[evlatests] Some amazing (good!) results
Rick Perley
rperley at nrao.edu
Tue Feb 9 17:20:06 EST 2010
Warning: If you're looking for bad news, don't read this posting.
Some days, things just seem to work ...
Michael and Ken took a short observation of the northern 'dot'
source 0217+738, and its nearby (6.5 degrees separation) companion,
0228+673. The flux densities of these two objects are 4.15 and 1.05 Jy,
respectively.
24 antennas were connected to WIDAR. Three subbands, each of 128
MHz, in dual polarization (RR, and LL, using the 'A' and 'C' IFs) were
provided. Channel resolution is 1 MHz.
The non-functioning antennas (all already known) are:
13: no receiver
25A: Provides integer zero visibility amplitudes
26C and 27C: Provides only noise.
The delays are very much better than last week: In RCP, most delays
are less than 10 nsec from the reference (antenna 5). Nine antennas
have delays between 10 and 90 nsec. In LCP, delays are typically 30
nsec, except for three antennas (14, 18, and 19) which are between 100
and 300 nsec.
A simple QUACK (with 25 seconds at the beginning of each scan) was
sufficient for 99% of the flagging needed for source motion. Two
antennas (9 and 23) are a little tardy in their travel... Easy to
handle. Other than this flagging, every antenna and every polarization
and every subband provided good, steady, reliable data.
There are a trivial number of 'perfect zero' visibilities (other
than those associated with 25A) -- amounting to 0.0028% of the total
number of visibilities. Easy to handle via CLIP.
There were no identifiable 'drop downs' which are not due to initial
records at the beginning of scans. I did a 'CLIP' (following
calibration) looking for low visibilities -- the 0.003% of the records
so found are certainly dominated by channels at the edge of the bandpasses.
The bandpasses are as perfect as I could possible expect.
There are eleven observations of the 4.15 Jy main source, each of
about 1 minute in duration. I ran BPASS to determine the bandpass
stability: Most antenna/IFs are stable over the 128 MHz BW and 28
minutes in time to the 0.01% level -- I can't see deeper than this due
to SNR. There are some exceptions: ea2 LCP has variability at the 1%
level at the subband edges, ea6 has ~50 MHz-wide structure at the 0.02%
level that varies in time, 14LCP is extremely noisy -- poor SNR?, 18LCP
and 19 LCP are also very noisy (but not as severely as 14). There are a
few other examples of possible variability, but all at the 0.1% level or
less.
Five, and only five, baselines have clear 'wobbles'. These are
common to both polarizations and all three subbands. In the
descriptions below, bear in mind that the noise is about 0.1 degrees in
phase, and about 20 mJy in amplitude (SNR of about 200).
1) 4 x 20 (W1 - N5): 290 mJy pk-pk in amplitude with 30 second
period in amplitude, 5 degree pk-pk in phase, with a 15 second period.
**** NB Note the phase period is half that of the amplitude!!!*** In
the complex plane, it would seem this baseline's visibility is tracing
out a figure 8, oriented along the real axis (the figure 8 is on its
side).
2) 6 x 18 (N3 - N9): 7 degree pk-pk in phase with 20 second
period. The amplitude oscillation is barely detectable (20 mJy), and of
the same period.
3) 2 x 6 (E2 - N3): 5 degree pk-pk in phase with 3 second period.
Amplitude oscillation is 80 mJy with the same period.
4) 7 x 21 (N6 - E1): 6 degree pk-pk phase with 6 degree period.
Amplitude oscillation is 100 mJy with the same period.
5) 8 x 24 (N1 - W5): 5 degree pk-pk phase with 2 second period --
but this one shows a clear 'beat' with the slower frequency having a 15
second period. The amplitude variations are barely visible.
No other baselines show any sign of these issues (but a faster dump
may reveal more ...)
Returning to good news:
The amplitude stability is essentially perfect. Over the 30 minute
duration of the test, no antenna-polarization channel changed its gain
by more than 0.5%. Nearly all are less than 0.1%.
The only exception to the above was the last scan -- no coherency at
all on any baseline.
Advice to observers: Use a dummy last scan.
Phase stability and connectivity between sources is as good as can
be expected from atmospheric effects -- changes in phase are at the 1 to
5 degree level, with the larger values associated with the longer
spacings.
*** EXCEPTIONS*
1) The initial scan had a deviant phase on all antennas numbered 19
or higher. (Deviant means tens of degrees). The deviancy was the same
for both polarizations and subbands for each antenna.
Advice to observers: Start with a dummy scan!
2) Antenna 5 had three scans (of a total of 22) where the phase was
60 degrees different than the others. This offset was the same for both
polarizations and all three subbands.
3) Antenna 26 RCP shows a consistent 10 degree phase difference
between the two sources!!! Unfortunately, with 26L not working, we
cannot determine if this is some gross error in (say) Bz, or something
peculiar to this polarization and antenna.
Following basic calibration, a visibility plot showed *no*
measureable deviations from the point-source visibility. This was
confirmed by running BLCAL to search for closure offsets. All values
were in the 0.1% level.
I SPLIT out the data to check the imaging. The principal source,
0217+738, provided an image with 160,000:1 dynamic range (in which a few
background sources are clearly visible). This is 'not too shabby' for a
ten minute observation! Removing the 'wobbly' baselines made no
noticeable difference to the image (as expected).
The 'target source', 0228+673, provided a 15,000:1 DR image with
just regular calibration (i.e. using 0217+738). After self-calibration,
I got 41,000:1 -- with loads of background objects. This source is
notably resolved, -- easily seen in the visibility amplitudes.
Bottom Line: Very, very good data.
More information about the evlatests
mailing list