[evlatests] K-Band system stability

[Rick Perley]

    A 6-hour 'holography dry run' at K-band was taken late last week 
under ideal weather conditions.  The data provide a good look into 
antenna pointing characteristics and system stability. 

    Referenced pointing (at K-band) was done hourly.  The target source 
was 3C84, which transits rather close to the zenith (8 degrees ZD), so 
we (inadvertently) got a look into antenna tracking capabilities near 
zenith.   Calibration was done every 15 minutes.   Over the 6 hours, the 
target source rose from 45 degrees elevation, transited at 82 degrees, 
and was at 65 degrees at the end of the run.  During this time, the 
azimuth rotated by over 130 degrees. 

    General Results:

    Phase stability was excellent throughout, with phase changes between 
scans (separation 15 minutes) nearly always less than 20 degrees. 
     Amplitude stability was very good for many antennas, following 
application of switched power.  The best were antenna 1, 3, 9, 14, and 
24, all of which show gain variations (amplitude) of less than 2% 
throughout, indicating pointing errors are less than 10 arcseconds. 
       
    But -- as usual, there are a number of oddities which indicate we 
haven't gotten all systematics under control.  Some of these are very 
peculiar.  Read on ...

    1) It has long been known that some antennas point poorly near 
zenith.  The following five show the effect  -- all were corrected by 
referenced pointing:  ea05, ea08, ea15, ea17, and ea23. 

    2) Early in the observation, when the elevation was near 50 degrees, 
a number of antennas showed a remarkable change in gain while on the 
calibrator.  Four consecutive calibrator observations were affected -- 
for each, the antenna gain changed smoothly over 10 to 20 seconds, by a 
large factor -- up to 15% in power!  Both polarizations were equally 
affected, and the effect was not seen by the switched power -- it is 
almost certainly due to antenna pointing changes.  The effect was seen 
only in four scans, and ended with a referenced pointing determination.  
(But how that could have ended a phenomenon that operated on a 10-second 
timescale is unknown to me).   Closer inspection shows that nearly all 
antennas showed this effect, however for the following ones, the effect 
is very strong:  ea06, ea08, ea11, ea13, ea14, ea15, ea21, ea27 and ea28. 

    3) Ideally, with 'set-and-remember', antenna electronic gains should 
not change during the observation.  Changes in fringe amplitude are then 
due to pointing, atmospheric absorption,  etc.  The switched power 
monitoring shows us how effective this is.  For most antennas, only a 
slow drift in electronic gain is seen, which is nearly always seen in 
the astronomical gains (and is nicely removed with application of the 
switched power).  Antennas with large electronic gain changes -- which 
are corrected by application of SW power -- are:
   
       ea2A and ea2C:  These two 'jumped' in gain by a few percent (in 
opposite ways and at different times) -- these changes were nicely 
repaired by the switched power.

       ea14C:  This one lost 50% of its fringe amplitude over a 90 
minute span -- the change was smooth -- then abruptly returned to its 
normal level.  All of this was precisely removed by switched power.   
ea14A was not affected, so this is definitely an electronic gain change. 

    ea27C:  This one showed a rapid 10% rise in power gain -- also seen 
in the visibilities, which was nicely removed.  The opposite 
polarization was not affected. 
  

    4) Finally, a number of antennas show poorer gain stability after 
all corrections have been applied.  In all cases, the instability 
(typically of 5% or so) are equal in both polarizations, indicating the 
origin is in the pointing.  Note that a 5% residual in amplitude 
indicates pointing errors of 15 arcseconds or so ...
    These antennas are:  ea04, ea06, ea10, ea11, ea20, and ea28.