[evlatests] Tests, May 2 2006

[Ken Sowinski]

                    EVLA Tests, May 2 2006

This is mostly a catalog of things done and curiousities noted during
SW time yesterday.

Exploring attenuator settings
Using a sample of one antenna and one band Walter and I tried to
verify that the apparant extra noise that can be introduced by
injudicious setting of attenuators in the T304 really affect more than
sampler statistics.  We looked at antenna 14 at X band; perhaps not 
the best choice.  IFs A, B, and C all require a total of 10 dB of
attenuation, and signal behavior is the same whether we put all 10 dB
at the input or the output.  IF D requires a total of 20 dB and here
a clear difference is seen.

The current database settings for the ea14-t304-d are 18 and 2 dB at
X band.  We redistributed the attenuation to both 20 and 0, and
0 and 20.  The latter was much improved, but did not provide enough
signal to the sampler.  The optimal setting seemed to be 0 and 17 dB.
The last line provides a number proportional to correlation coefficient
for the baseline of 6 with 14.
Summary:
Attenuation    (18,2)   (20, 0)    (0, 20)    (0, 17)
Sampler RMS     5.6       5.9        4.5        6.3
T5 Synch Det    2.2       1.4        2.2        2.5
Cor Coeff       6.1       5.4        6.5        6.5

The lessons are that we are probably hurting ourselves in some cases
by frontloading attenuation in the downconverter, and that the T5
synch det is more sensitive to power level, in the range of power
levels that we explored, than the correlation coefficient.


Holography
More errors were found and fixed.  The remaining serious problem is that
phase for EVLA antennas wanders for one pointing now and then.  Rick has 
already reported this in much more detail.


Antenna 18
The newsest version of the L302 installed there seems to work right.
However there is another problem, unlikely to be an L302 that has
been there from the beginning.  IF C, but not IF A, has a phase 
modulation with a very clear 10 second period.  The phase alternates
between two values about 100 degrees apart at five second ticks.
Vivek reported this last week and suggested the amount was 120 degrees. 
When looked at with enough time resolution amplitude seems to be about 
right.

We claim this cannot be the L302 because IF A is not affected.
We claim it cannot be the VLA sampler because it doesn't know about
time at this level and is most likely to create delay errors.  Because
amplitude is the same in both states it is hard to believe there is
a delay error involved.
We claim it cannot be the T304 because it doesn't know about time.
DTS or deformatter ????
Wiring between L302 and T304, or 4096 MHz to T304s?
Since this is X band there are not many more places to look.


Miscellaneous
Lots of BD IF pairs are dead at the high frequencies.  This information
has already been passed to Rob, but is included here for completeness
and whatever insight others may offer.  In all cases the L302 claims to
be in lock at the correct frequency.  Sampler statistics and T5 monitor
points (TP and SD) are all nominal, but there is no correlated signal
at all.
13 BD are dead at both K and Q band.
16 BD is dead at Q band. 
Antenna 16 has been doing this many days and somedays it works, others
it doesn't.


Interband Collimations
Q band was corrected a little.  We abandoned the thought of doing this
systematically because of the wind and Rick looked at holography problems
instead.  We should have a go at this on Thursday.
o