Re: To David Soderman: A Little Test

[Kennedy McEwen <rkm@xxxxxxxxxxxxxxxxxxxx>]

In article <FAEBJHPJNNGCAGDNGLNPKEAEEFAA.gsellani@accesscom.com>, gary 
<gsellani@accesscom.com> writes
>You need to consider that the LED illumination could have variance. If you
>have ever tried to design a dimmer for a CCFT, you would learn that they do
>not dim very well, while a LED has an exponential element to its light
>output versus current. You can send analog signals over led/phototransistor
>setups, something commonly done in the PCMCIA modem cards.
>
>I see no reason for the 3 line CCD to have more variance. The filter over
>each line is stable, the CCFT light is stable and there is no effect on
>position accuracy since the film is uniform in both dimensions.
>
>
Certainly the LED illumination can have noise, giving rise to a 
nonuniformity along the direction of the scan - but that can be designed 
out by using a sufficiently low noise drive.  The nonuniformity between 
CCD cells - in offset and response - is pretty much fixed and cannot be 
removed by design.  Instead it must be corrected, by use of calibration 
areas on the scanner.  No two CCD elements are identical in their 
response and offset and no two CCDs are identical in their distibution 
either - a variance of variance effect.

It was incompetent negligence beyond belief that a Nikon engineer 
expected the same calibration of one CCD to apply to all three to an 
accuracy of more than 14-bits!  (Since this is a systematic error 
extending over complete scan lines the correction necessary to ensure 
that mismatch is imperceptible is considerably greater than the noise 
floor of the scanner).

I still have trouble accepting that this aspect of the design wasn't 
done by the Nikon tea-lady!    :-)
-- 
Kennedy
Yes, Socrates himself is particularly missed;
A lovely little thinker, but a bugger when he's pissed.
Python Philosophers
-
Turn off HTML mail features. Keep quoted material short. Use accurate
subject lines. http://www.leben.com/lists for list instructions.