TGBT18B_502_08 - Additional LBW tests

Goals

  • Test switching signal control in all LBW modes
  • Regression-style testing of all HBW pulsar modes and all spectral line modes (if time)

Details

  • Note: Ryan working remotely today
  • Joe began switching things over shortly after 09:00
  • First set of scan to test switching signals submitted at 09:45. Will test incoherent and coherent modes just to be safe.
    • i0100x0512 and i0200x1024 modes properly controlled cal, but i0100x1024 did not
    • There was abort when trying to go into the first coherent DD mode (c0100x0064)
  • Ray and Joe did some investigating. They think it was guppi_daq Will try the i0100x1024 mode again and then the coherent modes.
    • Coherent modes aborted after scan started. Bank D was "hung in aborting".
  • Joe suggested looking at 200 MHz coherent modes while they diagnose. Submitted SS test scans for c0200 modes only at 10:15
    • Aborted again after first few seconds of scan
    • Ryan noted that OBSBW and OBSNCHAN are too small by a factor of 800/requested_bw again (i.e. a factor of 8 or 4, depending on 100 or 200 MHz mode)
    • It appears that changes made last week have been lost somehow. Ray is recompiling.
  • Trying coherent modes again at 10:32. First scan in this set is #8
    • c0100x0064 looks OK now. Cal is on.
    • Coherent modes all look OK
  • Moving on to test all HBW modes. Will use dual backend where applicable. First scan is #17
    • Incoherent HBW modes all seem to check out.
    • Coherent modes also seem to be behaving as expected.
  • Testing a fix to config tool so that GUPPI does not inheret stale VEGAS parameters when it is used on its own.
    • Joe's fix seemed to work. It sounds like it does a ResetConfig any time the 'backend' keyword changes.
    • I will need to make sure that this does not cause issues for anything else before releasing.
  • Pausing at 12:00 for lunch and to allow prospero and LO/Rx control computers to be rebooted.
  • Starting back up at 12:55. Will do TP scans of all spectral line modes using 4 banks. Will configure for L-Band at 1420 MHz, and will inject tone at 1423 MHz. First scan is #49
    • Frequency labels were off in gbtidl, almost as if it were still configured for S-Band.
    • Joe is restarting IF manager
    • New set of scans starts at #56
    • Still showing same issue but only in modes 1 -- 3. All other modes have correct frequencies. Tone shows up in correct channel reqardless.
    • Aborted during Mode 23 scan due to "compare error in sig/ref". Resubmitting starting with Mode 22 (scan #79). No issue second time.
      • Joe checked logs. Apparently this has been happening "a lot" since October. I guess it's not an issue with the LBW modes specifically...
    • Remaining spectral line scans finished. Other than Modes 1 -- 3 things look OK
  • Did another GUPPI scan to check config. No issues.
  • Doing a VEGAS config, AutoPeakFocus, then VEGAS config and scan to check config sequence.
    • No motion, so obviously no actual corrections, but procedures seemed to run OK.
  • Randy provided a new n_chan value of 5 to try for the switching signals in the incoherent modes. Giving that a try.
    • Still not working for i0100x1024
  • Wrapping up for today. Joe and Ray started switching back to release branch at 14:30
  • Checkout went OK. Frequencies were correct in this Mode 1 scan.

Conclusions

  • Switching control looks good in coherent modes. It looks like we still need n_chan values from Randy for some incoherent modes, but at least some of the incoherent modes are controlling switching correctly.
  • HBW modes all seem to behave as usual (both coherent and incoherent)
  • Need to understand and fix frequency labeling in spectral line Modes 1 -- 3

-- RyanLynch - 2019-01-30
Topic revision: r3 - 2019-01-30, 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