ADC blanking

23sep08

Links
17sep08: adc blanking with Faa, 1290, and  punta salinas radars. az=180,za=17 to  17.5

---

Intro.

    ADC blanking uses the digitized signal to determine whether to blank or not. If the digitizer exceeds the maximum value (ok maybe equals the maximum value)  then the current fft is marked as being blanked. All ffts marked as blanked will not be added to the accumulator.  Some properties of the adc blanking is:

• When blanking is done, all spectra,cross spectra for that integration are blanked as a unit.
  
• For each integrated spectra, the spectrometer returns the number of ffts accumulated so you can tell how many ffts have been blanked.
  
• The spectral samples are still spaced the the requested fft accumulation time they just have fewer integrations included. This is important for pulsar observations that need a constant sampling interval.
  
• As an example:  a 172 MHz bw, 1024  sample fft has a time duration of 5.9 seconds so you can include/exclude things at this time increment.
• See Ao fits file for pdev spectrometer: status Heap for a description of how the number of integrations are returned in the cima_fits fits files (psrfits to come..)
  

---

17sep08: adc blanking with Faa, 1290, and  punta salinas radars. az=180,za=17 to 17.5    

    On 17sep08 data was taken using the galfacts configuration. The setup and telescope motion was:

• az=180, za=17 to 17.5 in 20 seconds (about 1/2 slew speed).
  
• pdev cfr=1375 MHz. 172 MHz bands centered at 1300 and 1450 MHz
• Data taken at 2 millisecond sampling with 16 bits and 2048 channels in full stokes mode.
• The low winking cal was cycled at 20 Millseconds on and 20 milliseconds off (synchronized to the ffts).
• Data was taken  for 20 seconds with adc blanking off and adc blanking on. Both scans covered the same za range (although the radar phase was different for a given za).
  

     Dynamic spectra were made for each 20 seconds worth of data. The calOn+calOff cycle was summed to give a 40 millisecond sampling. PolA and polB have been averaged. The top plot is the 172 MHz centered at 1450 MHz while the bottom plot is the 172 MHz centered at 1300 MHz.
   Dynamic spectra 20 seconds with blanking Off (40 ms resolution) (.gif):

• at .65 and 12.65 seconds the FAA 1330 radar points at AO causing saturation.
• At 1.66 and 13.65 seconds the remy 1290 radar points at AO causing saturation.
  

   Dynamic spectra 20 seconds with ADC blanking on (40 ms resolution) (.gif):

• At 14.85 seconds the 1330 radar points at AO. There is no horizontal line of power and there is a dropout in thte 1330 power.
• At 16.1 seconds the 1290 radar points at AO.

    Dynamic spectra for 2,3 seconds where 1290,1330 radar are max.. 2 millisecond resolution (.gif)

• The top plot shows no blanking. The on/off pattern is the cal firing every 20 milliseconds.
• the bottom plot show blanking on.
  
• At 14.9 seconds you can see that the 1330 power is gone. All of the radar pulses during this peak were blanked.
  

Total power and spectra during the radar peaks (.ps) (.pdf):

•  The total power for 1 MHz about the 1330 and 1290 signals was computedand then plotted with 2 millisecond time resolution.
  
• About 60 spectra (2 millitsecond resolution) were plotted (with offsets) to show the spectra near the radar peaks.
• Page 1 blanking Off:
• The dashed green lines show where the spectra were taken
• There are lots of harmonics of the radar during saturation. They match what happens when you clip a sine wave (see clipping a sine wave).
  
• Page 2 blanking on:
• The dashed green lines show where the spectra were taken.
• The 1330 radar (red) drove the adc into saturation so blanking was enabled.
  
• The blanking  caused the 1330 MHz dropout near 14.85 seconds.
• The 1330 spectra have the 1270 image of 1330 near the parts that were not blanked. This is probably happening in the mixer.
  
• The 1290 radar (black) was not strong enough to cause adc blanking.
•  There is no drop in the 1290 power. Since the adc did not clip there are no harmonics from adc clipping.
• The 1310 image of the 1290 signal is present for the entire time the 1290 signal is strong. The bbm has probably gone into compression and not created i,q of the correct phase and amplitude.
  

Conclusions:

• When a radar signal  goes beyond the a/d limit you get a clipped sine wave that creates many harmonics in the band.
• The adc blanking can stop  the harmonics caused by the adc clipping to be included in the integrated spectrum.
  
• If the radar does not reach the adc limit, it will not be clipped. Since it hasn't reached the adc limit it will not create harmonics caused from adc saturation.
• The baseband mixer will go into compression before the adc clipping starts (see injecting sine wave into bbm). This occurs around an ADC count of 1500 (out of 2047). This can cause and image of the radar to appear.