[evlatests] EVLA long-term phase stability.

[vdhawan at nrao.edu]

                EVLA long-term phase stability.
                _______________________________

Summary:

This exercise started as an evaluation of the L352 round-trip phase
meters. The four units in the field are close to meeting spec, and
improvements are expected in the next 2 units to be deployed soon.
The fiber (at least the part that is monitored) is well behaved.
However, the RTP data do not account for the interferometer phase
behavior.

Two signatures are present, slopes with time and variations with
elevation. The RTP data account for the slope at X band. At C and L
bands the situation is not clear. The elevation dependent
interferometer phase is 10-30 times larger than the RTP phase scaled
to X-band (5-10deg). Temperature changes with elevation (i.e, in
30-minute cycles during the baseline runs) were found by Ken and are
being further investigated.  Obviously we re-test after the
temperature changes are fixed.

I go into a bit more detail below.

EVLA antennas were included in VLA 'baseline' runs to refine the
antenna locations.  The best behaved VLA antennas have residual phase
rms of 10deg at X-band (1mm).

EVLA antennas initially had upto 100mm position offsets, which Ken
improved to ~10mm by adjusting the longitude used for EVLA.  Since
then, several such runs were done as listed.  Only EA 18 can be well
fit, with position residuals of 1mm.

UT start,      dur.  band  EVLA ants.

May 21, 14:20  5hr    X   13 14 16 18

May 28, 00:00  5hr    X   14 16 18.

May 29, 17:40  6hr    L   14 16.

Jun 09, 22:40  3hr    C   13 14 16.

Jun 16, 05:40  5hr    X   13 14 16 18 24.

Jun 21, 00:00  2hr    C   13 14 16 18.

Jun 21, 02:00  5hr    X   13 14 16 18


1. Phase slopes with time.
__________________________

There are steady phase drifts of 20-40deg/hr with constant rate over
many hours during a run. These must be removed before position fitting.

At X band, the slopes are:
*  The same on all IF's A, B, C, D;
*  Pretty similar on all EVLA antennas.
*  Roughly equal to the RTP slope scaled to X-band, with factor 2 to
   account for LO and IF path. The slope changes sign from night to 
   day, as expected.

At C band, the slopes are nearly absent - lower than the C/X freq ratio.

At C and X band, the difference phases A-C, B-D, A-B, C-D are flat.

At L band, the slopes are:
* Higher than expected from the L/X ratio, by a factor of 2-3.
* Different at EA14 and 16 (the only ones with L-band baselines data).
* A-C and B-D  i.e. R-L phases are flat at both antennas.
* A-B and C-D phase diff has a slope of 30deg/hr on both antennas.
  (Is the L302 sideband different for A and B?)

Can one mechanism give the same slope on all antennas/IFs at X-band,
cancel at C band, and give different slopes on different IF's at L
band? I don't quite see how, but someone clever may be able to explain
all the above as the diurnal temperature slope acting in various
combinations on the fiber, the L300, L305, L301, L302.


2. Phase variation with elevation.
_______________________________

These amount to 50deg at EA13,16,18; 360deg at 14 and even more
at 24 (X-band).

Equipment where the IF signals take different paths can be excluded
since the IF phase differences (A-B, A-C, B-D, C-D) show < 5deg
elevation change at any band L, C, X. Thus e.g. amplifiers, freq
converters, some cables/fibers etc are exonerated (unless the
variations track within a few %).

Some remaining causes (common to all IFs) could be, e.g. 

* Change in the phase of the master LO with temp/elev.

   I retrieved the L305 temperatures and oscillator control voltages,
   and see variations with elevation. They are out of step with the
   F320 temperatures. I'll show some plots at the Monday test meeting.

* Antenna position offset:

    At EA18, the position fitting works well, both before and after
    the antenna move on Jun 16, and leaves it as good as the best 2
    VLA antennas. I think it unlikely that the elevation signature
    mimics a position offset for 4 data sets at 2 different positions.
    So EA18 is good but for the phase slopes.

    At EA13,14,16 and especially 24, I cannot find an improved
    position in the presence of the other variations. 24 is on the 
    master pad, whose position was updated some months ago but may 
    need another adjustment.

    At L-band, EA14, phase variations scale down by ~5 from X band.
    For EA16, the reduction in phase variation is only ~2.

    At C-band, the phases are a bit better than at X, as expected.
    No surprises, but not enough data to do a position fit.

* Strain or temperature change of fiber with elevation.

  RTP data show that this is small for the monitored part of the LO 
  fiber, 5-10deg at X band. It is unknown but probably small for the 
  IF return, and other cables/fibers in the antenna. 
  Any ideas on how to  measure this?

Conclusion: Blame it all on the temperature till its fixed.
___________________________________________________________________