Methodology and pileup definition

Analysis of the spectra was done using theTARA package developed at Penn State (v. 3.1). We accumulated events within a circle centered on the source and of radius 10.5 pixels for the brighter sources and 6.9 pixels for the fainter ones, excluding photons that were contained in the Charge Transfer Trail. We estimate that the latter are at most a few percent (2-3%) of the total flux. We inspected the light curves for flux variability, and found that, in general, the flux remained constant during the observations. In two cases (tests H-IAI-CR-1.003 and 1.009), during the exposure the source was moved to be centered in the middle of the spatial window; since apparent variability is produced in the light curve, we filtered the light curves for the longest stable exposure.

It is well known that, in the case of a monochromatic source, the primary spectral effect of pileup is producing secondary peaks at integer multiples of the primary peak energy, and with decreasing amplitude (e.g., Kastner et al. 1998).

We have previously given an empirical definition of pileup fraction. However, in order to allow a proper comparison between the data and the theoretical models, in the remainder of this section we will adopt a different pileup definition. We will define the pileup fraction as the the fraction of one or more events detected in the secondary peaks; in symbols, 

where N_i is the number of events in each detected peak, with i=0 being the primary peak. Note that the definition in eq. 6.14 differs from the definition  given elsewhere in the Calibration Report in the sense that the correction factors for the multiple-event peaks are neglected here. (This definition follows that used by Brian McNamara's work at the ASC described in Section 6.7.2.4.1.)

As pileup is dependent upon the grade filtering of the data(Kastner et al. 1998), our analysis will need to be performed for different grade selections to compare calibration data to flight data, or even between different sets of flight data if the grade selections are changed. At this time (pre-launch) we expect to use as a standard grade selection one to the two following selections: a) all events, excluding grade 255, and b) all events falling into ASCA-type grades: G0, G023, or G02346, for both FI and BI detectors.

The number of counts in each spectral peak was evaluated simply by counting the events falling in a given ``Region of Interest'' across the peak energy. This region may contain part of the adjacent continuum, which is formed by piled events: in fact, a portion of the charge produced from multiple events may not be recorded in the 3x3 event island resulting in tails below the pile-up lines. (See Section 6.7.1.)

For tests H-IAI-CR-1.001, 1.003, and 1.005, our analysis thus yields a higher number of events in each peak than in the work byKastner et al. 1998) who used much narrower spectral regions around the peaks, excluding the nearby continuum.