[evlatests] Some other notes from C-band holography test

[Rick Perley]

    A few, mostly minor, issues were noted from the short test this 
morning. 

    ea19A gave zero fringe amplitude, and was not flagged by the system. 
    ea19B, C, and D all gave good amplitudes, but have very large delay 
errors. 

    ea02  had stable delays this time. 

    ea23 is behaving badly:
        on IF 1, we have a single scan with a 150 degree phase jump.  
(All other scans were fine, and all other antennas showed beautiful 
phase stability).
       on IF2, we are seeing unstable visibilities and monitor data -- 
with slow (timescales of minutes) changes with amplitudes of ~10 
percent.  Tsys claims to change by this amount on these timescales, but 
I don't believe it! 

    ea24 was working fine on all IFs after it returned to service (about 
3/4 of the way through this run). 

    We have clear evidence that referenced pointing is useful at this 
band.  The run started about 8AM, after the sun had baked one side of 
the antennas.  The referenced pointing solutions were large -- typically 
1 arcminute.  And the correction was easily seen in the calibrator data 
following application -- amplitudes rose by typically 2%.  The lesson 
here is that, at least in daytime, users should consider doing 
referenced pointing corrections -- even at C-band -- if they are 
concerned about arcminute offsets. 

    The bandpasses were solved for every four minutes, with very high 
SNR.  Note that we did X-band referenced pointing, so these C-band 
bandpass solutions should show 'switch' effects.  There are some 
interesting conclusions from this:
    a) Most antennas are very stable, with pk deviations from the mean 
solution typically not greater than 0.5%. 
    b) Many antennas showed a small but clear change in their bandpasses 
*following* the first referenced pointing determination -- this is the 
first observation with referenced pointing applied.  I would normally 
ascribe this to the change in band from the X-band referenced pointing.  
But if this were the case, we should see a similar effect following all 
other referenced pointing solutions -- we do not.  The observed change 
is always sinusoidal, with a period of about 50 MHz.  The amplitude is 
small -- 0.5% or so. 
    c) A few antennas have rather poor bandpass stability (compared to 
others):
           - ea18D has large, sinusoidal variations (few percent) which 
change rapidly -- timescales of minutes.
           - ea21, on all IFs, shows a smooth sinusoidal drift in 
bandpass, with the 50 MHz period slipping by one-half period over one  
hour.  Easy to calibrate out, but it shouldn't be there, and no other 
antenna shows this. 
    d) I noted above that ea23 was unstable on IFs B and D -- this 
instability shows up in the bandpass solutions also. A good word for 
these is 'chaotic'. 
   
    Finally:  The switched power data were of outstanding quality 
(judged on the fraction of data which had to be removed).  The PDif 
compression problem is present, as always, with the same characteristics 
as noted yesterday.