Pulse intensity modulation and the timing stability of millisecond pulsars: A case study of PSR J1713+0747

TL;DR

This study investigates the intrinsic pulse shape variations in millisecond pulsar PSR J1713+0747 and their impact on timing precision, showing that pulse jitter dominates errors and can only be mitigated by averaging many pulses.

Contribution

It demonstrates that pulse shape changes are intrinsic and dominate timing errors in PSR J1713+0747, and evaluates correction methods that do not improve timing precision.

Findings

Pulse shape variations cause excess timing errors.

Timing errors are correlated across frequencies and days.

Averaging many pulses reduces timing error proportionally.

Abstract

Most millisecond pulsars, like essentially all other radio pulsars, show timing errors well in excess of what is expected from additive radiometer noise alone. We show that changes in amplitude, shape and pulse phase for the millisecond pulsar J1713+0747 cause this excess error. These changes appear to be uncorrelated from one pulse period to the next. The resulting time of arrival variations are correlated across a wide frequency range and are observed with different backend processors on different days, confirming that they are intrinsic in origin and not an instrumental effect or caused by strongly frequency dependent interstellar scattering. Centroids of single pulses show an rms phase variation \approx 40 microsec, which dominates the timing error and is the same phase jitter phenomenon long known in slower spinning, canonical pulsars. We show that the amplitude modulations of single pulses are modestly correlated with their arrival time fluctuations. We also demonstrate that single-pulse variations are completely consistent with arrival time variations of pulse profiles obtained by integrating N pulses such that the arrival time error decreases proportional to 1/\sqrt{N}. We investigate methods for correcting times of arrival for these pulse shape changes, including multi-component TOA fitting and principal component analysis. These techniques are not found to improve the timing precision of the observations. We conclude that when pulse shape changes dominate timing errors, the timing precision of PSR J1713+0747 can be only improved by averaging over a larger number of pulses.