[evlatests] C-band tests from 23 Feb 2006

[Rick Perley]

    I obtained slightly over one hour of useful test time last evening.  I
elected to observe only at C-band, on a single calibrator and single
blank field, primarily intended to check the sensitivity of the EVLA
antennas. 

    The test was done in spectral line mode 'PB', providing full
polarization in the B and D IFs.  This pair was chosen as earlier
data indicated the B IF had the full anticipated sensitivity on
antennas 14 and 16. 

    All three EVLA antennas worked well, although various 
interesting effects were noted. 

    A) Stability Issues. 

    With no change in frequency, no phase jumps were expected. 
Antennas 13 and 16 showed none.  Antenna 14 had, in both IFs,
a small (10 degree) phase change on one calibration observations,
which was not seen on the adjacent VLA antenna. 
    Short term phase stability is excellent on all three antennas. 

    Antenna 13 phase clearly shows a large position (baseline) error. 

    A curious amplitude event was seen in the middle of one of the
calibrator scans, on all three antennas at the same time (and on no
VLA antennas).  The characteristic is the same as earlier reported
for L-band data:  A sudden rise in Tsys (rise time of ten seconds or
less), followed by a slow exponential ('RC-like') decline in Tsys
back to normal values.  This decline is certainly due to the Tsys
integration mechanism, so the 'noise event' ends about a suddenly
as it arrives.  The duration of the event is a few, perhaps a couple
of tens, of seconds.  The effect on the visibility amplitude is as
expected from the Tsys-correction machinery in the Modcomps. 
Although the event was seen in all EVLA antennas, it was not
seen equally:  the effect was moderate on 13B and 13D, barely
visible on 14B, very strong on 14D, weak on 16B, and essentially
not visible in 16D. 

    B) Closure Issues. 

    A small (1.0%) closure error is seen on IF B's EVLA baselines.
No measureable closure error (max visible is 0.5%) seen on IF D
baselines. 

    C) Sensitivity. 

    Contrary to expectations, antenna 16B showed very poor sensitivity!
But the cause was quickly found -- the damned P-band receiver
installed on that antenna yesterday was found this morning to be
oscillating -- with a response in band.   I understand that arrangements
are being made to turn the receiver off. 

    Antenna 13B and 14B both showed outstandingly good aipsweights,
indicating sensitivity 30 to 50% better than the best VLA antennas.  I
checked the reality of this by plotting the histogram of the real part
of the noise for this baseline against two different VLA baselines --
one chosen to be 'average' (25 x 28), and one which is poor (4 x 12),
as judged by aipsweights.  The result of this is a resounding vote of
confidence in aipsweights:  There is a perfect correlation between
the predicted sensitivity judged by the weights, and the *actual*
on-cold-sky noise distribution of the data. 

    On the IF 'D' side, antenna 13D is easily the best of them all,
while 16D is only slightly better than the best VLA antennas.  I
think the problem with 16B may be system temperature -- it
seems to be about 30 K, a bit higher than expected. 

    Antenna 14D has poor sensitivity, however -- way down in
the sensitivity histotram.  It is also noted, however, that the Tsys
increments observed when going on-off source are factors of
a few too large (should be 0.6K, is more like 2.5K).  This
amplified increment was also seen going on-off Cyg A -- and
is also seen at other bands.  Clearly, this IF is in a very non-linear
state -- something is way out of spec. 

    All in all -- things are in a pretty good state!