[evlatests] Notes from July 30 test meeting

[Barry Clark]

EVLA Test meeting July 20, 2009

1.  WIDAR status.

M. Rupen reports hardware status:
Item              # installed    # in use    # coming   final #
Station Bds           24            24         60         128
Crossbar Bds          10             3         60          64
Baseline Bds          10             4                    128
Backend cluster nds    3             2                     33
The number coming column refers to a delivery from DRAO
expected next week.  The remaining station boards have been
delivered to DRAO, will be delivered to NRAO in November,
along with many baseline boards.  Final hardware delivery
to NRAO will be in March 2010.  With the delivery next week
we will have all the hardware necessary to support the VLA
emulator modes promised to the user community in first
quarter next year.  We will soon begum testing observing
in those modes.

Much has changed since the last serious attempt at making
images, last June.  MR now proposes starting from scratch,
pretty much ignoring results from then.  To better control
knowing the effects of future changes, he proposes to make
a weekly observation of a standard field with a standard
setup to serve as a regression test.

In the last few days, they have managed to observe in dual
polarization mode (RR and LL).  They hope to observe in full
polarization mode shortly.  It turns out that there are large
delay differences between the A and C IFs.  These delays were
tuned for the VLA correlator; it turns out that there are
substantial delays in fibers, splitters and deformatters on
the way to the VLA correlator.  The IF C delays also changed
from one day to the next; it is not known why.

The "wobble" problem (Hz rate changes in amplitude and phase)
has returned to the background for Socorro people - Dave
Fort is still investigating in Penticton.  Socorro seems to
have enough other stuff on its plate at the moment.

At the moment, the WIDAR phase convention is reversed for upper
sideband observations.  It is proposed to soon correct that,
in the process reversing the phase convention for the lower
sideband (mostly X band).  A more complete fix is being considered.

MR noted in passing that what WIDAR is currently producing is
not a real correlation coefficient, but the transform of lags
divided by data valid.  It should be normalized by the mean of
the powers of the two subbands.  Not a major practical consideration
at the moment, but something to be kept in mind.

One of the features detected in recent observing was a large
internally generated signal at 1404 MHz which is the eleventh
harmonic of the 128 MHz clock.  This is coming from the L301
synthesizer, and can be blocked by putting a filter on the
input to the T302 converter.  Electronics has requested a sweep
of the entire spectrum to locate any more of these internally
generated signals before they purchase the filters.

2.  Mapping.

R. Perley made an X band image of 3C 273.  This has a dynamic
range exceeding 10^4, perhaps as good as can be expected with
only 11 antennas.

He has data at L band on 3C 286, but is still in early days of
reducing it.  The spectrum of four subbands (500 MHz total) is
interesting.  The first subband contains signals from the
Albuquerque FAA radars, etc at the bottom, and the daily
GPS L3 transmission at the top.  The top of the second and
bottom of the third subband are overwhelmed with space
transmissions from Inmarsat, GPS L1, Glonas, and others.
The top subband appeared to be nearly interference free.

He made a map with a single 500 kHz bandwidth channel, and
finds near theoretical noise on the map, with a slightly higher
noise near 3C 286, and all the other sources in the field
appearing at their expected locations and strengths.

When he combined 20 such channels and made a map, the noise
in the map reduced by perhaps a factor of three, not the
expected factor of 4.5.  Cause is yet unclear - still much
to be learned here.

3.  Dual polarization tests

V. Dhawan started reducing a C band observation of 3C 48 with RR
and LL.  He combined a subband into a single 100 MHz channel
(after delay and bandpass corrections), and made maps.  Neither
RR or LL maps looks very good, but the LL map seems to have a
factor of two worse dynamic range than the RR.  Not very clear
why.  The map is not unreasonable - there is a source in the
field visible outside the point 3C 48.  But F. Owen pointed out
that there should be more than one, since that one has a strength
many times noise.

The antenna 3 L bandpass is strange, sloping dramatically to
the high end, an effect not seen in the wideband autocorrelation
spectrum for this IF.