Paul & Sabrina's cheap DIY 144v motor controller

[bennelson]

Originally Posted by MPaulHolmes

I need some sort of box to put the controller into. I need a good cheap supplier of aluminum fan heat sinks (to attach to the big dang aluminum heat spreader). I need to compare the h

Hey Paul, let me know exactly what you are looking for.

I have a very nice 8" x 6" aluminum heat sink with 45 very nice little tapped holes on the back of it. That's a fancy heat sink I bought, but never needed to use on my 72V Curtis controller.

Tom, who is building the 300V AC Neon has a never ending assortment of heat sinks available to him and may have a box that would be good for a controller enclosure as well.

Just let me know the exact sizes you are looking for.

[MPaulHolmes]

That 8x6 one looks perfect! Thumb's up, indeed! Nice picture, by the way.

I've been considering all the iterations of M-, B+, and B-, with one sided and 2 sided pcb.

On the power board, you can do

M-, B+, B-
M-, B-, B+,
B+, M-, B-.

Each choice has it's pros and cons, but the best way, it seems to me, is
B+, M-, B-. The smaller version was M-, B-, B+, which makes it easier to connect the capacitors, but needed those red wires, picture making it totally unacceptable in the high power case. Well, there are other possibilities, but B+, M-, B- seems to be the best.

[bennelson]

One other thing you might want to keep in mind is where the physical connections end up on the outside of the controller when you are all done.

On my Curtis controller, the big B- connection is right next to the 3 small pins for controller power and potentiometer -&+, which wouldn't be a big deal except that it's pack power going to the controller power pin.

That means that ALL PACK VOLTAGE is separated by about a centimeter.

I fried the controller power female spade connector right off the end of the wire taking it off one time. (I think power was disconnected to the main battery pack, but controller power was wired up separate, something like that)

Anyways, try to design the controller so that you don't need tiny hands and tweezers to keep from frying things when connecting and disconnecting.

[TheSGC]

And make sure that your MOSFETs are electronically isolated from the heatsink. Many people have destroy EV components because the controller B- was connected to the casing and it shorted with the car chassis.

I also have not found any mention of your freewheel diode. What are you using for the freewheel diode?

[MazdaMatt]

What exactly is a freewheel diode? Is that the method of ensuring that there is no back-current from the motor when you are coasting down?

[MPaulHolmes]

Freewheel diodes: 200v 60amp, PART # STTH6002CW

That's a good tip about isolating the drain from the heat spreader inside the controller. I was also looking into connecting the drain directly to the heat spreader, and then isolating the heat spreader from the heat sink that will be outside the controller. I think the latter would be best.

That's an awesome point Ben. I've been thinking about the connector locations a bit. Pin #1 on the Curtis, as you were saying is full pack B+. On this controller it will be +12v from the auxiliary battery. There's no reason at all it would have to be anywhere near the big B+, B-, and M- leads. The M- lead could be on the other end of the controller too. It might make it easier to have a connection to the motor on one side, and the B+ and B- on the other end. It would give plenty of space for the lugs. Also, I think I could get away with 3 pieces of copper bus bar that wouldn't have to be cut at all! How's that for open source! haha. Only holes need to be drilled in them. Everyone has a drill! Well, everyone that wants to build a controller, probably.

I found a Hall Effect Current Sensor that has a big enough of a hole so that 0.75" wide bus bar can be used. I think 0.75" is just about right for attaching lugs. The other current sensor had an opening that allowed for only about 0.5" bus bar, so that the bus bar would have to be cut from a wider piece, so the ends would be big enough to attach lugs, but most of the length would be about 0.5". I'm trying to avoid having a band saw be a part of the required tools for assembly!

[TheSGC]

A motor is essentially a massive inductor, and with PWM current is sent to the motor in Pulses, on and off, but with an inductor current always wants to flow, so during the time the MOSFETs are off, it needs somewhere to flow. A FreeWheel Diode "absorbs" the current during the time the MOSFETs are off, and these diodes need to be big, something like 1200 volts and at least 600 AMPs continuous.

Not having a FreeWheel diode is probably a reason why Paul is busting MOSFETs.

I have been developing my own controller for my EV, but had a FreeWheel diode failure and I rocketed down the driveway full reverse when the whole system blew. My diode failed and took everything else with it.

I would suggest reading up on PWM, freewheeling, smoothing capacitors and circuit theory before throwing a homemade controller into a car. While it may work on the bench, without proper caps, diodes, heatsinks, and current limiting it will probably fail in the worst way possible.

Also, I have a little side note about MOSFETs. I don't care what they are labeled as, always current limit them to HALF their rated value, and no MOSFET should go over 50 AMPs per module because that is the max the physical casing can take before it explodes, with a heatsink. Without a heat sink, 2 AMPs would probably kill it.

Just my $0.02 from me majoring in Computer Engineering.

[MPaulHolmes]

I had that exact question a few months back...

The current in an inductor has momentum. A freewheel diode allows the current to harmlessly go in a loop when a switch is suddenly open. Without them, it would destroy the mosfets when they switched off.

The motor has some inductance. Once the mosfets get switched off, the current from the motor would try to pile-drive its way through from drain to source. The voltage would be like 1000's of volts for an instant.

[TheSGC]

That diode is only good for a 48 volt 50 AMP controller, no way will it survive in a 144 volt controller. The reverse voltage can be 8 times that of the original source but the same current, so I would say a 1200 volt 600 AMP diode is needed.

Originally Posted by MPaulHolmes

Freewheel diodes: 200v 60amp, PART # STTH6002CW

That's a good tip about isolating the drain from the heat spreader inside the controller. I was also looking into connecting the drain directly to the heat spreader, and then isolating the heat spreader from the heat sink that will be outside the controller. I think the latter would be best.

That's an awesome point Ben. I've been thinking about the connector locations a bit. Pin #1 on the Curtis, as you were saying is full pack B+. On this controller it will be +12v from the auxiliary battery. There's no reason at all it would have to be anywhere near the big B+, B-, and M- leads. The M- lead could be on the other end of the controller too. It might make it easier to have a connection to the motor on one side, and the B+ and B- on the other end. It would give plenty of space for the lugs. Also, I think I could get away with 3 pieces of copper bus bar that wouldn't have to be cut at all! How's that for open source! haha. Only holes need to be drilled in them. Everyone has a drill! Well, everyone that wants to build a controller, probably.

I found a Hall Effect Current Sensor that has a big enough of a hole so that 0.75" wide bus bar can be used. I think 0.75" is just about right for attaching lugs. The other current sensor had an opening that allowed for only about 0.5" bus bar, so that the bus bar would have to be cut from a wider piece, so the ends would be big enough to attach lugs, but most of the length would be about 0.5". I'm trying to avoid having a band saw be a part of the required tools for assembly!

[MPaulHolmes]

Actually, I'm very confident that the mosfets were breaking because of a floating gate. I added a resistor from gate to source, and now it's fine. Turning on the power section before the control section creates a high voltage difference between drain and source. The mosfet was turned on because the floating gate was above 4v relative to source. Basically the mosfets got shorted, and the current limit was exceeded, causing them to fail in a shorted state.

Those 200v 60 amp diodes have a very fast response time (20 ns). I'm getting voltage spikes of only about 4 volts when I run the controller at 72v. I have been using freewheel diodes. I'll use about 10 or 12 of them in the car's version. That will allow for a current of 500 or 600 amps, and a voltage of around 144v, with voltage spikes of around 5v or so. The EVTech list has suggested the components I'm using, including filter caps, mosfets, diodes, etc..., and there are working 144v 600amp controllers made by people on that list that are using those parts.