Tackling artefacts in the timing of relativistic pulsar binaries: towards the SKA

Abstract

Common signal-processing approximations produce artefacts when timing pulsars in relativistic binary systems, especially edge-on systems with tight orbits, such as the Double Pulsar. In this paper, we use extensive simulations to explore various patterns that arise from the inaccuracies of approximations made when correcting dispersion and Shapiro delay. In a relativistic binary, the velocity of the pulsar projected onto the line-of-sight varies significantly on short time scales, causing rapid changes in the apparent pulsar spin frequency, which is used to convert dispersive delays to pulsar rotational phase shifts. A well-known example of the consequences of this effect is the artificial variation of dispersion measure (DM) with binary phase, first observed in the Double Pulsar 20 years ago. We show that ignoring the Doppler shift of the spin frequency when computing the dispersive phase shift exactly reproduces the shape and magnitude of the reported DM variations. We also simulate and study two additional effects of much smaller magnitude, which are caused by the assumption that the spin frequency used to correct dispersion is constant over the duration of the sub-integration and over the observed bandwidth. We show that failure to account for these two effects leads to orbital phase-dependent dispersive smearing that leads to apparent orbital DM variations. The functional form of the variation depends on the orbital eccentricity. In addition, we find that a polynomial approximation of the timing model is unable to accurately describe the Shapiro delay of edge-on systems with orbits less than 4 hours, which poses problems for the measurements of timing parameters, most notably the Shapiro delay. This will be a potential issue for sensitive facilities like the FAST and the forthcoming Square Kilometre Array (SKA); therefore, a more accurate phase predictor is indispensable.