30may19: 2380 Mhz rfi from lares tower
12feb18: 100ms period rfi
2389-2398 Mhz
18aug17: 2370-2375 Mhz starts 03may17
->present
25jan16: 2370-2375 rfi from at&t turned
off
19jan16: rfi at 2370-2375 and 2400 mhz
during road trip to hq/radio redentor towers.
18jan16: rfi at 2370 and 2400 Mhz from 12
meter site.
02dec15:2397
mhz rfi measured at 12meter coming from yiyi's tower
direction.
14sep15: 2380 Mhz Rfi
09aug15: sbn azswings: rfi:
2357-2367 2373-2390,2397-2412
25mar15: 2110-2155 proposed band reject
filter
03mar15: How sbw handles the 2110-2155 aws
rfi
Rfi at 2394-2403 Mhz
mar12: 2140-2150 Advanced Wireless Service rfi,
01nov11:
power
shutdown of visitors center.
12oct11: added monthly plots
of ism band from hilltop monitoring.
jan11 - rfi at 2380 MHz
22feb11 - rfi at 2380
goes off when transmitter turned off.
19jan11 - azswing
looking at 2380 rfi.
12jan11 - rfi appears at
2380 MHz
20sep01 - measuring
the power levels at the output of the sbw dewar.
190530 Rfi at 2380 Mhz
rfi at 2380 Mhz appeared on 28may19 around
11:45 (see hilltop monitoring 190528). It remained
constantly on with a bandwidth around 10 Mhz and 15 db
above the 2200Mhz hilltop band noise floor.
The sband radar experiments were affected by this signal.
How the offender was detected..
- On 29May19 luis and I went to the lidar lab and got a
bearing of 245(mag) for the offender (233 Tnorth).
- I downloaded the fcc antenna registration file for the US
and then extracted registered towers in pr.
- A .kml file was then created to generate google earth pins
for each tower west of ao (see x101/rfi/towers/).
- google earth was then used to see which towers were in the
233 direction from the lidar lab. The closest tower was
asr1282916 (south of lares on 129)
- On 30may19 luis and willie went to the 12meter and got a
signal that peaked in the tower 12,8 direction (they were
lined up). this is about 210T from google earth.
- Later that day luis and angel went to the asr1282916
tower. From there they saw that the signal had a
bearing of 150deg(M). This was toward the tower
asr101129 (farther south on route 129).
- At asr101129 they found the offending signal
- defective wifi link from vpnet. the tower belonged to
cbc (caribbean broadcasting co).
- angel called cbc and vpnet.
- the signal went away at 15:17 on 30may19.
pictures:
- google earth map1
blowup at AO. with 233 and 233 red lines (from lidar
lab) and 203,213 degrees from 12meter.. and the asr tower
locations:
- google earth map2
showing locations of the towers.
- tower pics: 1
, 2,
3.
Summary:
- vpnet was broadcasting illegally at 2380 Mhz (10mhz bw)
starting at 11:45 28may19.
- After being contacted, they stopped the illegal broadcast
at 13:47 30may19.
- They were broadcasting from tower asr 1011929.
- The tower is located (from the 12 meter) at
- 208.7 degrees(T) 221.4(M) from the 12 meter
- 8.1 miles.
- The tower coordinates are:
- lat: 18:14:45.5, long:66:48:38.1
- lat: 18.24597 , long:66.81058
- When taking bearing to rfi, the 12meter location gives a
lot better direction since there is less obstructions (when
not looking through the platform).
processing:x101/rfi/towerloc/
170818: 2370-2375 rfi appears 03may17.
The 2370-2375 Mhz rfi appeared on 03may17.
The hilltop
monitor for 03may17 shows when it came on (.ps) (.pdf)
- The spectra for the day are plotted with display offsets.
- midnite is at the bottom.
- End of day is at the top.
- each spectra is spaced by about 20 minutes.
- A signal at 2550 jumped to 2370 at 12:39 (about 1/2 way up
the plot).
- The signal stayed at 2370 through at least 18aug17 (today
it's still there)
Data was taken through the 305meter using the mock
spectrometer on 15aug17. The setup was:
- telescope sitting at : az,za(285,1.5) deg]
- sband narrow centered at 2380 Mhz was fed through the
750narrow band IF (60Mhz filter).
