[evlatests] P-Band status -- Lots of Issues
Rick Perley
rperley at nrao.edu
Tue Aug 21 13:02:36 EDT 2007
A 'fast-dump' continuum run was made last night at P-band, to check
system behavior. There are many issues to be addressed at this band...
Two scans were done, a long 'dummy' initial scan, followed by a
'real' scan, with the same LO and BW setup. Each scan was 5 minutes in
duration.
I'll try sort all the results by subject and my assessment of
severity.
1) Antennas not fringing:
* Antenna 4 gives no fringes in any IF, and is flagged correctly.
* Antennas 13 and 16 give no fringes (as expected), and are flagged
correctly.
* Antenna 26, in IF#2 (B and D) gave no fringes, and were *not*
flagged correctly. (Hence, I could not see the '26D shredding' reported
earlier by Ken).
2) Antennas late on source weirdness:
Antennas 2, 3, 9, 10, and 12 took a very long time to provide
fringes -- up to four minutes later than the rest of the array. All IFs
behaved similarly. Flagging did *not* catch this problem -- the
antennas apparently got on source along with the others, but for some
other reason, took a few minutes longer to get sorted out. Some of the
bad data were flagged, but most were not. There is no spatial
correlation for these antennas' locations -- other than they are all on
the E and N arms.
3) Weak Signals (poor SNR)
Antenna 11 is weak by about a factor of four in amplitude, on all
four IFs.
4) Other flagging issues
Immediately (< 1 second) after the start of the 2nd 'good' scan,
antenna 8 was flagged out by the system for the rest of the 5 minute
scan. But the data were good, with no change in gain or phase.
5) System Correlator Reorganization Trouble
One of the 10 to 20 second data gaps occurred in the run: Data from
all antennas suddenly go incoherent, until the next 10-second tick,
followed by a 10-second gap in the data stream, followed by good data.
None of the incoherent data were flagged. Ken explains this as a
'system controller reset' (or some such). This is easy to catch on a
calibrator, but impossible on a generic weak source. The frequency of
occurence is worrysome to me -- I think perhaps once per hour. Probably
not very important, but bears watching.
6) Stability Troubles.
* Antenna 14 is 'all over the map'. More specifically, both IF
pairs showed huge amplitude and/or phase instability for extended
periods of time. The two polarization in each IF pair behaved
identically. 14A and 14C were much the worse, with long (few minutes)
periods of complete phase incoherence (and the expected loss in
amplitude at the same time). The behavior in 14B and 14D was rather
different, where large phase ramps (rather than apparent random phase)
took place at the same times.
* Antenna 21 has two different amplitude states on all four IFs --
but unlike other multiple level issues, the four IFs all show different
characteristics. The SNR (as judged by stability of the amplitudes in
either state) is very very low. This problem has been around for a long
time, and Ken and I used to think it was symptomatic of absent Tcals.
But that should give random amplitudes, with a roughly 10-second
stability period (I think) -- what I'm seeing here is completely
different. Something else is amiss here. It is important to note that
the phases are *stable*.
* Antenna 18A and 18C showed bistable phase, about 75 degrees
apart, during much of the first scan. The second scan was fine. 18B
and 18D were clean.
* Antenna 28, (only!), showed a unique phase feature: *Exactly* 10
seconds before the end of the first scan, the phases on all four IFs
stepped uniformly by 90 degrees, in the following manner -- There were
eight steps, each exactly 1.25 seconds long, each step was about 12
degrees. All IFs showed identical behavior, and the amplitudes were not
affected. This behavior was not shown at the end of the second scan,
following which the correlator changed mode (to spectral line). (The
scan following the one with the 'steps' was an identical scan with same
frequency, BW, and mode).
7) Global changes between the two identical scans.
As noted above, the first two scans had identical setups:
* The EVLA and VLA suffered a global phase change upon the change
of scan. The change was about 100 degrees in IFs A and C, and about 45
degrees for IFs B and D. Sounds 'flukish' -- but I can't prove which
array 'jumped'.
* A number of antenna-IFs changed their amplitude gain upon the
change of scan:
- 17D rose by about 20%
- 19D rose by about a factor of 2.0
- 23A declined by a factor of about 5 -- however, prior to the
change, the amplitudes were bouncing back and forth between the two
levels. The latter (lower) level is clearly the right one, so some
large and variable signal was present during the first scan only, which
was 'sorted out' by the change of scan.
- 26C rose by about 20% -- very similar behavior to 17D.
Some of these troubles are related to bad Tsys measurements:
17D shows a drop in Tsys at this point by 25% -- since the
amplitudes were not correctly adjusted, the Tsys values recorded are not
correct.
19 has no Tsys values at all, so no Tsys adjustments can be
made. Evidently, the actually correlator coefficient changed by a
factor of two of 19D (only). It seems unlikely this is due to a change
in system temperature ...
23A has wild and crazy Tsys values during the period of amplitude
instability -- the unstable amplitudes are clearly due to erroneous Tsys
corrections being applied.
26C shows no change in Tsys. (But 26A changes by a factor of two
in Tsys between scans, with no change in amplitude! I can't offer a
rational, and believable, explanation for all this).
That's enough for one day at this band.
4-band report follows.
More information about the evlatests
mailing list