[evlatests] 180/90 degree phase jumps in holography

[Rick Perley]

    Another experiment was run today to close in on the 180 degree (and 
90 degree, as you'll soon read) holography phase jumps. 

    Because we had, in recent experiments, only seen the 180 degree 
jumps at 1275 MHz, and never at 1485 MHz, I set up the experiment to 
repeat this pair of frequencies, with AC at 1275, and BD at 1485 MHz, 
then with these IF assignments reversed. 
    And because we had a faint memory of 90 degree phase jumps being 
seen with a frequency of 1706 MHz, two other frequency pairs were also 
observed:  1275/1706, and 1485/1706 (AC/BD). 

    Data were taken in up-and-down holography, 11 points all within the 
main beam.  Six scans, each of two minutes duration, we made at each of 
the four frequency pairs.  Standard calibration was done at the 
beginning, in the middle, and at the end of each of the four groups.  
Data were dumped at 0.41 seconds.  Because I feared RFI at some of these 
frequencies, the BW was reduced to 6.25 MHz.  THIS IS A CRITICAL 
PARAMETER, as explained below. 

    The data, for all frequencies, and all combinations in all IFs, were 
filled with 90 degree phase changes.  Exactly 90 degrees.   Also seen 
were numerous `10-second' phase changes.  Either Barry's fix wasn't 
applied, or it didn't fix the problem at hand. 

    As always, all of the 180 and 90 degree phase changes occured 
exactly simultaneously for all EVLA antennas, on all IFs.  The 
'10-second' changes were not seen in all cases on all of the EVLA 
antennas. 

    This interesting behavior encouraged me to review all of my past 
holography tests.  One very clear correlation was found from this:

    90-degree phase changes are *only* seen at L-band with a bandwidth 
of 6.25 MHz. 
    180-degree phase changes are *only* seen at L-band with a bandwidth 
of 12.5 MHz
    No phase jumps of this type are seen with a bandwidth of 25 MHz. 
    No statement can be made concerning a bandwidth of 50 MHz -- I've 
not taken data with this BW at L-band.
    The only data taken at X-band was with 50 MHz.  No phase jumps of 
this type were seen. 

    There are no exceptions to the rules stated above.  There is no 
frequency dependence for these rules. 

    Because these big phase jumps are only seen on the EVLA antennas, 
and always on all of them at once, the question of whether the problem 
is with the EVLA or VLA remains open. 

    But there's more!

    A quick perusal of the phase plots encourages a conclusion that 
there is no predictability to when a phase jumps occurs.  But a more 
careful persual shows clearly that the pattern of changes is not 
random.  Some observations for the pundits to consider:

    1)  All calibrator phases repeat correctly. 

    2) The phase jumps one second *before* the end of each calibrator 
scan, to the value seen in the first holography point.  But -- in each 
case, the end of the calibrator scan is one second after a 'ten-second 
tick'. 

    3) The 180 or 90 degree phase jumps are clearly *associated* with 
the holographic scans, but the jumps seen only occur with 20 seconds 
after the start, and within 10 seconds of the end of each holographic 
scan. 

    4) The pattern is always the same at the beginning of a hologrphic 
scan:  the first 10-second point is at one phase, the next shows the 
'ten-second' jump, and the third point at a third phase, which may or 
may not be 90 or 180 degrees away from the first point.  The next 8 
points are always at the same phase.  The last point (or the 11-point 
scan) is always at a phase different by 90 or 180 degrees. 

    5) Exception to the above rule:  When a holographic scan is followed 
by a calibrator, the phase jump at the 11th point does not happen -- the 
phase is the same as the 3rd to 10th point.  The phase, however, does 
jump to the preceding calibrator phase, if that is different than the 
holographic scan phase. 

    6) When the 180 or 90 degree phase jumps occur, they are always 2 or 
3 seconds after a ten-second tick. 

    7) And finally:  All four frequency pairs used the same holography 
pattern -- 6 scans, each of 11 points.  And the phase jump behavior of 
each was identical!  To clarify -- the relative times of the jumps are 
all the same.  Whether the jump was +90, - 90, or 180 may not be the same.