CARSR-3 Punta Borinquen radar (1274,1332
A new CARSR (Common Air Route
Surveillance Radar) is being installed at punta borinquen (remy
airport, end of the runway). It is a Westinghouse ARSR-3 radar. We
first saw the radar appear on 01apr13 (around 14:00) in the hilltop
monitoring data. The measured freq are::
A similar radar will replace the faa radar. It will broadcast at
- 1274.58 . smaller signal at 1269.24
- 1332.66. smaller signal at 1327.54
The punta borinquen radar is undergoing acceptance testing. It will
be accepted on 15may13. At that time they have promised to blank in
our direction. We need to see if they can do that during acceptance
- 1257.6 with weaker pulse at 1252.41
- 1349.6 with weaker pulse at 1344.41
09may14: punta borinquen blanking and sidelobe
Punta borinquen carsr radar specs.
04oct13: punta Borinquen blanking re enabled
after faa radar comes back on the air.
29apr13: measuring the radar blanking.
23may13: first look at 1274,1332 MHz radar.
on some "other" radars.
- 03oct13: blanking back on. but about 1.4 deg too early in
- assuming old blanking of 107-111 i asked them to move
- may13: turned off blanking in ao direction while faa being
New punta borinquen carsr-3 1274,1332
MHz radar specs: (top)
The radar specs listed below were measured on 24apr13
total pulse length 279Usecs
- F1=117 usecs
- 2usec off
- F2=117 usecs
- 2 usec off
- F2n=19 usecs
- 2 usec off
|ipp pulse sequence
60secs, 23usec spc
||about 1 MHz
||29apr13: blanking 4 degrees in the direction of AO
|distance to ao
||az beam width : probably about 1.4 degrees.
09may14: punta borinquen blanking and
Test data was taken on 14may14 to check
the FAA and punta borinquen radar blanking. The data looks
at the blanking region as well as the sidelobe height over the 12
second rotation period.. The 09may14 data took 3 sets of 25 second
data (2 rotations per set). The data sets were spaced by many
minutes so plane echoes would not be stationary.
- lbw, linear pol
- az=80, za=18 (this was a peak for FAA compression for alfa
data taken a few years back).
- mock spectrometer 16 bit timedomain data.
- 3.2 Mhz bandwidth
- center frequencies of each 3.2 Mhz: 1257.5,1349.5,1274.5, 1232.5 Mhz
- the first two are punta del este (FAA) radar frequencies
- the last two are punta borinquen frequencies.
- 3 scans were taken each with 25 seconds of data.
- Start time for each 25 second scan
- 15:36:17 ast
- 15:38:17 ast
- So each scan was separated by an integral number of 12
The plots show the radar power vs
- Each 25 sec scan was input and processed
- Power was computed for each 3.2 Mhz sample
- power was averaged to 100 usecs. this is close to the 117 usec
- polA and polB was averaged.
- The total power time series was normalized to when the radar
was blanked (pointing at ao).
- This make the vertical scale Tsys units for AO.
- The normalized time series was then converted to db. 0db is
not Tsys at AO.
- Page 1: over plot 6 12 second rotations (with offsets
- black red were the first two rotations starting at 15:36:17
- green,blue are the two 12 sec rotations starting at 15:38:17
- purple,dark red are the two 12 sec rotations starting at
- The gap at 1.1 seconds is when the radar blanks in the
- Page 2: average the 6 12 second rotations.
- Top: average 12 second period
- red dashed line shows the peak value (with ao blanking
- green dashed line shows close in sidelobe peak.
- blue dashed line shows the far out sidelobe peak
- The far out sidelobe is about 10db below the blanked peak.
It is 16db below the close in sidelobe peak.
- bottom: blowup around AO blanking.
- purple vertical line: blanking specified in dod/dhs memo:
4.25 deg (.142 seconds)
- The light blue vertical line shows that the near in
sidelobe is about 18 degrees from the center of the blanking
radar power versus rotation phase for 1274.6 Mhz radar (.ps)
- The far out sidelobe is 6 db above the blanked peak.
radar power vs rotation phase for the 1332.6 Mhz radar (.pdf)
- the far out sidelobe is 9db below the blanked main beam.
