TGBT15A_915_101 - 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. Also test coherent 32-channel modes.

Details

Session begins at

• Session began at 14:15
• Switched to Justin's VPM branch
• Set iniitial fftshift to 0xffffffff and vegas.scale to 1000. Will use GUPPI when possible to verify data. Initial guppi.scale set to 0.125.
• Using CoherentFFTSHIFTTest SB
  • SB submitted at 14:35
  • Scan #1 is c0800x0032
    • Got a balance failed error but tried to keep going
    • Scan aborted.
    • Trying to resubmit
    • Same result
  • Will skip 32-channel modes and try other modes.
  • Resubmitted SB at 14:42
  • Scan #3 is c0800x0064, vegas.scale = 1000
    • Balance failed but scan executing
    • Saturated. Need to significantly reduce vegas.scale
  • Scan #4 is c0800x0128, vegas.scale = 500
    • Balance failed but scan executing
    • Not saturating but need to decrese vegas.scale anyway
  • Scan #5 is c0800x0256, vegas.scale = 1000
    • Balance failed but scan executing
    • Not saturating but need to decrese vegas.scale anyway
  • Scan #6 is c0800x0512, vegas.scale = 500
    • Balance failed but scan executing
    • Not saturating but need to decrese vegas.scale anyway
  • Scan #7 is c0800x1024, vegas.scale = 1000
    • Balance failed but scan executing
    • Not saturating but need to decrese vegas.scale anyway
  • Scan #8 is c0800x2048, vegas.scale = 100
    • Balance failed but scan executing
    • Levels look good
  • Scan #9 is c0800x4096, vegas.scale = 1000
    • Not saturating but need to decrese vegas.scale anyway
  • Scan #10 is c1500x0064, vegas.scale = 300
    • Not saturating but need to decrese vegas.scale anyway
  • Scan #11 is c1500x0128, vegas.scale = 300
    • Not saturating but need to decrese vegas.scale anyway
  • Scan #12 is c1500x0256, vegas.scale = 300
    • Not saturating but need to decrese vegas.scale anyway
  • Scan #13 is c1500x0512, vegas.scale = 150
    • Not saturating but need to decrese vegas.scale anyway
  • Scan #14 is c1500x1024, vegas.scale = 300
    • Not saturating but need to decrese vegas.scale anyway
  • Scan #15 is c1500x2048, vegas.scale = 800
    • No subints written on Bank D
  • Scan #16 is c1500x4096, vegas.scale = 1000
    • Scan aborted. Many messages about illegal values in several parameters.
  • Will try to use some lower fftshift values.
  • SB submitted at 15:22
  • Scan #17 is c0800x0064, vegas.scale = 100
    • Balance failed but scan executing
    • Still high. Maybe go down by a factor of 5--10
  • Scan #18 is c0800x0128, vegas.scale = 100
    • Balance failed but scan executing
    • Still high. Maybe go down by a factor of 2--3
  • Scan #19 is c0800x0256, vegas.scale = 100
    • Balance failed but scan executing
    • Pretty good. Could maybe lower by a factor of 2.
  • Scan 20 is c0800x0512, vegas.scale = 100
    • Balance failed but scan executing
    • Perfect.
  • Scan #21 is c0800x1024, vegas.scale = 100
    • Balance failed but scan executing
    • Good. Could even by 1.5 times higher but not essential.
  • Scan #22 is c0800x2048, vegas.scale = 200
    • Perfect.
  • Scan #23 is c0800x4096, vegas.scale = 500
    • Pretty good. Could even by 1.5 times higher.
  • Scan #24 is c1500x0064, vegas.scale = 100
    • Still too high by a factor of about 10.
  • Scan #25 is c1500x0128, vegas.scale = 100
    • Still too high by a factor of about 8.
  • Scan #26 is c1500x0256, vegas.scale = 100
    • Still high. Decrease by a factor of 2--3.
  • Scan #27 is c1500x0512, vegas.scale = 100
    • Pretty good but could decrease by a factor of about 1.5
  • Scan #28 is c1500x1024, vegas.scale = 100
    • Perfect.
  • Scan #29 is c1500x2048, vegas.scale = 100
    • Bad bandpass.
  • Scan #30 is c1500x4096, vegas.scale = 100
    • Aborted again.
• Switching to release branch at 16:48
  • Looks good
• CheckMode1 SB submitted at 16:52
  • Looks good
• CheckMode29 SB submitted at 16:53
• Session ended at 16:55

Conclusions

I was able to find some pretty good vegas.scale values but there are a few that need refined. 32-channel modes did not work. Balancing issues on many of the c0800 modes. c1500x4096 kept aborting.

-- RyanLynch - 2017-07-13