TGBT15A_915_125 - Check Spectral Line Configuration and Balancing

Goals

  • During tests last week we noticed that Banks E-H were not properly balanced in one polarization. One of the spectral windows on Bank A also seemed to be cut off at low frequencies. Today we will see if we can recreate this issue and investigate the IF configuration to make sure there aren't any incorrect filters or LO frequencies being used.

Details

  • Session begins 13:25
  • Joe previously switched versions to regression test candidate
  • Will use Richard's runRegressXTPTest SB
  • Will use X-Band, but there was work on the turret drives, so it is not in focus. This shouldn't impact IF balancing.
  • SB submitted at 13:37
    • Scans #1/2 are On/Off pair, Mode 1, 8 IFs
    • None of the banks seem to be balanced properly.
    • Noticed that the Balance() command was commented out of the SB. Resubmitting with it uncommented.
  • SB submitted at 13:45
    • Scans #3/4 are On/Off pair, Mode 1, 8 IFs
    • This time the balance was successful. All channels are within about +/- 1 dB (or less) of nominal
    • Balance() command is still printing out a bunch of diagnostic info. This is left over from Ray's debugging.
    • Baselines in Pol1 are not great, but at least seem to mostly have zero mean. However, IFs 4-7 have a high frequency ripple across the band. Pol2 has about a 0.5 -- 1 K offset across nearly all IFs, except for IF 4 which has a low frequency ripple and a - 5 K slope across the band.
    • These definitely do not seem like good baselines, but with X-Band out of focus and work being done on the turret, it's not entirely clear why.
  • SB submitted at 13:59
    • IF config seems mostly correct. IF manager reports the right RF but it says the bandwidth at VEGAS is only 320 MHz. But GBTIDL has all the correct values. I've never really understood the IF manager values when using VEGAS...
  • Next set of scans will be On/Off pairs for modes 4, 10, 20, and 29.
  • SB submitted at 14:07
    • Scans #5/6 are On/Off pair, Mode 4, 8 IFs
      • Input IF and filter snaps all seem to have nominal power levels
      • Baselines seem mostly OK. There may be a < 0.5 K offset in some windows but it is not large.
    • Scans #7/8 are On/Off pair, Mode 10, 8 IFs
      • Input IF and filter snaps all seem to have nominal power levels
      • Baselines seem pretty good. Again, slight offset (< 0.5 K) in some IFs, but generally look good.
    • Scans #9/10 are On/Off pair, Mode 20, 64 IFs
      • Power levels mostly seem OK, although channel 1 on Bank D was a bit under-powered.
      • Baselines all consistently have a +0.5 K offset but are otherwise OK. IDL reported "Blanked spectra: ingored 1 integration" for IFs 16-23, 48-55, both pols.
      • No apparent issues with data from Bank D
    • Scans #11/12 are On/Off pair, Mode 29, 64 IFs
      • Power levels still seem mostly nominal. Bank D has slightly low power again, but a couple dB
      • Baselines all seem OK, though every IF is reporting the "Blanked spectra: ignored 1 integration" warning
  • Turret work is done, so will do one more round with X-Band in focus.
  • Will do On/Off pairs for modes 1, 4, 10, 20, 29.
    • Scans #13/14 are On/Off pair, Mode 1, 8 IFs
      • IF balancing seems good to within +/- 1 dB or less
      • Bank B, pol 1, and Bank E, pol 2, seem to have very slightly high power, around -18.5 dBFS instead of the nominal -20 dBFS. I don't think this is too much cause of concern.
      • Baselines seem similar to above - few tenths of a K offset, some baselines structure, including higher frequency ripples in IFs 4, 5, 6, pol 1. Pol 2, IF 4 has same lower frequency ripple, and other IFs in pol 2 have a noticeable slope across the band. It doesn't seem quite as severe as in the last scans in Mode 1 with X-Band out of focus.
    • Scans #13/14 are On/Off pair, Mode 1, 8 IFs
      • IF power levels once again seem close to nominal.
      • Baselines all look the same as before, i.e. mostly flat with a few tenths of K offset in some IFs.
    • Scans #15/16 are On/Off pair, Mode 4, 8 IFs
      • IF power seems OK
      • Baselines are same as before - most flat with a slight offset of a few tenths of a K in some IFs
