Contents of this directory:

HEXTE RESPONSE MATRICES AND ARFs:

These have been produced in a 2-stage process using version 8.4.2 of our
response matrix calibration code at UCSD, which applies both ground and
in-orbit calibrations to a Monte-Carlo model of our scintillation detectors.

They supersede the HEXTE matrices produced earlier (1996 Nov 18) and
represent improvements in our code's energy resolution, which has resulted
in less scatter in residuals to Crab Nebula spectral fits (and a best-fit
spectral index in closer agreement with the canonical value of 2.08). The
"c" after the date denotes that the RMFs have been adjusted to remove this
residual structure from the Crab Nebula spectrum, which provides a
first-order correction for the elements of our response which we don't
understand yet.

In addition, for display purposes in XSPEC, the discontinuity in the HEXTE
energy/channel relation has been smoothed, though this change has no effect
on spectral fits.

For example, hexte_97mar20c_pwa0.rmf is the Crab-fudged response for Cluster
A, PhosWich detector 0 (PWA0). This should be used in conjunction with the
ARF hexte_97mar20c_pwa0.arf when performing spectral fits to data from
this detector.

We have also provided the summed matrices for the clusters as a whole:
hexte_97mar20c_pwa.* - detector-summed response for cluster A
hexte_97mar20c_pwb.* - detector-summed response for cluster B before 1996 March
6 11:27:12 UT
hexte_97mar20c_pwb013.* - detector-summed response for cluster B AFTER 1996
March 6 11:27:12 UT (i.e. after the loss of spectral capability from detector
PWB2 - this response includes the sum of detectors 0, 1, and 3 only).

FITTING THE CRAB AND OTHER SPECTRA:
We recommend the following:

1. Be sure to IGNORE channels 1-10 and 250-256, for which the spectral
response cannot be determined. Depending on the lower energy bound for your
observation, you may need to ignore channels above 16 too. This will be
obvious when plotting the count rate spectrum.

2. Keep the model fit normalization a free parameter, since our detectors'
absolute geometric areas are still preliminary. However, detectors within a
cluster have a good RELEATIVE calibration.

Also, these matrices assume that the source is perfectly aligned with each
detector's axis, i.e. no boresight correction (an Ftool, HXTARF, will apply
this 2-8% correction to the .arf files provided here).

The HEXTE response matrices are still evolving. Specifically, before the
"Crab corrections" we still find significant (~3%) residuals in our fits to
Crab data below 30 keV. Above the Iodine K-edge at 33 keV, however, the
response appears to be accurate.

SOFTWARE NOTES: 
These matrices will work corrrectly with XSPEC 9.02, and the Ftools 3.6.1
tasks RBNRMF and ADDRMF. However, the released version of MARFRMF has
insufficient memory to handle them, and produces a fatal error. (This can be
avoided by editing the source code marfrmf.f, setting maxgrp =10 in the
subroutine DO_MARFRMF, and recompiling the Ftool).



For questions regarding HEXTE calibration, contact Dr. Philip Blanco 
at UCSD (pblanco@ucsd.edu).