[MetroMPG]

Another thought, Tom:

Since the PWM controller isn't 100% efficient, another means of obtaining yet more speed (if you frequenly drive with your foot on the floor) would be a controller bypass. Based on controller efficiency figures I've seen tossed around on the EVDL, I think you'd see at least another 10% speed increase. Plus you could still use field weakening on top of a bypass circuit.

The challenge would be setting it up in such a way that didn't damage the controller from back EMF on switching the controller back into the circuit.