    • Scans #17/18 are On/Off pair, Mode 10, 8 IFs
      • IF power seems good to within about 1 dB, as with other scans
      • Baselines look pretty good
    • Scans #19/20 are On/Off pair, Mode 20, 64 IFs
      • IF power seems goo to within about 1 dB. as with other scans. Bank C and E differ by about 1.5 dB between the two polarizations in scan #20. Again, not sure if this is normal.
      • Frequent approx. 0.5 K offset in baselines by most IFs, both pols. Many GBTIDL messages about blanked integrations.
    • Scans #21/22 are On/Off pair, Mode 20, 64 IFs
      • IF power levels seem OK.
      • Consistent approx 0.5 K offset in baseilnes for most IFs, both pols. Many GBTIDL messages about blanked integrations.
  • So far I have not been able to recreate the issues from last week. I will try one more thing: Joe will switch from Matrix spectral line HPC to guppi_daq pulsar HPC. I will take one scan in the pulsar modes, and then have Joe switch back to the Matrix HPC, to see if something about having previously configured for pulsar modes triggers the bad balancing and bandpasses in the spectral line modes.
  • Joe switches from Matrix spectral line to guppi_daq pulsar at 15:22, finishes at 15:23
  • Will use IncoherentModeTests SB for pulsar scan
  • SB submitted at 15:25
    • Scan #23 is i0800x0512, L-Band (not in focus), VEGAS/GUPPI
    • Had some issues with vegas.subband being stuck at 8 from previous spectral line scan. Needed to specify vegas.subband = 1.
    • Balancing looks good.
    • Scan ran without issues
    • Data look OK
  • Called Joe to switch back to Matrix spectral line at 15:31, finished at 15:34
  • Telescope seems to be in motion starting around 15:33, I think for a film crew. Hopefully this doesn't cause any issues...
  • Will try another set of On/Off scans at X-Band for modes 1, 4, 10, 20, 29
  • SB submitted at 15:36
    • Scans #24/25 are On/Off pair, Mode 1, 8 IFs
      • IF balancing once again seems mostly OK. The only issue is slightly low power (-21.5 dBFS) on Bank C, channel 1
      • 2 K offset in baselines, but they actually look better in terms of ripples and slope.
    • Scans #26/27 are On/Off pair, Mode 4, 8 IFs
      • Realized that the power monitor was not updating on Bank C. So the note above about low power may be spurious. Restarted the power monitor and things seem OK on this scan.
      • There is a bout a 0.5 -- 1 dB offset between the two polarization channels on Banks E-H, as well as Bank A. Pol 1 is a bit high on Banks A, E, and G, while on Banks F and H it is Pol 2. Still, it's not that big of an offset.
      • Baselines are awful, but could be do to telescope motion...
    • Scans #28/29 are On/Off pair, Mode 10, 8 IFs
      • Levels are a little more jittery on scan #28, but don't deviate from the nominal -20 dBFS by more than 1--1.5 dB. No consistent offset between polarizations on certain banks.
      • Baselines look better, similar to other scans in this mode. Telescope motion may have stopped.
    • Didn't have time for Modes 20 and 29
  • Called Joe to switch back to 16.3 at 15:

Conclusions

  • I was unable to reproduce the same balancing issues seen last week. The banpasses for Mode 1 showed a lot of structure and were definitely not great, but I didn't see the type of roll-off at the band edges that we saw previously. Power levels would sometimes fluctuate by up to +/- 1 dB from the nominal value on same banks, but I never saw any deviations larger than this, and nothing consistent among Banks E--H like we saw last week. Both input IF power and output power after internal VEGAS balancing seemed generally OK. Bandpasses in Mode 4, 10, 20, and 29 sometimes had a few tenths of a K offset but otherwise seemed OK.
  • I'm not sure if some of the small power fluctuations and baseline offsets I see here are normal, but they don't jump out to me as glaringly bad. Since I was unable to recreate the issues we saw last week, it's tough to say what caused them.
  • The only thing I didn't try was configuring for a coherent pulsar mode and then switching back to spectral line. I guess we could try that at a later date.
-- RyanLynch - 2017-09-28
Topic revision: r1 - 2017-10-03, 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