Hawkesbury Radio Astronomy Observatory

Vela Pulsar Observations

Observing PSR B0833-45 with a Small Aperture Antenna


First Results

Obtained with a fixed pointing (transit mode) circularly-polarised (CP) antenna - theoretical equivalent aperture to 2.8 m diameter dish @ 50% efficiency.

Note that - according to expert opinion - that the theoretical gain @ 436 MHz for that size dish is not reached in practice due to the proximity, in wavelength terms, of the feed to the dish surface.

Data acquired and processed as described in 'System Design'.

PLEASE NOTE:  These are the 'first light' results.  It details initial results as well as verification activities.  After verification of these first results the system has been upgraded to perform automated daily observations which can be viewed via the 'Observations' tab. Note also that the format of the images and various designations on this page (which details first results only) may be different to the latest image formats and designations on other pages due to system upgrades in hardware and software.

Index


First 'Light' of the New CP Antenna on B0833-45

After many unsuccessful attempts with other antennas at detecting Vela (B0833-45), the first light with the new 42-element circular-polarised 436 MHz antenna provided a positive result.

In each of the graphics below the pulse peak has been rotated to the centre.

1st May 2017 (B0833-45 - Vela Pulsar)

Observation Parameters
Start MJD........ 57874.2879
Frequency....... 436 MHz
Bandwidth...... 2.4 MHz
Duration......... 120 minutes

RTLSDR Dongle Settings
RTLSDR Gain... 29.7 dB
Digital AGC..... Enabled

Display Settings
Fold Period...... 0.0893993840 s

3rd May 2017 (B0833-45 - Vela Pulsar)

Observation Parameters
Start MJD........ 57876.2824
Frequency....... 436 MHz
Bandwidth...... 2.4 MHz
Duration......... 120 minutes

RTLSDR Dongle Settings
RTLSDR Gain... 29.7 dB
Digital AGC..... Enabled

Display Settings
Fold Period...... 0.0893994580 s

5th May 2017 (B0833-45 - Vela Pulsar)

Observation Parameters
Start MJD........ 57878.2804
Frequency....... 436 MHz
Bandwidth...... 2.4 MHz
Duration......... 110 minutes

RTLSDR Dongle Settings
RTLSDR Gain... 29.7 dB
Digital AGC..... Enabled

Display Settings
Fold Period...... 0.0893995273 s

6th May 2017 (B0833-45 - Vela Pulsar)

Observation Parameters
Start MJD........ 57879.2742
Frequency....... 436 MHz
Bandwidth...... 2.4 MHz
Duration......... 120 minutes

RTLSDR Dongle Settings
RTLSDR Gain... 29.7 dB
Digital AGC..... Enabled

Display Settings
Fold Period...... 0.0893995601 s

Verification Testing

An important and essential part of presenting scientific results is verification testing.  For systems with large antennas (large apertures) and strong pulsars, such verification is simplified because the signal-to-noise ratio (S/N) is high and a positive result is clearly demonstrated.  In small aperture systems however, where the S/N is low (even marginal), special care is needed to ensure the result is valid.

As the pulsar signal is a periodic increase in random noise within the received bandwidth, other terrestrial signals can readily mimic a pulsar signal, especially in marginal S/N scenarios.

The preceding paragraph cannot be over-emphasised and the failure to appreciate that statement leads to false results being taken as real.

Fortunately the pulsar signal has a number of characteristics which can used to reliably distinguish it from terrestrial signals...

Familiarity with the results of epoch-folding data over a range of periods (say 100 ppm) in small steps (say 0.01 ppm) will reveal that almost always there will be a period or two where, purely by chance, even in the absence of any pulsar signal, noise will sum to produce a 'false pulsar profile' of reasonable S/N.  Typically these false profiles only hold over a very small range of periods, while a true pulsar profile will hold over a larger range.  Also, false profiles will fail the 'two pulse' test where the data is folded at twice the period.  A true pulsar profile will show two pulses, a false profile will typical show no pulses.

From the above discussion a range of tests should be performed...

Folding at Predicted Period

To perform this test the sampling clock frequency must be calibrated to a high order of accuracy (< 0.1 ppm) - or alternatively the clock error determined to the same level of accuracy to allow a correction to be applied.Clock Calibration

The system here uses a 0.5 ppm TCXO which is expected to hold within 0.1 ppm over the temperature range experienced at the in-house observatory desk.

