[Andy Brockway]

I am glad to see that you have taken the plunge to share your design. Your design has pushed the edge for weight and gear load. I design equipment that has to run 10 years in production so I err on the side of strength. Whose design is better? A good design performs its job as intended, if it lasts 10 times longer than needed than possibly too much was spent on it. If it blows up 10 seconds after it completes its mission than it is still a success.

Having said all that, I would like to add some of my experiences and thoughts to help you have a reliable gear box.
-Do not run the FP 32 pitch and Bosch 0.7M to the same gear. The bending moment on the intermediate gear teeth as they run from 32dp to 0.7M will quickly take them right off. Those that ran a 32 pitch gear to the 0.7M actually produced a new tooth pattern.
-The bearings you are using are low grade. This means that there will be some slop as they run. Our 2003 single speed used these bearings on the output and had about a 1/16" play, measured at the end of the output shaft, at the end of the season. Consider what this may do to the 12 tooth gear when making your final choices.
-I recommend using the hex and plastic kit coupling on your output shaft and then having a self supported sprocket/gear output shaft. The reason is that the moment resulting from chains or gears at the distance you are showing from the gearbox bearings may cause movements that are detrimental to the gears. Do not just add one more bearing in-line, this is hard to align and causes a bending action in your shaft.
-Plastic gears: Team 571 ran these successfully in 2004 using Bosch and FP motors. I will be testing this fall using the plastic gear with the Bosch.
-Welding is positive. It also can cause shafts to bend. We have not used welding due to lack of equipment. Last year I used keys and Loctite #609. Loctite advertises that I really did not need the keys (see my note above about how I design)!
-My gut feeling warns me about the 1/8" plates and the small gears. The only way to truly know is to build and test. Our team fights weight every year, this may be a solution.

Build it! When all is said and done the only way to know is to try it.