[evlatests] Oddities at L-band (pointing, Tcal, Tsys, Psum, etc.)
Rick Perley
rperley at nrao.edu
Tue Aug 13 12:05:40 EDT 2013
In preparing for a long L-band run, I've been looking at some test
L-band data. Good news is that all antennas are providing apparently
usable switched power calibration. But there are other curiosities,
perhaps worthy of note:
1) I tried L-band referenced pointing, on 3C147. (Don't try this on
your standard 1 Jy calibrator -- background confusion causes apparent
offsets larger than the true pointing error). This was a great success
-- ten consecutive solutions provided exactly the same solutions to
within a few arcseconds. Three antennas have significant pointing
errors -- (16, 20, 25) -- more than 1.5 arcminutes.
At Ken's request, I did an X-band reference solution, followed by an
L-band solution -- the differences give the L-band collimation error.
This was done only once, so the results cannot be considered definitive
-- but there is a strong indication that a half dozen antennas have a
L-band collimation error of about 0.5 arcminutes in elevation. (It's
harder to tell in azimuth, due to the large (2.5') beam squint).
2) Although all antennas give plausible switched power, one antenna
-- ea04 -- had the PDif and PSum changing in steps. The PDifs for RCP
and LCP stepped in opposite ways (mirror images), by large (tens of
percent) amounts. PSum also changed oppositely. But I could find no
change in the visibility amplitudes, telling us that the gains did not
change, so the switched power changes are bogus.
3) I used a central subband (#8 -- 1480 MHz, nice and stable) to
look at the PSum values. They should all be reasonably close to each
other if 'set and remember' is working right. I'm prepared to accept
that 'reasonably close' means with 50% of some median value. But this
is not the case. The median value for the subband appears to be about
10 counts. In LCP, all antennas are within 50% of this. But in RCP,
there is a very different story: ea02R, ea07R and ea13R are all at 1
count, or less! That's a factor of 10 too low!!!
4) The claimed Tsys varies rather more than I would hope. This can
be due to either a noisy receiver, or a bad value of Tcal. Some
notable outliers (all from subband 8, in the middle of the band):
ea08R claims to be 16K! I wish. The Tcal value looks normal
(1.6K).
Antennas with claimed high (>50K) system temperatures are:
ea07R and L, ea08L, ea03L, ea19R and L, ea24L, and ea28L. All of
these have low visibility amplitudes and normal(ish) PSums, so the
indication is that they indeed have high Tsys values.
5) In perusing the Tcal tables, some curiosities are found:
a) ea07 has a Tcal about 8K -- 5 times higher than any other
antenna! Is this right?
b) ea11, ea14 and ea17 all have incredibly high Tcal values listed
for the first two subbands (values exceeding 55K for ea11!!!), but
normal (1.5K) values for all other subbands. Can this be real? I
didn't think the noise diodes could have output powers 30 times higher
than the mean in the bottom 200 MHz of the band...
More information about the evlatests
mailing list