Date: Tue, 28 Apr 1998 12:26:26 -0400 (EDT)
From: "Kathryn A. Flanagan" <kaf@bodicea.mit.edu>

Here are notes of conversations I have had with various HRC team
members.  This information may already be known to most of you,
but I thought I'd pass it around for completeness. 

DETECTOR UNIFORMITY:

	I spoke with Almus Kenter about correcting our grating results
for detector uniformity and bias angle.  The HRC-I was quite uniform. 
Almus recommends using the "post-fix" spatial uniformity maps and
normalizing the abolute QE by the ratio of pre-fix to post-fix values.

BIAS ANGLE:

	I spoke with Ralph Kraft about the variation of QE with pore
angle. Recall that in the case of HRC-I, the bias angle is not along z,
but at about halfway between the y and z axes.  Since the HEG and MEG
dispersion axes are at +/- 5 degrees relative to the y-axis, the problem
is complicated. There are some angle dependency curves on the HRC web
page, but there are lab tests currently going on which will soon
replace these curves.
	Since I am concerned about comparing + and - high orders for
MEG and HEG, I spoke with Mike Wise about the possibility of employing MARX
to explore the effect of angle on the result.  This may be possible to
incorporate.

FILTERING EVENTS:

	Steve Murray spoke with me about our event selection
and its effect on our grating dispersion images. When the pulse
height is 255, this represents saturation, and produces a "ghost" image.
The image is not randomnly skewed, but is skewed to one side along y.
Around 15-20 percent of the photons were like that. This does not depend
on energy and is random, so that we can filter them out and then
renormalize according to the ratio of total events to "good" filtered events.
	This does not clean up the problem entirely: The real problem
depends on the amplifiers at wire level.  I am working with the level 1
qp file, which contains the six individual amplifiers. 
We wish to reject events that have an amplifier value of 4096, for example.
The HRC team uses the following filter: amplifier value below 4000,
PHA value below 250.  Scale the result according to the raw/filtered ratio.
	About 0.5 to 1% of events are slightly misplaced (by 1/2 tap
or 800 microns or 128 pixels.  This is the "real" ghost image, but will
be very slight. They will be clustered, with diagonal displacement (along
y or z - he thinks along z).  
	Note that the HRC web page now has "pre-fix" calibration results,
in particular, the QE of HRC without the gratings for XRCF conditions.