[evlatests] EVLA tests September 1

[Barry Clark]

with appologies for the inclusion of a lot of work-sheet type numbers -
this message was just a handy place to write them down.

Set up for observing at 6cm.  Spent some time getting levels set right
to get about 8-12 rms on the link.

Fringes in all four IFs from antennas 14 and 16.  However, they were pretty
weak in general.  Only 14A and 14B have sensitivities within a factor of
two of VLA antennas.  Delays were quite wrong for 14 C and D, but even
after adjusting that, the statement remains true.

Tried a quick peak of pointing.  Representative powers (from F display)
shown below (off is 15' away):

     -Az    On     +Az     Off    -El    On    +El
14  1.8    7.9     5.8     0.3    2.7    7.1   3.35
16  0.51   0.92    0.31    ---    0.24   0.94  0.58

This suggests at antenna 14 correction for mispointing is about 1.06,
mainly due to the azimuth error.

Next, trying to read source / system at the T5.  Only Antenna 14, IFs
A & B seemed worth trying.  Turned off T5 ALC.  Considerable fiddling
to find best way of doing this, which turned out to be to set the 
antenna brakes, let it drift off source, then watch the step when it
comes back on source.  
Antenna 14 T5 dectctor
     Off         On
A      B      A    B
6.78 5.52    7.52 6.19
6.79 5.58    7.52 6.22
6.81 5.60    7.56 6.26
6.89 5.67    7.61 6.28

This gives a source/system ratio of 0.11 in IF A and 0.115 in IF B

VLA antennas cannot be done the same way, because of the F4 ALC.  The 
equivalent is the reduction in the sync detector due to the added source 
noise.

Typical antenna - antenna 10 Sync det
        Off                   On
 A    B    C    D      A    B    C    D   
3.07 3.25 4.32 3.66   2.86 3.02 3.90 3.32
3.05 3.24 4.30 3.67   2.85 3.00 3.87 3.30
3.05 3.20 4.24 3.64   2.82 3.03 3.84 3.26

Source to system ratios in IFs A, B, C, D  are 0.070, 0.066, 0.12, 0.099

Just to be sure, I redid antenna 14, using the sync det
   Off         On
 A    B      A    B
2.98 2.76   2.64 2.45
2.98 2.74   2.66 2.46
2.95 2.74   2.66 2.45

Source to system 0.12, 0.11, pretty nearly the same as before.

Then to compare sensitivities most directly, I observed suppressing 
Tsys.

Antenna 14 amplitudes for 14 A, B are 10.1, 10.9.
Antenna 10 amplitudes for 10 A, B, C, D are 12.4, 12.4, 16.8, 16.8.

Rechecked pointing:

     -Az    On     +Az     Off    -El    On    +El
14  2.7    10.1     7.1     ---    4.8    10.6   4.1

This suggests a 3% correction for mispointing.

Bottom line:  14 AB has a Source/system better than 10 AB by a factor of
1.68, but has a correlation SNR worse than 10 AB by a factor of 0.85.
14 AB has a Source/System closely equal to that of 10 CD, but has a 
correlation SNR worse than 10 CD by a factor 0.62.

This difference may come from two sources:  malformation of the bandpass
(this was all done with 50 MHz bandwidth, and with the delays optimized
for that bandwidth), or from a loss of coherence.  The former is easy
to check out by using spectral line mode, but I was getting tired.

I then went on to trying to run VLA antennas with the Executor.  There
appeared to be timing problems of some undetermined nature.  The walsh
functions did not line up using the system offsets from UTC that I had
been using previously.  I found empirically that changing the time by
two waveguide cycles caused antenna 10 to line up nicely, but antenna
6 showed no fringes at all with that alignment.  Easiest explanation is
that the alignment was off by 1 2/3 seconds.  (Antenna 10 pattern repeats
after 1 2/3 sec, antenna 6 inverts.)  Also, the timing is not entirely
stable.  There was an interval of about 40 seconds when the amplitude
with antenna 10 suddenly dropped by a factor of about two, most easily
explained if the walsh function had slipped by one cycle during that 
time.