based on experience with my 
own test data) that the visibilities will look just fine. 

    Is there an explanation for the problem noted?  And is there a fix?