Intro

Setup:

• The telescoped was stationary at az=84,za=18 degrees.
• Data was recorded with the mock spectrometer
• 6 x 160 MHz bands were recorded  covering 1114 to 1755 MHz.
  
• spectra were dumped at 64 useconds  with 625 KHz resolution.
  
• A 5 second calOn,calOff was done  to convert from spectrometer counts to kelvins.
• data was taken for 37 seconds (a little more that 3 12 sec radar rotations).
  

Processing the data:

• The calon,off was used to compute the scaling factor from spectrometer counts to kelvins.
  
• The average spectra was computed for the 5 second  calOn and calOff spectra.
• an attempt was made to exclude frequency channels with rfi by fitting to  calOn/Caloff and excluding freq channels residuals greater than 3 sigma.
• the spccntsToKelvins was computed by using the cal value for the center of each band and the mean value of (calon-caloff).
  
• outliers (from step 2) and the bandpass edges were excluded in the average.
  
• the same channels were used in the actual spectra when converting  spc counts to kelvins.
• A band pass correction (to remove the IF bandpass) was computed using the calon-caloff spectra
• a robust 13th order harmonic fit was done to calOn-caloff.
  
• let bpIF(frq) be the IF bandpass.
  
• assume Tsky,Tcal is constant in freq over the 160 MHz bands.. then:
• CalOn-calOff = (Tsky + Tcal)*bpIF - (Tsky)*bpIF  = Tcal*bpIF(Freq)
  
• the band pass correction is then Tcal*bpIF(freq)/mean(Tcal*bpIF) .. (where the average is over the non rfi channels).
• Each 64 usecond spectra was then converted to kelvins  by multiplying by spcCntsToKelvins/bpc
• PolA and polB were then averaged together (note: the first 3 bands had trouble with polB so only polA was used).
  
• To convert from kelvins to dbm:
• dbm= alog10(degK * resolutionBandWidth*Kboltz)*10 + 30 .. (Kboltz =1.3806e-23)
• The 6 x 160MHz spectra were interpolated to a single spectra covering the entire band.
  

The results:

The average and peak hold spectra:

• The average and peak hold spectra  in degrees Kelvin (.ps) (.pdf):
• The vertical axis is degrees Kelvin.
  
• Top Frame The black trace is the average spectra. The red trace is the median spectra
  
• 2nd Frame: the peak hold spectra for the 37 seconds (using the 64 usec spectra).
  
• This kept the highest value in each frequency channel (625Khz resolution) for the 37 seconds.
• Bottom Frame: rms/Mean  for each frequency channel for the 37 seconds.
• If we had just noise, the radiometer equation would give sqrt(1./(625Khz*2pol*64e-6secs) = .11
• You can see a jump around 1380 MHz. below that only polA was used.
  
• The average and peak hold spectra in dbm (.ps) (.pdf)
• The average spectra in degK was converted to dbm using the 625Khz channel resolution.
• Top Frame The black trace is the average spectra. The red trace is the median spectra
• bottom Frame: peak hold spectra.
  

The total power vs time

    The total power vs time (.pdf)

• The total power (1114-1755 MHz) was computed for each 64 usecond spectra.
• the 37 seconds is a little more than 3 12 second radar rotations.
• the carsr radars pulses are:
• 117 usecs freq 1, 117 usec freq 2, 20sec freq , 20 usec freq4.
  
• the 64usec spectra should sample this correctly.
  
• power ranges from about -96dbm to -83 dbm
  

 Plotting some radar bands vs time.

    The power in some radar bands was plotted to see which was affecting the total power the most.

• Only 13 seconds  was used for the plots (a little more than 1 radar rotation).
  
• Top frame: lbw total power
  
• 2nd frame: the 4 frequencies used by the FAA carsr radar (more info)
• The 4 frequencies are plotted in different colors.
• You can see when the faa radar blanks when it points toward AO.
  
• The two edges of the blanking are the same height, so their blanking direction is aligned with the AO direction.
  
• 3rd frame: puntaBorinquen radar (more info)
  
• This is the same type of radar (carsr) as the faa.
  
• The strongest peak in lbw total power is coming from the sidelobe .5 seconds (15 degrees) after the center of the beam.
• this sidelobe is a known problem with the punta borinquen dish.
  
• It is 10db higher than the blanked beam. (should we encourage them to fix it??)
  
• bottom frame: the punta salinas frequency agile radar (more info)
• It was running in ModeA.. using only 4 frequencies. 
• I only plotted 3 frequencies since one of them overlapped with gps.
• punta salinas  has a short pulse/ipp: 51/1500usecs =3.5% and a long pulse: 409/2600usec 15%.
• We are probably seeing the long pulses sticking out in the plots.
• the punta salinas blanking period is much wider than the other two radars.
  

The radar bands total power vs time (blowup around the strongest total power) (.ps)  (.pdf)

• This has the same radar plots as above, with the time range limited to  6 to 7 seconds.. during the strongest total power spike.
• You can see that this is coming from the punta borinquen sidelobe.
  

Plot showing the onset of compression in the if/lo or backend (.pdf)

• Page 1 : total power vs time for 13 seconds.
  
• top: the lbw total power.
• bottom:the total power in the 1390 - 1420 MHz band during this same time period.
• The negative going spikes show the system going into compression.
• The strongest compression is around 6.5 seconds (the punta borinquen signal).
• there is also some compression around 12 seconds (the faa radar pulse).
• Page 2: .15 second blowup around the strongest compression
  
• Top: lbw total power vs  time.
• The spikes are the punta borinquen rf pulse (every ipp).
• The red lines show where the system went into compression.
• Bottom: The 1390-1420MHz total power
• You see the radar pulses sending the system into compression.
• The red dashed line is close to the onset of compression.
• The radar pulse for this time had about -86.7dbm.
• We don't know what part of the system went into compression.
• the 1400 MHz band also has the radar close to it, so it could have caused this to happen in say the digitizers.
• I also check the total power at 1700 MHz. It also showed compression from the punta borinquen pulse.
• The 1700 MHz part of the band has gone through a separate IF filter than the pb radar.
• So the compression is probably not in the digitizers.

Summary

• the total power vs time  was computed for the lband wide (1114-1755 MHz).
• spectra were dumped at a 64 usec rate.
  
• the power varies by about 13 db in the 64 usec  averaged spectra.
• Sitting at az=84, za=18, the strongest signal came from the punta borinquen radar sidelobe.
• looking at the 1390-1420 MHz total power, the system went into compression around -86.7 dbm
• This was probably before the mock mixers/digitizers
• the compression was also seen at 1700 MHz. 
  
• The 1700 MHz band uses the 1750 IF filter.
  
• this filter should cut down the radar signal by at least a few db..
• There is no guarantee that the az,za used will have given the strongest radar signal.
  
• The maximum levels will be at least this strong.
  

home_~phil