TGBT15A_915_90 - Test fftshift values i0800x0512 and obtain ADC snap blocks

Goals

  • Test different values of fftshift for the i0800x0512 BOF file and obtain some ADC snap blocks using the Archivist to aid in diagnosis of the pulsar mode U/V wrapping/saturation issues

Details

Session begins at approximately 09:00
  • Switched VEGAS to pulsar branch
  • Will use the 03-ADCSnapBlock scheduling block. This sets up the Archivist to record adcpwr1 and adcpwr2 snap blocks from the active bank, in this case Bank A. All scans with i0800x0512, looking at noise diode.
  • Changed fftshift to 0xffffffff and vegas.scale = 100
    • Scan is #1
    • Bandpass is all zeros. Need to increase vegas.scale
  • fftshift=0xffffffff and vegas.scale = 300
    • Scan is #2
    • Bandpass is all zeros. Need to increase vegas.scale
  • fftshift=0xffffffff and vegas.scale = 1000
    • Scan is #3
    • Bandpass is all zeros. Need to increase vegas.scale
  • fftshift=0xffffffff and vegas.scale = 15000
    • Scan is #4
    • Bandpass starting to become visible but there is quantization/suppression and power is low. Need to increase vegas.scale further.
  • fftshift=0xffffffff and vegas.scale = 65000
    • Scan is #5
    • Bandpass shape looks better but levels are still low, and there is still some step-like features that may be quantization issues. We are near the maximum value of vegas.scale anyway. Need to change fftshift.
  • Changed fftshift to 0xaaaaafff and vegas.scale = 500
    • Scan is #6
    • Bandpass is all zeros. Need to increase vegas.scale * fftshift=0xaaaaafff and vegas.scale = 15000
    • Scan is #6
    • There is still quantization/suppression and power is low. Will go through different values of fftshift.
  • Changed fftshift to 0x555557ff and vegas.scale = 15000
    • Scan is #8
    • There is still quantization/suppression and power is low
  • Changed fftshift to 0xaaaaabff and vegas.scale = 15000
    • Scan is #9
    • There is still quantization/suppression and power is low
  • Changed fftshift to 0x555555ff and vegas.scale = 15000
    • Scan is #10
    • Bandpass starting to look OK but levels are still low. Will try to increase vegas.scale.
  • fftshift=0x555555ff and vegas.scale = 50000
    • Scan is #11
    • Bandpass starting to look OK and levels are OK, but we are still uncomfortably close to the max value of vegas.scale. Will keep trying different fftshift values.
  • Changed fftshift to 0xaaaaaaff and vegas.scale = 15000
    • Scan is #12
    • Bandpass looks OK but levels are still low.
  • Changed fftshift to 0x5555557f and vegas.scale = 15000
    • Scan is #13
    • Bandpass looks OK and levels are good but it would be nice if we could use a smaller vegas.scale. Will keep going.
  • Changed fftshift to 0xaaaaaabf and vegas.scale = 15000
    • Scan is #14
    • Bandpass looks OK and levels are good, but still using a high value of vegas.scale.
  • Changed fftshift to 0x5555555f and vegas.scale = 15000
    • Scan is #15
    • Bandpass is saturated. Will decrease vegas.scale.
  • fftshift=0x5555555f and vegas.scale = 1000
    • Scan is #16
    • Bandpass and levels are OK. Maybe starting to see some overflow artifacts, but not entirely clear.
  • Changed fftshift to0xaaaaaaaf and vegas.scale = 1000
    • Scan is #17
    • Must have been a hiccup. No data.
  • fftshit=0xaaaaaaaf and vegas.scale = 1000
    • Scan is #18
    • Bandpass and levels are OK. Overflow artifacts don't seem to be much of an issue.
  • fftshit=0x55555557 and vegas.scale = 1000
    • Scan is #18
    • Bandpass and levels are OK. Overflow artifacts don't seem to be much of an issue.
  • Time is up. Switching back to release branch.
  • Checking out spectral line modes with CheckMode1 scheduling block.
    • Scan is #20
    • Looks OK
Session ended at approximately 13:00

Conclusions

All f's will not work for incoherent modes. I was able to find an fftshift/vegas.scale combination that seemed to produce acceptable results for i0800x0512 but I will need to test other modes. Need to watch to make sure overflow artifacts don't become an issue again as we go to less aggressive values of fftshift. ADC snapblocks seem to have been recorded successfully. I'll need to analyze those offline.

-- RyanLynch - 2017-02-28
Topic revision: r1 - 2017-03-07, RyanLynch
 

This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding NRAO Public Wiki? Send feedback