- increasing the blanking by a small amount would drop the
blanked peak by about 6db
- the punta borinquen blanked radar is about 23 db above
the AO system temperature
- in a 3.2Mhz total power bandwidth.
- the current blanking is .142 seconds (4.25 degrees)
- there is a large sidelobe 6db above the blanked main beam 12.6
deg from the main beam boresight.
04oct13: Radar blanking re-enabled.
03oct13 before radar azimuth alignment
power plots vs time (.ps) (.pdf)
of total power vs rotation phase for 79 rotations (.gif)
04oct13 after radar azimuth
power plots vs time after blanking direction updated (.ps) (.pdf):
of total power vs rotation phase for 17 rotations (.gif).
The punta borinquen radar had blanking enable
on 29apr13 ( more
info). The pico del este radar (FAA) was upgraded to a
carsr radar during the summer of 2013. During the upgrade, the
punta borinquen radar was used by the FAA. The blanking was
disabled sometime during this time period. On 03oct13 the blanking
was re-established. On 04oct13 the blanking region was moved by
1.4 deg in azimuth to better line up with the observatory
Data was taken on 03oct13 and again on
04oct13 to measure the blanking. The setup was:
- lbw receiver centered at 1330 MHz with circular polarization.
- 5 MHz IF filter followed by square law detector with 200
usec time constant
- radar interface datataking at 100 usecs/sample
The plots show the current state of the blanking:
03oct13 blanking before 1.4 deg move in azimuth
power plots vs time (.ps) (.pdf)
- Top: Total power vs 12 second rotation phase for 10
rotations. Offsets have been added for display. The dashed red
line shows the start of blanking.
- The blanking is not symmetric about the beam as it sweeps
- this probably had the same azimuth blanking as 29apr13:
azimuth 107 to 111 degrees in sweep.
- middle: plot of a single 12 second sweep (the y axis is a
log scale). Looks like the square law detectors (or something
upstream) saturated at the end of the blanking.
- Bottom: A total of 79 12 second rotations were recorded.
- The black trace shows the maximum value found .1 seconds
after the end of blanking for each rotation.
- The red trace shows the same value for .1 secs before the
start of blanking.
- The blanking is not centered on the ao direction. The offset
is about .047 secs or .047/12 *360=1.4 degrees.
of total power vs rotation phase for 79 rotations (.gif)
- 5 MHz total power about 1330 MHz was computed for each 100
- The data was then made into an image with the xaxis equal to
1 rotation period (12 seconds)
- at the radar rotation period: 12 secs + 2 usecs.
- Blanking occurred between 7.3 and 7.3 seconds.
- The rotation rate should be 12 seconds. The 15 ipp pulse
train (47,620.9usecs) divided evenly into 12sec +2 usecs. The
residual slope comes from the 100 usecs sampling.
- Each stripe in the image is a sidelobe passing through the
- You can see the blanking jump in units of 5 sidelobes (about
15 milliseconds in time). I think this comes from:
- The ipp sequence is 5*ipp1 then 5*ipp2 then 5*ipp3 giving
the 15 ipp sequence
- Each set of 5 ipps lasts for about 15 milliseconds.
- the 15ippsequence divides into the 12 second period
251.990 times. There is a .01*47.62ms=.476 millisecond
residual after each 12 second rotation. After about 30
rotations, the residual is equal to 15 milliseconds. They
then jump a set of 5ipps.
- Because of this, i think the blanking will drift by about
30 or 45 milliseconds.
04oct13 blanking after blanking region moved by 1.4 degrees.
- The blanking region was 107 to 111 degrees of azimuth. I
asked them to move it to 108.4 to 112.4 degrees to try and
line the blanking region up with AO.
- On 04oct13 the blanking had been moved. We then took
some more data:
power plots vs time after blanking direction updated (.ps)
- Top: Total power vs time. 10 12 second rotations are
plotted. The plots are blown up around the blanking region.
- The beam amplitudes before and after the blanking are
- You can compare this with the 03oct12 data by looking
at the sidelobe .5 seconds after then main beam.
