DIY Electric Car Forums banner

Open Revolt with IGBT driver blew IGBTs

46K views 100 replies 14 participants last post by  Stiive 
#1 ·
Hey guys, for the past few months I have been putting together an Open Revolt controller with a VLA500 IGBT driver and 3 Fuji 2MBI300L-060 IGBTs. I've read about a few other people doing the same thing and I've read the documentation on driving IGBTs and minimizing inductance. However it seems that I've done something wrong because today we tested the controller and 2 or 3 of the IGBTs blew. We first tested it with no load or battery hooked up, just to test the IGBT driver. The signal looked fine, with the low Base-emitter signal being -8v, and the high being 16v. We then hooked it up to a 12v battery and a 30-ohm burner, and it ramped the current up and down just fine from 0 to .5 amps. I did not notice anything weird about the gate drive signal on my oscilloscope. Then we hooked up our 144v traction pack and when we opened the throttle up, a high whine and then a pop was heard, and current went up to about 7 amps through the burner.

Here are some pictures if anyone can spot any stupid mistakes I might have made. The gate resistor on each IGBT is 5.6 ohms, as per the IGBT spec sheet. Unfortunately, I didn't bother to implement any voltage clamping diodes on the gate wires. Could that have been my mistake?






Any advice or comments are appreciated! Thanks!
 
See less See more
4
#2 · (Edited)
This might be not very important but shouldn't be the current sensor on M- busbar? Looking at pictures it's on B-, please correct me if I'm wrong. It doesn't sense current flowing trough freewheeling diode (E1->C1) during switchoff phase of PWM.

edit: it's hard to see, are E1-G1 connected on each module? are gate resistors parallel to g-e?
 
#3 ·
Thanks for the response!
Maybe you are right. I just assumed that the current sensor is supposed to measure battery current.
As for the resistors, there is a 10k ohm resistor between E1 and G1 and another one between E2 and G2 on each module. The 5.6 ohm gate resistors are between the driver and G2 on each module. The E2's are all connected to the driver.
 
#14 ·
I almost hate to post because I won't be able to follow this thread much over the next week or two...

The E1-G1 needs to be a jumper wire to keep the FWD side IGBT turned OFF.

The wiring needs tended to as well. Your drive signal from the control board to the driver board is hanging right over the top of your bus bars.
You might end up with an oscillator :eek:

The common reference signal from the control board to the driver board appears to be one of the twisted pair wires used to drive the IGBT.
It should be from the control board and twisted tightly with the PWM FROM the control board.

The current sensor should be on the M- bus bar.

Your link cap is a decent film cap and should work fine.

You are almost there! (and I think that there is some sort of rule that you have to blow stuff up first anyway...)
 
#4 ·
Unfortunately, I didn't bother to implement any voltage clamping diodes on the gate wires. Could that have been my mistake?
Any advice or comments are appreciated! Thanks!
can't find a spec sheet but if you don't have FWD across the Source and Drain of each, the inductive EMF will be many time 144 volts. I am guessing about 1,440 volts.

Wire dressing to keep inductive pick up.
 
#5 · (Edited)
Across the gate wires? wouldn't adding a diode create a short circuit during negative gate turn off? I think he is using the FWD of the upper IGBT across C1 and C2E1

By the way, what switching frequency are you running those at?

I've got the same IGBTs, I use them for a low frequency induction heater, I don't run them above 10 kHz.

Also with those crazy gate connections you've got going on, I would recommend increasing the Rg (gate resistor) to about 10 ohms. You'll get alot of voltage buildup between the gate driver board and the IGBT gate terminal with that low of a resistor.
 
#6 ·
motor amps * pwm duty = battery amps. So, if you tried, say, 20% of full throttle, the pwm duty would ramp up until the current feedback was 20% of the max current. I'm not sure what the software was, but on a 0.75" x 0.375" bus bar, the full range of that current sensor (assuming the melexis HB) is 0 to 1200 amps or so. That would mean you commanded 20% of 1200 amps from the Battery pack. The motor amps can be very very large when the battery amps are small. So, while the pwm was ramping up, in the vain attempt to see 20% of 1200 amps (or whatever), the motor amps would have been very large near zero rpm.
 
