Signal compression from the faa radar
01aug04
Pulsar data was taken on 01aug04 (p1944) using the alfa
receiver centered at 1420 Mhz with a 100 Mhz bandwidth. Data was sampled
at 64 usecs for 67 seconds tracking a search position (p1944.G212.06-00.23.wapp1.53218.0052).
The total power was used to look for any compression of the signals in
Beam 0 and Beam 1 of the alfa receiver (wapp1). The plots
show the results:
-
Fig 1: The top plot is beam 0, the bottom plot is beam 1. The data was
stacked in 12 second sections (the rotation period of the faa radar). The
5 colored lines are each spaced by 12 seconds in time. They have been offset
for display purposes. The system temperature was close to 1. on this scale.
The plots show 10.65 thru 10.95 seconds of each 12 second section. The
negative going spikes are where the system is going into compression. It
lasts for about 80 milliseconds (the time it takes the radar beam to sweep
thru arecibo's direction). The compression is drifting by about .033 milliseconds
every rotation (which means that the rotation period of the radar is a
little longer than 12 seconds). The maximum compression is about 25 %.
-
Fig 2: The entire data set was searched for negative going spikes and then
the difference between adjacent spikes was computed. This gives the radar
ipp that created the spikes. The plot lists the usecs between adjacent
compression dips that were found (in usecs). The red lines are the known
ipps of the FAA radar. They match pretty well considering that the radar
pulse lasts for 5 usecs (times 2 freq) while the sampling was done at 64
usecs.
The measured data is being compressed by the FAA radar.
This radar transmits at 1330 Mhz and 1350 Mhz using staggered 5 usec pulses
(see measured
lband radars for further info). The compression could be in the iflo
system or in the wapp digitizers. The wapps have a 100 Mhz filter IF filter
that was centered in this experiment at 1420 Mhz. There was probably not
much roll off of the filter from 1370 to 1350 Mhz (so the culprit is most
likely the sampling). One way to test this would be to move the cfr up
to say 1450 or 1470 to see if this improves the situation.
For people who are going to cover the 1330,1350 band
they will be stuck with this compression until we get a back end with more
bits (assuming the compression is not in the iflo).
processing: usr/p1944/rfi01aug04.pro
home_~phil