[evlatests] 1420.3 MHz birdies!
Rick Perley
rperley at nrao.edu
Tue Apr 3 19:05:50 EDT 2007
I have done some more investigations on the bizarre RFI reported by
Lynn Matthews, and can now provide some much better clues as to what is
going on.
The latest test used mode 1A, with 1.56 MHz BW. I tuned in 1 MHz
steps from 1416 to 1424 MHz. This was quite sufficient to show what is
going on.
The RFI is *strictly* bounded in frequency: It occurs between
1419.9 and 1420.9 MHz. The comb has a spacing of exactly 40 kHz --
therefore, there are *exactly* 25 'lines' in the RFI spectrum. It looks
like the originating signal has passed through a 1 MHz wide filter.
The tuning I commented on in the previous message is a red herring. The
setup used by Lynn had a .78 MHz BW, with tuning separated by 1 MHz.
The middle of the three she used lay in the 'forest', the other two were
outside. Amazing.
The RFI affects the VLA antennas much more than the EVLA antennas.
This is most clearly seen in the autocorrelation spectra, where the 25
lines are prominent in three antennas only: 22, 6, and 11. These are 3
of the four outermost antennas on the west arm. They are all VLA
antennas. EVLA antenna 16 lies in between 11 and 6, but is (as we all
know) currently off line. The next antenna, heading east, is 17 -- this
shows no sign of the comb. No EVLA antennas show the comb in total
power. A few other VLA antennas do -- but weakly: 3 (E2), 8 (W1), 20
(E1).
The X-power spectra are as expected, from the above descriptions.
Antennas 6 (W8) and 22 (W9) show the comb on all baselines. Antenna 11
(W6) has the comb only on some VLA baselines, but not to any EVLA
antennas. EVLA Antennas 17 (W5) and 24 (W3) do not show the comb on any
baselines, but VLA antennas 4 (W4) and 8 (W1) show it on baselines to
other VLA antennas.
This band-limited comb acts like an external radiator, located near
to antennas 6 and 22, *except* that it seems to prefer VLA antennas over
EVLA antennas.
I hope this curious set of observations can jostle loose a good idea
as to the origin. We need to get rid of this -- whatever it is.
More information about the evlatests
mailing list