Setting up the Artificial Pulsar with the IBOB

Artificial Pulsar Test Results: July 31, 2008

Marc Eimers

For the past couple of days, I have had an artificial pulsar machine hooked directly into the Parspec back end in an attempt to begin to narrow down the possible causes of the problems stated in the previous update. This document will present my initial test results and serve as a reference guide for those who wish to reproduce these results.

Setting up the Artificial Pulsar with the IBOB

The name of the Parspec design that was loaded onto the IBOB for this testing was parkes200MHz08May26. The connectivity settings and initial startup parameters were set by using the “Reset” button on the Parspec Datataking Control GUI. The commands that this button sends to the IBOB are listed below:

regwrite reg_ip 0xc0a80307

regwrite reg_10GbE_destport0 4001

write l xd0000000 xffffffff

setb x40000000

writeb l 0 x00000060

writeb l 4 xdd47e301

writeb l 8 x00000000

writeb l 12 xc0a8030a

writeb b x16 x0f

writeb b x17 xa1

writeb l x3038 x00000060

writeb l x303c xdd474d0e

writeb b x15 xff

write l xd0000000 x0

regw qscale 30

regw uscale 30

regw vscale 30

The accumulation length and sync period were also set by the GUI. The commands sent were:

regw reg_acclen 49

regw reg_sync_period 2560000

The signal from the output of the artificial pulsar is a baseband signal reaching a maximum of about 120 MHz, so the machine must be hooked directly into the spectrometer. To do this, simply connect the outputs (A and B) on the back of the artificial pulsar machine into the two data inputs on the IBOB. Then flip on the on/off switch on the artificial pulsar and set the right most knob on the machine to “Pulsar”. Throughout my testing, I had the “Output Attenuation” knob set to 0. I started off with the “Pulse Attenuation” knob set relatively high at 23, and then adjusted it at various points in the testing.

The function generator that was used with the artificial pulsar was the Agilent 20 Mhz Function/Arbitrary Waveform Generator. Connect the output from the function generator to the TTL input on the artificial pulsar, then turn on the function generator. Select the pulse option and set the desired pulse frequency, amplitude, offset, and pulse width. The scroll dial on the function generator doesn't seem to work too well, so set the values using the number pad. For my testing, I set the amplitude to +15dBm and the offset to 0 V. I changed the pulse frequency on occasion as I proceeded through my tests, but I consistently kept the pulse width set to about 3% of the pulse period. Start off with a low pulse frequency to check that the function generator is connected right. If you set the frequency to 1 Hz, you should see the “Pulsar” LED on the artificial pulsar flashing once per second. Once you have verified this and that the artificial pulsar is indeed connected to the IBOB, you should be ready to observe.

Observing the Artificial Pulse

For observing the artificial pulse, I used the Parspec Datataking Control GUI that Duncan Lorimer and myself wrote. Instructions for how to operate the GUI will be documented in my final paper. To check that the IBOB is actually receiving data from the artificial pulsar, I used the “Monitor” option in the GUI which runs /data2/glangsto/guppi_monitor_plotonly, a real time monitor that displays the power levels for the different polarizations. Below is a screen shot of the GUI and the monitor display.

The top plot displays the power of Stokes parameter I. Here the bandpass shape of the signal from the artificial pulsar can be seen. I currently have the scaling coefficients for polarization 1 and 2 set all the way up to 2000, however I tried different scalings throughout the testing. This scaling is different from attenuation in that the higher the value is set, the stronger the signal is. Now Parspec is ready to take data.

Results

The majority of my observations were 60s in length. I first started with the “Pulse Attenuation” knob on the artificial pulsar set to 23 and the scaling coefficients in the GUI set to 2000. I then proceeded to observe at different settings on the “Pulse Attenuation” knob ranging from 23 down to 12 to see if the negative dips would show up. I set the frequency of the pulse to be 3 Hz and the pulse width to be 10 ms. Below are two diagnostic plots of folded data from converted filterbank files, the first with the “Pulse Attenuation” knob set to 23, the second set to 12.

It can be seen from the plots how the S/N ratio is better in the observation with the lower “Pulse Attenuation” setting. The negative dips on either side of the pulse that were previously seen in the integrated pulse profiles do not show up in either case. Next I observed with different scaling coefficient settings at intervals of about 400 while keeping the “Pulse Attenuation” set to 12. The only difference was a very slight decrease in S/N, again no dips. Then I observed with the pulse frequency set to 3 Hz, 30 Hz, and 300 Hz. At 300 Hz, I observed an exponential tail in the integrated pulse profile, here shown below.

This effect may be produced by the function generator, though I myself cannot say for sure. Still, there are no negative dips. Finally, I increased the observation length to 5 and 10 minutes, coming out with the same results as the previous observations.

These test observations seem to show that the problem with the system does not lie within the Parspec machine or the GUI itself.