[evlatests] WIDAR mysteries
Rick Perley
rperley at nrao.edu
Thu Jun 4 19:08:49 EDT 2009
One hour of data on the strong, unresolved northern calibrator
0217+734 was taken yesterday afternoon. There were four IFs, in RCP,
each of 128 MHz, spanning 4.5 to 5.0 GHz in C-band.
General data quality was excellent, as noted earlier, beyond the
issues already identified.
Bandpasses look very stable. SB0 (IF#1) has a strongly sloped
attenuation through it. Although this calibrates out well, the analysis
below excludes this IF.
After bandpass, delay, and gain calibration, the single-channel
(125 kHz wide) histograms were computed. These were of exactly the
right width in both real an imaginary parts in all four IFs. A mild
increase in width was seen towards the edge channels, as expected.
Images were made of various numbers of channels. Single channel,
and Nchan up to about 100 provided maps near the noise level. However,
images made with wider bandwidths clearly showed 'lumpy' backgrounds
which are well above the noise levels. As these residuals were 'lumpy',
and not 'swirly', they point to non-closing effects which are limited in
time.
Investigation of the individual baselines revealed a remarkable
effect seen only in phase: antenna 19 with 2 and 24 are giving
sinusoidally varying phases. The shorter baseline (19 x 2) shows a 4
degree (pk-pk) phase oscillation with variable period, with a stationary
point about 10 minutes after the start of the run. This stationary
point is *not* coincident with the u = 0 point (which is where the phase
rate drops to zero) -- not even close. Something else is driving this.
The amplitude of this error signal remains the same throughout the
entire one hour. All sub-bands show the effect. Baseline 24 x 19 also
shows this oscillation, but with 2 degree amplitude, and no stationary
point visible. No other baseline shows a similar effect.
Excluding these baselines made no different to the sensitivity/lumpy
problem noted above -- as expected, since these phase oscillations are
fast and continuous throughout.
I made images in 10-minute snapshots. All show lumpiness above the
noise.
I tried using the 5 antennas which show the square-wave jumps, and
the 5 which do not. Both have the lumpy problems.
At this point, I'm out of ideas. Time to rest. We are taking a 5
hour integration, with a calibrator, a blank field, and a resolved
object. The 40 GB database should be ready for analysis tomorrow.
More information about the evlatests
mailing list