Date: Mon, 10 Apr 1995 17:35:49 -0400 (EDT)
From: MURCHIE@s1pvx7.jhuapl.edu
To: near-s-msinis@aplmail.jhuapl.edu
Message-Id: <950410173549.37800226@S1PVX7.JHUAPL.EDU>
Subject: more updates

To:	MSI-NIS team
From:	Scott Murchie
re:	more updates

(1) I have updated the files of measurements of the calibration lamps
using calibrated photodiodes, imcorporating measurements through all
the neutral density filters.

(2) There have been more questions about hte nature and cleaning of the
lubricant contaminant on the CCD.  The mass spectrum most closely matches
that expected for the solvent of an oil-based paint.  There is no component
of the detector array or any associated hardware that should produce
such a substance.  The likely candidates (so far) are a C-clamp on the
inside of the FPU characterization chamber and the vacuum pump.  Results
appear negative so far for both.

When this contamination was discovered, all surfaces exposed in the
chamber were cleaned following manufacturer's instructions with pure
ethanol, and then the whole unit was baked out at 80C for 30 hrs.  This
procedure should insure that there are no remnants that volatilize again
later, since the CCD was the lowest-temperature surface inside the chamber
at the time of measurements, and it would have acted as a cold finger.

(3) We think we know the source of the "unfocused" images through the 
450 nm filter.  You will recall from my last update that the other color
filters have spot sizes of about 1 pixel, whereas for the 450-nm filter it
is closer to 2-3 pixels.  

Recall that each filter acts to correct a chromatic
aberration of the camera, using the thickness of the filter to bring light
into focus on the CCD for the center wavelength of the filter.  The aberration
is more severe over broader wavelength ranges, and at shorter wavelengths.
(This is the reason why the clear, broad filter was not going to be as
well focused.)

Our theory is that the larger blur circle for the 450 nm filter is the
result of the broadness of this filter (100 nm) and its short wavelength.
Terry Harris's ray trace program in fact predicted (which I didn't realize)
that this filter would not have as good a focus as the very narrow ones.
For a narrower 450-nm filter (50 nm), the blur circle is predicted to be 
about 1 pixel in size.

Joe has been made aware of these facts, and asks for the following course of
action.  First, Terry is now analyzing the results of the focus test to
demonstrate that his predictions about blur circle sizes are corroborated
by the test results for the other filters.  This corroboration would be hard
evidence that we have identified the source of the poorer focus in the
violet filter.  Second, he will predict what the improvement would be in 
the narrower 450 nm filter (50 nm wide).  Based on these data to be supplied
to Joe, the team can make an informed decision about what to do about this
filter.  Third, pending that decision, the engineers are going to switch
to the narrower 450 nm filter and do some quick focus tests to acquire
empirical evidence on the effect on focus of the filter width.

(4) I've discussed the issue of the scratch on the CCD at some length
with Joe.  In a nutshell, it is about 100 pixels long in the lower
left of the field.  It probably formed during the cleaning of the CCD
prior to detector-level calibration measurements.  It shows up
relatively well in these data because the parallel rays cause
shadowing.  However it is quite small, below the level of what can be
seen with the naked eye.

Ed Hawkins did some calculations and believes that in converging rays 
its effects will be spread over about 30 pixels, such that its effect 
on brightness levels will be attenuated by 30 times.  So if in the parallel 
rays it appears, say, 30% darker than surroundings, mounted in the camera
its effect will be 1%.  Right now we don't have really enough data to
determine to what extent and how it will affect images.  Joe is requesting
of Rob Gold a little time back in the OCF to image a variety of targets to
acquire more information.  I have tracked down 1 image, of graph paper,
where there is some illuminated target with a regular pattern in the area of
the scratch.  So far, nothing I can do in the way of edge enhancements, etc.,
brings out anything at all.