Paul & Sabrina's cheap DIY 144v motor controller

[MPaulHolmes]

I'm using 200v mosfets, and at 500 amps, the voltage spikes are around 13v. So, for a 144v battery pack, the spikes go up to around 157v or so. That's with a power section that consists of a capacitor bank from B+ to B-, freewheel diodes from B+ to M-, and mosfets from M- to B-.

[princeton]

Originally Posted by MPaulHolmes

I'm using 200v mosfets, and at 500 amps, the voltage spikes are around 13v. So, for a 144v battery pack, the spikes go up to around 157v or so. That's with a power section that consists of a capacitor bank from B+ to B-, freewheel diodes from B+ to M-, and mosfets from M- to B-.

Very interesting. Based on my testing with a small inductive load, I would have expected a lot higher voltage spikes. Maybe the low resistance of the motor is dampening the spikes? Would you be able to post the O-scope waveforms? Also, what type of motor are you using? I assume it is under a load if you're pulling 500 amps? Is there any way you can measure the peak current through your diodes? (Maybe putting a fractional ohm resistor in series and measuring the voltage drop of the resistor?) I'm trying to decide between mosfets and IGBT's for my controller. If I can safely keep a comfortable margin between voltage spikes and max source-drain voltage, I may go with mosfets. If there's any question of a safety margin, I'll go with IGBT's.

P.S. -- It's great to have someone with an actual motor for testing. Thanks for your help.

[ElectricZX2]

Originally Posted by princeton

I am sorry for your loss (said with head hanging low).

I doubt it will cost too much for a few mosfets to replace the blown ones. How are your soldering skills?

I can sweat pipes . I don't have a ton of experience soldering small parts but do have experience soldering wires, rc components etc. I've been kind of afraid to break the seal on the controller so far hoping it would somehow magically come back to life...but I'm going to have to bite the bullet I guess. Thanks for your remorse!

[jyanof]

Originally Posted by princeton

Very interesting. Based on my testing with a small inductive load, I would have expected a lot higher voltage spikes. Maybe the low resistance of the motor is dampening the spikes? Would you be able to post the O-scope waveforms? Also, what type of motor are you using? I assume it is under a load if you're pulling 500 amps? .

http://ecomodder.com/forum/showthrea...tml#post122874

See the post above for scope video. the motor is an impulse 9. I also set up a peak-voltage-measuring device (a diode and cap in series across Drain and Source, and voltmeter) and tested on the road at 500A. That's where the numbers that paul quoted came from.

The spike is low due to the physical arrangement of the parts and the modest turn-off time of the mosfet. We're driving the gates with 20ohm gate resistors. The slower you turn off the mosfets, dI/dt decreases, more time is available for the diodes to forward-conduct, and the voltage spike decreases. The downside is that switching losses go up and the mosfets run hotter. It's a fine balance of heat generation and voltage spikes.

The voltage spike is a function of current, (parasitic) inductance between the diode and mosfet connection, (parasitic) capacitance of the drain pin relative to source (make sure you isolate the back of your mosfet from something that could increase the parasitic capacitance, like a large heatsink), mosfet turn off time, and diode "turn on time".

One method of combating voltage spikes is using snubbers. In some applications, they sacrifice physical arrangement of the parts so they can fit everything in a smaller package. This results in large voltage spikes that can be mitigated with snubber circuits. I'm not sure if it'd be practical to use them in a high power application.

Search for RCD snubber circuit for details.

[MPaulHolmes]

Oops, I guess it was 20-25v spikes in the case of the oscilloscope. and around 13v for the peak detector.

[princeton]

Originally Posted by MPaulHolmes

Oops, I guess it was 20-25v spikes in the case of the oscilloscope. and around 13v for the peak detector.

It sounds like the margin of error might be getting a bit slim if using 200V mosfets with 144 battery voltage. 144V + 25 = 169 V. It's still 31 volt leeway to take into account an errant spike here and there, though. Does anybody know if there have been any problems with the 200v mosfets blowing? I'm considering 250V mosfets vs IGBT's. Anybody know anything about paralleling IGBT's?

[jyanof]

Originally Posted by princeton

It sounds like the margin of error might be getting a bit slim if using 200V mosfets with 144 battery voltage. 144V + 25 = 169 V. It's still 31 volt leeway to take into account an errant spike here and there, though. Does anybody know if there have been any problems with the 200v mosfets blowing?

It's worse than that... fresh off the charger, my 144V nominal pack of floodies drop from ~178V to about 162V. They'll stay there for hours. However, one "advantage" with floodies is the heavy sag they experience under load. For my setup, a 2nd gear start will peg the motor amps at 500A and the battery amps at about 70A. That 70A from the batteries will drop the pack voltage about 7V. So, 155V + 25V = 180V peaks.

So, I think we can safely say that the controller will survive 500A output with an actual pack voltage of 155V for brief periods of time. Once the surface charge wears off, I typically run somewhere between 126V and 150V depending on current draw and state-of-charge.

I'm at 7k miles on the controller and just over a year with no problems with the power electronics. (i'll attribute the most recent problem a few pages ago to assembly procedures and structural hardware).

[MPaulHolmes]

Originally Posted by jyanof

(i'll attribute the most recent problem a few pages ago to assembly procedures...).

Quit picking on me! hahaha.

[princeton]

Have you incorporated a way of making the mosfets turn off slowly without affecting the turn on time?

[MPaulHolmes]

You could do that no problem, but whether they are turning off or on, if they are in the transition phase, it's bad for them. They get really hot. You could use a "diode and resistor in series" in parallel with another resistor to make the turn off time longer, but it's already about 2 times longer than the turn on time anyway.