2400 Mhz rfi:

UPDATE: JUN12: This was coming from KS computers using the 2400 ism band for intenet wireless communications. They have moved to the 5ghz ism band. 

 Mike nolan reported 2395 rfi on 12oct11 during an sband transmitter run. This is outside the ism band that goes from 2400 to 2500 MHz.:

Measurements:

• 30apr12: measure 2400 Mhz rfi coming from yiye avilar tower (roncador).
  
• oct11: hilltop monitor data, az dependence using the telescope

30apr12:Measuring the 2400 Mhz rfi at yiye avila's tower (roncador).

 The setup used to take data was:

• Sband helix antenna
• anritsu MS2713E spectrum analyzer:
• rbw: 0.3 Mhz, avg 50 sweeps, preamp on.
• Angel's gps receiver said we were 3 miles from the tower (but the line of site was probably much less).

• The rfi goes from 2394.5 to 2403.4 (about 9 Mhz bw).
• When we pointed away from the tower, the signal strength decreased.

• Pointing the antenna up at the tower, the signal was about 15 db above the noise floor.
• When we pointed the antenna away from the tower, the signal got much weaker.

Summary:

• The 2400 Mhz rfi is coming from yiye avila's tower (on roncador).
• Looking at the hilltop monitoring data, the rfi has been sitting at 2400 Mhz  for a few months without moving.
  

oct11: hilltop monitor data, azimuth swings. (top)

Intro:

Summary:

• before 08oct11 a 9 MHz band centered at 2401.5 MHz was present. It was often there but would occasionally go away for a few days (see the hilltop monitor data).
• 08oct11: The band moved its center freq (cfr) from 2401.5 MHz to 2396.5 MHz.
  
• 12oct11: azswings showed that the cfr had moved up to 2399.0 MHz.
• The azimuth swing has the source peaking toward az=180 degree. it is a true azimuth dependence since the cw and ccw spins overlaid each other.
• So the rfi is probably not rock solid stable in frequency (we probably wouldn't see this shift in the hilltop data with its coarse frequency resolution).
  
• A helical antenna with the portable spectrum analyzer from on top of the antenna test shack showed that the rfi was coming from the direction of yiyi avilar's tower (or a tower close to it).
• The signal looks like an ofdm signal with
• 52 subcarriers spaced by 155 Khz.
• every fourth subcarrier is stronger than the rest.
  
• Looking back through the hilltop monitor data may11-oct11 the 2405 Mhz band started jumping around on 11sep11. The table belows shows when the jumps occurred.

• 2405 channel jumps

  date	freq
  1may11-11sep11	2405
  11sep11	2420
  26sep11	2430 then 2440
  29sep11	2435
  30sep11	2440
  03oct11	2405
  08oct11	2400

• Who could it be?
• 111017. jumped 2400->2510-2520
• 111018. 2510-2520 -> 2395 -> 2405
• looking through the uls database:
• wlx661 educational broadcasting ana G mendez. jayuya, 2512->2518 Mhz
• Tower at: 18.1677,-66.57679  of pr 143 in jayuya.
• Need to check if the direction lines up with this, or with roncador.
• from google maps:
• 18.1677, -66.57679  jayuya tower
• 18.23273, -66.75917 yiyi's tower on roncador.
• 18.344, -66 753  AO (from google maps)
  

Hilltop monitor plots of the ism band:

• (see 2011 oct plots)  especially 08oct11 when the shift occurred)
  
• Notes on hilltop monitor data:
• The hilltop monitor uses a spectrum analzyer that has drifted off freq. To correct the data, i  aligned the 2380 tx signal to 2380 MHz.
• The spectrum analyzer uses a 3 MHz rbw in the 2200 MHz band.
  
• To accurately locate the frequencies, i used the data taken on the telescope (which is tied to the hydrogen maser).
  
• The hilltop monitor data shows that a 9 MHz band centered at 2402   MHz shifted down to 2397 MHz on 08oct11.
• The band shifted down around 19:33 ast on 08oct11
• The band remained at this lower frequency (at least thru 14oct11 when this was written).
• Looking through the plots for 2011, this was the first time in 2011 that this dump occurred.

2400 rfi seen through the telescope.

• Spectra taken before and after the jump shows the frequency change of the rfi (.ps) (.pdf):
• Page 1: x111 sbw data 08oct11,09oct11
• rbw=50Khz after hanning smoothing. black is polA, red is polB
• The top plot shows a  spectrum averaged for 60 seconds before the jump (16:20 ast 08oct11)
  
• The rfi is centered at 2401.5 MHz with a 9 MHz bw. (this band spills outside of the ism band which starts at 2400 MHz).
• the az=285deg, za=8.5
• The bottom plot shows the rfi after the jump (07:59 on 09oct11).
• the rfi is centered at 2396.5 MHz.
• az=72.7, za=8.835. This is a different az,za than the top plot.
• Page 2: average spectra from the az swing done 12oct11 using the mock spectrometers
• The rbw is 20Khz. The azimuth swung from -90 to +270 deg at .38 deg/sec with the dome at za=15
• Top: the spectra averaged over the 20 minutes azimuth swing
• red is black is polA, red is polB
• Bottom: Averaged polarization A and B.
• You can clearly see multiple small carriers spaced 155 Khz apart with a stronger carrier every 4 channels.
• Dana though this looked like an ofdm (orthogonal frequency division multiplexing) signal
  
• 12oct11 azimuth swing showing position depedence of rfi (.ps) (.pdf)
• The azimuth was swung from az=-90 to az=+270 at .35 deg/sec (cw) and then -.38 deg/sec ccw. The dome was at za=15 deg.
• The frequencies 2394.5 to 2403.34 were averaged and then plotted vs azimuth:
• Page 1: rfi power vs azimuth. Black in CW spin, red is CCW spin
• top PolA . peaks at 62 and 182 degs az.
  