- The mock spectrometer took 1 sec averages, 8192 channels,
86Mhz bw, polA, polB
The plots
show 30 1 second spectra (.ps) (.pdf):
- Top frame: 30 1 sec spectra overplotted.
- black: polA, red: polB,
- green dashed lines:
- ATT 2350-2360 band
- 2400-2410 iq image of att band (probably from mock
complex mixer).
- blue: 2370-2375 rfi
- The vertical scale is db above Tsys.
- Bottom frame: blowup in frequency of a 1 second
integration,
- the frequency response looks like an OFDM transmission
(carrier in center of band, lots of subbands).
SUMMARY:
- a signal jumped from 2550 to 2370 on 03may17 around 12:39.
- It has remained at 2370 till the present time (18aug17).
- The band is about 4.5 Mhz wide centered at 2372.
- It looks like an OFDM spectra.
- We had similar problems with 2370-2375 Mhz rfi back
in jan 2016 (more info)
- That signal was coming from the HQ tower in Utuado
- This signal is contaminating sband radar observations that
use 20 Mhz bw.
- We need to go to the 12meter site to see if this signal is
also coming from the HQ tower direction.
processing: x101/170815/sbnrfi.pro
25jan16: 2370-2375 mhz rfi turned off
Rfi at 2370-2375 appeared around 30dec15. This is in
the sband transmitter band. The hilltop monitoring showed the
approximate time when it appeared:
- the hilltop
monitoring plots for the end of December (.ps) (.pdf)
- page 1: 28dec15: there is a band from 2345-2350Mhz .
this has been there for a long time.
- page 2: 29dec15 morning: the 2345-2350 birdie is gone
- page 3: 31dec15: the 2345-2350 birdie is gone, but
we now have a 5 mhz band from 2370-2375.
- looking at the x111 data for 17:00 30dec15 shows
that the 2370-2375 was present.
On 19jan16 dana and i went on a road trip looking for
the source of the 2370 rfi. We found that it was coming from
the HQ tower above utuado (more
info)
Angel called the HQ tower engineer and
complained about it.
Looking at the rfi hilltop monitoring, it looks like the 2370-2375 Mhz birdie was
turned off on 25jan16 (.ps) (.pdf)
Question:
- At&t spent lots of money buying this frequency band.
It's been in use for a long time.
- For about 1 month, it was broadcasting 20 Mhz above where
it was supposed to be.
- How were all of the users of this service able to follow
this jump?
- It's hard for me to believe that all of the users didn't
have access to the service for about 1 month..
- If this was true they would have complained.
- Since it stayed at 2370-2375, i'm assuming that the
users followed the base station up to 2370 Mhz
- Does the base station tell the users to follow it up in
frequency?
- It would really be nice if there was some way that the
users would not follow the base station out of band.
- If this was true, the user complaints would get this
fixed a lot quicker than 1 month.
processing: x101/160125/x111_2370.pro
19jan16: 2370, 2400 Mhz rfi during
trip to HQ tower.
On 19jan16 dana and i drove out to the
hq/radioRedentor towers to check the 2400Mhz and
2370-2375Mhz rfi. (we also checked
the rfi below 5150 mhz).
The 2400 Mhz rfi is illegal because it
spills below the ism band by about 2mhz.
The 2370-2375 Mhz rfi is illegal, and
it falls within the 20Mhz band of the sband transmitter.
We made two stops before arriving at the
radio redentor/hq towers:
- First stop was the intersection of 6612 and routine 10
- coordinates: 18.31576, 66.68288
- This is west,northWest of the towers.
- We looked at hq,radio redentor towers at cband and 2380
MHz..
- 2nd stop was across from colmado torres up R611 on the way
to the HQ tower
- coordinates: 18.28576, 66.68934
- We looked out at yiyi's tower, and the tower to the east
of yiyi's tower using 5100 and 2380
- The final stop was at the base of the HQ/radio Redentor
towers:
- coord: HQ: 18.2887, 66.6606, radio redentor: 18.289535,
66.660439
The 1st plot shows the
2400 Mhz rfi when we looked at yiyi avila's tower from the
colmado torres location (.ps) (.pdf)
- This was looking from colmado torres parking toward yiyi
avila's tower.
- It's possible that this could have been coming from the
tower to the east of yiyi's tower. when we took this data,
it was cloudy.
