[evlatests] P-Band Performance Check
Rick Perley
rperley at nrao.edu
Fri Nov 13 17:56:54 EST 2015
I've been reducing old P-band data for the past two weeks -- with
excellent results. This encouraged me to ask for a few minutes of test
time yesterday to check the current performance.
Observations were of 3C295 and 3C405. The setup included 4-band --
those results are a story for a different day. Here I report on P-band
only.
Four antennas were out of the array at the time: ea01, ea02, ea06,
and ea13.
Vivek was kind enough to arrange the RQ power settings -- this has
the big advantage that the calculated gains are then inversely
proportional to the antenna's effective system temperature --
Tsys/Efficiency. So antennas or SPWs with poor SNR will have high
gains, with all antennas on the same scale (presuming that the RQ
settings worked as expected). In fact, I checked the SPW powers -- all
were near 14 when on 3C295 -- as expected.
After flagging RFI, bandpasses were found, and gains determined.
The basic results are given below:
1) *No Fringes* ea18 provides essentially no fringe power on
either polarization. This situation has I think been with us most of
this year (as my science data were taken in spring, and showed similar
issues).
2) *Weak Fringes: Two antenna-IFs give very weak fringes:
ea14X: Weak by about a factor of 4 to 10 compared to the
others, with the sensitivity improving with higher frequency. In other
words, it's low by a factor of 10 at 220 MHz, by a factor of 4 at 330
MHz, and about equal to the others at 450 MHz. (Here I'm using
effectively Tant -- the power from the sky compared to the power from
the receiver).
ea23Y: Weak by a factor of 10 to 30 in all spectral windows.
3) *Typical Behavior* All other antennas behave nearly identically
-- both in sensitivity and frequency response. A summary is:
The sensitivity has a broad minimum ('good') between 270 and
350 MHz, followed by a significant degradation between 350 and 385 MHz
(of about 50%), then another minimum between 385 and 460 MHz. This
minimum is about 20% worse than the 350 -- 385 minimum. Below 270, and
above 385 MHz, the sensitivity rises (worsens) sharply. Curiously, in
the highest SPW (centered at 472 MHz), the 'Y' response is nearly always
much more sensitive than 'X'. The lowest SPW, centered at 232 MHz, is
about a factor of 10 poorer in sensitivity than the optimum part of the
band.
4) *Switched Power* Switched power was working on all antennas.
Computed system temperatures were 'reasonable', except for 14X, and 23Y
-- where it is clear (from item 2, above) that the receivers or feeds
are not functioning correctly. ea20X's switched power was highly
variable, with no visible change in antenna gains. ea23X's gain, and
switched power, changed continuously throughout the short run.
5) *PDif Compression?* Cygnus A approximately quadruples system
power (typically raising it from 150K to 600 K). We can use this to
determine if the infamous 'PDif' compression is operating in this band:
The answer is 'yes', but it is far less severe than at all the other
bands. Most antenna/IFs show no measureable compression (limit about
1%). Those that show it at a level exceeding 3% are:
ea4X (by far the worst, at 15%), 7X, 8X, 21X, 26X, 28X,
7Y, and 25Y.
ea27Y is 'special' -- it shows 'PDif Expansion' -- the
PDif's are larger on Cygnus A (by 7%) than when on cold sky.
6) *Bandpasses* Very few antenna/IFs show unusual bandpasses.
Those exhibiting ripple, or other weirdness greater than 1 dB (in power,
pk -- pk) are:
ea16Y. No easy description for this. The bandpass varies
by 10 dB (!!!!) with a frequency scale of ~ 10 MHz -- but it is not due
to a reflection, as there is no dominant frequency scale.
ea17X. A strong reflection, with 8 dB pk-pk power, with 4
MHz period -- about 40 meter effective (free space) travel distance.
ea28X. A weak reflection -- same period as above, of
about 1 dB amplitude.
7) *Cross-polarization* Signals in the 'XY' and 'YX' channels are
far higher than I had expected. Typical values are 5 to 10%. Some
antennas, notably 15, 16, 17, 19, 20, 22 and 27, have much higher values
(up to 50% !!!) on some baselines.
More information about the evlatests
mailing list