- on 04oct13 it is bigger than the edges of the blanked
- on 03oct13 the .5 sec sidelobe is a least 100 times
smaller than the edge of the blanked beam
- Bottom: the maximum value before and after blanking for
the 17 recorded rotations.
- Red is from the start of blanking. black measures the
end of the blanking.
- The two levels are equal. So the alignment is pretty
of total power vs rotation phase for 17 rotations
- The dynamic range of the data is much smaller. The
vertical stripes in the image are now showing the individual
ipps from the radar (2.5 to 3 millisecs).
- The ipps before and after the blanking look similar in
131004/pntBorBlanking_03oct13.pro .. 04oct13.pro
- The punta borinquen radar blanking was re enabled around
03oct13. It was blanking 107 to 111 in azimuth. The blanking
region was not centered on AO.
- On 04oct13 the azimuth blanking was changed to 108.4 to 112.5.
This was not centered on the AO direction
- The blanked main beam is not smaller than the sidelobe at +15
deg in azimuth.
- The radar blanking jumps jumps by units of about 15
milliseconds. The jump occurs after about 30 rotations (30*12)=6
- I think this comes from the15 ipps sequence not dividing
evenly into 12 seconds.
29apr13: measuring the radar blanking (top)
On 29apr13 i spoke with MS Sgt Bezos about the
radar blanking in our direction. He had the radar blank 107 thru
111 degrees of azimuth (as it points at the AO). Spectra were
taken before and after the blanking. Total power data was taken
after the blanking was in place.
total power data:
After the blanking was turned on data was
taken with the lbw receiver was centered at 1274.5 MHz. The
plot shows the 5 MHz total power at the 1274.5 MHz pulse for 4
complete rotations (.ps) (.pdf):
- lbw centered at 1274.5
- The signal was passed through a 5 MHz filter and then detected
with a 2 usecond time constant.
- The detected signal was sampled at 1 MHz with the ri (12
- This was done for 48 seconds (4 radar rotations)
second plot is a blowup around when the radar was blanked (.ps)
- The data was sampled at 1 usec but then smoothed,decimated to
40 usecond resolution.
- Each rotation is plotted in a different color (with offsets
- The top frame is plotted vs time. The blanking occurs around
- The bottom plot is plotted vs azimuth (of the radar).
They told me they blanked from 107 to 111 degrees of azimuth.
- For the azimuth, i used the first signal dropout to be
az=107 degrees. This is where they told me they started
- I don't see very strong backlobes of the radar.
- The 4 rotations have been over plotted (using different
- top frame: plotted vs time
- the radar is blanked from 3.565 to 3.717 = .152 seconds
- bottom frame: plotted vs azimuth
- the radar is blanked from 107 to 111.62 degrees (assuming
107 is the first drop off).
plots shows 12 1 second spectra before and after blanking (.ps)
- The interim correlator used 9 level sampling, 1 second
- 12 1 second spectra show a complete radar rotation (with
offsets for display)
- The white traces are with no blanking.
- The red traces are with blanking turned on.
- The traces with blanking on actually look to be the same
strength as when blanking is turned off.
- The limited number of bits may have been clipping the signal
at a level outside the blanking region
- A better test would have been to look outside the radar
freqeuncy and see if the system was compressing less when the
blanking was on.
Rfi hilltop monitoring:
rfi hilltop monitoring data at the radar frequencies was plotted
for 29apr13 (.ps) (.pdf):
- The black trace shows the 1274.5 Mhz peak. The red trace is
the 1332.5 Mhz peak
- The green line shows where the blanking was enabled.
- The max value prior to blanking was -14dbm.
- the max value after blanking was -25dbm
- So the blanking lower the signal by about 11db..
- The punta borinquen radar began blanking the radar when it
pointed at AO (on 29apr13).
- The radar blanked for azimuths 107 - 111 degrees of azimuth.
- Using the interim correlator, the radar spike remained about
the same (before and after blanking).
- The hilltop monitoring saw the radar signal decrease by about
11db when blanking was enabled.
- We probably need to see if the radar is compressing the system
with it's current blanking window.
- After the radar is accepted we may try to get them to increase
the blanking window.
