• lband wide receiver
  
• interim correlator with 4 sub correlators:
  
• sbc 1,2 : 25 Mhz polA,polB each with 2048 channels (12.2Khz channel resolution)
  
• sbc  3,4: 6 Mhz polA,B coverage center of sbc  1,2
• 1 second integrations
• 180 second on , off  position switching followed by a 10 sec cal on,off.

• find all a2811 files for Nov13: 271
• 271 files
  
• each file had 4 scans:
  
• 180 secs on source
• 180 secs off source
• 10 secs cal on
• 10 secs cal off
• Only  keep the  180 secs on, or off sources (542 scans)
• find all scans that include 1381.05 Mhz
  
• 218 180 secs scans
• for each scan
• extract the channel closest to 1381.05 Mhz (25 Khz after hanning smoothing)
  
• normalize each 180 numbers for a scan with the median value.

• The 218 scans were split into 4 sets (to see them a bit better).
• The x axis holds the 1 second data samples of the 1381.05 Mhz channel
• The y axis is in units of Tsys (since i divided by the median value).
  
• each scan of 180 seconds is plotted in a different color
• I defined GPS L3 as anything 2% away from Tsys.
  
• The dashed horizontal lines show this clipping level.
• This algorithm overestimates the amount of gps L3 rfi if the rfi is on for > 50% and < 100% of the 180 seconds (the median will use the gpsl3 value if it is on > 50% of the time.
• There were 32240 1 second samples.
• There were 1720 secs with gpsL3 rfi
• GPS L3 was on : 1720/32240. = 4.38% of the time

• ALFA receiver
  
• mock spectrometer covering 1250 to 1520 Mhz (spectral resolution of 21Khz).
• The frequency bands are fixed (they don't change scan to scan like a2811).
  
• 1 second samples

• find all of the A2754 ALFA beam 0 files (590)
• each file had 197 1 second spectra
• that last 17 secs of each  file had 8 secs cal on, 8 secs cal off.
• for each rec (second) of each file compute:
• the average power [1380.29 to 1382.559] Mhz
• Compute a baseline using
• .5 Mhz average center at +/- 2 Mhz from 1381.
• Divide the average 1381 Power the the average of the two baselines
• This does a better job a estimating the gps l3 rfi since it doesn't rely on the median value over the entire scan.
• This wasn't used with the a2811 data since the 1381 Mhz  band occasionally fell at the very edge of their 25 Mhz band.

• Top: 1381 power vs date.
  
• The + show the  1 second samples
• The y axis is Tsys (after removing 1).
• Middle,Bottom frames:  1381 Power vs 1 second samples
• I've plotted each days data with a separate color
• the dashed horizontal threshold (1.5% of tsys) was used define gps l3 rfi.
• There were 115594 1 second samples
• There were 3936 samples with gps l3 rfi
• So GPS L3 was on for 3.41 % of the time.

SUMMARY

• The a2754 measurement is probably more accurate because is didn't use a median bandpass for each scan
  

page up
home ~phil