[evlatests] Holography Oddity

[Rick Perley]

    We ran L-band holography this morning,  with the goal of chasing 
down our peculiar L-band sensitivity on frequency.  As usual, other 
issues arose ...

    The file attempted to drive the antennas in a 17 x 17 raster, on two 
different sources (at different elevations, to check if the stepsize is 
indeed adjusting for the elevation).  For each source, three 
calibrations were made, to measure system gain and phase.    Two 
interesting characteristics emerged from this:

    A)  For the first calibration, which did *not* follow a holographic 
scan, the four working EVLA antennas (13, 14, 16, and 18) arrived on 
source over 5 minutes after the VLA antennas.  The three all arrived at 
slightly different times (differing by a few seconds), so it seems the 
issue relates to slew -- perhaps they were all in the wrong wrap, or the 
EVLA script got started late.  Fortunately, the first calibration 
duration was very long, and good calibrator data were obtained.  There 
was one other calibration which did not follow a holographic scan, and 
for this one, all data are normal. 

    B) All the remaining calibrations took place following a holographic 
cut, and for all EVLA antennas, the data show a peculiar effect that 
I've never seen before, and which must be due to the script.  All four 
EVLA antennas arrive 'on axis' (i.e., on the calibrator) about 2.5 
minutes after the VLA antennas.  This is not a slew issue -- the 
amplitudes of the data during this 2.5 minutes show the source drifting 
through the main beam or sidelobes!
    After some further sleuthing, the cause of this odd behavior becomes 
clear.  For this file, I had added about 20 seconds to each of the 
holographic scans, figuring that perhaps the EVLA antennas were more 
confused than the VLA antennas, and needed the time to catch up.  But 
the holography programs are pretty clever -- they pay no attention to 
the stop times in the observe file (news to me!), instead, the use any 
extra time left at the end of any scan as a 'down payment' for the next 
holographic scan, so that if the non-cognizant observer (like me) adds 
unnecessary extra time to the scans, the extra time is accumulated, to 
be used later. 
    And therein lies the resolution of the oddity.  For the VLA, the 
extra time accumulated is used for the next non-holography scan -- for 
my data, the calibrator.  Instead of the requested 50 seconds of 
calibrator data, I got 210 seconds.  However, the EVLA behaves 
differently.  For it, the extra time is used to simply 'sit', and wait 
for the time actually requested for calibration to arrive.  So,  at the 
end of the last holographic scan prior to a calbration, the VLA antennas 
promptly headed off to the calibrator (nearly 2 minutes early), while 
the EVLA antennas took a (well-needed?) rest.   The holography source 
then drifted through the beam or sidelobes for the two minute period. 
   
    It would be useful to adjust the EVLA code to bring it into 
synchronism with the VLA.