Paul & Sabrina's cheap DIY 144v motor controller

[MPaulHolmes]

Awesome Jack. Yes, I think it's probably easier to use the intrinsic diode as a freewheel diode since it is matched to the device already. Better than going out and getting a 1200v 800 amp diode. What size gate resistor are you using? You could probably get away with something really small.

[jackbauer]

The igbt i'm using at the minute is only 600v 150a. I have 2 1200v 800a units on order from ebay. I may end up ordering more and just use 2 units as freewheel diodes with the gates tied to emitter. I'm actually using one of the gate resistors you supplied in the kit. 30R i think? I'm still trying to wrap my head around the drive equations. The 800amp igbts have a 90pf gate capacitance. So i need to figure out what current i need to swing 180pf (2 devices) on and off at 15.5khz.I think this could be the basis for a high performance option if i get the design right.

[MPaulHolmes]

Jack: Right now, for an instant it's 12/22 amps, and then drops to 0 amps once the gate gets to 12v, because the gate resistors are 22 ohms. If you use 2 igbts in parallel, I would use maybe 2.4 ohm to 4 ohm, 1 watt gate resistors. That would bring the instantaneous current draw to about 6 or 10amps from the 12 amp driver. The RC time constant of the gate capacitance would be much shorter, so you could pile-drive through that miller knee where all that wasted heat is being generated.

Also, I believe it takes 2 times as long to turn the IGBT off as it takes to turn it on, so you could have a (say) 4 ohm gate resistor in parallel with "diode and 4 ohm gate resistor in series", so that when the current goes the other way, the gate resistance behaves like about 2 ohms, so it turns off in the same amount of time as turning on. Isn't that sneaky!? Fran suggested it to me.

You would have larger spikes from the lower gate resistors, but who gives a crap, you have TONS of room.

[jackbauer]

Thats a smart idea. I'll setup with two devices on a heatsink tomorrow and run the starter a bit longer with the scope. Datasheet on the fuji 800amp part states typ ton of .32us and typ toff of .40us so not too bad. I'm wondering would it be any harm to add a snubber across the device?

[jackbauer]

Is the circuit setup for a linear or a log pot? I seem to have only a very small are of movement between zero and full throttle with a linear pot.

[jackbauer]

the following takes place between 11am and 12pm:
We're now running 3 150amp igbts on 10R gate resistors.

http://www.youtube.com/watch?v=Y_VMPA-oO3w

[sawickm]

Hi All,

Doing some research on IGBT controllers I found the “Kilovac Controller”. It was used on a 3 – wheeled EV called the Sparrow. The company has since gone out of business, but info on the Kilovac controller is still available. The “Kilovac” uses low side PWM switching of the IGBT’s, same as the ReVolt’s Mosfet switching.

I have attached a link to the Kilovac schematic: http://enviromotors.com/wiki/uploads...agsKiloVac.pdf

I found it interesting that the Kilovac uses a SOT-227 IGBT package, I am working on a Power PCB design that also uses SOT-227 for the Mosfet packages. My proposed Power PCB design is similar to Adam Brunette “Sandwich Power PCB contruction”. The Mosfet’s are attached directly to the heatsink, with bussbars, driver PCB, and ReVolt controller PCB above them. The old DCP “Raptor” controller uses that same construction technique.

- Mark

[MPaulHolmes]

Originally Posted by jackbauer

Is the circuit setup for a linear or a log pot? I seem to have only a very small are of movement between zero and full throttle with a linear pot.

It's set up for linear 5k POT. But also set up so that there will be smooth acceleration with a 9" diameter motor assuming 144v, and full 500amps (or thereabouts). It's really not the right software for practicing with a small motor at very low currents (low compared to 500 amps) and low voltages (less than maybe 72v or so).

I have no idea why it was behaving like it was in the video. I actually found this software to be completely unusable on my desktop with the oscilloscope. It acted so weird, that I didn't even bother to try to figure out what was going on. I just drove with it. hehe.

Maybe I should mail you 2 new ATMEGA8s: one where throttle is only proportional to current, and one where throttle is only proportional to pwmDuty. You are doing some awesome stuff, and I think you need some different software for the bench testing. Actually, it would be even better if you could somehow get an AVRISP make 2.

And it doesn't matter if you have never programmed before. If you were able to get that, everything else is free. All the software and development environment, etc... It would just be a matter of putting the .HEX file onto the chip. Maybe that could be a more long-term solution. I just need to mail you a couple more chips for now, but it took forever last time to get to you. Man! Ireland is far away! And I don't get paid until the 10th! Dang it!

IS THERE ANYONE NEAR IRELAND RIGHT NOW READING THIS THAT HAS THE TOOLS TO PROGRAM AN ATMEGA8 THAT CAN MAIL HIM A PROGRAMMER????????????????????? Or that could reprogram the chip and mail it back to him much quicker than I could?

I think your approach has so much potential for SO MANY DIY'ers, who may not have the tools and such for assembling a mosfet based kit. We have to see this to completion!!!!! before it's too late! What do we have now, like 22 hours or something?!

[MPaulHolmes]

Awesome Mark!!! I can't wait to read more about it! I've been checking out your links. Very informative!

[jackbauer]

Paul i noticed you have tracked the board for an icsp port. Can i build a cable and program the device in circuit? If not i can build up a simple programmer on a piece of stripboard like this:
AVR Programmer

I think my problem is the starter is not drawing enough current to cause the micro to recognize the lem so it just keeps increasing the pwm duty. I have another monster forklift motor that i call the "warp13" which i'll use for the next batch of experiments when i get the more powerfull igbts. That will draw enough current. I am quite good at pic micro assembler code just never made the leap to c. Are the atmega devices flash or otp? The clock is ticking!