[evlatests] Characteristics from L-band, using 30 hour run
Rick Perley
rperley at nrao.edu
Wed Feb 2 16:10:39 EST 2011
Reduction of the 30-hour flux densities run for the L-band data has
revealed a few interesting issues:
1) Switched Power. As noted in earlier reports, the script had to
be aborted about 9 hours into the run. Upon return (40 minutes later),
many of the receivers at many bands changed attenuator setting,
typically by 1 dB (about 25% in power). This is not too surprising.
But at L-band, the change in attenuator was often greater than at any
other band -- often by 3dB or more!. This is unreasonably (even
ridiculously!) too much. Unless one of the two settings at which the
power was measured was at an elevation of less than 10 degrees, it's
hard to see how the actual power of the L-band system can be so
different. The change inserted must be erroneous -- the total power
also changed (so the attenuator change was not to compensate for some
remarkable change in system temperature), and in some cases, one
polarization changed by 3 or 4 dB, while the other polarization didn't
change at all.
Only a very few antennas had no change in power at this moment, and
for these, the diurnal change in power noted at other bands was seen.
However, the amount of change varies remarkably widely: A list showing
the ratio of maximum to minimum PDif, for those antennas in which no
jump was seen:
1LCP 1.5%
3LCP ~1%
6LCP 3%
7LCP 5%
11LCP 2%
13 LCP and RCP: No diurnal effect noted, but rather a steady
decline of 12%
21LCP 8%
22LCP <2%
25LCP 4%
Note that only LCP was jump-free. Every antenna (except 13) showed
a power jump in RCP. Something fooled the system ...
Application of the PDif did not provide us with <1% amplitude gain
stability, however (unlike S, C, and X bands). Residual gains of 2 --
3% (in amplitude, so ~5% in power) were seen on many antennas. This is
likely due to errors in the PDif measurements, but I have not yet proven
this is the case.
As noted earlier, we have some evidence that the SY tables are not
correctly labelled,, as subband 4, which contains the strong GPS signal
provides beautifully smooth power measurements. Some effort to prove
this assertion is needed.
2) Bandpass. L-band is infamous for its interference. But this did
not prevent excellent BP solutions from being generated, using the ~30
observations of the four key calibrators used in this project. Only
the satellite downlink subbands (1520 through 1620 MHz) failed to get a
good solution!
3) A fit for elevation gain dependence was made -- no antenna showed
a dependence greater than ~1% over the range of elevations employed (18
through 80 degrees).
4) R-L phase: Good polarimetry *requires* at least one antenna
with stable R-L phase (unless you like doing painful transfer of the
phase difference from a strongly linearly polarized source). Plots of
R-L phase show a distressing variance amongst the antennas:
a) many antennas showed a change after the script abort mentioned
above. This is bad for those who would like to utilize solutions from
other observing sessions ...
b) many antennas showed significant changes in differential phase
for different scans. Notable are: ea02: 25 degrees, ea13: 6 degrees,
ea20: 15 degrees (single scan only), ea21: 10 degrees, ea23: 25
degrees, ea24: 15 degrees.
5) Polarimetry results: These are quite stunningly excellent! I
have now the run of polarized flux and position angle for 3C48, 3C138,
and 3C286 (assuming the P.A. of 3C286 is fixed at 33 degrees). (TECOR
was run to help remove ionospheric Faraday rotation -- I can't claim it
helped, since I don't know the correct answers in advance, but it
certainly didn't hurt!). Notable is that good polarimetry was
accomplished within *all* subbands except around 1240 MHz (GPS), and
between 1520 and 1630 MHz (satellite 'zone of death'). In particular,
no problems were encountered in the DME zone 1030 -- 1130 MHz!
Antenna polarizations are high below 1200 MHz, but this does not
appear to have prevented reasonable results (albeit on strong sources).
One significant oddity: Antenna 5's polarization solution for RCP
shows a strong resonance in both amplitude and phase -- the period is 5
MHz, and the pk-pk amplitude is nearly 10% (!). This resonance is
completely absent in LCP. The amplitude increases with increasing
frequency -- it's barely visible at the bottom end of the band, and
completely dominates above 1600 MHz.
6) There is a 25-MHz-wide slot, centered at 1015 MHz which is
completely free of any RFI. As a test, I extracted these data for
imaging, with excellent results (other than SNR, which is low since most
of these antennas do not yet have their wideband OMTs installed).
6) The narrow window between
More information about the evlatests
mailing list