Highly accurate timing of radio millisecond pulsars permits a wide variety of fundamental astrophysical and gravitational experiments. Examples include the study of neutron stars, the interstellar medium and binary system evolution, and stringent tests of the predictions of general relativity and cosmology.
The chief obstacle to high timing precision is dispersion of the pulses during their propagation through the ionized interstellar medium. This phenomenon results in delays of lower-frequency radiation components relative to the higher frequencies, and across a typical observing bandwidth can amount to many (often hundreds or more) times the intrinsic pulse widths. With a traditional filter bank system, which removes the dispersion between channels after the pulse is detected, residual smearing remains within the individual channels. A superior approach is pre-detection or "coherent" dedispersion, in which the raw signal voltage is convolved with the inverse of the transfer function of the interstellar medium. This technique was first used more than two decades ago, but only recently have computer speeds and the cost of data storage approached levels allowing its widespread use over bandwidths of several MHz or more.
We have built a new data acquisition system designed to implement this coherent-dedispersion technique in software. The "Mark IV" system digitizes the complex signal voltages in two orthogonal polarizations before detection, Nyquist-sampling either a 10 MHz bandwidth with 2-bit quantization or a 5 MHz bandwidth with 4-bit quantization. The data stream is stored on a combination of magnetic disks and tapes for off-line processing.
Data reduction is computationally intensive: each hour of observations requires 10-12 Gflop-hours to reduce. We will process our observations at Manchester Computing, using the CSAR Cray T3E-1200E service and/or the Kilburn SGI Origin 2000 service.
Our immediate scientific goals centre around using the 76-m Lovell telescope at Jodrell Bank to study a number of neutron-star--white-dwarf binary pulsars.