Next: Modeling AXAF Obstructions with the Generalized Aperture Program.
Previous: Simulated AXAF Observations with MARX
Up: AXAF
Table of Contents - Index - PS reprint
R. A. Zacher, A. H. MacKay, B. R. McNamara, and L. P. David
SAO/ASC, Cambridge, MA 02138
We are developing computer models to simulate the focal plane detectors of the Advanced X-ray Astrophysical Facility (AXAF). The models, being developed as part of the AXAF Science Center, are being used to aid in calibration planning and to characterize the performance of the AXAF observatory. Scripts have been developed to configure and run the simulations automatically from a test database. Depending on the output mode selected, the results can be viewed directly, sent through telemetry processing, or fed into higher-level analysis pipelines in the Data System.
Much of our work has been focused on developing high-fidelity simulations of the two main Scientific Instruments located in the focal plane of the telescope. These are the AXAF CCD Imaging Spectrometer (ACIS) and the High Resolution Camera (HRC). In addition to these, we have integrated gratings modules and we use output from other simulators which model the sources and telescope mirrors (see Figure 1).
Figure: Simulation Schematic. Original PostScript figure (8kB).
The SHUTTERS module simulates 16 separately configurable shutters
behind the mirror assembly. The GRATINGS module simulates the High,
Medium, and Low Energy Transmission Gratings. The FILTERS module
simulates optical blocking filters in front of the detectors. The
detectors are ACIS ( a charge coupled device) and HRC ( a microchannel
plate).
ACIS is a charge-coupled device optimized for X-ray
detection. Its 24µm pixel size offers ½´´
resolution in the AXAF focal plane. The field of view is
16×16´ for the imaging array and 8×48´ for the
spectroscopic array. The chip is modeled as a multilayer
structure. The incident X-ray photons create a charge cloud
whose position and size are determined by the silicon
absorption depth, the photon energy, the photon position and
the dopant concentration. The charge cloud then drifts to
the surface of the chip under the influence of a layer
dependent electric field, which we model using a Monte Carlo
method. At the surface, the charge is mapped onto a 3×3 pixel
array. The
functional dependence for the number of electrons (ne)
created by an incident X-ray of energy Ex
is given by ne~nx(Ex/
E).
Here, nx is a
function calculated by the Monte Carlo program and =3.65eV
is the energy required to liberate a
charge carrier. Additional features modeled include read
noise, charge transfer inefficiency, bias, gain, and layer
thicknesses. The algorithms used in the CCD simulation are
based on the program XRAYSIM developed by Lumb
et al. (1994), which in turn was based on analytical
calculations by Janesick (1987, 1988) and Hopkinson (1987).
The simulator can be operated in two modes. The Event List Mode outputs the abovementioned 3×3 pixel array to a FITS event list. This mode is designed for high throughput and does not model effects which arise when two photons hit the same location on the chip in a given integration period. The Full Frame Mode embeds the 3×3 pixel array in a much larger rectangular array in memory. This mode includes the effects which arise when multiple photons hit the same location in a given integration period. The Full Frame Mode has two output formats available. The events can be extracted from the array in memory using the same algorithm used to detect events in real ACIS frames. The extracted events are output to a FITS event list. Alternatively, the full arrays can be written to FITS image files, one for each integration period. These image files are similar to those produced by the physical chips before event extraction. Event detection in the array in memory yields substantial performance gains over detection of the events in the FITS image files.
The HRC is a microchannel plate (MCP) detector that provides
a spatial resolution of less than ½´´. The field of
view is 31×31´ for the imaging array and
7×97´ for the spectroscopic array. Resolving power
is limited, with E/
E~1. The
simulation models the UV Ion Shield (UVIS), the MCP itself,
and the wire charge grid. The UVIS is modeled using a
generalized filter program that statistically simulates
photon absorption by applying a transmission curve to the
input photon energy. The MCP surface is modeled as a surface
of circular pores with a diameter of 0.0125mm and spacing of
0.015mm. A model of quantum efficiency
as a function of incident angle is also applied. The wire
grid charge resulting from the charge cloud produced by the
MCP is modeled by a scaled Lorentz function. Events are
passed into a telemetry simulator that models dead time
induced by telemetry bandwidth limitations. Output modes are
raw telemetry, FITS event list, and QPOE image formats.
The HRC simulator has also been adapted to simulate a similar
instrument called the High Speed Imager (HSI) which is used for
telescope mirror calibration.
The mirror simulation's raytrace output can be projected
directly on to the model detectors, or diffracted by the
gratings module before projection on to the model
detectors. The dispersed gratings spectrum provides a
resolving power of E/E ~ 10
.
The High Energy Transmission Grating is
typically used in conjunction with the ACIS detector and the
Low Energy Transmission Grating with the HRC.
Figure: Simulation Control Hierarchy. The hierarchy of
program control is depicted. Using database entries
describing the test to be performed, parameters describing
the configuration are set. The ASCDS pipeline
controller initiates the raytrace and monitors program
execution. Original PostScript figure (21kB).
The simulator control hierarchy is depicted in Figure 2. The simulators run as a set of UNIX processes, each of which represents a physical component being modeled. These processes are started, monitored, and stopped by the ASCDS Pipeline Controller. The simulators utilize common ASCDS libraries where possible, such as the IRAF parameter interface. Events (photons) are passed from one process in the pipeline to the next with each process performing some necessary action on an event before passing it along. The action may be to alter the event or to decide not to propagate it. The simulators are implemented primarily in C, with some supporting code written in Perl.
The software is designed to easily accommodate modifications and enhancements. The modular pipeline approach has facilitated the interchange of modules as we continue to upgrade the fidelity and capabilities of the simulations. In order to further improve flexibility and provide access to data in the simulator pipelines, a C++ Application Program Interface to the simulator data stream is being prototyped.
This project is supported by NASA contract NAS8-39073 (ASC). We would like to thank the following individuals for their contributions to the simulations. Terry Gaetz, Diab Jerius, and Dan Nguyen developed the mirror and source models. John Davis, Dan Nguyen, and Mike Wise developed the gratings model. Diab Jerius developed the HRC pore surface model. Dave Plummer developed the HRC Telemetry dead time simulation. Adam Dobrzycki developed the HRC charge grid algorithm.
Janesick, J. R., Elliot, T., Collins, S., Taher, D., Campbell, D., & Garmire, G. 1987, Optical Engineering, 26, 2, 156
Janesick, J. R., Elliot, T., Bredthauer, R., Chandler, C., & Burke, B. 1988, SPIE, 982, 70
Hopkinson, G. R. 1987, Optical Engineering, 26, 8, 766
Lumb, D., Townsley, L., Nousek, J., Burrows, D., & Corbet, R., 1994, personal communication
Next: Modeling AXAF Obstructions with the Generalized Aperture Program.
Previous: Simulated AXAF Observations with MARX
Up: AXAF
Table of Contents - Index - PS reprint