Regarding "is higher voltage always more efficient?": It depends. In general, I'd go with what's been said above, to go with the highest voltage you can carry, with a caveat:
--if the higher voltage ends up forcing your system to be too fast to use full throttle for the majority of your riding, then controller internal losses due to switching/etc at partial throttle may cause controller heating that could be a problem, depending on your climate and usage. I've experienced this already, damaging capacitors (fortunatley not FETs) in a controller while riding on a very hot day here in Phoenix, trying to run slower to keep the motor cooler and make my battery last longer, on a long-ish Freecycle pick-up run. Doing the exact same thing at full throttle did not have the same controller heating problems (although it certainly did heat the motor a lot more, as it had more load on it, and used battery faster).
If you're in colder areas, it may never be an issue, or you can ventilate the controller, with perhaps a temperature-controlled fan.
When running a mid-drive setup, it may not be an issue at all, as you can always just gear down the motor more before it enters the drivetrain, to compensate for the faster RPM it will have at a higher voltage.
--One other gotcha is that at higher voltages, the same controller FETs will have higher losses to RDSon/etc than at lower voltages, but this inefficiency may be made up for by the overall system efficiency at the higher speed--running the motor faster but geared down more means it will probably be spending a lot less time at low (inefficient and current-hungry) speeds, before reaching speeds at which it pulls much less current to do the same work.
pdf wrote:Well, color me dumbfounded. I am not sure why people have problems with the Stokemonkey setup.
Remember that my powerchair motor setup is not quite like a SM setup, as it's not driving the pedal crank directly, but rather a jackshaft that the pedal's chain happens to also be connected to (but still no FW on either one, so both drive each other). Also, the PC motor is capable of (at least) short bursts of power in the 150A+, 3500W+ range (possibly a lot higher, but that was perhaps the only folded-chainring event that I was able to record the current spike for). I can't even recall just at this moment if that was with the lower-power 300-350watt motor, or the bigger 600-650W one.
My pantleg issues are probably more because of it's semi-recumbent seating than anything else, as there's no chain guard on it and it runs for some length next to my leg. Problem was usually with wind catching it and pulling it in just below the knee area, on the seam, even with the end of it tucked into a sock or with a tie strap.
I daresay my experience is an exception rather than the rule, but I wanted to put the warning out there just in case.