[evlatests] Notes from test meeting of Feb 25

[Barry Clark]

EVLA Test Meeting February 25, 2010

1.  Widar status.

M. Rupen reports on progress on achieving the OPT->Archive path.
The OPT does reasonable things with OSRO modes (currently on webtest
only; will make a new production release shortly).  Conversion of
the OPT model to a script has been does by a test program.  Having
the scheduler do it is believed to work now, but hasn't been tested.
Script is interpreted correctly by Executor.  VCI (virtual correlator
interface) commands are picked up by Executor and passed to
ConfigurationMapper (CM) and to MCAF (metadata capture and format).
CM passes commands to station boards and to baseline boards, it is
believed correctly.  Baseline boards produce data as expected.
CM also passes configuration information to the correlator backend
(CBE).  The handshake with the CBE is working, but there remain
problems within the CBE in writing BDF (binary data format) files -
believed to be fixable shortly.  MCAF receives the configuration data
and uses it to fill in the SDM (science data model).  Last time it
was tried, there were minor errors, now believed fixed.  After hand
correction of these, CASA filler apparently did the right thing.
The automatic path into the archive hasn't been tested, because of
these minor glitches, but no problems are expected.  It is still
possible that we will have a real science observation Monday.

Observing in RSRO modes will at first be done only by hand editing.
Micheal will provide one or two VCI setups.  Observers must make
scripts, extract them from the OPT, and edit them to change the
VCI setups to those provided.

2.  Set & Remember test
A test was run of switching between C and X band, 30s each, as a
test of the "set and remember" modes for the T304 attenuators, since
the lack of switched power data being recorded means calibration
is at the mercy of the attenuator settings.  The data were looked
at by R. Perley and S. Myers, who verified that the attenuators
always returned to the values initially set (except for antenna 26,
which had the feature disabled).  However, a number of interesting
phenomena showed up in the data reduction, not related to attenuator
setting.

C band worked well - amplitudes stayed stable to a few tenths of a
percent, for the most part.  X band had more interesting behavior.
Some IFs were OK, some were bistable, with power levels differing
by a few percent (not an attenuator step).  Some IFs wander around
at the 1% level.  Antenna 5 had a particularly clear bistable behavior.
R and L polarizations changed at the same time - the problem lies
in front of the T304, in an oscillator or switch between the T304
converter and the front end.  The gain usually remained constant
for the 30s scan, but jumped randomly between scans.  The exception
was a three minute scan at the beginning, where the gain changed
during a scan.  This sort of behavior has not been noticed in
long observations of short scans all at X band.

Phases also showed glitches on some antennas.  Antenna 3, at C band,
had bistable phases, differing by 60d or so; X band was also bistable,
but the difference was only 15d or so.  Antenna 7 had a few odd
phases, again sticking for a 30s scan.  Antenna 26 (which had the
attenuator set-and-remember feature disabled) had a large phase
slope with time at C band, with a much smaller one at X band.  But
the X band exhibited two state behavior, with a 50d difference between
the two states.  This sort of phase behavior is especially distressing
because it is not fixed by the application of the switched power
measurements, which should take care of the amplitude problems.

3.  Polarimetry.

R. Perley, as reported in an EVLA memo, has a method of absolute
polarization calibration, by rotating one receiver.  He has redone
these measurements recently.  At C band, where we mostly have
receivers in final configuration, the polarization leakage is
about 5% over most of the frequency range, as was specified.
The feed orthogonality tends to be good - that is the Left to Right
leakage differs in phase by 180d from the Right to Left leakage,
giving small power out of the RL and LR correlators.

L and S bands have only a few of the final configuration receivers,
but these also have polarization impurities of about 5% as well.

RP also difference the C band results with the smaller set he had
from April, and sees differences mostly less than 0.5% - these
things are very stable.

The 5%+ numbers seen in the polarization purity are primarily due
to the hybrid in the receivers.  It is believed that this comes
from detuning as the hybrid is cooled.  It may be possible to get
better hardware, optimized at cryogenic temperatures.

4.  Widar first science.
C. Chandler has a list of the first projects to be observed.
AH1004 - Polarimetry at C band, several tunings, but only one per SB.
AV317  - (OSRO2 mode), spectroscopy
AR704  - L band continuum polarimetry
AK723  - Search for CH lines in L band spectrum of 1830-210.
AU129  - C band images of galactic sources

5.  Odds and ends
V. Dhawan reports that he has, with Michael's help, imported flags
to AIPS from the SDM, without going through CASA at all.  Currently
not a very practical route - associating flags with antennas is
labor (and knowledge) dependent.