[evlatests] System Check today
Rick Perley
rperley at nrao.edu
Wed Apr 25 19:35:56 EDT 2007
The intention was to compare Modcomp-on and Modcomp-free data, but
as IDCAF was not operational, I have only the Modcomp-driven data to
review.
Global results:
The 150 degree phase jumps seen over the past couple of weeks on
14C, at P, L, and occasionally C-bands, is now gone. I hope this was a
result of somebody finding and repairing the problem!
The multiple amplitude levels reported recently from antenna 26D are
also gone. Same comment given above applies.
Antenna 21 worked well now that it points properly. Fringes were
found on all IFs for L, X, and K bands. Still no fringes at P-band, and
there is no receiver at C and Q bands yet.
The last 1 second of data from EVLA antennas, at all bands, was bad,
but was not caught by the flagger.
Antenna 23B and D have 'ramp gains' -- gain changes changing
linearly, with a 'step' back to the original gain. Typical timescale is
40 seconds, amplitude of ramp about 15%. B and D are not ramping in
synchronism. This happens at all bands.
Band-Dependent results:
I filled the data with flags off, and with flags on, to see what the
system is catching, and not. In general, the EVLA flagger is overdoing
it, taking out more data than warranted. There are some curiosities ...
1) P-Band. Antennas 14 and 17 fringing nicely on all IFs. Nothing
from 21. All others have no receivers yet.
Most antennas take about 30 seconds to settle down after the
frequency change (from L-band in this case). The VLA is catching this
nicely, so there is little initial bad data to worry about. The EVLA is
taking out 70 seconds at the beginning -- quite a bit too much. In
addition, 17A was flagged entirely, but in fact the data are good.
2) L-Band. All EVLA antennas worked well, except 23, which was
stowed to get its Q-band receiver installed.
The last 20 seconds of data -- from all antennas, was garbage.
This was missed by both VLA and EVLA flaggers.
The antennas take typically 20 seconds to settle down after band
change. The VLA's flagger caught the first 12 seconds. The EVLA took
out 42 seconds -- much more than required.
3) C-Band.
The settling time and flagging characteristics are the same as at
L-band.
In addition, the EVLA flagger is letting pass some data very early
in the scan -- clearly bad data.
4) X-Band.
Only 23 failed to fringe -- it was stowed. 16C had a ramp-like gain
(similar to 23, described above) -- but only at this band!
Curiously, at this, and at K and Q bands, there is essentially zero
settling time for the antennas. This is known to the flaggers, which
flags very little initial data -- correctly. However, there are curious
exceptions on the EVLA side:
Some antenna-IFs taken nearly 30 seconds to 'come into
equilibrium' -- and this is not caught by the flagger. Antenna-IFs
affected are a strange lot: 13A, C, 14A, 16-all IFs, 21B and C, and
26A, B, and D. All other antenna-IFs not listed showed no effect --
gains were right, from the start.
But -- the EVLA flagger for some reason tossed out 10 seconds of
(generally good) data, on all antennas and IFs, about 12 seconds into
the scan.
5) K-Band
All EVLA antennas worked, except 23 (stowed).
Here there is essentially no 'settling' time after band change. And
the flaggers did the right thing. (This band was first in the file).
Many antennas (both VLA and EVLA) have oddly variable gains in the first
30 seconds -- these look nothing like the changes commonly seen at the
beginning of the low-frequency band scans.
6) Q-band
All antennas outfitted worked, except 26. (Is its receiver gone?)
Like K-band, there is virtually no 'settling' time when changing to
this band (from K-band, in this case). Nearly no initial flagging was
needed, and the flaggers flagged essentially nothing at the beginning of
the scan.
However -- all EVLA antennas had 10 seconds of good data flagged,
after the initial 10 seconds of data (mostly good data) were not flagged.
More information about the evlatests
mailing list