TGBT15A_915_87 - Investigate Coherent Pulsar Mode Artifacts.

Goals

Ryan discovered artifacts in coherent VEGAS pulsar mode data taken as part of astrophysical acceptance testing. Specifically, there is extra noise in certain frequency channels, which seems to repeat with two spurs offset by 20 MHz, with the spurs repeating every 50 MHz. This does not appear to be affecting pulsar fluxes but it is obvious in observations of the noise diode. We will inject pure tones at a variety of frequencies to see if we can excite these artifacts and potentially map any relation with injected frequency.

Details

Session begins at 13:06
  • Switched VEGAS to pulsar branch
  • Performing initial scan using full IF and L-Band to make sure artifacts are present today. GUPPI is also used.
    • Using 00-CoherentModeArtifactTesting scheduling block
    • SB submitted at 13:17
    • Scan is #1
    • SB ended at 13:19
    • Artifacts are present
  • Performing second scan using full IF and L-Band but using 1500 MHz mode. GUPPI does not support this mode.
    • Ryan wants to see if the presence of the artifacts is BW/ADC sampling frequency/BOF file dependent
    • Using edited 00-CoherentModeArtifactTesting scheduling block
    • SB submitted at 13:27
    • Scan is #2
    • SB ended at 13:29
    • Difficult to tell if artifacts are present. Diode is not clearly visible. Levels too low?
  • Performing third scan using full IF and S-Band with 800 MHz mode. GUPPI is also used.
    • Using edited 00-CoherentModeArtifactTesting scheduling block
    • SB submitted at 13:30
    • Scan is #3
    • SB ended at 13:32
    • Artifacts are not immediately evident.
  • Will now disconnect Bank A from IF and try tone injection. Ryan hypothesizes that strong RFI could be exciting ADC spurs and suggests injecting tones near the frequency of the brightest RFI in the L-Band data.
  • Performing scan using test tone injected at 400 MHz baseband. This should correspond to 1600 MHz RF.
    • Using 00-CoherentModeArtifactTesting scheduling block
    • SB submitted at 13:40
    • Scan is #4
    • Scan aborted on Bank B "VEGAS HPC program taking too long to be ready"
    • Power cycling managers.
    • Resubmitted SB at 13:44
    • Scan aborted with multiple "failed to start VEGAS HPC subprocess"
    • Ray recommends doing a manager stop/start through task master on vegas-hpc3
    • Resubmitted SB at 13:48
    • Scan still aborted. Attempting systemstop/start.
    • Resubmitted SB at 13:52
    • Scan is #7
    • SB ended at 13:54
    • Tone is not readily visible. Some artifacts may be present, though.
  • Moved tone to 500 MHz baseband. This should correspond to 1400 MHz RF
    • Using 00-CoherentModeArtifactTesting scheduling block
    • SB submitted at 13:59
    • Scan is #8
    • Scan ended at 14:01
    • Tone is now visible at expected frequency. Low level artifacts are still present.
  • Moving tone to 380 MHz baseband. This should correspond to 1620 MHz RF. There is strong RFI at L-Band present near this frequency.
    • Using 00-CoherentModeArtifactTesting scheduling block
    • SB submitted at 14:07
    • Scan is #9
    • SB ended at 14:09
    • Tone is visible but the RF was wrong. It was 1520 MHz. Should have given Jason a baseband frequency of 280 MHz. Artifacts are present, though.
  • Moving tone to 280 MHz baseband.
    • Using 00-CoherentModeArtifactTesting scheduling block
    • SB submitted at 14:15
    • Scan is #10
    • SB ended at 14:17
    • No data on Bank D. Redoing scan.
    • SB submitted at 14:18
    • Scan is #11
    • SB ended at 14:20
    • Tone is visible. There are still artifacts visible.
  • Moved tone to 200 MHz baseband
    • Using 00-CoherentModeArtifactTesting scheduling block
    • SB submitted at 14:25
    • Scan is #13 (not sure what happened to scan 12)
    • SB ended at 14:28
    • Tone is visible. There are still artifacts.
  • Taking a pause so Jonah can do a PF checkout.
  • Jonah finished at 14:37. Jason is going to hook up the wideband noise source. We should see pure Gaussian noise. Will see if there are any low-level artifacts still present.
  • Taking scan with wideband noise source. VEGAS power is a little low at -24 dB but we are going to go ahead anyway.
    • Using 00-CoherentModeArtifactTesting scheduling block
    • SB submitted at 14:40
    • Scan is #15 (maybe Johah's checkout was scan #14)
    • SB ended at 14:44
    • Difficult to tell from scale but it looks like there are some artifacts in the bandpass.
  • Jason is going to combine the noise source and tone generator to inject a tone on top of the noise source. VEGAS power is at -22 dB
    • Using 00-CoherentModeArtifactTesting scheduling block.
    • SB submitted at 14:53
    • Scan is #16
