Paul & Sabrina's cheap DIY 144v motor controller

[snakub]

I will post pictures of the soldering and the current sensor but i have it on M- and the white part is facing outwards toward the double O wire from the motor.Is there anything else that could be wrong besides the NAND?

[MPaulHolmes]

It sounds like the LEM was hooked up right.

Possible things to cause overcurrent not to work:
D3 in reverse. D3 bad. R12 not 10k, R15 not 330 Ohm, R11 not 1k, Bad U8 (comparator), U8 in reverse, D8, D9 in backwards or bad, R9 not 2k.

These would be really important: R14 not 750 Ohm 1%, R18 not 3k 1%. Not 500 Ohm pot for VR1.

[jackbauer]

Just looking at the datasheet for that transistor. Its an npn darlington not an igbt or fet. I would imagine (not sure) that it would require a different driver stage. Also it looks to be quite a slow device in terms of switching ability. What frequency are you using on the arduino? The revolt default is 16khz which i would say is too high for those parts.

[snakub]

Thanks paul I will check over all those parts tomorrow and here are the pics for you I just couldn't get a good pic of the soldering though the light just wasn't right but I assure it is adequate.

Just looking at the datasheet for that transistor. Its an npn darlington not an igbt or fet. I would imagine (not sure) that it would require a different driver stage. Also it looks to be quite a slow device in terms of switching ability. What frequency are you using on the arduino? The revolt default is 16khz which i would say is too high for those parts.

Hmm I had it up to 8 k at one time and it worked ok but 1.5k was the normal driving frequency I used. When I had this car up on jack stands I was testing the throttle for the revolt to watch the wheels turn it seemed to keep up with the switching time it turned the wheels. Just when I put the car down it did some funny stuff. It seemed to pull so much current that there was no electrolyte left in my batteries I had to refill them today!! How could it have done this?? Vaporized the electrolyte?? The data sheet says 3us for switching time whatever that means maybe you can elaborate?

[jackbauer]

Regards the batteries i have no idea thats being honest. Something ain't right. 1.5khz would be fine for those devices. Anything higher will cause the waveform to become trapezoidal. 3us is the time it takes the device to go from fully off to fully on with the right driving current into the base. My old igbts switch in half that time. Remember also that a darlington is basically a current driven current amplifier whereas a fet or igbt is a voltage driven device. My guess would be that the revolt driver stage does not like the transistor and vice versa.

Spinning the wheels requires only a very small current so you "get away" with a non ideal driver. when you then load the motor and it draws more amps the fun starts. my 2c.

[The Toecutter]

I had thought the batteries were the problem at first; we found one bad one, removed it from the string, and tested the car on the lift with the bad one removed, starting at 24V and gradually stepping the voltage up. In the lift, with 156V, the ammeter read ~20A just to spin the wheels.

The batteries are flooded Exide Nautilus type.

The Arduino board gave no problems and the truck drove around fine on that using the same exact power stage setup.

Aside from one bad battery that he removed(it was originally 168V nom), all of the other batteries didn't even register a resistance on the multimeter(it only goes down to 0.1 ohm), while the bad battery registered ~6.5 kOhm. Removing the bad battery took us to 156V nominal. I then thought we had solved the problem.

Before the final on-ground test, their water level was fine, and the 156V pack was reading ~160V. We even checked all of the connections before running the truck off of the lift, along with the internal resistance of each battery.

When we used the ReVolt control board with the truck on the ground, it reversed back a few feet just fine. Forward in 2nd gear, the truck lurched for about 1/3 of a second with very light throttle and the IGBTs blew up. There was no spike in the ammeter reading(it probably didn't react fast enough). The batteries quickly gave off a smell of rotting eggs once the IGBTs blew. Acid/water sprayed all over the battery box.

[Weisheimer]

Originally Posted by The Toecutter

The Arduino board gave no problems and the truck drove around fine on that using the same exact power stage setup.

Those darlington modules require far more POWER to drive than FET's
or even IGBT modules with a similar power rating.
Perhaps as much as 80 times more power!
They also have "turn off" issues due to their tendency to saturate.
A "baker clamp" in the drive circuitry would help with that.

