INTRO:

First trip out to the transmitter site:

• pictures of:
  
• site id  on tower fence (.jpeg)
• construction permit on tower fence (.jpeg)
• fcc antenna structure registration (.jpeg)

• Looks like they have 5 sites in PR. #2 is closest to AO.
  
• siteIds, lat,long, alt, atnModel,type, eirp, fccCall (.pdf)
• Look at page 3 of file..
  
• Omni directional antenna (TYPE:750 10074) used at sites 2-5.
  
•  specs (.pdf)
• beam pattern (.xlsx)
• Directional antenna (AP-19) used at side #1.
• Uses 3 antennas
  
• specs (.pdf)
• beam pattern (.xlsx)
  

TMUSPRSNJN0001

TMUSPRSNJN0002

TMUSPRSNJN0004

TMUSPRSNJN0005

TMUSPRSNJN0006

• 56dbm eirp = 26dbw eirp = 400 watts. (max allowable from fcc =2kw peak)
• antenna gain of 13 dbi --> amp outputs 400watts/13dbw -> 20 watts.
• chris computed antenna 2 distance to be 11.5 miles from AO.
• Free space propagation loss for 10km is 116db (at 1670) (from spectrum wiki)..10miles is about 4db more
• we should see a loss of 120db
• The federal rules say out of band emission must be less than 43 + 10log(P)db below the in band power (P) (i think they mean to do it in watts...)
  
• if 26dbW eirp, then the out of band would be eirp of -43dbW (using 5MHz bw) or -13dbm eirp.
  
• The itu recommendation for out of band emission is  -251 dbW/m^2/Hz
• I'm assuming this is at AO. using 11.5 Miles we can compute the out of band eirp
• 11.5 mile sphere has 4.3e9  m^2
• sphereAreaM^2 * energyDensity/Hz=4.3e9*(10^-25.1) = 3.4e-16W/hz=-154dbW/hz
• a 5MHz band would have: -154dbW/Hz + 60 = -94dbW  = -64dbm  eirp out of band in 5MHz.
• So this looks to be about 50 db less than the federal spec for our distance..
  

fcc related rules

• fcc allocation table for 1670-1675 (.ps)
• title 47 (telecommunications) subsection 27 (misc wireless devices) specs for 1670-1675
• 27.50 power limits and duty cycle
• (f) The following power limits apply to the 1670-1675 MHz band:
• (1) Fixed and base stations are limited to 2000 watts EIRP peak power
• (2) Mobile stations are limited to 4 watts EIRP peak power.
• 27.53 Emission limits
•  (k) For operations in the 1670-1675 MHz band, the power of any emission outside the licensee's frequency band(s) of operation shall be attenuated below the transmitter power (P) by at least 43 + 10 log (P) dB. Compliance with these provisions is based on the procedures described in paragraph (a)(4) of this section.

Terrain modeling output for sites

• terrain modeling path loss: site numbers are transmitter id number's last digit.
  
• site  1 (.pdf)
  
• site  2 (.pdf)
• site  4 (.pdf)
• site   5 (.pdf)
• site   6 (.pdf)

150708: telescope data taken with wapps

    On 150708 project s2141 took lband data with the wapp auto correlators. The setup was:

• cfr 1667 to 1670
• 50 MHz bandwidth, 2048 channels (24Khz rbw),  9 level sampling, 1 second integrations
• 5 minute scans (300 1 second samples). Looks like they were doing on/off position switching so there are two 300 second scans over the same part of the dish.
  
• The 1670-1675 was present during this observation

• the average spectra was computed for each 300 second scan.
  
• the spectra were hanning smoothed so the chan width is 2*24=48kHz.
  
• Page 1: linear plot of average spectra.
• I normalized the spectra by averaging over 1655 to 1659 MHz.
• Top polA, bottom: polB
• Each color is a different 300 second average
• it gets to about 1000 Tsys in the 48KHz channel widths.
• Page 2: Blowup of linear plots.
• you can see the falloff of the spectra close to the 1670,1675 band
• It's hard to tell if the bandpass ripple is from the filters, telescope, or from the rfi.
• Page 3: average spectra as db plots
• the average rfi strength is 25 to 35 db above Tsys.
• Page 4: rms/mean for each freq channel
• The rms/mean was computed for each frequency channel in the raw 1 sec records.
• This should not depend on the bandpass shape (as long as it is stable).. Tsys variation with za during the scan will increase the rms a little.
• You can see that the rms is normal to about 1665 MHz, and the it starts to rise.
• We don't know if this is transmitted by the rfi tower, or a problem with our if/lo / spectrometer.
• The red horizontal is what i estimate the rms should be:
• expRms=9levelFactor/sqrt(bw*time)
• bw=24kHz *hanningFactor*sinx/XFactor
• HanningFactor = 1.5*1.5 = 2.25
• sinx/xFactor is from the sinx/x filter shape in the auto  correlator (1.208)
• 9level factor: 1.02
• (wonder whether i left out a sqrt(2) on top?....
• The raw spectra had the average spectral value close to .04 away from the rfi block. Normally the value should be 1 (measured/optimum 9level value)... So this may have increased the rms a bit.
• Page 5: plot average power in 1670-1675 vs az,za ...
• top: average power in rfi band for the 10 scans
• 2nd: average azimuth for the 10 scans
• 3rd: average zenith for the 10 scans
• bottom: the az,za for each scan with the average power written next to it.
• The highest power is at an az of 145degrees. I think this has the azimuth arm perpendicular to the tower location.

