At 11:43 AM 1/16/01 +1100, Julian wrote:
>If that is true, then because you only have one sensor, you can engineer it
>to greater tolerances, and read a smaller spot size and thus get better
>resolution. Because you only have one sensor too, you can design the
>amplifier and subsequent circuitry in a more expensive way and thus get
>better performance - or at least you don't need the switches that you would
>need to read an array of sensors. And because you only have one sensor you
>don't have the problem of matching the response of thousands of different
>sensors and their associated switching circuitry etc., as you do for CCD
>array scanners.
>
>Hope this helps, or elicits more accurate information,
Correct -- there's only one sensor element in a
drum scanner, and that's the PMT tube.
But in both cases -- drum or CCD -- the data arrives
at the A/D as an analog-serial data stream.
Matching the responses of each CCD element is
the responsibility of the scanner's calibration
system, and is typically done by scanning a
white reference just before each scan. The
math is pretty straightforward, but has to be
done for each pixel, of course.
I'm actually a bit puzzled, myself, as to why
drum scanners are (were?) so damned expensive.
Aside from the rotating drum, the mechanical
considerations aren't that tough, it seems to
me. Nothing a decent South Bend lathe can't
handle. I think it's a matter of a very
limited market.
rafe b.
