It should be simple enough to model,
Yep. Was no problem to feed the information into a plug-in and get an accurate gear.
...unless you have a high end one with a small head ....
You can get a 0.1mm nozzle for most heads relatively cheap (<$20). A good hot end is only about $100, so it can be done on a small budget.
...only able to use two types of plastic...
A high-heat hot end will allow a lot more plastics. I've heard of people using nylon, faux-wood, conductive ABS, HDPE, etc. Again, it just takes the right extruder set-up.
The major thing that makes this impracticle is printing time. To get the resolution you would need (0.1mm width nozzle, 0.1mm height) takes an incredible amount of time, even for a small part . (Figure about 5 hours per gear) That's the trade-off: higher quality == more print time
I'm talking about sending it off to Shapeways.
You would almost have to. Be warned, though: That's probably why they are $20 a gear - shapeways is expensive if you want high-quality and low volume. We checked for the base plates, and it was outrageous.
......................................
<off-topic>
I have a Makerbot...
Which one? The machine we're using is based on the v2, with a steel frame, iirc. I modded it for all bearings on x-y carriage, and fishing line instead of the belts, so it's got good accuracy and reproducability. In the process of upgrading to an e3d end, so hopefully I won't have as many ruined prints due to clogging. Unfortunately, that mean upgrading the power supply, and possibly the controller board. We will see how it goes.
My ultimate plan is to add an azteeg controller with the 1/32 stepper drivers and swap out the motors for ones that can actually do that micro stepping. Then maybe I'll try the gears again
Of course, by then there will be something better to play with
(Now, does anyone know of a good, reliable, cheap injection molding set-up?)