TGBT15A_915_100 - Detailed fftshift and scale tests for incoherent modes

Goals

  • Find best combination of fftshift and vegas.scale to eliminate/minimize FFT overflow artifacts and get good output levels from VEGAS.

Details

Session begins at
  • Session began at 13:00
  • Switched to Justin's VPM branch
  • Set iniitial fftshift to 0x5555555f and vegas.scale to 1000. Will use GUPPI when possible to verify data. Initial guppi.scale set to 1.5.
  • Using IncoherentFFTSHIFTTest SB
    • SB submitted at 13:15
    • Scans are aborting for unknown reasons. No helpful error messages.
    • Eventually got help from Joe. Had to manually kill guppi_daq
    • SB resubmitted at 13:42.
    • Scan #2 is i0800x0064
      • Levels too low. Increasing vegas.scale
    • Scan #3 is i0800x0064 with vegas.scale = 5000
      • Levels are still low.
    • Scan #4 is i0800x0064 with vegas.scale = 10000
      • Levels look OK here.
    • Scan #5 is i0800x0128
      • Levels are OK but high by a factor of about two. Will vegas.scale to 5000 on next round.
    • Scan #6 is i0800x0256
      • Levels look OK
    • Scan #7 is i0800x0512
      • Levels are OK but high by a factor of about two. Will change vegas.scale to 5000 on next round.
    • Scan #8 is i0800x1024
      • Levels are OK
    • Scan #9 is i0800x2048
      • Levels are way too high. Will decrease vegas.scale to 1000 on next round.
    • Scan #10 is i0800x4096
      • Levels are OK but could be higher by a factor of two. But I don't think we have the room in vegas.scale to do so.
    • Scan #11 is i1500x0064
      • Levels are too low by about a factor of 2.5. Will change vegas.scale to 3000 on next round.
    • Scan #12 is i1500x0128
      • Levels are too low by about a factor of three. Will change vegas.scale to 3000 on next round.
    • Scan #13 is i1500x0256
      • Levels are too low by about a factor of 3.5. Will change vegas.scale to 4000 on next round.
    • Scan #14 is i1500x0512
      • Levels are too low by about a factor of two. Will change vegas.scale to 2000 on next round.
    • Scan #15 is i1500x1024
      • Levels are too low by about a factor of three. Will change vegas.scale to 3000 on next round.
    • Scan #16 is i1500x2048
      • Levels are too low by about a factor of 10. Will change vegas.scale to 10000 on next round.
    • Scan #17 is i1500x4096
      • Levels are too low by about a factor of 10, maybe more. Will change vegas.scale to 10000 on next round.
    • Running again with new levels. Will do a better job of checking GUPPI, too.
    • SB submitted at 14:16
    • Scan #18 is i0800x0064
      • Levels are sort of low. GUPPI levels are also a bit low.
    • Scan #19 is i0800x0128
      • Missed this scan.
    • Scan #20 is i0800x0256
      • Levels look OK on both. Maybe a little low on VEGAS. Things are acceptable in psrplot.
    • Scan #21 is i0800x0512
      • Both look OK.
    • Scan #22 is i0800x1024
      • VEGAS OK, GUPPI is too high. Changing guppi.scale to 1.0
    • Scan #23 is i0800x2048
      • VEGAS is OK. GUPPI is high. Will change guppi.scale to 1.0
    • Scan #24 is i0800x4096
      • Levels are OK.
    • Scan #25 is i1500x0064
      • Levels are OK. Changing vegas.scale to 2000
    • Scan #26 is i1500x0128
      • Levels are high by a factor of two.
    • Scan #27 is i1500x0256
      • Levels are OK but could be a bit lower. Will change vegas.scale to 3000.
    • Scan #28 is i1500x0512
      • Levels are OK but could be a bit lower. Will change vegas.scale to 00.
    • Scan #29 is i1500x1024
      • Levels are spot on.
    • Scan #30 is i1500x2048
      • Levels are a bit high. Decreasing vegas.scale to 2500
    • Scan #31 is i1500x4096
      • Levels are spot on.
    • Last run was mostly OK. Will do one more with new levels.
    • Scan #32 is i0800x0064
      • Levels are OK
    • Scan #33 is i0800x0128
      • Both GUPPI and VEGAS are a little low, but OK.
    • Scan #34 is i0800x0256
      • Acceptable on both, but a little low.
    • Scan #35 is i0800x0512
      • Still a little lower than I would like, but OK.
    • Scan #36 is i0800x1024
      • GUPPI still a bit high.
    • Scan #37 is i0800x2048
      • GUPPI still a bit high.
    • Scan #38 is i0800x4096
      • Levels are OK.
    • Scan #39 is i1500x0064
      • Levels are OK.
    • Scan #40 is i1500x0128
      • Levels are OK
    • Scan #41 is i1500x0256
      • Levels are OK.
    • Scan #42 is i1500x0512
      • Levels are OK.
    • Scan #43 is i1500x1024
      • Levels are spot on.
    • Scan #44 is i1500x2048
      • Levels are a still a bit high. Decreasing vegas.scale to 2000
    • Scan #45 is i1500x4096
      • Levels are spot on.
    • These scans looked OK. Will try the i0800x0032 and i1500x0032 modes.
    • Scan #46 is i0800x0032.
      • VEGAS levels too low. GUPPI levels OK. But balance also failed on VEGAS. Will try vegas.scale at 15000 just for testing purposes.
    • Scan #47 is i1500x0032
      • VEGAS levels are a bit low, but could be acceptable. Will try vegas.scale at 6000.
    • Scan #48 is i0800x0032.
      • Balance failed again and I accidentally hit abort.
    • Scan #48 is i0800x0032
      • VEGAS levels are still low. But the data seem maybe OK?
      • Balance failed again and I accidentally hit abort.
    • Scan #49 is i1500x0032
      • VEGAS levels seem OK. Bandpass seems to be shifted (i.e. peak is not occurring in expected channels).
  • All done for today. Reverting back to release branch at 15:21
  • Submitted CheckMode1 SB at 15:26.
    • Balance failed on first attempt. Second attempt was fine.
  • Submitted CheckMode29 SB at 15:29, just to be safe.
    • Had receiver set to NoiseSource. Oops. Ran again with Receiver set to Rcvr2_3
    • Looks good.
* Wrapping up at 15:31

Conclusions

I think this choice of fftshift is too low. We are uncomfortably close to maxing out vegas.scale. Having said that, the levels were in an acceptable range. The i0800x0032 mode looks OK. The i1500x0032 mode seems to have a shifted bandpass, though. -- RyanLynch - 2017-07-06
Topic revision: r2 - 2017-07-06, 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