[evlatests] L-band phase jumps, and more.
Rick Perley
rperley at nrao.edu
Mon Nov 26 17:51:34 EST 2007
A one-hour 'stress' test was run at lunchtime, attempting to
encourage bad phase behavior.
Short (1.3 minute long) observations at L, C, X, K, and Q bands,
with 50 MHz BW and 0.418 second integrations in continuum. Observations
were 'paired' for L, C, and X bands, (meaning, repeated with no change
of source or band) to see if phase jumps could be generated between
consecutive pairs.
Three cycles of the 5 frequency bands listed above were done. A
single P-band observation, using 3 MHz BW and spectral line mode was
then made, followed by one more cycle of the standard high-frequency
bands. The purpose of this was to see if 'global' phase jumps are
uniquely associated with change of BW.
Results from the L-band observations:
1) The only 'global' phase jump occured following the single P-band
observation.
2) No phase jumps every occurred between consecutive observations of
the same band and source.
3) There were some antenna-dependent phase jumps seen on one of the
paired observations:
a) Antenna 13, all IFs, dropped by 100 degrees on
the last scan,
b) Antenna 15 on A and C dropped by 80 degrees on
the 2nd pair of observations,
c) Antenna 18A and C rose by 80 degrees on the on
the third pair of observations.
Other problems noted:
1) The usual end-of-scan change in amplitude and phase occurred,
about 10 seconds following the 'official' end of the observation.
2) Antenna 13 failed to 'lock up' (gave zero amplitude fringes) on
one scan, on all four IFs.
3) Antenna 11, 14, 18, 19 and 25 are slow to stabilize in amplitude
following a band change, on one or both of IFs A and C only. It is
hoped the new, faster attenuator change algorithm will solve this problem.
4) Antennas 16 and 18 were slow to stabilize in phase, following a
band change, on all IFs. A tardy subreflector?
For both (3) and (4), 'slow' means 5 seconds or so after all
other antennas are firmly stable.
Notably poor sensitivity is seen on 16B (low by a factor of two in
G/Tsys), 21 in RCP (low by a factor of three), and in LCP (low by a
factor of 1.5). All others are reasonable.
More information about the evlatests
mailing list