- We did not see the 2370-2375 Mhz rfi in the direction of
yiyi's tower.
The
2nd set of plots shows
the 2370-2375 Mhz rfi (.ps) (
.pdf)
- The top frame show the 2370-2375 Mhz rfi from the first
stop (intersection route 10 and R 6612).
- We were looking wNw toward the hq/radio redentor
towers.
- rbw .1 mhz, avg 5, preamp on.
- We we see a little leakage below 2400Mhz.. so there may
be some of this coming from these towers (or it could be
sidelobes from yiyi's tower).
- The bottom frame is taken between the HQ and RadioRedentor
towers (spaced by about 100m).
- .1 Mhz rbw, preamp off.
- This data was when we pointed up at the HQ tower.
- When we pointed at the RadioRedentor tower, the
signal was about 15-20 db when we pointed at the
RadioRedentor tower.
The 2370-2375 Mhz rfi.
- It looks like it is coming from the HQ tower.
- the hilltop
monitoring plots for the end of December (.ps) (.pdf)
- page 1: 28dec15: there is a band from 2345-2350Mhz .
this has been there for a long time.
- page 2: 29dec15 morning: the 2345-2350 birdie is gone
- page 3: 31dec15: the 2345-2350 birdie is gone, but
we now have a 5 mhz band from 2370-2375.
- looking at the x111 data for 17:00 30dec15 shows
that the 2370-2375 was present.
So it looks like sometime between 29dec15 and 30dec15 the
2345-2350 Mhz rfi jumped up to 2370-2375 mhz,.
The
2345-2350 Mhz rfi is band D of
the WCS bands.
- it was part of the fcc wcs #14 auction on 25apr97
- In PR it was bought by coloma wireless, which was
then acquired by at&t
Summary:
- we have 2 mhz spillover below 2400 Mhz coming from
yiyi's tower, or the tower to the east of it.
- the 2370-2375 Mhz rfi is illegal
- it is coming from the hq tower
- It looks to be the at&t band at 2345-2350 that
jumped up to 2370-2375 around 30dec15.
- We need to talk to the engineer of the HQ tower.
processing: x101/160119/rifvan.pro
18jan16: measuring 2400, 2370 Mhz rfi
from the 12 meter
dana and i went to the 12meter site to check on
the 2400 Mhz rfi on 18jan16. The setup was:
- sband helix antenna
- anritsu spectrum analyzer with:
- cfr 2380 Mhz, span 100 Mhz, rbw 1 mhz, vbw .3 , preamp
on, avg 5 sweeps, and peak hold
The plots show what we saw....
The
plots
show the rfi around 2400 Mhz (.ps) (
.pdf)
- The top plot shows 2330 to 2430 Mhz.
- black in an average of 5 sweeps, red is a peak hold.
- the 2370-2375 Mhz signal peaked up when pointing at
139Deg(Mag). This is the direction of the hq tower.
- the 2400Mhz rfi maximized when pointing in the direction
of yiyi avila's tower.
- middle frame: blowup showing the 2370-2375 rfi when
pointing at hq/radioredentor tower
- bottom frame: blowup showing 2400 rfi peak hold when
pointing at yiyi's tower.
processing: x101/160118/rfi12meter.pro
02dec15: measuring 2400 Mhz rfi from
the 12 meter
dana and i went to the 12meter site to check
on the 2400 Mhz rfi. The setup was:
- sband helix antenna
- anritsu spectrum analyzer with:
- cfr 2380 Mhz, span 100 Mhz, rbw 1 mhz, vbw .3 , preamp
on, avg 10 sweeps, and peak hold
We looked for the 2380 Mhz rfi, but we could
not see it with the setup (probably too weak.. we had
barely seen it fro the platform).
The plots show what we saw....
The
plots show the rfi
around 2400 Mhz (.pdf)
- The top plot shows 2330 to 2430 Mhz.
- The bottom plot blows up the horizontal scale
- the black trace is an average of 10 sweeps.
- the red trace is a peak hold (of 30 sec to 1 minute).
