Modeling the periodicity's caused by the aerostat
radar
may05
Plots:
Comparing
the periods from the model and P2030 data (.ps) (.pdf)
aeroPer.dat...file
containing aerostat periods.
The aerostat is a
tethered
balloon radar that flies above lajas puerto and is used for drug
interdiction
(more
info). A model was made of the aerostat radar total power
output
received at the observatory. The model was then used to determine the
periods
that it could generate in pulsar observations. The model used the
following parameterization of the radar:
-
rf pulse length: 320 Useconds (this is actually 2 160 usecond pulses
that
are sequential in time).
-
7 ipps of 3771., 3504., 3076, 2809., 2903.5, 3289., and 3676. useconds.
-
blanking of the radar signal for 42 degrees of azimuth centered on the
observatory direction.
-
Sidelobes of the radar at -108 and +115 degrees azimuth relative to the
observatory direction.
-
Noise level - 1 sigma= 5 counts
-
radar on pulse level (not pointed at ao) 10 counts
-
Gaussian peak for when radar points at AO: 1e6 counts
-
Gaussian width for when radar points at ao: Set width so that edge of
blanking
region is equal to the sidelobe levels (12db beneath peak).
-
SideLobe gaussians. Make them 12 db less that peak when radar points at
ao.
The model creates a 12 second stretch of radar data
(total power vs time) at 1 usecond time resolution. This is 1
rotation
of the radar with 12000000 samples.
Comparing the model data with the pulsar data
The periodicity's in the aerostat model data
was
compared with the periodicity's found in the total power time series of
p2030 pulsar data (100 Mhz bw centered at 1420 Mhz with 64 usecond
sampling).
12 seconds of data was fourier transformed and then then the magnitude
was computed. The comparison was made with the full model, and a
version
of the model with no blanking, sidelobes, or gaussian beams.
The aerostat radar was outside the rf band of p2030,
but the data was taken when the 10db amplifier for the filter bank was
installed (08apr05 thru 12apr05). This amplifier caused the system to
go
into compression when the radar pulse was on.
The plot compares
the spectra computed from
the total power time series for the model and p2030 data (.ps) (.pdf):
-
Top plot: Black.. This is the fft of the p2030 total power time series.
The strongest birdie is at 304 hz. The red plot is the fft of the
aerostat
model over plotted. You can see that the aerostat periods all line up
with
the pulsar data.
-
Bottom plot. This is a blowup around 304 hz (the strongest period). The
black line if from the pulsar data, the red line is the full model, and
the dashed red line is the model with no blanking, sidelobes, or main
beams.
The blanking causes sinx/x nulls spaced at .17 hz. The sidelobes
cause nulls spaced at .46 hz. The first nulls lines up with the pulsar
data, the second null of the full model doesn't. This is probably
because
the pulsar data is only seeing compression of the system and the
sidelobes
may not be strong enough to drive the system into compression. But the
pulsar data is definitely seeing the blanking.
List of periods from aerostat data.
The bottom plot shows the periods and relative
amplitudes
from the model for periods with amplitudes down to 10% of the
maximum.
Multiple periods caused by the sinx/x have been grouped with the peak
amplitude
near that region. You can get a list of these periods from the file aeroPer.dat.
The model can be asked via the idl routine: mkaerostatrot.pro
processing: usr/p2030/rfirdr/aero/pltaerofit.pro
home_~phil