WVU / NRAO / U. C. Berkeley Giant Pulse Statistics Project

WVU, NRAO and U.C. Berkeley are developing a pulsar search and full sky mapping data acquisition system.

As a thorough checkout of the observing system, we are preparing a dual frequency band Crab Giant Pulse observation, using Peter McMahon's Parspec design, modified for full stokes observations.

For these observations, we used the GUI designed by Duncan Lorimer and Marc Eimers, his student during summer 2008.

WVU/NRAO Parspec GUI interface

The Parspec gui is initiated on the data taking computers. In our case, we are using tofu.gb.nrao.edu and cicada2.gb.nrao.edu. The setup for data acquisition is fairly simple.

  • ssh tofu
  • cd /data/scratch/glangsto/parspec
  • python ibob5.py
  • gui5+

This yields a main screen controlling the file label names, obs duration and data monitoring. The image below shows a screen capture of a VNC session for observations.

Parspec GUI during monitoring of the data quality. At right are the 4 stokes parameters
I, Q, U and V. The solid lines are the average data in the last second and the top and bottom dashed
lines are the maximum and minimum spectra in the last second. At left is the GUI for starting data

Quick instructions for looking at data

(A) put the following aliases in your .cshrc

  • alias psrfits2fil ~dlorimer/bin/psrfits2fil
  • alias bandpass ~dlorimer/bin/bandpass
  • alias pgplotter ~dlorimer/bin/pgplotter
  • alias process ~dlorimer/bin/process
  • setenv PSRCAT_FILE /users/dlorimer/psrcat.db

(B) convert the first few dumps of the fits file into filterbank format and look at the bandpass

  • psrfits2fil parspec_MJD_SOURCE_SCAN_SUBSCAN.fits bandpass
  • eg: psrfits2fil parspec_54903_PSRB1937+21_0002_0001.fits

this will produce a file called "parspec_MJD_SOURCE_SCAN_SUBSCAN.fil" which is ready for processing by sigproc tools... the first one is

  • bandpass parspec_MJD_SOURCE_SCAN_SUBSCAN.fil -numerate | pgplotter
  • eg: bandpass parspec_54903_PSRB1937+21_0002_0001.fil -numerate | pgplotter

look at the output PGPLOT window and decide what your first channel and last channel will be. These are referred to as "first" and "last" below.

(C) decide whether you need to change the header parameters

by looking at the output from psrfits2fil shown above, you will see what center frequency and sampling time are assumed. These might not be correct! Figure out the sky frequency and sampling time needed (ask Glen for help on this!) and let's call these "fcent" and "tsamp" in the next step.

(D) convert the fits file and trim off the unwanted channels

  • psrfits2fil parspec_MJD_SOURCE_SCAN_SUBSCAN.fits first last 0 fcent tsamp

here the "0" means no flipping of the bandpass, if you know that the band is inverted (i.e. lowest frequency is written out first) then replace "0" with "flip". If not, it will be obvious at the next step.

(E) take a quick look at the data

  • process parspec_MJD_SOURCE_SCAN_SUBSCAN.fil
  • eg: process parspec_54903_PSRB1937+21_0002_0001.fil

this will produce a PGPLOT file and also a png file "parspec_MJD_SOURCE_SCAN_SUBSCAN.png". Take a look at this and if the pulse is arriving first at the lowest frequency, you will need to rerun psrfits2fil using "flip" instead of "0".

bandpass.gif process.gif
B1937+21 800+/-160 MHz band pass Folded B1937+21 pulsar data pulse
period versus frequency.

The data in the plots above are on the NRAO computer tofu.gb.nrao.edu and are located in directory


If you are not seeing anything at all - contact Duncan.Lorimer at mail dot wvu dot edu !!

Observing Log

2009 February 28; Initial tests of 43m and WVU GUI

We performed a few hours of testing of the 43m and the observing system, and detected a few crab giant pulses, after some work in understanding the configuration.

Duncan Lorimer called in and checked on the data acquisition system via VNC from Morgantown, WV. Dong, Shilpa Bollenini and Glen Langston came to the GUPPI development area, in the Reber building of NRAO in Green Bank.

We initially had trouble with setup. At some point we accidentally changed the clock frequency from 900 MHz to 700 MHz. This caused the folding to give strange results. After hooking up the fake pulsar, Dunc was able to confirm the system was working but it was likely the timing information was wrong. We identified the clock frequency change as the cause of our confusion, and and were able to fix the clock, then start data taking. The configuration was 900 MHz clock, 56 accumulation per dump and 512*2 channels, yielding a dump time of 0.000063715555 seconds or 63.7155555 microseconds.

Dunc was able to fold some data on PSR0329 and was able to detect at least 1 giant pulse from the crab (See figures below). These observations started on MJD date 54890.

Scan log follows:
  • Scan 7: PSR0326+54 Good 5 minute scan
  • Scan 8: Crab pulsar: Good 1 hour scan
  • Scan 1: Crab pulsar: Good 1 hour scan (Date changed to March 1, 2009 == MJD 54891)

0329.png crab1.png
Folded pulse profile for Pulsar 0329+54Example Crab Giant Pulse

2009 March 1: Update frequency, 8 hours of Crab obs

The start-up was smoother for the second set of observations for the weekend. Glen Langston updated the clock frequency to 896 MHz and the number of integrations to 56. This combination, including real and imaginary parts of 512 channels, yields a dump time of exactly 64 microseconds.