- We were pointing at yiyi avila's tower when we did this
measurement
- As we moved the antenna away from yiyi's tower, the
signal got weaker.. so it was definitely coming from the
direction so yiyi's tower
- This rfi should be 2400 Mhz and above.
processing: x101/151202/rfi12meter.pro
25mar15: 2110-2155 band reject filter
Dana has been working on getting a quote
for an band reject filter to get rid of the 2110-2155
rfi. It will be placed in front of the current sbw rf
filterbank.
The first plot has the
quote and the wideband band reject filter performance
(.pdf)
- The blue line is S21.. what the filter allows through
- The red line is S11.. what the filter reflects back to the
input
The 2nd set of plots shows
a blowup of
the filter response as well as the location of the rfi (.ps)
(
.pdf)
- Page 1:Filter response 2000 - 2300 Mhz
- Top:Black line: S21 (transmitted power), Red: S11
(reflected input power)l
- the black line goes below -110 db (at least in the
simulation)
- Bottom: S21 (angle) This show the phase change across
the filter.
- Page 2: Blowup of filter response 2090 to 2170 Mhz
- Top S21 (black) S11 (red)
- Page 3: Blowup of filter response an the rfi overplotted
- Top: filter response (red) and measured rfi (black)
- The black trace is an average spectra taken with the
mock spectrometer.
- I set the minimum value to 1 (2165 Mhz) so the scale
should be close to units of Tsys.
- There is probably some compression in the spectra.
- The location of the filter and spectra are arbitrary,
i just wanted to show the relative locations of the
filter and the rfi
- Bottom: Filter and filter*rfispectra
- The black spectra is the filter response,
- The red trace is the fitler response * the
rfiSpectra (plotted in db scale)
- If the filter simulation is correct, the 2155 rfi edge
will barely reach Tsys.
- The red bump at 2100 is a 3rd order intermod occurring
in the rfi spectra. This should go away when the band
reject filter is installed (as long as the intermod is
no in the lna).
Summary:
- We can get a pair of band reject filters ($1386 each) to
remove the 2110-2155 aws rfi.
- We will probably put this in front of the current sbw
filter bank.
processing:
x101/150319/sbwBandReject.pro
rfi
measurements home
home_~phil
03mar15: how sbw handles the 2110-2155
AWS rfi (top)
The Advanced Wireless Services (AWS)
transmissions covers 2110 to 2155. This lies with the sbw
receiver.
An rf filter bank is available with sbw to filter out some
rfi. Its passband filters are:
- 1700 to 1800 Mhz (filter 1)
- 2040 to 2400 Mhz (filter 2)
- 2600 to 3100 Mhz (filter 3)
An or'ing network allows the user to select one or more of these
filters.
The 2110-2155 AWS RFI falls inside the passband of filter 2.
The first set of plots shows
an average mock
spectrometer covering 2040 to 2200 Mhz (.ps) (
.pdf)
- 60 seconds, 8192 channels over 160 Mhz (20 khz resolution)
- filter 2 and 3 had been selected.
- The upstairs attenuator setting was:
- rfattn: 6,6 if1 attn 10,10
- The y axis is spectrometer counts (in db).
- Top frame: 160Mhz band.
- The dash red lines show the edges of the AWS band
- bottom frame:
- blowup in frequency, labeling the largest signals
- The dashed purple and blue lines are intermods caused by
2*F0 mixing with F1 and F2
Adjusting the sbw power
The if/lo system has a set of adjustable
attenuators and amplifiers to adjust the power levels in the
system. There are also two power meters.
The setup is:
- rfAttenuators: after the dewar, filterbanks, before the
first mixer.
- if1 Attenuators after the first mixer and before the fiber
optic transmitter that sends the signal downstairs
- power meter before fiber optic transmitter
- it has a 20db pad inline
- it is after the if1 attenuators
- if1mp is the command to measure this power
- typical values are -40db for 200 Mhz bands, -32dbm for 1
Ghz bands
- Fiber optic transmitter
- Fiber optic receiver
- downstairs attenuators and gain
- after the fiber optic receiver. 11 db of attenuation, 30
db of gain in 10 db steps.
- downstairs power meter after the adjustable gain,attens
- if2mp will measure this power level.
There are commands to automatically measure and set the power
levels for the upstairs, downstairs, and backend power
adjustments.
