TGBT18B_502_01 - LBW tests round one

Goals

• First round of testing of modifications to config tool and VEGAS M&C to support low bandwidth (100/200 MHz) modes.
  • Will test incoherent and coherent dedispersion modes
  • Will set up for cal and search mode scans
  • Where supported, will test GUPPI and VEGAS in dual backend mode

Details

• Session begins at 09:30
• Joe took care of software version switch over
• Ryan edited vegas.conf with a section for c0200x0064
• Ryan edited scripts with legal values of tint/nchan (need to check that the config tool limits are appropriate)
• Had to sort out some script typos. Nothing major.
• NOTE: first few scans have a source name from last project. First scan number will be 14
• Trying i0100x0512 with full Stokes parameters
  • The VEGAS mode isn't recognized. Script aborted.
  • Joe had to restart Grail
  • On second attempt GUPPI did not configure properly, and LO2 seemed to be tuned for VEGAS LBW group 2 instead of 3.
    • Joe edited vegas.conf to use group 2
    • VEGAS is also warning that Valon frequency is not equal to that set, although data seem tuned properly...
  • On third attempt GUPPI seemed to configure and data is flowing!
    • VEGAS levels look low (scale set to 100) but configuration seems OK. GUPPI data look OK at first glance.
• Edited vegas.scale to 10000
  • Levels too high now, but data successfully recorded for cal and search mode, as well as with total intensity only
• Noticed that VEGAS isn't balancing. Joe suggests changing the order in the Backend keyword. This has no effect.
• Adding a manual balance and trying again.
• Manual balance worked. Levels are a bit high with vegas.scale = 2000 in inco modes, but going to forge ahead to test other modes.
• Search-mode data folded OK and cal is there. Bandpass has a strange dip in the middle, though, and there seem to be some periodic low-level artifacts to investigate.
• guppi.scale is too high in i0100x1024
• Cal not seen in i0100x1024.
• Levels too high in GUPPI for i0100x4096. Data are zeros in VEGAS.
• i0200x2048 scan aborted. Bank A had a fault in the scale_inco parameter.
  • It is properly set to 2000, though I notice that scale_[iquv] are not set. Not sure why the manager is complaining.
• Signal only appearing in upper 50 MHz of band for i0200x4096. Input IF is tuned to 300 MHz but if we are in group 2 it should be 500 MHz.
  • Same thing in i0200x8192
  • Also not seeing cal in the part of the band where there is a bit of signal
• Trying c0100x0064
  • VEGAS is not configured properly. It seems to be set up for a c0800x0064 with only one bank (A) active, instead of c0800x0512 (the c0100x0064 equivalent).
  • Cal is not showing up in GUPPI. GUPPI does seem to be configured correctly.
  • Cal is there in VEGAS although the square wave is noisy and there appear to be regular artifacts in the cal phase space
  • Frequency labels in VEGAS are wrong. Clearly it doesn't recognize what set-up it is in.
  • Bank A is active and should be getting an input IF of 250 MHz. I think this means that it should be in center of the observed band...
• Same VEGAS config issue in c0100x0128. Cal is there in GUPPI this time, but only in one small part of the band...maybe it's a scaling issue?
• Same VEGAS issues in c0100x0256. GUPPI no longer shows any signs of cal signal.
• Same hint of cal signal in GUPPI in c0100x0512. VEGAS issues still the same.
• Probably issues with vegas.scale in coherent modes, but tough to tell given set-up issues.
• Same issues in 200 MHz BW modes (both VEGAS and GUPPI)
  • Cal signal does look stronger in VEGAS
• Joe began switching back to release branch around 13:10
• Did check outs and ended session at 14:00

Conclusions

• 100 MHz inco modes seemed to configure well. Other modes had some issues.
• Outstanding issues/items:
  • Check config tool limits on tint/nchan combinations
  • Make sure GUPPI configures properly every time
  • Fix Balance() behavior to automatically include VEGAS in dual backend mode
  • Look into VEGAS bandpass shape and possible artifacts
  • Clear up Valon warning in VEGAS
  • Need to do a test-tone injection to verify frequency tuning (though it seems to be OK by-eye)
  • Zeros in VEGAS bandpass in i0100x4096
  • Cal not seen in VEGAS for i0100x1024
  • scale_inco parameter seems to be illegal in i0200x2048
  • Fix coherent set up in VEGAS
  • Investigate artifacts in cal signal in coherent modes
• Will investigate further and test again next maintenance day