Pulsar Observing Advice for the NRAO Skynet 20-meter telescope

Links UNC Skynet 20m skynet interface Log of 20m results. Latest observation NRAO Skynet main

Details for various kinds of observing projects
Main Advice Mapping/Imaging Advice Spectrum Observing Advice Pulsar observing

Pulsar Observing

After the observation, the system will attempt to run Scott's "prepfold" on the data, if it can find a "par" file, or if it can find the period and DM in a pulsar catalog. The resulting plots from prepfold will be put on the web in the "Skynet quick look" web area: http://www.gb.nrao.edu/20m/peak/latest/

If it cannot find information about the pulsar it will just list the name of the "psrfits" file, which you can copy and process offline.

Processing off-line (advice for pulsar experts)

The pulsar fits files will be on "pork" at /raid/scratch/cyborg/SkynetData/
(or if running GBO mode: /raid/scratch/cyborg/GBOdata/ )
Any Green Bank login account should be able to access these files.
Just do: "ssh yourlogin@pork"
The quick-look web page "http://www.gb.nrao.edu/20m/peak/latest/" will show information similar to the following:

         Pulsar: B0329+54
     psrfits file available on
The path designation following "pork:" is the path where the FITS file is located.
The last line (starting with "Skynet") is the name of the FITS file.

Processing with prepfold:

  source /home/pulsar64/pulsar.bash
  prepfold -help
  prepfold -n 128 -nsub 128 -timing parfile fitsfilename
       parfile is /users/sransom/parfiles/psrname.par
       or /users/pdemores/tzpar/psrname.par 

If you know the period and DM:
  prepfold -n 128 -nsub 128 -nodmsearch <dmvalue> -p <period> fitsfilename

  (the period is in seconds)

Sample time weirdnesses

The rate at which the pulsar data is sampled depends on the "integration time" that you enter in the Skynet web interface, but is not exactly what you enter.
  • The sample time (i.e., time from one sample to the next) is 2^n (1024/5.0e8) seconds.
  • The maximum allowed by the present software setup is n=11, i.e., tsample = 4.19ms.

    So if you set the integration time to, say, 0.5 seconds, tsample is 4.19ms. In fact, if you set the integration time to anything greater than 4.19ms, the sample time will be 4.19ms.
    For shorter integration times, the system will set tsample to the closest power of two submultiple. For example if you set Tint = 0.001 second, what you get is 1.05 ms (i.e., 4.19/4). The smallest tsample that works reasonably well is 0.1ms; if shorter than that the data acquisition cannot keep up with the spectrometer, and most of the data gets lost. So as a practical matter, do not use Tint less than 0.0001 second. (actually 0.131 ms) The system is rigged to prevent Tint shorter than that.

    Assuming we should have at least 50 time samples per pulsar period, then we can set a limit on the shortest pulsar period that it would be useful to observe. P ~ 50*(0.0001) = 0.005 seconds, hence observe pulsars with longer periods than 5 ms.

    [Frank D. Ghigo, NRAO-Green Bank, April 2014, rev March 2015, rev June 2019]