#7 ·
@bjfreeman: if you mean freewheel diodes across the motor, I am indeed using the upper half of the IGBT module for this purpose.

@subcooledheatpump: I'm not sure what exactly you mean by "crazy gate connections." I will try switching out the resistors for ones with higher values.

I am using the firmware modified for a hall throttle and 8khz PWM.

Thanks for answering, paul! The thing about the motor current is, we were testing the controller with a stove burner, which is basically the same as a resistor as far as I know. The 30 ohm resistance of the burner we were using limits the current that can pass through it at 144v to about 5 amps. Is it possible that the heating coil has enough inductance to blow the IGBTs?
 
#37 ·
@bjfreeman: if you mean freewheel diodes across the motor, I am indeed using the upper half of the IGBT module for this purpose.
If you are using the upper half of the IGBT module for freewheel diodes, then the motor must be connected between B+ and the E1-C2 connection, while E2 is at GND. So you are modulating the bottom half with the PWM? I don't believe it's a good idea to have the motor float at the top of the battery pack. Even if the frame is isolated from the M+ and M- connections, there is still some leakage, and the possibility of metal or carbon particles to create a conductive path to the frame. :eek:

And swoozle also confirmed that the hot plate was connected from B+/M+ to M-. But the Open Revolt controller apparently has the motor connected from the common sources of the MOSFETs to the motor, and the diodes across the motor to GND. But I don't know if there is a reason to reverse those connections when modifying the circuit for IGBTs. I'd love to see a complete circuit diagram! ;)
 
#13 ·
Ya, we need to tighten those up. The curious thing is, we had the power side hooked up to a 12V source first and ran the PWM to 100% just like we did when the IGBTs blew. It was fine that first time. Doesn't that imply that the driver circuit/wires are good and that whatever is wrong is in the high-voltage circuit?

Yes, it's the only cap. Not trying to throw Paul under the bus(s), but that cap was Paul's recommendation.

In retrospect we should have looked closely at the driver and high voltage waveforms before charging ahead so quickly. Any suggestions on the best stepwise test sequence moving up to full pack and motor is welcome. Is it better to go straight to pack(full or partial?)/motor than to try and jury-rig some low current load?
 
#16 ·
I agree the film cap should be enough for the local bus for the IGBTs, but really a few larger caps would offer some peace of mind, especially when switching hundreds of amps.

Definitely short out G1 to E1. I always use a short copper wire and solder it to the terminals when using the IGBT as a chopper
 
#17 ·
Just to add to what others have said. When driving multiple igbts in parallel you need emitter resistors. these should be connected between the kelvin emitters and the driver board.Should be about 10% of the gate resistor value. Have a look at some of my youtube videos on building the liquid cooled controller.
 
#18 ·
Thanks for all the responses!

okay, so I'll set up kelvin emitter resistors, put a jumper on the FWD gate, shorten the gate wires, put the current sensor on the motor bus bar, and twist the pwm signal wire together with the controller ground. Hopefully this stuff will keep it from blowing up again!
 
#26 · (Edited)
OK, new IGBTs and some experiments. Please take a look and help us understand what's going on.
These are all one IGBT with a 12V battery source running a starter motor.
All have the same time scale and V scale except the last trace (V scale increased to show the driver signal, time scale the same).

