Timeconv improved

The ASCA ftool to carry out barycentric correction, timeconv has been modified, so that accuracy of the event arrival time is improved by up to +/- ~130 microsec. The change is effective for timeconv v1.53 (January 1999) onward. The executable binaries for major platforms as well as the source code are obtained from http://heasarc.gsfc.nasa.gov/ftools/ftools_update.html.

This change will not significantly affect timing analysis of most sources, except a few very fast pulsars (see below). The correction is in the calculation of the satellite position at a given MJD; in the new timeconv, the leap-seconds since the launch of ASCA (February 1993) are taken into account in the calculation, while it was not in the previous versions. In 1999, not taking account of the leap seconds, you will get a time difference of five seconds. This will lead to difference of the satellite position by (2xPIxR/P)x5sec= 38 km, where R=6931 km and P=5745 sec are the orbital radius and period respectively. Thus, event arrival time difference could be up to +/-38km/c = 130 microsec depending on location of the source.

In Figure 1, we compare difference of the event arrival time calculated with the old timeconv (v1.52) and the modified one (v1.53). The source is the 1.6 msec pulsar PSR1937+21 observed on November 16, 1997. You can see that the arrival time difference modulates at the satellite orbital period. The green curve is the best-fit sinusoidal curve with P=5745.428 sec and amplitude=65.08 microsec. Figure 2 shows the periodgram (chi2 variation for trial periods) with the old timeconv and the new one. It is seen that the new timeconv gives a higher significance of the pulsation. Actually, Figure 3 (the folded light curve at the best-period) indicates that the pulse profile is much sharper with the new timeconv.

Please refer to IAUC 7030 and Shibata et al. ApJ Lett., 1999 for precise results of the PSR1937+21 observation.

ASCA GOF acknowledges Dr. S. Shibata, PI of the PSR1937+21 observation, for allowing us to use the data before the publication. We also thank Drs. Y. Saito and M. Hirayama for finding the problem in the old timeconv.