    • Scan ended at 14:55
    • Tone is present. There is now a ripple in the bandpass and a stronger artifact at 600 MHz baseband.
  • Set up is the same but tone has been turned off.
    • Using 00-CoherentModeArtifactTesting scheduling block
    • SB submitted at 15:01
    • Scan is #17
    • SB ended at 15:03
    • Ripple is still present but no strong artifact at 600 MHz. Ripple may be due to extra cabling.
  • Jason is hooking VEGAS back up to IF.
  • Trying a scan in coherent dedispersion mode at L-Band but with fctr=1400 MHz instead of 1500 MHz. This is being done to determine if the frequency of the RFI at baseband has any impact on the amplitude of the artifacts.
    • Using 00-CoherentModeArtifactTesting scheduling block.
    • SB submitted at 15:10
    • Scan is #18
    • SB ended at 15:12
    • Artifacts are still present though the amplitude appears different.
  • Trying a scan in incoherent dedispersion mode at L-Band. Back to fctr=1500 MHz. All other parameters are the same.
    • Using 00-CoherentModeArtifactTesting scheduling block.
    • SB submitted at 15:15
    • Scan is #19
    • SB ended at 15:17
    • Artifacts are present.
  • Taking an incoherent scan with tint=40.96e-6
    • Using 00-CoherentModeArtifactTesting scheduling block.
    • SB submitted at 15:18
    • Scan is #20
    • SB ended at 15:20
    • Artifacts are present.
  • Oops. Just realized that vegas.scale needs to be different for incoherent modes. Trying again. Same setup as scan #20 but with vegas.scale=100
    • Using 00-CoherentModeArtifactTesting scheduling block.
    • SB submitted at 15:23
    • Scan is #21
    • SB ended at 15:25
    • Artifacts still seem to be there.
  • Taking an incoherent scan with tint=10.24e-6 but with 256 phase bins.
    • Using 00-CoherentModeArtifactTesting scheduling block.
    • SB submitted at 15:26
    • Scan is #22
    • SB ended at 20:27
    • Artifacts are still present.
  • Taking an incoherent search mode scan with tint=40.96e-6
    • Using 00-CoherentModeArtifactTesting scheduling block.
    • SB submitted at 15:31
    • Scan is #23
    • SB ended at 15:33
  • Taking an incoherent cal mode scan. Shifting the center frequency up to 1550 MHz to see if RFI near 1100 MHz might be the cause.
    • Using 00-CoherentModeArtifactTesting scheduling block.
    • SB submitted at 15:34
    • Scan is #24
    • SB ended at 15:36
    • Artifacts still present. And I'm not sure if the spike near 1100 MHz is actually RFI at sky frequency. It seems to just be at 1200 MHz now. But maybe it is something different.
  • Taking incoherent cal scan at S-Band, fctr=2000 MHz.
    • Using 00-CoherentModeArtifactTesting scheduling block.
    • SB submitted at 15:43
    • Scan is #25
    • SB ended at 15:45
    • Well, now it looks like there might be something present at certain pulse phases but all frequencies. GUPPI scales were too low to compare.
  • Redoing scan #25 but with guppi.scale=5
    • Using 00-CoherentModeArtifactTesting scheduling block.
    • SB submitted at 15:49
    • Scan is #26
    • Stopping scan. GUPPI levels are too high.
    • Re-running with guppi.scale=2
    • Scan is #27
    • GUPPI levels look better
    • SB ended at 15:53
    • Looks like the phase-constant artifacts are there at a lower level in VEGAS. They are also there in GUPPI. So at least that is nothing specific to VEGAS, I guess.
  • That's all for today. Switching back to release branch.
  • Taking a spectral line Mode 1 scan to check things out.
    • Using CheckMode1 scheduling block
    • SB submitted at 16:00
    • Scan is #28
    • Scan aborted. I forgot to take the Antenna out of the scan coordinator after the software switch. Trying again.
    • SB submitted at 16:01
    • Scan is #29
    • All looks fine.
  • Session ended at 16:02

Conclusions

The artifacts (maybe we should just start calling them spurs) are present. Spurs are even present at a low level when the tone generator or noise source are hooked up. Whether these spurs show up at the same frequency as the artifacts we see in real L-Band data is TBD. At first glance, changing the frequency of the tone did not have a huge impact on the bins that the spurs appeared in. We did see artifacts at L-Band in incoherent data this time. I am at a loss to explain why they didn't not show up during acceptance testing back in December. Right now it really seems like we are seeing ADC artifacts that are dependent on something at L-Band (one presumes the RFI environment). I really hope this is not the case because I don't know what we are going to do about it. Ryan is very frustrated and depressed.

-- RyanLynch - 2017-02-08
Topic revision: r2 - 2017-02-08, 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