Pileup Effects in HETGS Spectra, Dec.'99

Back to HETG TechNotes

Summary
ACIS pileup occurs in the HETGS dispersed spectra as in all ACIS observations. For a point source the dispersed events are in narrow linear spectra and an estimate of pileup effects along a spectrum is given by the event rate per pixel island; that is by the rate:

                number of events in 3 pixel bin along the dispersion axis
 Island Rate =   --------------------------------------------------------
                           number of ACIS readout frames

The value of this rate indicates the magnitude of pileup effects: for values of 0.01 and less pileup effects are negligible; for values of 0.01 to 0.1 pileup will produce changes in event rates less than 10%; for values much above 0.1 pileup can be severe and the data quality dubious.

Pileup plots indicating this rate are presented for several calibration sources and show the varying levels of pileup in the HETGS data sets. Similar plots can be made from simulated data sets to check the expected pileup level for an observation.

Pileup Plots
Pileup Plots for several calibration targets are available below. The plots were made with HAK code version 1.3 with this updated version of obs_anal.pro.

The Y axis shows the number of events summed over 3 pixels - specifically, "AX" is binned to 1 pixel bins and then convolved with a box-car kernal, [1.,1.,1.]. The resulting histogram gives the numerator of the island rate as a function of the AX (~ TDETX) value. Note that for simplicity all events in the cross-dispersion are included as the bulk of events are concentrated within a pixel or two of the dispersion axis. No ACIS ENERGY selction is applied so that all orders are included.

The number of ACIS readout frames, the denominator of the island rate, is shown in the plots by the solid red line. Thus this level corresponds to an island rate of 1 event per frame; dashed and dotted red lines indicate island rates of 0.1 and 0.01 per frame. The number of frames used here is the number of frames in the input data set for all CCDs - in the high event rate case (e.g., Cyg X-2 below) the actual number of frames is a function of the CCD and can introduce a factor of 2 error as the red line is approached.

Mkn 421, Obsid 457
Mkn 421 shows very low island rates of less than 0.005 per frame for HEG and MEG at all energies.
Capella, Obsid 1318
Capella has generally low island rates (0.01 or less, the dotted red line) except for some of the brightest emission lines which have observed rates up to 0.2 events per frame. A MARX simulation of Obsid 1318 produces very similar island rates. Note that unlike real data, simulated island rates can exceed 1.0 if pileup effects are not included in the simulation.
Crab Pulsar and Nebula, Obsid 168
The Crab pulsar continuum is slightly piled up in some spectral regions. (The values shown are upper limits since some of the nebula is included as well.) As an example of using the s/w on an extended source, the island rate for a nebula-only region of the Crab was made. Spectra were extracted in a 3 pixel wide (CD_WIDTH) region with the zero-order offset by -50 pixels from the pulsar. As shown here the rates are low and so pileup is small/negligible in the extended nebular events.
Cyg X-2, Obsid 1102
Cygnus X-2 is heavily pileup along most of the MEG and HEG spectral regions with values of 0.1 to 0.6. S0 and S5, however, have rates of order 0.01 to 0.03



Please send any comments to Dan Dewey at dd@space.mit.edu.