My last post sounds like I might have been trying to attach you, DaveyJ. I wasn't, I was just conserving words because I wasn't on a keyboard.
So did you get a D800, then? Congratulations if you did. It has been a bad year for farmers, hasn't it? Where I live we've been in drought for a while now so it's kind of just business as usual, but I guess the midwest has been hammered by this dry heat.
After reading the paper carefully now (I thought it was fun), they didn't explain their definition of quantum efficiency. So I thought I would explain my quantum efficiency quote now that I'm on a proper keyboard. Feel free to not read this. Most people define quantum efficiency as the number of photoelectrons created per photon. I've posted here about photoelectrons before (http://nikonrumors.com/forum/topic.php?id=23&page=4#post-56114 ), and I'll paste in part of that:
Here's an analogy that might help someone understand. Someone sets up a lot of beautiful women (applied voltage) along one wall in a room (a silicon diode) full of sleeping men (electrons). You take off the roof (shutter) and let golf balls (photons) fall from the sky (outside the camera) into the room. Some golf balls miss but others hit the men and wake them up (excite them across the band gap). Some of the men that are awoken notice the women (positive charge) along the wall and start drifting that way to get closer to them. Once there are so many men that have gone to the female wall that the newly awoken men can't see the women and so don't have any reason to move, they just start going back to sleep (this is called saturation of the photodiode). The roof (shutter) is closed, the number of men on the wall initially lined by women are counted (Analog to Digital Conversion), and it gives you an idea of how many golf balls were lobbed into the room.
What I didn't mention then is that the falling golf balls (photons) actually have enough energy to promote more than one electron across the band gap—in other words, some can bounce off of one guy and wake him up and then bounce off another and wake him up, too. To give you a feel for this, visible light photons have roughly 2 or 3 electron-volts of energy and the bandgap of silicon is 1.1 electron-volts. That means a blue light photon, which has ~3 eV of energy, can create two electrons to be counted. So a QE of 50% for blue light would actually mean that the sensor absorbs 25% of the photons but because they each create two photoelectrons, you get an average of half an electron per photon that hits the sensor.
This still isn't the whole story (is it ever?), but it gets you closer to being able to understand quantum efficiency. For those who actually want to look at that graph in the paper, to get the energy in each photon of a given wavelength, you take 1200 and divide by the wavelength in nm and it's very close to the number of electron-volts per photon. Red (~650 nm) would have a little less than 2 eV per photon, then, and blue (400 nm) would have almost exactly 3 eV.
By the way, thanks for your kind words, Squamish. I'm sure some people just wish I would shut up.