• plot txid4 az swing total power data
• plot txid1,5,6 az swing total power data
• look at linearity tests
• plot short az, za swings for txid4
• put anritsu plot from platform on web.
  

Summary:

• we saw an average level 25 to 30 db above the system temperature. There is not guarantee the these measurements remained linear.. so the level may actually be higher.
• adjusting the 9 level power in the 50 MHz band resulted in an average level of .04 outside the rfi block. This should be the ratio of the measured/expected power for 9 level sampling
• looking at the rms/mean by channel
• the rms was higher than the expected value by about sqrt(2).
• the rms started to rise above the constant value at 1665 MHz.
• az,za position of maximum signal strength
• looks like azimuth perpendicular to tower gives the strongest signal (about az=145).

151006: on,off testing with lightsquared

    On 06oct15 we did tests with lightsquared. We tested transmitter id's 1,,4,5,6. light squared stated prior to the test that they will not be using  Transmiter id #2 (the closest to AO)  since it's coverage was redundant.
    The tests started up 9:00 am ast.

The setup was:

• lband wide receiver with 1370 hipass and 1550-1820 band pass filters in.
• rf center frequency 1660 Mhz
• We used 3 mock spectrometers.
•  All had 8192 channels running at 160Mhz bw with 1 second integrations.
  
• mock 1: cfr 160 Mhz, bandwidth 160 Mhz
• mock 2; cfr 170 Mhz, bandwidth 80 Mhz
• mock 3: cfr 170 Mhz, bandwidth 16 Mhz
• mock 1,2 used the digital mixer to set the a/d dc at 160Mhz and the center of the band at 170 Mhz.
• I set the adjust power level so 1 sigma rms in a/d input was 60 counts.
  

The datataking sequence was:

• files written to /share/pdataN/x101.201511006.bMs1g0.mmmmm.fits
• N=1..3 for mocks 1..3, M= 0..2 for mocks 1..3, mmmm = 000 to 1500 (in steps of 100).
• file:0000  testing. tried noise shift in spectrometer. Got lots of overflows.. (was coming from 1680 birdie). I then switched to radar shift for then rest of the runs (this is the butterfly shift.. see mock pdev page).
• Adjusted power levels with rfi off
• rfattn : 0,0  if1attn: 0,0  if2gain:12,10. mock gains: (8,8),(4,4), (8,7)
• These atten levels were left for the entire run..except the the rf attn 3db step during linearity test
• files: 100,200  linearity test rfi off
• F100 47sec  rfattn 0,0,
  
• F200 47sec  rfattn 3,3
  
• Then rfattn set back to 0,0
• transmitter ID 4 turned on (closest in use to ao, line of site).
  
• file 300: azswing;
• domeza=18, ch=stow, azimuth 200 about 700 at .35 deg/sec.
• Found 1670-1675 peak at 299.14 degrees
• moved to az=299.14 degrees, za=18.
• file 400 to 700 linearity test txId: 4 on
• f400  60 secs rf attn 0,0
• f500  60 secs rf attn 3,3
• f600  60 secs rf attn 0,0
• f700  60 secs rf attn 3,3
• rf attn back to 0,0
• file 800 to 1100. 300 sec txId 4 on, off
  
• f800: 300 sec txid4 on
• f900: 300 sec txid4 off
• f1000: 300 sec txid4 on
• f1100: 300 sec txid4 off
• tx4id off
• file 1200: monitor txid1,5,6 coming on
• took 30 to 60 secs for a transmitter to come all the way up.
  
• txid1 up (before file started)
• rec 6: txid5 coming up
• tec 36: txid 6 coming up
• file 1300: az swing with txid1,5,6 up
• za=18
• az 660 to 190 at -.35 deg/sec
• saw sun sidelobe, no obvious az dependence of the others.
• file 1400: 300 secs td1,5,6 on
• az=299.14,za=18
• file 1500: small azswings, step in za with txid1,5,6 on
  
• tiltaz 295 305 -3 .35 6
• az swing 295 to 305 at .35 deg/sec
  
• start at za=18, step -3deg za each swing (starting on alternate sides)
• do this 6 times (down to za of 3). data was taken continuously

• spectrum analyzer with preamp on.
• peak signal around bearing of 278 D magnetic.
  
• subtracting 12.9 deg gives a true bearing of 265.. this is essentially the same that google earth gives for the bearing from ao: 266 degrees.
• We saw the signal about 6db above the noise floor (get image from analyzer...)