The determination of the sampling clock error was done by generating a 10 Hz narrow noise pulse derived from a Rubidium Frequency Source (RFS).  A data run of 4 hours was done and the deviation from 10 Hz noted and used as a correction.  The best period was determined by searching for the narrowest pulse width.  This method can be used because the S/N is high for this test pulse.

The previous correction from 6 months ago was +1.147 ppm, the new correction is +1.271 ppm - a change of 0.122 ppm.  Note that the above pulse profiles and the two below were generated with the old clock correction - i.e. an error of 0.122 ppm.

Pulsar data from each day (1/5, 3/5, 5/5 and 6/5) was folded at the predicted period and the results are shown above.  Each result shows a pulse at a received S/N consistent with the calculated S/N of 8.5±3 dB (done using this calculator). So the results PASS this test.

Results outside the calculated S/N range would indicate a problem with system sensitivity if the S/N was poorer, or a false result if the S/N was significantly higher, e.g., 6 dB higher.

Folding at Double the Predicted Period

Many false profiles can be eliminated by folding at double the predicted period.

The data for the 3rd May was folded at twice the predicted period...

0.0893994580 x 2 = 0.1787989160 s

... and the presence of two distinct pulses was confirmed awarding the result a PASS for this test.  Note that the two pulses are of almost identical amplitude and the S/N has dropped by about 1.3 dB - in reasonable agreement with the expected 1.5 dB drop due to each pulse having half the number of integrations.

Compare De-dispersed with Un-de-dispersed Results

Pulsar signals are dispersed, that is, lower frequencies in the bandwidth arrive later than the higher frequencies.  This means that when total power detected the pulse is 'smeared' in time (broadened) - reducing the S/N.

Pulses of terrestrial origin are not dispersed and so can be eliminated from the results.  To reverse the time smearing of the pulsar pulse signal, an inverse dispersion process is applied.  This narrows the pulse back to its original shape restoring S/N.  It also has the benefit of smearing narrow pulses of terrestrial origin - further improving discrimination.  The dispersion measure of Vela is DM=68, causing about 16 ms delay between the lowest frequency (434.8 MHz) in the 2.4 MHz bandwidth @ 436 MHz, w.r.t. to the highest frequency (437.2 MHz).

Note the same data that gives a clear pulse after de-dispersion (left hand profile above) produces no significant pulse when analysed in its un-de-dispersed form (right hand profile above) - confirming that the observed pulse is dispersed by the same measure as Vela's signal is.  Therefore, the result is awarded a PASS for this test.

Plot Daily Observations of Best S/N Period against Predicted Period

Another test method for discriminating against terrestrial signals is to plot the observed period against the calculated topocentric period.  For Vela the Earth's orbital doppler component affecting the observed period is sinusoidal with a period of one year.  The magnitude of the annual doppler effect swing is of the order of ±47 ppm, large enough to allow a comparison by plotting the calculated topocentric period vs the observed period to verify the signal has the Vela Pulsar's unique annual period change curve. Click on graphic to view larger size.

In this case the period is not set to the predicted period and folded, but instead a search is done for the folding period which gives the best S/N. After 14 days elapsed time, it is clear that the observed period is tracking the calculated topocentric period. Therefore the result is awarded a PASS for this test.

The upper and lower curves in the graphic plot the +1 ppm and -1 ppm values either side of the predicted period curve.

Updated 13/05/2017: the plot was redrawn after the sampling clock correction was updated as determined by a 4 hour run with a Rubidium Frequency Source reference and improvements made to the 'best S/N period' search algorithm.  Also an up-to-date ephemeris supplied by Jim Palfreyman (UTS) allowed a more accurate prediction curve to be plotted.

Visibility Test

Doing a data run when the pulsar is below the horizon should show a negative result.  This test eliminates the possibility that a terrestrial signal is mimicking the pulsar.  A 2 hour data run was started at MJD = 57880.845 - when Vela was below the horizon.  Analysis of the data failed to locate any significant pulse signal...

 

...therefore this test is awarded a PASS.

Repeatability Test

Repeatability of results is a strong test as it eliminates the false profiles due to random noise spikes.  The results are repeated over a number of consecutive days (except when PC reboots and power cuts prevent data collection) and so this test is awarded a PASS.

Processing by Professional Software

By passing an example observation data set to 'PREPFOLD' (part of the SIGPROC software suite) a further verification is obtained...

Click image for a larger view...

Conclusion

All the above tests are awarded a PASS which is a strong verification of the results.