Still, there are other people using that module, and you've been driving
with them for a year, so the Revolt controller can be made to work with them.
It will take a bit of driver circuit fiddling to get there.
Edit: Also might want to slow the PWM down to 1.5 khz

Does the power stage still use 4 or 5 of these modules?
The Revolt controller might drive a single one of those modules directly with it's driver,
but it won't drive any more than that without adding additional drive circuitry.

What exactly is the driver circuit when driven by Arduino?
Is it still Solid State Relays? and if so, what is the exact base drive circuit?

Does the Revolt controller drive SSR's or is it driving the modules through
a base drive resistor?
What is the base drive resistor value, type, and power rating?
It should probably be on the order of 1 or 2 ohms, non-inductive, and 10 watt power rating.
There should be a separate base drive resistor for each module.

Clearer pictures and schematics would be helpful if that is possible.
It does look like there might be one substitution cap
used in the build of the Revolt controller, though that is not very
likely an issue here, only a very minor detail.
These old eyes just can't make out much else because of the focus.

Did you check the modules to see if the there was a short to the case after they failed?
From what I have been able to find, the truck build looks well done.
Are there "solid state rated" fuses in the battery loop?
If so, what is the current rating? Did it blow when the batteries went thermal?
I am at a loss as to where all of that battery current went.
It certainly did not flow through the Revolt control board or it would
have been vaporized!
And was it the power modules that failed?
FETs and IGBTs have been cited as blowing, so I am a bit confused as to
whether a driver component failed, or did the power module(s) fail?

Do you have a battery resistance meter, or were you measuring the
resistance of the batteries with a multi-meter?

Yes, a lot of questions.

I'm thinking that the answers will allow us to help with the solution.


Mark

[thingstodo]

Originally Posted by The Toecutter

...
The batteries are flooded Exide Nautilus type.
...
The batteries quickly gave off a smell of rotting eggs once the IGBTs blew. Acid/water sprayed all over the battery box.

I read through the posts, but didn't catch what your maximum current draw was expected to be. I don't know where all those amps went, but to boil the electrolyte in 1/3 of a second, it sure went there FAST! 156V to 80V in 0.33 seconds, 12V to 6V on the batteries. There had to be a dead short there, or as close as the small resistance on the connections would allow.

Fuses above 250 Amps seem expensive, but not compared to the devices they protect. The commercial and industrial controller/inverter designs that I've seen (read taken apart because they failed) ALL have fusing. I know that the current-limiting hardware was intended to be better and faster than a fuse could be, but can you REALLY be too paranoid?

There was a mention of Electrical safety standards, Arc-flash, and NFPA 70E in one of the other threads. One of the easiest ways to limit the current (and the arc-flash) is to put in a fast-acting fuse. This is not specific to DC, but a DC arc does not stop by itself. The DC Arc, once it starts, keeps making gases and plasma until the something stops the current or all of the metal connections have changed from solid to gas (at least, that's the working theory last time I checked)

Has anyone installed DC fuses before the ReVolt? (or any of the other controllers, I guess)

thingstodo

[MPaulHolmes]

I use a 300amp slow blow DC fuse in series with the battery. It doesnt' blow up with brief jaunts above 300 battery amps. They are only like $4.99 on ebay.

The arduino was switching at 1.5kHz? One possibility is that they were spending too much time in the transition between off and on at 16kHz. And under load that would be a ton of heat. Also, the gate resistors that are on the wiki are 22 Ohm. Probably way toO big for this. I tried 10 Ohm resistors with some IGBTs and they almost caught on fire. The fact that the hardware overcurrent didn't stop anything was probably because the overcurrent circuit pulls the gate (oops, base) low when an overcurrent event occurs. I think the transistors were already failed shorted by the time the overcurrent started happening. The controller has no safety built in for failed-shorted transistors.

[bennelson]

I'm also using a 300-amp fuse.

(Mostly because they are cheap. You can go over 300 amps briefly. I would have to PAY GOOD MONEY to buy a 500 amp fuse. 300 amp and less can generally be scrounged. I think I have blown fuses twice.)

Yep, you always want to have a good fuse around, especially high-power stuff!

Another possibility is a high-amperage DC-RATED circuit breaker. They are harder to find and more expensive. If you scrounge one though, great! You never have to replace a fuse that way!

(You could also have it under your hood and run a hood-release type cable to it, now you have a battery-disconnect/emergency stop!)