TGBT15A_915_137 - VPM Scale and Filter Tests


  • Test smaller fftshift values with 950 MHz IF module filter
  • Find better vegas.scale value for c0800x0512 coherent search
    • Incoherent modes: 0x555557ff, 0xaaaaabff, 0x555555ff, 0xaaaaaabf, 0x5555555f
    • Coherent modes: 0x55555fff, 0xaaaaafff
  • Verify Ray's patch for frequency precision in shared memory
  • Test Luke's new BOF file i0800x2048 mode


  • Session begins at 13:30
  • Editing vegas.conf with new i0800x2048 bof file
  • Submitted IncoherentModeTestsCal SB at 13:40 running i0800x2048 to test new BOF file
    • Didn't pick up new BOF file. Restarting VEGAS
    • Resubmitting at 13:47
    • Scan is #1
    • Scan successful, no obvious issues with BOF file.
  • Will move on to testing fftshift values. Going to stick with this mode. First will try 0x555557ff
    • Submitted SB at 13:58
    • Scan is #1
      • Levels are too low. Maybe I went in the wrong direction with the fftshift value?
  • Will try 0x55555555
    • Submitted at 14:02, scan #3
      • Data are maxing out as expected. Will lower vegas.scale to 5000 (50% reduction)
    • Submitted at 14:04, scan #4
      • No obvious effect. Lowering by 50% again to 2500
    • Submitted at 14:08, scan #5
      • Still too high, but not maxing out now. No sign of overflow artifacts. Will decrease scale by 50% again to 1250
    • Submitted at 14:13, scan #6
      • Data seem mostly OK, but will decrease scale by 70% to 875 to see how that looks.
    • Submitted at 14:18, scan #7
      • Looks encouraging
  • This looks pretty reasonable, but I'm curious to see what it looks like with an fftshift of 0xaaaaaaaa
    • Changed vegas.conf and dropping vegas.scale to 10
    • Submitted SB at 14:30, scan #8
      • Not enough power. Will increase vegas.scale to 100
    • Submitted SB at 14:32, scan #9
      • Still too low. Increasing back to 875, scan #10
    • Submitted SB at 14:34
      • Levels were OK but I think we are seeing the overflow artifacts again. Going back to 0x55555555
    • Will try some other incoherent 800 MHz modes.
      • Scans #11--16 will be i0800x0128 -- i0800x4096, with vegas.scale values of 875 or 437.5
      • Submitted SB at 14:43
      • Levels look OK on scans #11, 13, 15, with no obvious overflow artifacts
      • Levels too low on scans #12 (i0800x0256), 14 (i0800x1024), 16 (i0800x4096)
      • Obvious overflow in i0800x1024
  • Need to look at these data more carefully, but want to move on to the scale setting for the c0800x0512 mode.
  • Submitting CoherentModeTestsSearch SB for a c0800x0512 scan with vegas.scale = 400. No change to fftshift at this time.
    • SB submitted at 14:58, scan #18
      • Levels were still a bit low. Increase scale factor to 800
    • SB submitted at 15:14, scan #18
      • Levels look OK, but will compare with GUPPI
      • Still a bit low compared to GUPPI. Will increase scale to 1015
    • SB submitted for scan #19
      • This looks good.
    • Taking cal scan with this vegas.scale setting to see if it is also OK
      • In agreement with GUPPI
  • Ended a little early instead of trying to squeeze in the coherent mode fftshift tests. Restored vegas.conf to original settings and asked Ray or Joe to commit it.
  • Session ended at 15:50


  • The i0800x2048 BOF file seemed to be OK
  • The lower incoherent mode fftshift values seem usable, but more work is needed to understand why some modes show lots of overflow artifacts while others seem to show none.
  • The search-mode coherent mode scale for c0800x0512 now looks good. I need to do this exercise for other modes but should be able to use past tests as a guide.
  • Didn't get to test the coherent mode fftshift values.

-- RyanLynch - 2018-02-20
Topic revision: r3 - 2018-03-16, 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