The log of observations follows:

  • Scan 5: PSR0329+54: 60 second observation, but had wrong number of accumulations, 55 not 56
  • Scan 6: PSR0329+54: 5 minute obs. Edited GUI to fix roundoff in calculation. Have 56 integrations per dump.
  • Scan 7: Crab pulsar: 5 minute obs. ON source at all times
  • Scan 8: Crab pulsar: 5 minute obs. During a petal (Peak) scan, 1 degree radius, should be able to determine gain with this data.
  • Scan 9: Crab pulsar: 5 minute obs. During a petal (peak) scan, 2 degree radius, for gain calibration

Intensity Versus Time plot for 200 and 800 MHz petal observation of the Crab SNR.

  • Scan 10: Crab pulsar: 4 hour continuous obs
  • Scan 11: Crab pulsar: started 4 hour obs, but interupted for high wind. Deleted all off source data
  • Scan 12: Off source, all data deleted.
  • Scan 13: Crab pulsar: 4 hour continuous obs.

2009 March 7: Change bandwidth, Mapping and Crab obs

We attempted to start dual band (700-900 MHz and 150-300 MHz) obs this weekend, but had trouble configuring one of the ibobs. Currently suspicious of one of the analog to digital samplers. We did have success with 10 GBe with both IBOBs 5 and 8. (The IBOB numbers are the last digit of their IP addresses, or 8.)

We had been running the sampling clock at a higher speed, 896 MHz, than the build allowed. To try and reduce reliablity problems we channged the sampling clock to 640 MHz, so now have 320 MHz bandwidth. We kept the accumulation time at 64 micro-seconds. There 40 accumulations per data dump.

Log of observations:
  • Scan 1: Mapping test on Cygnus A for calibration, but bad source position, data deleted. (Crab had not yet risen).
  • Scan 2: 5 minute Petal Scan of Cygnus for calibration. Simultaneous Petal data taken with 43m.
  • Scan 3: half-hour map of Cygnus, to test spectral line mapping. Simultaneous Map with 43m.
  • Scan 4: 3 minute observation of W51, to see if molecule CH is visible at 722 MHz in absorption.
  • Scan 5: half-hour map of W51.
  • Scan 6: 1 minute obs of 0329+54
  • Scan 7: 5 minutes obs of 0329+54
  • Scan 8: 2.1 hour obs of the Crab, stopped due to high wind. Off source data removed.
  • Scan 9: 4.1 hour obs of the Crab, stopped due to end of disk space. All data on source.

2009 March 9: Tests with dual polarization, dual iBob operations.

After investigation by Jason and JD, we found that by jiggling the input cables to IBOB8, we were able to get good signals to both Ibobs. We attempted data acquisition and found that we could not have two monitor programs running on the same linux host. However the monitor data looked completely reasonable.

Second Ibob, frequency range 0-320 MHz from 43m.
Sideband is flipped and 320 MHz is to the left.

2009 March 13: Single Parspec, but with correct observing frequency in file

Found that previous data had correct bandwidths, but wrong center frequencies in the files. Corrected that by making changes on the 43m.

Log of observations:
  • Scan 2: 1 minute observation of strong test pulsar B1937+21; Folded data looked good!

Topic attachments
I Attachment Action Size Date Who Comment
0329.pngpng 0329.png manage 3 K 2009-03-01 - 23:45 GlenLangston PSR0329+54 profile
b0329bandpass.gifgif b0329bandpass.gif manage 3 K 2009-03-11 - 10:10 GlenLangston bandpass plot
bandpassP1937.gifgif bandpassP1937.gif manage 4 K 2009-03-13 - 16:24 GlenLangston band pass obs of PSR1937
crab1.pngpng crab1.png manage 5 K 2009-03-01 - 23:45 GlenLangston Crab Giant Pulse example
crabPetal2d.pngpng crabPetal2d.png manage 13 K 2009-03-01 - 19:01 GlenLangston 4 peaks of a petal observation of the Crab SNR
ibob0-320MHz.gifgif ibob0-320MHz.gif manage 10 K 2009-03-09 - 17:16 GlenLangston Second Ibob, frequency range 0-320 MHz from 43m.
parspecB1937bandpass.gifgif parspecB1937bandpass.gif manage 4 K 2009-03-13 - 14:11 GlenLangston B1937+21 800+/-160 MHz band pass
parspecB1937process.gifgif parspecB1937process.gif manage 19 K 2009-03-13 - 14:10 GlenLangston folded B1937+21 pulsar data pulse period versus frequency.
processP1937.gifgif processP1937.gif manage 15 K 2009-03-13 - 16:24 GlenLangston processed PSR1937+21
tofuGui.gifgif tofuGui.gif manage 39 K 2009-03-16 - 14:22 GlenLangston Capture of interface GUI
Topic revision: r12 - 2009-10-19, CarolynWhite
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