- adjpwrif1:
- Will try to adjust the rfattn, if1attn to give a
spectral density high enough to be well above the noise
floor of the fiberoptic transmitter and still give the
maximum amount of head room for rfi... This assumes
that we are adjusting things with no strong signals
in band
- For 200 Mhz bands, if1mp will typically adjust to -40
dbm
- For 1 Ghz bands, we shoot for -32 dbm
- (note that the actual power level is 20db higher because
of the 20db pads before the power head).
- adjpwrif2:
- will adjust the levels downstairs so the backends have
enough power..
Auto adjusting the power with AWS in band does not work.
The auto adjust power algorithm tries to
set the minimum power into the fiber optic transmitter
and still have the spectral density high enough so that the
fiber optic transmitter does not contribute much to Tsys. The
idea is to give the most dynamic range to the system without
affecting Tsys. Unfortunately we need to measure
spectral density and we're using a power meter....
On 24feb15 data was taken with sbw to see
how the AWS inband rfi was affecting the system. The setup
was:.
- rf band centered at 2570 Mhz
- sbw filters 2 and 3 inserted (2040 to 2400) and (2600
to 3100)
- The 10 Ghz upconverter was used for the first mixing stage
- the mock spectrometer took 7x160 Mhz bands covering the
range
- 8192 channels over 160Mhz for each band
- .1 second integrations
- A hardware winking cal was also used during the datataking
3 60 second data sets were taken:
- set 1: use the autoadjust commands to adjust the upstairs,
downstairs gain, as well as the mocks.
- This gave
- rfattn: 11,11 (max)
- if1attn (10,10) almost max
- gainif2: (3,2)
- mocks set for rms of 30 counts at the digitizer.
- Set 2: Allow 6db more power into the fiber optic
transmitter.. (after the 1st mixer)
- rfattn: 11,11
- if1attn: 6,6
- gainif2: (8,7) so that the same power went into
the mocks
- Set 3: Allow 6db more power into the fiber optic
transmitter (before the 1st mixer)
- rfattn: 5,5
- if1attn: 10,10
- gainif2 (8,7)
- Comparing set 2 and 3 would show whether the mixer was
compressing.
The plots
compare the
measured system temperatures derived from the 3 sets (.ps)
(
.pdf)
- Page 1: screen shot of spectrum analyzer
showing IF1 band after downstairs gain stage,
- The dashed red lines show the frequency covered by the 7
mock bands.
- the notch and 2500 Mhz is where filter 2 and 3 do not
intersect.
- You can see that there is less than 10db from the noise
floor to the signal at 2600 Mhz.
- (the spectrum analyzer noise floor is 10'sof db's below
this.
- The fall off at the right edge (3100 Mhz) is the 1 ghz
bandwidth filter in the 1st IF.
- Page 2: raw spectra
- Top calOn spectra is black, cal off spectra is red
- The 7 mock bands will have different gains
- bottom: calOn - calOff in spectrometer counts
- Page 3: Tsys computed for each set of data
- Top PolA, bottom PolB.
- The colors are:
-
rfAttn
|
if1Attn
|
if2Gain
|
notes
|
11,11
|
10,10
|
3,2
|
Auto adjust Power
|
11,11
|
5,5
|
8.7
|
6db more power after mixer
|
6,6
|
10,10
|
8,1
|
6db more power before mixer
|
- You can see that the Tsys for the black line is higher
than the green,red lines
- This is not a compression problem since there is less
power in the black line
- the green and red lines overlay one another
- This shows that the mixer is not compressing.
- The dashed purple lines are resonances in the sbw omt.
- Page 4: blowup of Tsys 2600 to 3100 Mhz
- The black lines is 2 to 5 k above the red and green line
- you should ignore the right part (2950 and up since
the IF filter is cutting off here.
Summary:
- The AWS rfi is causing 3rd order intermods in the sbw
system
- The autoajust power algorithm does not work when there is
strong rfi in the band
- after total power auto adjust, the spectral
density away from the strong rfi is too low and the fiber
optic transmitter causes Tsys to go up
- We need to increase the auto adjust power level for sbw
by at least 6db (but only when filter 2 is in use)
- The 1st mixer is not compressing even with 6db more power
into it.
- If it had compressed, the cal deflection would have been
smaller. This would have increased Tsys..
- But we see the red and green lines overlay one another,.
- We are looking into a filter (either two band pass
filters, or 1 notch filter) to remove this rfi.
processing:
x101/150224/sbwmocktsys.pro
