Low-DM Dispersed Transients Found
Because the initial premise of the HawkRAO analysis was based on the possibility that the archive data may contain dispersed transients of unknown DM, the search technique adopted covered a wide range of DMs. That is, it was not a targeted search for Crab Pulsar Giant Pulses (GPs).
Normally testing over a range of DMs is computationally expensive - but a fast search algorithm was developed which is nearly two orders of magnitude faster.
Low-DM Dispersed Transients in HawkRAO Archive Data
As more files were processed it became clear there were many dispersed transients in the HawkRAO archive data displaying dispersion measures ranging from 40 to 70.
Some examples at DM=52.9, 53.1 and 44.3 are shown below.
It has been suggested that the dispersed transients are Crab GPs.
Certainly there is evidence from elsewhere that the Crab Pulsar can produce super giant pulses.
These can be of the order of MJys !!!
However, all attempts to get a DM=57 result have failed in the files processed first.
PRESTO's 'single_pulse_search.py' returns DM=52 with a clear peak as shown on the right - which agrees with my own single pulse search code - but there is no peak near DM=57.
It should be noted that pro software would not be optimised for such a narrow bandwidth (2.4 MHz), low number of channels (32) and low DM. For example PDMP (like it does for period search) returns varying 'best DM' results (i.e. jumps around) depending on what DM the archive file generated from the filterbank file was de-dispersed at. I will write some DM-search code which is tailored to narrow bandwidth.
The other possibility for the DM offset (i.e. 52 versus Crab's 57) is that the filterbank files I generated from my RTL_SDR IQ data has some sort of error in mapping RF frequencies into channels. To double-check this I labouriously fed an RF carrier into the front end and examined the resultant filterbank file in a hex editor to see which channels that carrier ended up in. I found a small offset in the absolute frequency mapping (half a channel: the highest channel maps to 437.125 MHz actual: the FB has 437.1625 MHz) - but the step between channels is exactly right - and it is the step size which affects the DM measurement. The small offset in absolute frequency would cause a 0.03 % shift in DM - not the ~10 % difference between 52 and 57.
So - these dispersed pulses look like Crab pulses - except for the DM.
Code was written to examine this in more detail.