23apr13: First look at the radar. (top)
Data was taken on 23apr13 to characterize this
new radar. The setup was:
spectra of 172 MHz bw centered at 1300 MHz (60 secs of data)
- za=2,az=286. Telescope was stationary during the
- lbw receiver, linear polarization, no filters.
- mock spectrometer
- file:000 - cfr=1300 MHz, 172 MHz bw, .1 second sampling of
spectra. Take data for 60 seconds. starting around 11:45.
- file:100 - cfr=1271.5, 8.6 MHz bw, time domain, 8 bit
sampling, 25 seconds of data.
- The dynamic spectra shows which birdies are coming from the
- The radar rotate with a period of 12 seconds. When the radar
points at ao, the birdie gets stronger
- Common radars:
- Punta salinas: 1232.65, 1241.60, 1247.77, 1256.59
- punta Borinquen: 1269.44, 1274.48, 1327.52, 1332.56
- FAA Radar: 1330,1350
- You can see that the 1274/1332 radar is not rotating the same
as punta salinas or the FAA radar.
The first set of plots shows
the average spectrum and low resolution power vs time (.ps) (.pdf)
The next set of data used the high resolution time domain data:
- Page 1: 60 second average of the 172 MHz bandwidth centered at
- Top full band. the new radars are flagged in color
- the 1269 and 1327 birdies are much weaker than the other
- Middle,bottom:blowups around the new radar
- The bandwidth is about 1 MHz
- The signal probably saturated when it pointed directly at
- Page 2: Total power times series (.1 second resolution).
- top: 60 seconds of data. red is 1274.75, green is 1332.7
MHz. Blue vertical lines are placed every 12 seconds (the
rotation period of the radar).
- Bottom: blowup when the radar pointed at AO.
- the signal was I/Q sampled at 8.6 Msamples/second
centered at 1271.5 MHz. This contained th 1274.5 and the 1269
- The total power was then computed for each sample (polA and B
was then averaged).
power vs time folded at cumulative ipp of radar (.gif)
The final plot shows the
pulse duration (when not pointed at AO) (.ps) (.pdf)
- The total power data stream was searched for the radar ipps
- There is a sequence of 5*ipp1 , 5*ipp2,5*ipp3. So the
pattern repeats every 15 ipps.
- The image shows the total power data plotted at this 15
- For the image the data was smoothed/decimated by 100 (to a
resolution of 11 usecs).
- the 15 ipp period is 47.62088 milliseconds.
- You can see the 3 times when the radar pointed at the AO (the
- The narrow line to the right of the radar may not be real.
- The 8 bit samples saturated. squaring to get the power gave
a constant output value for a few samples.
- Top: total power showing pulse duration. This .11 usec
- The first pulse lasts for 117 usecs
- there is a gap of about 150 useccs
- then a narrow pulse of 19 usecs.
- The power included the 1274.5 and the 1269.4 birdies.
- The blank space is probably when the 1332 and 1327
mhz pulses were active
- So then entire pulse period is about 300 usecs.
- Bottom: 25 seconds of data folded at the cumulative radar ipp:
- The 3 different ipps are flagged in color
- the punta borinquen carsr radar on 01apr13 around 14:00
- The radar frequencies are:
- freq: 1274.58,1332.66 with smaller birdies 1269.24,1327.54
- The large birdies are separated by 58 MHz
- The smaller birdies are 5.12 MHz below the larger birdies.
- Bandwidth : about 1 MHz
- 3 ipps: 2755.4, 3150.7, 3627.5 usecs
- Each ipp is repeated 5 times, then it steps to the next ipp
- The 15 ipp period is : 47.62088 milliseconds.
- Rotation period of 12 seconds.
- pulse length and order:
- 117 usecs 1274.5, 117 1332.5, 19 usecs 1327, 20 usecs 1269
- The bandwidth at each freq is about 1 MHz.
- The radar will finish acceptance testing on 15may13.
- I think the 1332 MHz radar is interfering with the radar
- The radar blanker generates a total power time series with
the 1330,1350 radar folded together.
- Unless there is a pretty narrow filter, the 1332 radar will
end up in this time series.
- when the blanker searches for the FAA radar, it may get
confused by this other set of pulses.