@paul_wallich the Sharp sensors are just that (they include an IC with compensation stuffs). You can get them either as a digital (on/off) version or as an analog one, and they are surprisingly tolerant to different surface materials, they are even reading some translucent materials correctly. Even if it’s not too accurate.
The sharp IR sensors are actually a bit more clever than just measuring the amount.
http://www.acroname.com/articles/sharp.html http://developer.mbed.org/forum/mbed/topic/1048/?page=1#comment-5065
But even the inductive sensors need ‘calibrating’(By adjusting the distance the sensor triggers relative to the nozzle) as different built platform configurations change the distance.
@Electra_Flarefire
Wow, thanks! That sounds as if you could indeed calibrate pretty closely. If the internals allow, you might also be able to mess with the alignment of the two parts and get more useful close range.
I did a little messing around with a sharp sensor I had lying in the basement, and it looks as if it might be plausible, especially in the high-slope region right around the maximum output. With a max count of about 600 (3v nominal) I was getting a change of 100 counts with roughly 1cm of motion, which would give you roughly (ahem) on the order of 0.1mm. The optics can definitely be messed with, so an optimized setup might get you down to half that. It’s not small, but I don’t know that it really needs to be if properly positioned.