TGBT19A_502_02 - VEGAS LBW Pulsar Modes Release

Goals

  • Release new M&C version to support VEGAS LBW pulsar modes and some new HBW BOF files
  • Test BTL configuration, C-Band RRL set-up, and Argus as a sanity check

Details

  • Ray and Joe started the version switch over at 13:05
  • Switch complete at 13:23
  • Going through standard set of checkout scripts first.
    • Tried a dual GUPPI/VPM config but got an abort. Astrid reported BAD DEVICE: >>LO1:testTonePowerLevel<<
    • Worked on a second configuration
      • i0800x2048 dual backend scan looks fine
      • Going to run coherent mode again. Seems like auto-plotters weren't updating.
        • Doesn't look like it took data
        • Second attempt looks better.
    • VPM only scan ran fine
    • Tried running a spectral line scan with 4 banks but Astrid complained that it couldn't route 8 paths (that would be one for each polarization channel) to VEGAS. That is odd.
      • Single bank configuration seems to be working. Bandpass seems fine.
      • Joe restarted Turtle and now it seems to be OK. Joe thinks the cabling file needed to be regenerated.
      • But balance failed on first attempt. Trying again.
      • Balanced fine the second time and now taking data with 4 banks.
      • Taking a scan with 8 banks. Balance failed on several converter modules again, but worked on second time.
    • Checking GUPPI alone
      • Looks good
  • Calling Amber to help with a BTL set up
    • Running an OREO BTL script
    • Initial (fake) pointing and focus scans executed OK but BTL manager crashed. Ray is looking into it.
  • In the meantime, just running through configuration of C-Band RRL setup
    • VEGAS Banks A and C are complaining about a number of invalid parameters
    • Trying to cycle VEGAS off/on
      • That didn't help
    • Ray rolled back the BTL messages update.
      • Still getting faults in some parameters. Ray is going to restart the whole rack
      • The faults seem to have cleared but Bank A seems to be losing connection to the power samplers. Joe is restarting the manager and the accessor.
      • Ray powered off the entire rack plus the network switch. That seemed to allow us to configure for C-Band and run an L-band Mode 1 scan. * Going to do an OREO configuration again.
    • Amber is happy this time
  • Running some LBW pulsar modes
    • Seem to be working alright
  • Things seem to be OK now. Going to wrap this up.
  • Back to do some on-sky tests at 16:24. Will do a C-Band RRL, Argus, and KFPA scan, and finish up with a quick test of GUPPI raw mode for the UCLA obs tomorrow.
  • Starting off with a copy of 05_S206_C from AGBT17A_205. It will do a peak/focus and then an on/off on an HII region.
    • Scans are running but GFM says pointing scan is not handled by any plugin.
    • Scan numbers are also starting back at 1. Going to restart Astrid.
    • *Note that session has been changed to 3*
    • Trying again.
    • Ah, OK. The script I am using does not configure during the AutoPeakFocus, so the DCR was not in the scan coordinator and no data was being written. I modified the script and it looks better now.
    • Peak scans worked but then VEGAS complained about not being able route paths to more than one bank, even when I tried using my own scripts. After I did a ResetConfig it worked. Not sure why this would happen.
    • Recomb line detected. I am a little concerned about the VEGAS routing issue. I am going to run the pulsar LBW config that I ended with last time and then try the C-Band scan again.
      • Going from pulsar to multi-bank spectral line is definitely causing a problem. Doing a ResetConfig between avoids it, so it must be in config tool. I am pretty confident this sequence worked fine on the HBW pulsar branch. Joe is going to look at some config tool logs to see if he can figure out what is wrong.
      • We can definitely trigger this by going from coherent mode to a multi-bank spectral line script. Going from coherent mode to incoherent mode (both with one or two banks) to multi-bank spectral line mode works. A script that goes straight from coherent to multi-bank spectral line doesn't even validate.
      • We are going to roll back to the old, non-LBW branch.
      • Things work the way they should on the release branch. We are leaving it here.

Conclusions

  • We did not release. Need to figure out the config issues.
-- RyanLynch - 2019-04-25
Topic revision: r2 - 2019-04-25, 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