[evlatests] Phase Spinning, and more.

[Rick Perley]

    This is a continuation of the short report, sent out last weekend, 
of the
EVLA antennas' behavior from the 4 hours of test time observations
taken early Friday morning. 
    Some more details of the experiment are relevent:  Alternating
observations of 3C48 (the calibrator) and 3C84 (the target), in both
fast continuum, and fast line mode.  There were 7 observations of
each source, in each mode.  The 3C48 observations were short -- 1 minute,
the 3C84 observations long -- 16 minutes. 

    This is a very long report.  I'll give the short form first. 

    Antenna 14:  No fringes, but we are confident it was pointing. 

    Antenna 16:  Good fringes, but rapidly spinning phases on
some of the scans.  Significantly, no loss of correlation is seen on
the scans with phases spinning -- indicating the problem is introduced
following correlation. 

    Details:

    1) Antenna 14. 

    As noted earlier, there was no correlated power seen from this 
antenna. 
    The back-end Tsys data give us a clear indication that the antenna was
pointed correctly -- the change in system temperature expected when
going onto 3C84 from 3C48 was clearly seen each time -- about 2K. 
The system temperature reported by 14A is about 58 degrees -- twice
what it should be.  This IF is, however, the only one giving sensible data.
According to 14B, the antenna temperature is  anywhere from 600 to
2000 degrees.  14C and 14D jump back and forth between two
different states with a ratio in Tsys of 1.75 for
14C, and 2.15 for 14D.   The times of the jumps, and the structure in
the Tsys values are the same for both IFs. 
    The 'low' Tsys for these IFs is 80 and 120 K, respectively. 

    2) Antenna 16. 

    Contrary to what was reported in the Monday morning test meeting,
the antenna phase was not spinning all of the time.  The situation is much
more complex:

    a) The phases *never* spun for observations of 3C48.   Except for
16A (see special description below), the phases within any one scan are
very good -- similar to a VLA antenna.  However, each observation of
3C48 gave a significantly different phase -- far, far larger phase changes
than the adjacent VLA antenna.  3C48 was observed at low elevation --
from 25 to 13 degrees.  However, no reasonable pattern, attributable
to elevation effects, was seen. 

    b) The phases *sometimes* spun for observations of 3C84.  Specifically,
we had spinning phases for the first three observations, stable phases for
the next three, and spinning phases for the last scan.  As for 3C48, the
stable phase states have different antenna phase -- changing by -150 degrees
between the first two stable observations, and rising by 50 between the
last two.  These observations were made at high elevations (from 55 degrees
to 83 degrees).  Again, no sensible pattern with elevation is evident. 

    The phase spinning rates are different for each of the four spinning 
scans. 
It is 175 degrees/sec for the first, 125 degrees/sec for the second, 78 
degrees/sec
for the third, zero for the next three (as noted above), and -120 
deg/sec for
the last. 

    The rates are nearly the same for all four IFs.  They are all doing 
it, or not
doing it, together.  But there are  differences.  Within the first scan,
the phase rate is decelerating by about 1 degree/sec^2.  This rate 
change is identical
for IFs A and B -- it is slightly, but clearly different, for IFs C and D. 

    c)  You would expect that there should be a reduction in the 
correlation
coefficient during those scans with the phase winding, compared to those
where the phase is not winding.  *** THIS IS NOT THE CASE!!!*** 
The 400 msec amplitudes are the same whether the phases are rotating
or not.  Hence, the phases are not really rotating during the 400 msec
integration.  This suggests that the problem may be generated within
the Modcomps, rather than the EVLA antennas.  Ken can imagine a
scenario wherein this can happen for IFs B and C, but not A or D. 
But the lack of correlation degradation is seen in all IFs.

    d) The short amplitude dropouts are seen in most scans.  Interestingly,
but probably misleadingly, there are *no* dropouts on 3C84 when the
phases were not spinning.  That this is a coincidence is suggested by the
abundance of dropouts during most observations of 3C48, for which the
phases were never spinning. 

    f) 16A is a special case.  The amplitudes are swiching back and forth
between two states, differing by 2.5%, (or -0.25 dB in power) regularly,
throughout the entire run.  The system spends 30 seconds in the 'up' state,
then 17 seconds in the 'down'.  The transitions between these states is
less than 1 second.  The 'up' state appears to have two levels
of its own, separated by 1%, with 10 second durations.  A discerning,
or at least an imaginative, eye sees a ~2.5 second sinusoidal periodicity
throughout, of about 1% amplitude.

    g) The remaining IFs for antenna 16 have quite different pk-pk noise
in the antenna solutions made with 400 msec averaging:
    16A   5% (due to the switching effect noted above).
    16B   1.8% (with an apparent 2.5 second periodicity).
    16C   0.6%.  (Easily the best, with no apparenty periodicities).
    16D  2.1%.  A 'switching pattern' similar to, but much faster than that
seen in 16A is apparent.  (Period 12 seconds, amplitude ~1%).