[EricVanWyk]

Ether helpfully pointed out that the first sentence of my previous post makes a few too many leaps. The remainder of the post supports only that higher frequency motor controllers produce less waste heat. This in turn is based on the premise that torque is proportional the algebraic mean of current, but power is proportional to the geometric mean.

The comment on using Current Mode was unsupported, so let me take a whack at that now. If a PID is not well tuned, it can oscillate. These oscillations can create efficiency losses just like having a low frequency motor controller can. A pathologically mis-tuned loop can have efficiency as bad as a Victor.

Using CAN allows the Jaguar to do the PID itself, and runs at 1kHz. Running the PID on the cRIO introduces some communication time, so the loop runs a bit slower.

Using current mode removes several variables from the equations, and makes the effective loop quicker: The loop isn't waiting on anything mechanical, and can therefore respond on an electrical time line.

It is quite possible to achieve the same stability with other control modes, I just find it easier in current mode generally. For your application, CAN Position Mode might be easiest.