Intro: (top)

• cband receiver was used. freq range recorded was 4-5 GHz.
• Previous azimuth swings tended to cover 4700_5500 MHz.
  
• The mock spectrometer was used to record the data.
• 7 160MHz bands were used (with a 10mhz overlap on each edge).
• 8192 channels / band (20Khz freq resolution) and 1 second integrations.
• When generating images, the images covering the entire band have 600 Khz resolution , while the individual band images have 120 Khz resolution
  
• 6 360 azimuth swings were done: 3 Clockwise (CW) and 3 Counter clock wise (CCW)
• The rfi looked like it might be time as well as azimuth dependent, so multiple swings were done.
• The gregorian dome was at 18 deg.
  
• The azimuth moved from 270 to 630 degrees and back at .35 deg/second.
• The weather  was clear during the swings.
  
• The data was taken  16:15 to 19:45 AST.

Data from the rfi hilltop monitor. (top)

    The rfi  hilltop monitor uses filters, amplifiers, switches, and a spectrum analyzer to monitor the rfi environment from DC to 10Ghz.
The frequency range is broken up into 20 separate frequency bands.

•  A 1 minutes peak hold on the spectrum analyzer is done at  each band.
  
• The system returns to the same band every 20 minutes.
• The 4-5 GHz band has 401 points over the 1 GHz bandwidth.

The first plot shows the 1 minute peak hold spectra during the 6 azimuth swings (.ps) (.pdf)

•  there are 9 spectra over plotted.
• The vertical scale is dbm and measured by the spectrum analyzer (after the amps and filters).
• The dashed green and red lines mark parts of the spectrum where the average power was computed over time (in the following plots). 
  

Plots of total power vs date for the various rfi (.pdf)

• The average power for each of the 4 bands flagged above, was computed and then plotted vs time.
• the bands were:
• 4220-4340
• 4450-4700
• 4630-4700
• 4750-4800
  
• The red color is the average power over the rfi band
• the black color is the median bandpass over the 4-5 ghz band
  
• Page 1: Average power vs time 2012 - 2019
• each frame is a separate frequency band.
• The steps in the total power around aug2015 is probably the amp/atten changes to the monitoring system.
• Some of the narrow spikes may be 430 Mhz transmitter harmonics getting into the system (i tried to remove the data when the sband or 430 Mhz transmitters were on).
  
• Top: 4220-4340. this is part of the aeronautical radio navigation assignment
  
• The power levels  changed toward the end of 2013.
  
• it was on,off during the day. after 2013 It changed to always on.
• You can see the drop around sep17. this was hurricane maria leveling the transmitters (or planes?)
• The bottom 3 frames seemed to have gotten stronger toward the end of 2018.
• Page 2: blowup sep18 - feb19
• the bottom 3 rfi bands are plotted for sep18-feb19
• The 4450-4700 and the 4630-4700 rfi bands got a lot stronger around oct18.
• there changes tracked one another. They may be coming from the  same source.
  

The azswing data processing for each board (top)

• input the 1 second spectra
  
• average pola and b.
  
• interpolate az positions to the center of the data records.
• For the dynamic spectra two sets of data were processed:
  
• combine all frequency bands into 1 spectra.
• for each freq band  a median spectra was computed over all data for that band and then used as the bandpass correction.
  
• interpolate each freq band spectra onto a fixed spacing grid that covers all the bands
  
• these images had lower frequency resolution (because of the limited number of pixels on the screen).
  
• Individual dynamic spectra were made for each frequency band from the first az swing.
• this had higher frequency resolution than the combined images.
  

The az swing dynamic spectra: (top)

All frequency bands in 1 image:

• the top frame is the clockwise az swing.
• The bottom frames is the following CCW az swing.
• The first column has a high sensitivity image (where some things are saturated). The 2nd column has a lower sensitivity version of the  image where the stronger signals are not as saturated.

1 freq Band per image

•  Each row is a separate frequency band (as taken by the mock spectrometer).
• each column is a different azimuth swing.
  
• In each image the top frame is the clockwise spin while the bottom frame is the ccw spin.
  

Peak hold Spectra for the azswings (top)

    For each CW or CCW azswing

• a peak spectra was computed (max value in each freq channel for the  swing).

the plots show the peak hold spectra for each swing (.ps) (.pdf)

• Each page is a separate az swing.
• Black is the peak hold spectra for the CW swing. Red is the spectra for the CCW swing
• The y axis is a log scale.
• the top frame shows the full scale.
• the bottom frame blows up the vertical scale.
• 4200 MHz is the upper cutoff for the micro wave point to point link band.
  

Average birdie power vs azimuth. (top)

• The table below summarizes the 5 frequency ranges.
• The wide bands  may include more than 1 transmitter.
  

Summary/ what they could be. (top)

• the rfi hilltop monitor showed rfi at:
  
• 4035-4152 .. 30db above noise floor
  
• 4216-4326 ..  27db above noise floor
  
• 4451-4547 ..  10db above noise floor
  
• 4628-4700 ..  10db above noise floor
• 4752-4802 ..  1db above noise floor
• The levels are peak hold over 1 minute.
  
• 6 azimuth swings were done covering 4 to 5 GHz.
• Some of the rfi was extremely repeatable in azimuth.
• this implies the signal is stationary on the ground
• it is not time variable.
  
• The following rfi needs to be resolved:
• FCC frequency allocation table  for 3500-5000 MHz (.pdf)   (downloaded 27feb19)
  
• 4000-4200 MHz: this is part of the point to point microwave link band. (more info)
  
• The gregorian dome az of the rfi was:  -55 to +12.
• peaks at -48 and +11 gregorian azimuth.
• peaks at [132,191] if you add 180 degrees.
  
• prtc tower east of jayuya is the only FCC licensee i found in pr  covering 4000-4200 MHz. (location)
• it has a bearing of 121.5 degrees from AO (via google earth).
• This is about 11 degrees from the gregorian az + 180.
  
• The peak at gregorian az bearing of -48/132
• 132  is close to the cerro punto tower (or the 3 towers east of cerro punta). (location)
  
• the FCC data base does not show them as using the 4000-4200 MHz band.
  
• 4250-4350 MHz . rfi that does not repeat in azimuth
• The 4250-4350 plane radio altimeter band could be the source of this rfi (more info).
• You would expect it to not repeat in azimuth (since the planes are moving).
• Looking at the hilltop monitor spectra, it looks like the 4000-4200 Mhz rfi.
• We need more investigation to determine if this is a land based xmiter or not.
  
• 4500-4700 MHz.
  
• occurs at az of -30 and -8 deg. adding 180 degrees gives  az 150 and 172.
• the closest tower  to 172 degrees is the tower east of yiyi avila's tower (it is at 176 deg) (location).
• The FCC non federal allocation is fixed-satellite (space to earth). I don't think this  would have a fixed azimuth location.
• I wonder if this could be an  modulation production of some other microwave link transmissions (6GHz, 11GHz)..
• 4750-4800 MHzMhz.
  
• Also peaks at -8 deg gregorian az.
  
• May be part of the 4500-4700 signal,
• It is much stronger than the other signals in the dome signal. It did not appear as strong in the the rfi hilltop monitor data.
• This was also seen in all of the previous azimuth swings done in 2016 (links)
  
• 4910-4920 MHz. 
  
• 10 Mhz band that peaks around 130 deg azimuth.
  
• Narrow band signals like this have been seen in previous measurements.
• To do:
• the 4500-4800 MHz rfi should be further investigated.
  
• I don't see any FCC licenses for these freq range in p.r no federal (except space to earth)
•  It also seem to be stable in time and direction... it's been around since at least 2016