[evlatests] X-Band Results from Saturday

[Rick Perley]

    I've looked more carefully at the Saturday data, which did not suffer
the odd 'low-amplitude' problems of Sunday. 

    The primary goal of this test was to look for long-term subtle effects
in gain, stability, closure, system temperature, etc.  This was not 
possible,
due to the extreme winds.  The gain solutions, made on a 3-second time
basis, show all antennas *except one* have up to 10% amplitude variations
which are highly correlated between IFs of any antenna, but are not
correlated in detail between different antennas.  The sense of the gain
solutions shows that in all cases, the variations are in the downwards
direction -- the correlated amplitudes drop and recover, never rise and
recover.  The single antenna which did not show this effect is antenna
13, which we note is proximate to the antenna barn, and is nicely
shielded from SW winds.  So we ascribe the variability to wind. 

    The phases are better behaved than the amplitudes, but show the
expected variations due to the long baselines. 

    The bandpass solutions look normal, as defined by what we have
seen before.  All solutions show the 'Gibbs Ringing' on the zero-frequency
side.  What is notable is that the EVLA antennas show much stronger
ringing the VLA antennas.  I formed an EVLA-only subarray, and
found that the bandpass solutions for this subset show much less
ringing than the solutions from the entire dataset -- indeed, they look
like the VLA-only solutions.  This would indicate that VLA-EVLA
baselines have a much stronger amplitude ringing than either VLA-VLA,
or EVLA-EVLA baselines in the bandpass.  I don't understand why this
should be the case.  Informed opinions are welcome! 

    Closure effects are interesting.   From the full dataset, EVLA-EVLA
baselines show the largest closure effects -- 1 to 2.5% in amplitude, even
after bandpass calibration.  However, the EVLA-only dataset shows
very much lower closure -- 0.5% following bandpass calibration.  This
level is about the same as VLA-VLA closure levels.   (There is negligible
phase closure for any combination). 

    Sensitivity.  A wide spread is seen.  Antenna 13 is easily the msot
sensitivity X-band antenna in the array, as judged by aipsweights.  Antenna
16 is about equal to the best VLA antennas.  Antennas 14 and 18 are both
terrible, especially on the LCP side (IF C) -- where they are two to three
times noisier than antenna 13.  (In other words,  they have an effective
G/Tsys worse than antenna 13 by a factor of 2 to 3).  Proof that this effect
is real comes from actual amplitude distributions -- the noise distribution
for 16 x 13 (RCP) is two times or better less than 14 x 18.   Part of this
may be due to pointing (as Ken has noted earlier), but I suspect most is
due to troubles with the electronics or efficiency. 

    Plots of the system temperature over time were also interesting.  
This is
a sensitive indicator of performance, and I think we haven't made enough
of them.  The dominant effect, seen on nearly all antennas, is a slow rise
in Tsys, due to the source setting over the 2.5 hours length of the 
run.  The changes
due to going to blank sky (once at the beginning, and once at the end)
are easily seen on most antennas -- 0.5K, which is about right for the 4-Jy
source I used.  The changes when antennas 2 and 5 kept autostowing (due
to wind) are easily seen.  All kinds of disparate phenomena are noted -- too
varied to note here.    I'll comment here only on the EVLA antennas' Tsys
trends. 
       Ant. 13.  Looks very good -- well behaved with the expected changes
due to elevation and source clearly visible. 
    Ant. 14.  The RCP side looks good, but the LCP side is terrible, 
with a high
value (58 K), and variations of 2 to 3 degrees over 30 minute timescales 
which
are not correlated to anything I can think of.  The 0.5K change when 
going off
source (to cold sky) is completely absent. 
    Ant. 16.  The behavior is normal, but the noise in the Tsys 
measurements is
at least twice that of good antennas.  Yet antenna 16 shows good overall
sensitivity, so I wouldn't expect the noise in Tsys to be due to a high 
Tsys. 
    Ant. 18.  The situation is the same as for antenna 16 -- normal 
behavior,
but high variations in the measured Tsys. 

    I'll note here that the noisy Tsys measurements on 16 and 18 are not 
unique
to them.  Many VLA antennas show the same problem.  In all these cases,
I have used the backend Tsys. 

    I will likely repeat this test next weekend, presuming the wind is down
and things like pointing, power levels, etc. are all tweaked up.