[Ken Patton]

Adam-

You are right to be doing some detailed calculations like the JVN calcs as a way to figure out the potential performance of your drive system. Even if you don't end up using the design, the calculation exercise is still educational. So don't give up on that approach - just keep refining the input assumptions as you get more info or data.

My first inclination is to suggest that your assumed efficiencies (you quoted 60% overall, I think) are too low. This definitely will drive your top speed to a lower value and your currents (at a given speed) to a higher value. The efficiencies posted earlier in this thread are pretty accurate, and if you do a good design job (rolling elements, get rid of sliding contact wherever possible, minimize misalignment under load) you can hit the high end of those values.

One other source of efficiency loss is the tire-to-carpet interface. This is maybe a 93-95% type of number. Soft sticky tires are lower (90-93% maybe).

I think a good system with a simple spur gear drivetrain (no planetaries, no belts, just gears and chains and wheelchair wheels) should be able to be in the mid-80's efficiency-wise.

Advanced topic:
For people who are doing more than a closed-form calculation (i.e., they are using calculus to derive a simple equation that they can numerically integrate - the type of stuff you would do in 1st year of engineering school) I would suggest adding a loss term that is a constant with respect to speed. For a *very* efficient gearbox I use a 0.2 Nm spin loss applied to the output of the gearbox, and a 12 N drag loss applied to the robot mass.

Hope this helps,
Ken