[evlatests] 'Global' Phase Jumps, continued ...

[Rick Perley]

    The presence of both Barry and Ken at the WIDAR correlator meeting 
meant there was a long period of Software time available during the 
afternoon yesterday for stability tests.    I used 4.3 hours of this 
time to watch a single source, at X-band. 
    The primary purpose for this experiment was to measure the bandpass 
phase and amplitude stability.   Data were taken in a single IF ('A'), 
with 25 MHz BW, with 32 channels.  All EVLA antennas were included, 
although 24 was available for only 30 minutes at the end, and 26 was out 
of the array for slightly more than an hour at the beginning.   I 
utilized 3C345, a 7 Jy nearly point source calibrator.  To remove 
residual effects due to  its structure, an accurate map of the source 
was utilized in the calibration procedure.

    This report focuses on the antenna phase stability.  The bandpass 
characteristics will be reported upon later. 

    Overall phase stability for both VLA and EVLA antennas was 
remarkably poor, with minute-to-minute changes of up to 60 degrees 
visible throughout.  Despite the magnitude (and the apparently stable 
and dry weather), these are very likely due to tropospheric 
fluctuations, as the magnitude grows with baseline length, and the 
variations are smooth on the sampling timescale of 3.3 seconds. 

    A number of VLA antennas have large phase slopes (over 100 degrees 
over 4 hours):  2, 7, 20, 21, and 27.  All of these are I believe 
recently moved -- it would seem that a baseline run is needed. 

    Six 'global' phase jumps were observed --- where 'global' means that 
the jumps can only be seen on VLA to EVLA baselines.  All six jumps 
occured *in the middle of scans* -- none were in any way identifiable 
with scan boundaries. 
    The six jumps come as three matched pairs -- the phase 
discontinuously (within 3.3 seconds) changes by a large value (the same 
for all EVLA antennas w.r.t. the VLA, or vice versa), then returns to 
the original phase some fairly short time later.   The durations of 
these jumps were ~ 24, 14,  and 14 minutes each. 
    Most curiously, the 'return jump' was of a smaller magnitude than 
the original -- yet the phases returned to the original level!  This 
apparently contradictory statement is true because the phase between the 
VLA and EVLA drifts during the 'jump' period!!!
(If this isn't clear, come see the plots I have).