drchrisbarrett@xxxxxxxxxxxxxxxxxxxx writes:
<< For this ray to 'miss' the CCD well I can calculate the depth
the well would need to be. It is Tan(26)*8microns, or about 4 microns
deep. I'm going to check with my colleagues (who design CCDs for our
applications) but I think they are going to say this a most unlikely
structure.
>>
I am not sure it is that unlikely a structure as the surface layers need to
include quite a few layers including a passivation layer, at least one
metalization layer to bring in connections (possibly 2) at least one layer
for the filters and possibly three seperate layers (1 for each color) and
maybe even passivation layers between each of those. In a a non RGB sensor as
is used in a dichroic beam splitter design the pixels are proportionately
larger and don't need all the filter layers so the signals are larger by a
factor of 3 and they don't have such deep wells. For analytic applications as
your coleagues are probably using (?), the same thinner design applies,since
they don't need on chip filters.
When calculating these things the pixel sizes are reduced also by the
metalization area. A typical value for CCDs seems to be at least 20%. It
maybe that as the pixel size decreases this loss also increases since the
metalization needs to be some minimum size for RC speed/current etc reasons.
There is a long Kodak paper on this. Here is the abstract:
The Advantages and Disadvantages of Small Pixels
by Russell J. Palum
of Eastman Kodak Company, Rochester, New York
Abstract:
Digital cameras whose imagers are comprised of small pixels provide the
following advantages when compared to cameras whose imagers are comprised of
large pixels.
Small pixel imagers use shorter focal length lenses than large pixel imagers
to obtain the same magnification.
Short focal length lenses are generally faster than long focal length lenses.
Short focal length lenses yield greater depth-of-field/focus than long focal
length lenses.
Shorter lenses and smaller imagers greatly influence the compactness of a
camera.
A small pixel imager requires less silicon than a large pixel imager requires
for a given resolution.
The one disadvantage of using an imager with small pixels is that the optics
of such a short focal length camera requires very tight manufacturing
tolerances. This factor may outweigh the benefits mentioned above.
There is an absolute lower limit, based on physical optics, for the size of a
pixel used to capture an image in visible light. If one specifies two pixels
per optical spot size and uses an f/0.5 lens to produce an optical spot size
of 0.5 microns, that pixel size limit is 0.25 microns. (An aperture of f/0.5
is the smallest possible and probably not available at any price, however.)
A second limit is based on the noise caused by variation in the number of
photons reaching each pixel. Imager noise caused by thermal effects and/or
clocking is ignored.
A single expression for determining a "best" pixel size has been developed.
It includes illumination level, angular resolution, lens aperture, and
exposure time. This expression can be used to optimize system performance and
cost by evaluating the cost of focusing tolerances, other camera parameters,
silicon costs, etc.
Regards,
Tim Hughes
>>Hi100@xxxxxxx<<
< This message was delivered via the Olympus Mailing List >
< For questions, mailto:owner-olympus@xxxxxxxxxxxxxxx >
< Web Page: http://Zuiko.sls.bc.ca/swright/olympuslist.html >
|