[evlatests] `10-second' phase jumps in Holography

[Rick Perley]

    Presuming the large phase jumps are now a non-issue, we have only 
one remaining annoyance -- the `10-second' phase jump. 

    Holography is done by scanning the beam along some direction, with 
of a number of ten-second pauses.  Antenna motion takes up about 5 
seconds, leaving about 5 seconds of good data.  What we have found in 
the preceding tests is that, on occasion, perhaps one of these 10-second 
positions along one scan has suffered a phase jump  of some tens of 
degrees.  The offset always disappears on the following stopping point. 

    The two tests yesterday have been closely analyzed to provide more 
detailed information on this last (?!) remaining phase stability issue. 

    I find:

    1) There is, on average, one `10-second' phase offset per scan 
consisting of 11 stopping points.  A few scans had none, one had two, 
and all the rest one.  (There were a total of 18 scans, with 4 antennas, 
and two IF pairs -- 144 samples). 

    2) Not all EVLA antennas partake on any given jump.  But all 
antennas that do jump, jump forwards and back at exactly the same 
time.   There is no preference amongst non-jumpers -- antennas 13, 14, 
16, and 18 are equally likely to not jump.  (But by far the usual 
activity is for them to jump). 

    3) The A and C IFs always jump by the same value, as do the B and D 
IFs.  The A/C and B/D values are nearly always different.   The values 
of the jump are different amongst the four EVLA antennas.  The values of 
any particular jump are not 'remembered', and do not repeat on the next 
instance for that antenna/frequency.  I can see no trend, repeatibility, 
or predictability in the jump values.  Phase tracking is as good within 
the 'jumped' data as for the non-jumped. 

    4)  There is no connection between this phenomenon and the 90/180 
phenomenon discussed earlier.  The 10-second effect is not sensitive to 
frequency or bandwidth. 

    5)  The precise times at which the jump occurs is very curious, and 
indeed shows that there are in fact two different effects:
          a)  Most jumps begin 3 to 4 seconds after the beginning of a 
holographic scan, and last until 1 second after the beginning of the 
next stopping point.  In other words, the phase jumps 3 to 4 seconds 
after the first holographic position, and jumps back 1 second into the 
second holographic position. 
          b) But a minority of these jumps begin 11 seconds after the 
beginning of a holographic scan (that is, 1 second into the 2nd point of 
the scan), and last exactly ten seconds -- to 21 seconds into the scan.  
The phase behavior of this class is quite different -- the first two 
seconds are at one phase, and the remaining 8 seconds at quite another 
(which can be the original, correct phase). 
       Note that the jump duration is usually less than 10 seconds. 

    6) There were a total of 18 holographic scans in the two tests 
described above.  In these, there were about 18 `10-second' type phase 
jumps.  Only one did not meet the description given in the preceding 
point.  This one variant occured in the middle of the scan (the 6th 
point out of 11), and was of the variety described in 5a, above (9 
seconds long, single phase).  Within this same scan, there was another 
'class 5a' phase jump, at its standard position, 4 seconds after the 
beginning of the scan.  These two phase jumps, despite their similar 
behavior, were of very different values.  They were separated by 50 
seconds of time. 

   
    I hope this generates another 'aha!' thought out there -- holography 
(and accurate antenna metrology) needs stable phase!