One question I do have is how rapidly can the torque command be updated?
I’m planning on programming a full vehicle simulation mode whereby the load on the test motor is varied to match that of the rear car (including inertia effects, and the load vs speed profile).
I can’t imagine I’ll need to update this faster than say 5Hz as eco-marathon vehicles run at very low accelerations to minimise losses.
In excess of 1kHz via USB, a little less via ASCII RS232, probably >1kHz via native RS232 or CAN 
Is that one of those Sensor Technology TorqSens transducers? I used one on a similar rig
don’t break it! 
Sweet, should have no issues then (in theory 
).
It is yeah. It’ll be treated like a new born child believe me.
Did you find anything from using/implementing it thats worth sharing?
Yeah, be careful about couplings, mounting and alignment. Any radial force due to mounting misalignment will be seen as a spurious torque reading.
That’s why we ended up using the expensive MDW double-disc couplings that I mentioned earlier. They can take quite a lot of misalignment and are extremely stiff, which we needed for our measurements (torque ripple & friction of a motor and gearbox)
Also, their software was pretty horrendous. Written in some kind of LabView/VB/ActiveX abomination. 
But they do have an API, which for me was easier to use.
We also had some ground loop issues on the USB, which we solved worked around with a USB isolator.
The sensor itself is very nice though. Almost zero friction, completely non-contact “Surface Acoustic Wave” thing. I think it works by searching for the resonant frequency of the rotor element, which changes depending on load.
Don’t over-torque it though. I’d advise to fit fuses in line with your motor phases, at a current corresponding to a torque equal or below the sensor’s full-scale torque. IIRC the sensors can withstand no more than 3x their full-scale torque before they are permanently borked, and that applies if they are on or off.