[MPaulHolmes]

I wasn't using the FetchRTData() at all at the moment. And thingstodo is right. I just blocked out all the throttle code during the debugging just so he could eliminate the "throttle out of range" error that was being caused by a bad POT.

I"m just about done with a simple to assemble (only a hand drill required) TO-247 all through-hole (with optional 1206 surface mount) AC controller that should be good to about 50kW assuming you can push 100amp through a high end TO-247 mosfet. And it has a boost front end. The cost (excluding inductor) should be around $250 or so for everything. The control board is 2 layer (phew!) so it will be a lot cheaper. No wiring really. The TO247 mosfets or IGBTs solder directly to the driver board AND the power board and the driver board solders (or gets plugged in) to the control board. At the moment I have it as AN0 = CH1 = resolver 1, AN1 = CH2 = current 1, an2 = ch3 = current 2.

Then, the all flexible CH0 can be
AN3 = index/resolver2
an6 = brake
an7 = temperature
an8 = throttle

So, it's a commitment to do resolver1 even if you aren't using a resolver. But that's OK. brake, temperature, and throttle are all "slowly" changing quantities. You could do "an0, an1, an2, an3" 7/8ths of the time, and then, 1/8th of the time switch between an6, an7, and an8. something like that.

I have never had any trouble changing ch0 on the fly (I've done it quite a bit on other projects for my job). I think it's designed for that. Similarly, you can change the PWM period every single cycle without any glitches too. I would be interested in knowing if it could cause a problem. I know that it's pretty standard practice to change A/D channels on the fly. The brushed DC controller does that and hasn't ever had any problems, and that's with the lowly 8 bit atmega168.

It's meant to be a ghetto version that just gets the boosting done, but isn't a total solution for everything. There's no voltage monitoring. Just a pot that you set the multiple of the input voltage. The extra pwm can be run at a higher frequency, since it's a different microcontroller. It's just a dspic30f2010 or something like that, so only a couple extra dollars. And it doesn't need a crystal, because it has no communication between it and the main microcontroller. The only message is a low signal from the main micro to the small cheap micro means there is a fault somewhere. At startup, the moment the fault clears signals the message to the small micro that it can go ahead and ramp up the boost voltage to the multiple specified by the POT. If there's a hardware fault, the boost and all the phases get disabled anyway, and the AC motor starts to freewheel, adn the bus voltage just stays where it was before the fault (so no sudden open circuit causing the boost to destroy anything).