[evlatests] Final(?) Word on Fluke-induced Phase Jumps

[Rick Perley]

We had the opportunity yesterday to do one, hopefully final, test to 
prove the origin of ALL the phase jumps which occur upon frequency 
changes are due to the Fluke synthesizers. 

Short answer:      They are.

Details:


In this test, we employed the same 5-frequency observing script, which 
cycles the array through five different L-band frequencies, with one 
minute spent at each.  Thus, every 5 minutes we are back at the same 
frequency.  What has been noted in the past was that upon each return, 
the phase between VLA and EVLA antennas advances (or retards, depending 
on IF) by 72 degrees.  On top of this regular advancement, on occasion 
an additional 180 degrees is seen.  There is no apparent regularity in 
the 180 degree phase change. 

One hour of Ken's software time was split into two halves.  For the 
first half, the Fluke synthesizers were put into manual mode, and the 
appropriate frequencies,  for the four IFs, for one of the five 
frequencies, were dialed in.  This means that the four synthesizers were 
'immune' from all computer generated commands, and remained fixed at a 
single setting.  (This also means that all the VLA antennas will be 
incoherent with EVLA antennas for the other four frequencies, and that 
the EVLA antennas will continue to fringe with each other).   If all the 
phase jumps noted upon frequency change are due to recommanding the 
Fluke synthesizers, then the phases between the VLA and EVLA antennas 
will remain constant, (for the one frequency selected).  If, however, 
one or both of the two phenomena noted (the 72 degree, or the 180 degree 
changes) originate with the EVLA's system, they will continue to be seen. 
For the 2nd half of the test, the Fluke synthesizers were returned to 
computer control, to confirm that the basic phase changes seen before 
remain. 

Results:

Completely as expected:

1) With the Flukes in manual, and set to the values needed for 1240 MHz, 
there are no changes in phase between the VLA and EVLA upon returning to 
the frequency on IF 'A' or IF 'D'.  (Due to some blunder, the Fluke 
synthesizer settings for B and C were miscomputed or misdialed.  No 
matter -- these have been shown many times over to track the jumps seen 
at A and D). 
2) Upon return to computer control, all the phase jump phenomena upon 
frequency changes reported in the past are seen. 

Short Discussion:

 From these tests, it seems very clear that all the phase jumps noted 
upon frequency/scan change originate with the Flukes, and that the EVLA 
electronics/software are exonerated.  The phase jump problem becomes a 
'transition issue', and will gradually go away on its own as VLA 
antennas are converted.  (I am told that it's possible the EVLA's 
executor will not generate this problem when it controls the Flukes 
after the Modcomp shutdown, but it seems unlikely, and therefore unwise 
to pin any hopes or plans on it.) 

My own opinion is that nothing more be done chasing this problem down 
(for example, trying to figure out the rules by which 72 degree, or 180 
degree, or whatever degree) changes occur.  Users should be reminded at 
every opportunity that if they change frequency, they must include a 
phase calibrator at that frequency.  As noted by Barry yesterday, the 
'Barry rule extension' by which the VLA phase does not change if the 
Flukes are incremented by a multiple of ten MHz, is an interesting clue, 
but of little practical value for observers.