Digital BW, The Print

Again I'll say it, check the details on the Gretag iSIS XY scanner. 
That is the closest thing to what HP has installed. The LED industry 
has certainly taken off in the last couple of years, output is up by 
a huge factor, so is the spectrum that can be covered. I don't 
remember seeing a single high output LED that covers the whole 
spectrum, but with 2 high output LEDs, you might be able to do it. I 
think that is what Nichia suggests, two LEDs with certain 
characterists. Remember that you also need a certain amount of light 
to read the darker patches to get the signal to noise high enough for 
accuracy. This goes back to the dmax greater than 2.5 readings that 
are sometimes claimed. I would have to look up the exact amount of 
reflected light at 2.5D, but it is extremely small (I'm thinking 
around 0.01 percent reflected but could be wrong). It takes a lot 
more light or designs that minimize noise is the sensor to resolve 
much lower than that. And that might be the reason for the longer 
time on the darker patches, the LEDs may not have the same amount of 
light output as a Xenon lamp used in just about all the older spectro 
designs.

Going back to the Datacolor instrument, the reason they were probably 
able to cut the number of colors is from the increased spectrum in 
the LEDs that they did use. That's just speculation, and I would love 
a real technical breakdown of how and why, with differences compared 
to the old Colormouse CM2C (if that doesn't break any trade secrets). 

Probably best to handle that part off list. In fact, we are probably 
getting close to getting yelled at for getting too far off topic 
already. Maybe we should move this somewhere else, I know Andrew 
Rodney has looked at the iSIS, and that he posts at DPReview, so 
maybe I'll start a thread over there.