[evlatests] Holography Phase Behavior

[Rick Perley]

    Two tests were performed yesterday, to check if the 180 and 90 
degree phase jumps seen in past holography data can be removed by 
appropriate settings of the Flukes. 
   
    To review:  It has been shown that at certain frequency settings 
(chosen arbitrarily), there is a remarkable bandwidth sensitivity of 
these large phase jumps:
       a) With bandwidths of 50 or 25 MHz, no phase jumps were seen.
       b) With a bandwidth of 12.5 Mhz, 180 degree phase jumps were seen.
       c)  With a bandwidth of 6.25 MHz, 90 degree phase jumps were seen. 

    All EVLA antennas do this together, without exception, while VLA 
antennas never do.    But, since the behavior is 'global', it is quite 
possible that the behavior is due to the VLA. 

    Barry, in a weekend 'aha!' message, suggested that the problem lies 
within the VLA's Fluke synthesizers, and that settings of a multiple of 
10 MHz (or perhaps 5 MHz) would cause the array-differential phase jumps 
to disappear. 

    The first test done yesterday was to check on this prediction.  
Frequency/BWs of
    1464.9/50, and 1385.1/50 (which use Fluke settings of 100 and 200)
    1443.025/6.25 and 1363.225/6.25; (which also use Fluke settings of 
100 and 200)
    1453.025/6.25 and 1368.225/6.25  (which use Fluke settings of 110 
and 205)

    were used.

    Sure enough, all 180 or 90 degree phase jumps were absent.   So this 
is strong evidence in favor of a Fluke-origin, and that the EVLA is 
exonerated. 

    However, Barry's formula is not sufficient.  In reviewing the old 
data in which the 180, 90, or 0 degree p hase jumps were present, it was 
immediately noted that the 25 MHz data, which showed no phase jumps in 
all past experiments, were run with a fluke synthesizer setting not a 
multiple of 5 or 10 MHz.  (In fact, they were 112.4, and 212.4 MHz).  In 
reviewing the Fluke setups for all the previous runs, an interesting 
association was found:

    If the Flukes are an integer multiple of 100 kHz, no phase jumps are 
seen.
    If the Flukes are an integer multiple of 10 kHz, 180 degree phase 
jumps are seen.
    If the Flukes are an integer multple of 1 kHz, 90 degree phase jumps 
are seen. 

    This rule gives a (so far) 100% correct prediction of Fluke phase 
behavior.    I offer no explanation of why this rule works. 

    The 2nd test yesterday was to check this, and also to try Barry's 
conjecture that the problem is only with Flukeset #1.  The experiment 
was the same as before, EXCEPT the third frequency setup was

    1465.0/6.25 and 1385.0/6.25,   which uses Fluke settings of 
121.9750, and 221.7750 MHz. 
   
    From the rules given above, we should see 90 degree phase jumps with 
this combination, but not with the 2nd (1443.025/6.25 and 1363.225/6.25, 
which used 100.0 and 200.0 Mhz Fluke settings). 

    And indeed, the 90 degree phase jumps returned for that specific 
Fluke setting.  And, as this was done with Flukeset #2, it seems to be a 
generic characteristic. 

    From my point of view, this problem is now merely a curiosity.  
Although we have one remaining issue to solve (next email), we can 
easily avoid the catastrophic 90 and 180 degree phase changes (which 
completely destroy any chance to use holography for antenna metrology).