• middle polB
• Bottom: average polA and polB
• The peaks for both the CW and CCW swings line up so it is a real azimuth dependence (and not a time dependence).
  
• Page 2: blowup around az = 60, 180, and 300 degs.
  
• These are the azimuth positions that are perpendicular to the sides of the triangle.
• black is polA, red is polB
• There are peaks close to 60 and 180.
• polA at 180 has a peak at 182 degs. The triangle apex (north) actually points at 2.87 deg east of north. So perpendicular to the 180 side would actually be at 182.87 deg.. maybe this is just a coincidence for polA.
• dynamic spectra (172 MHz bw) showing the spectra vs azimuth.
• The receiver was also seeing the cell phone band, so some of the weaker rfi might be ghosts (intermods).
• The vertical scale is azimuth but it also maps to time.
  
•  CW spin .gif
•  CCW spin .gif
• You can clearly see the 2400 mhz rfi.
  
• The 2435 rfi peaks at an azimuth close to 90 degs.

From the antenna test shack.

RFI at 2380 MHz. (top)

22feb11 - 2380 rfi goes off when tower transmitter off.

• Each strip is a 1 minute peak hold from the hilltop monitoring system.
• Strips are measured once every 20 minutes.
• the 11:41 AST measure showed the rfi jumping from 2380 to 2437 MHz. This is around the time the tower operator said the signal was turned off.
• I'm not sure why the signal jumped in frequency rather than just going off. This same type of jumping occurred when the rfi first occurred.
• was there really only 1 xmter at 2400? if multiple transmitterd exist,  then maybe the load was too large for something causing the frequency to fail. Or maybe the frequency became unlocked.
• Maybe this frequency communications got picked up by another xmiter when the "bad one " was turned off.

19jan11 - azswing looking at rfi

12jan11 - rfi appears at 2380 MHz

• On 18,19jan11 data was taken on the telescope to examine the rfi more closely.\
• On 18feb11 we made a trip to a radio tower in utuado owned by radio redentor.
•  The rfi is coming from this tower (probably one of the parabolic directional antennas mounted on the tower.).
  
• There is a clear line of site from the tower to the platform.
• Tower coordinates: 18:17:22.1N, 66:39:37.3w
• Tower registration number: 1011545
• owner:
• radio redentor inc
• attn: jesus rivera
• po box 29404  sanjuan pr. phone:728 7751-6318
  
• There is a 2nd tower belonging close to this tower that did not show the 2380 rfi
• probably 1036516 HQ 103 inc at 18:17:27.0N,66:39:31.0W
  

Hilltop monitoring:

• Hilltop monitoring dynamic spectra 10jan11 - 15jan11 at 2200 MHz (.gif):

• each frame is one days worth of data at 2200 MHz.
  
• 1 minute peak hold taken every 20 minutes, 3 MHz rbw
• center left is 12jan11 where the jump occurred.
• The 2380 rfi appeared at 1:42 am.
  
• It has remained at this position.
• Hilltop monitoring dynamic spectra for 12jan11 at 2200 MHz (.gif):
• This shows the hilltop dynamic spectra for 12jan11.
• The rfi moved from around 2380 in about 3*20 = 60 minutes.
• Individual spectral plots around appearance of 2380 rfi (.ps) (.pdf)
• Top: 1 minute peak hold spectra about jump.
  
• Each spectra is spaced by 20 Minutes (with time moving from bottom to top).
• There are offsets included for display.
  
• The jump occurred at the purple spectra around 1:42 am.
• Initially there was rfi at 2420 and 2480
• something appeared at 2460 for about 2 spectra (40 minutes).
• It then moved to 2380, 2410 (stronger) and 2440 (weaker).
• rfi remained at 2480 but it's strength had decreased by about 10 db.
• bottom: over plot 5 spectra around jump
• There are no offsets so you can see how the amplitude changed.

20sep01.Measuring the power levels at the output of the sband wide dewar. (top)

• Figure 1 top plots the spectrum 1000 MHz to 2950 MHz for polarization A. The resolution bandwidth was set to 3 MHz. The system temperature is about 40 K. The red line averages 100 sweeps of the spectrum analyzer. The black line is a peak hold for greater than 30 seconds (I don't believe that the black line is a peak hold, it looks like it was left on averaging mode). Integrating the spectra gave -23.5 dbm.
• Figure 1 bottom plots the spectrum 1000 MHz to 2950 MHz for polarization B.  The red line averages 100 sweeps of the spectrum analyzer. The black line is a peak hold for greater than 30 seconds  Integrating the spectra gave -11dbm (peak hold) and -24.3 dbm (100 sweeps). The peak hold values where there is no interference should be about 3 db larger than the averaged version (not 13 db). So something went wrong in the measurement.

 

The large interference at 1940 MHz are from cellular phones. The large birdie at 1350 MHz is the san juan airport radar. I don't think that there is any filter between the dewar and the post amps. This needs to be corrected.  We also need to decide what to do about the birdies at 1940 MHz. Do we cut them out with a bandpass filter or try and use a notched filter?  Our spectral line list does not have any lines between 1720 (OH) and 2661 (HC5N).

rfi measurements home
home_~phil