FWIW, I've been spending some time staring @ the cumpooter, playing "what if" scenarios based on the MGR. Since there are no real tests ( like ORNL's ) on the MGR, I'm trying to accurately simulate the performance, based on the 2010 Prius data.
The big question in my mind is "what if you don't have 650V to power this thing?" For example, my 48 Nissan Leaf modules will put out 393V max.
I know there are some big compromises involved - there was a guy who put one of these into a Saturn and was not happy with the results. From what I know now, it was pretty much doomed to fail. Not to rip on him - he certainly forged new ground with the conversion; it's more to learn and hopefully make a better setup.
1) He used a relatively low voltage battery pack - something less than 300V.
2) He used a standard motor controller designed for BLDC motors.
I'm not sure if he had control of field weakening - since he used a good quality new controller, I'll assume he did. However, the controller did not give the ability to control reluctance torque, which can increase the output in the MGR by about 60%!
Even with field weakening and control of the reluctance torque, the numbers ORNL found for the 60kW Prius motor when using a 225V "link" voltage are pretty pathetic.
The "link" voltage is basically the DC bus voltage - in our case, the battery pack voltage. Their tests ( graph on pg 66 of the 2010 Prius test ) show a maximum torque of 120 N-m (88 ft-lb) and a maximum speed of 5000 RPM. The constant torque section ends at 1000 RPM.
So, the maximum power output at 225Vlink is 88ft-lb*1000RPM/5252 = 16.75 hp. I know this doesn't make sense - you have to convert the link voltage to 3phase rms voltage, then the #'s start to work.
You can't simply increase the current to make up for this. These tests were pushing the stator windings to their limits - we're talking about 200A - and that can only be sustained for 10 seconds while using 10*C coolant! Obviously, the passively cooled MGR is going to require serious cooling improvements before getting close.
I'm getting close to something that will allow me to put in different link voltages and calculate different torque-speed values. Hopefully this will allow good motor control and help with some realistic expectations.
-e*clipse