From: "Jerry" 
To: "'James Lerch'" ,
        "ATM List" 
Reply-To: "Jerry" 




        I suggest an alternative method that may avoid the problem.

        This is a mental picture. You do the programming. When I say
"graph"
it is for the visualization. If you graphed the light intensity returned
from the left and right side of the mirror under test, and lets say the
knife is set for somewhere between 0% and 100% zone, the left side of the
mirror would be dark where the right side is bright and bright where the
right side is dark.

        Graph with 0% zone on the left and 100% zone on the right.
Understand that I am saying that the 0% zone of both left and right side of
the mirror are on the left of the graph and the 100% zone of the left and
right sides of the mirror are on the right of the graph.

        Graphed this way the resulting lines on the graph for the left and
right sides will cross at some zone making an X. No mask is required and
probably not advisable. A band of pixels five or ten wide across the mirror
should be enough pixels at any particular zone to average to avoid problems
with hot or numb pixels skewing the measurements.

        This method could be carried out in practice by stepping
longitudinally by some distance and noting the zone that is
"nulled". Lets say you have a mirror that will have .2 inches of
knife travel from 0% to 100% zone, .002 inch steps would give 100 reference
points. That should be enough, and fewer would be ok for most.

        I would adjust exposure time to be just below saturating the pixels
with the knife completely out of the beam. Dark frames and flat field
exposures would probably improve results, but maybe not enough to make it
worthwhile.

        Keep in mind that occasionally you will have an air current in the
light path while testing just as when testing visually. So a series of
tests will still be advisable. Perhaps that could be programmed also.... to
detect the changing pixel intensity patterns changing from turbulence and
wait for it to dissipate.

Jerry

--- BBBS/NT v4.00 MP