The data:

    Transmitter TxID 4:

    Transmitter id #4 is closest tx to AO (26 miles) and is line of sight (0 terrain shielding)

The first set of plots show the 300 second on/offs with txid 4 (.ps) (.pdf)

• I plotted the 160Mhz,80Mhz, and 16 Mhz bandwidths on separate pages
• Top frame: the 2 on/offs are over plotted. black is polA, red is polB. Vertical log scale
• 2nd frame: blowup in vertical scale (+/-3% of Tsys).
• bottom frame: average 2 on/off's and then average polarizations.
• I measured the rms between the green dashed lines. It is printed as well as the expected rms:
• expectedRms=1/sqrt(bw*time)
• bw=bw/8192channels
• time=300secs*2.
  
• The extra sqrt(2) in the numerator cancels with the sqrt 2 in denominator from adding pols.
• The negative going spikes are spikes in the tx Off.
• Page 1: 160Mhz bandwidth
•  the signal is 5 to 30Times Tsys
• frame two shows an image signal at 1645-1650.
  
• This is an image from the mock analog mixers (mixer cfr is at rf=1660)
• Using the peak values:
• polA; 12/.02= 600 : 28 db rejection
• polA: 28/.029=965 : 30 db rejection (I used the right part of peak since true peak cutoff in middle plot)
  
• Page 3: 16 Mhzband
• bottom frame shows steps in spectra (blue  lines). These are spaced by 1024 channels in the fft
• I think this comes from underflows in the butterfly stages of the mocks.

    Transmitter TxIDs 1,5,6:

    Transmitters with txIds 1,5,6 have terrain shielding that varies fromj -57 to -77 db. The total losses from each of these transmitters is -192 to -208 db total path loss. For these transmitters we did the following:

• put the telescope at az=299.14 , za=18
• txId 4 was turned off, txIds 1,5,6 were all turned on together.
• We did not do a transmitter off. So i used the transmitter off from the 2nd on,off of txId4
• most offs were 5.4 minutes after the ons.
• the txid1,5,6 off was 35 minutes before the on  

 The first set of plots show the 300 second on/off with txid 1,5,6 (.ps) (.pdf)

• I plotted the 160Mhz,80Mhz, and 16 Mhz bandwidths on separate pages
• Top frame: the 2 on/offs are over plotted. black is polA, red is polB. Vertical log scale
• 2nd frame: blowup in vertical scale (+/-2.5% of Tsys).
• bottom frame: average 2 on/off's and then average polarizations.
• I measured the rms between the green dashed lines. It is printed as well as the expected rms:
• expectedRms=1/sqrt(bw*time)
• bw=bw/8192channels
• time=300secs*1.
  
• The negative going spikes are spikes in the tx Off.
• Page 1: 160Mhz bandwidth
•  the signal is .5% to 1% of Tsys in the 1670-1675 band.
• The ripple across the band is about 1 Mhz.
• This is probably from using an off that is 35 minutes from the on. In this time, the sun probably moved enough so that standing waves created by the sun did not cancel.

    Transmitter TxID 4 offs.

 The first set of plots show the 300 second Tx4off1/tx4Off2 with (.ps) (.pdf)

• I plotted the 160Mhz,80Mhz, and 16 Mhz bandwidths on separate pages
• Top frame: the 2 on/offs are over plotted. black is polA, red is polB. Vertical log scale
• 2nd frame: blowup in vertical scale (+/-2.5% of Tsys).
• bottom frame: average 2 on/off's and then average polarizations.
• I measured the rms between the green dashed lines. It is printed as well as the expected rms:
• expectedRms=1/sqrt(bw*time)
• bw=bw/8192channels
• time=300secs*1.
  
• The negative going spikes are spikes in the tx Off.
• Page 1: 160Mhz bandwidth
•  the signal is .5% to 1% of Tsys in the 1670-1675 band.
• The ripple across the band is about 1 Mhz.
• This is probably from using an off that is 35 minutes from the on. In this time, the sun probably moved enough so that standing waves created by the sun did not cancel.

• M/E is measured/Expected rms
  

• the txId1,5,6 rms ratio is high. This is probably from the txId4 off that was used was 35 minutes earlier. The solar standing waves probably changed over this time period.
  
• 
  

Summary

• for za=18 degrees the peak signal from the strongest (txid4) peaked at az=299.14
• we saw no variation in the signal from txid1,5,6 for the az swings (1 sec integrations).
• txId 4 on,offs:
• In 20Khz channels the txid4 signal ranged from 5 to 20 time tsys.
• it covered the entire 5 Mhz 1670 to 1675
• the mock  mixers show an image of the 1670-1675 at 1645 to 1650.
  
• this is a mixer image rejection of 28 to 30 db,.
• txId 1,5,6 transmitting simulatenously
• We saw the 1670-1675 band signal at a level of .5% to 1% of tsys in spectral density.
• the large ripples in 1660 to 1669 were probably from the sun.
• The off for this measurement was from the txId4 off (35 minutes earlier)
  
• the expected rms's across the oh band was within a factor of 2 of the measured value.
• Using the 2 txId4 off's (no transmitter) the ratio of measured/expected rms was 1.4,1.5, 1.04 for the 3 bandwidths.
• doing 5 minute on, offs during 9:30 to 10:45 am looks to have standing wave ripples from the motion of the sun relative to the structure.
  
• the 16Mhz band mock band showed spectra steps at 1024 channel steps.

page up

home_~phil