An Empirical Method for Improving the Quality of RXTE PCA Spectra

TL;DR

This paper introduces a calibration method, { t pcacorr}, derived from RXTE PCA spectra of the Crab, which significantly enhances spectral fit quality and sensitivity for high-count observations of various X-ray sources.

Contribution

The paper presents a new empirical calibration tool, { t pcacorr}, that improves spectral fitting accuracy and sensitivity for RXTE PCA data, based on residual analysis of Crab spectra.

Findings

Application of { t pcacorr} dramatically improves fit quality for spectra with >10^7 counts.

The Crab residual spectrum is stable over the mission but varies among detectors.

Using { t pcacorr} and a 0.1 ext{ extperthousand} systematic error increases sensitivity to faint features.

Abstract

We fitted all of the several hundred {\it RXTE} PCA spectra of the Crab individually to a simple power-law model; the total number of counts in the composite spectrum is $>10^9$. We then used the spectrum of residuals to derive a calibration tool, called {\tt pcacorr}, that we apply to large samples of spectra for GX~339--4, H1743--322, and XTE J1550--564. Application of the tool improved the quality of all the fits, and the improvement is dramatic for spectra with $\gtrsim10^7$ counts. The Crab residual spectrum is somewhat different for each of the five PCA detectors, but it was relatively stable over the course of the mission. We recommend that {\tt pcacorr} be routinely applied to spectra with $\gtrsim10^6$ counts and that one include a systematic error of 0.1\%, rather than the 0.5--1\% value that has customarily been used. We expect that application of the tool will result in an increase in sensitivity of the PCA to faint spectral features by up to an order of magnitude.