C-E trace, no snubbers, increasing but low PWM, amps probably around 20
Oscilloscope Electronics Technology Screen Electronic device


One snubber added.
Oscilloscope Electronics Screen Technology Electronic device


Two snubbers added (though one had long leads)
Oscilloscope Technology Electronics Electronic device Display device


same with stable PWM somewhere around 50%, current at ~35amps
Oscilloscope Technology Electronic device Display device


The drive signal with 12V source and starter motor load.
Oscilloscope Technology Electronics Electronic device Display device


The snubber. They came with the used IGBTs and we can't find any info on it.
Passive circuit component Electronics Capacitor Technology Circuit component
 
#27 ·
So we don't really understand why our voltage traces don't look like those shown in other people's pictures of their oscilloscopes when testing their controllers. But we're pretty confident that the snubbers are helping. The first snubber that we added, with extra short leads, helped a lot. The second one, which had the standard leads that were about .75" longer, helped less. I looked around for better snubbers and I found these:

http://www.digikey.com/product-detail/en/C4BTHBX4500ZAFJ/399-6240-ND/2783740


2.9 mOhm ESR, 29 amp ripple current rating, 5 microfarads. These have about 5x the capacitance of the ASC snubbers that we have, but I don't know if their ripple current rating/inductance/resistance is any better. Do you guys think that these will quell the voltage spikes to acceptable levels?
 
#29 · (Edited)
Yup 60 volts or higher seems to be the spike level when the motor is starting. We haven't added any bulk capacitance, just stuck with the round 400v 380uf cap and added the snubbers. The gate resistor is up to 11.2 ohms + .56 ohms on the emitter. Hmm what does a non-saturated IGBT look like? The curve downwards after the spike?
 
#30 ·
The C-E should look roughly like the gate drive signal

The IGBT Should be on for the duration of the pulse, that should look like a flat line. Then it should almost instantly fall to zero, another flat line. Just like the gate signal.

Personally, I would add some big capacitors to that, keep the snubber, but add some 6000 microfarad capacitors, rated at 400 volts. change the gate resistor to a slightly lower value, 10 ohms, then try again with the same setup. You really shouldn't be spiking that high with only 12 volts. It should really not spike above 15 volts.
 
#31 ·
OK, I am just wondering what are the signs of partial saturation? I know what the C-E signal is supposed to look like, but I just assumed that ours looked strange because of inductive spikes.

I will look at some big electrolytics to add to the DC bus, thanks for the tip.
 
#32 ·
It would really help to have a detailed schematic and a set of specifications for this. I looked at the BMW conversion website and watched some of the videos but there were still many details missing. I looked up the Open Revolt controller and found this:
http://www.paulandsabrinasevstuff.com/HelpFileMcontrollers.html

The schematics and the PC board and components look OK, but it is still unclear if there are any external snubbers to take care of voltage spikes on the MOSFETs. And if you are using IGBTs then you must have a modification to the circuitry, as you show in the video, but a schematic would be a lot more helpful.

Also, I don't know what sort of motor is being controlled, although I assume it is a PM DC motor, or perhaps a series wound motor. If it is a PM motor, I think it should have some dynamic braking or regen or load dump capability.

But if you were testing this on a resistive hot plate, there should be no significant inductance, and the waveforms should be very clean rectangular waves. If you are blowing IGBTs on a resistive load, and if you are getting messed-up waveforms as shown, then something is very wrong.

Is it possible that the scope is set for AC input? That would explain same of the curved waveforms, and what appears to be some negative and positive portions.

Good luck, and HTH :)
 
#33 ·
We're controlling a car starter motor and we are using IGBTs. The necessary modifications to the control board to use IGBTs have been applied. I don't think the scope has an AC input mode, it's a very simple portable oscilloscope. All the signals have been above 0v so far, ground is down at the bottom of the screen.
 
#35 · (Edited)
The last two traces are the only ones that make any sense. And it seems that the gate drive does dip below ground (as I think is recommended). But the C-E trace above that shows the voltage well above ground and then passing through 0V and going negative. If you are measuring across the upper IGBT then it shows desaturation and a lot of power being dissipated when it should be solidly ON. If you are measuring to GND then you are seeing the supply rail. And if you are measuring across the lower IGBT, which is used as a commutating (or free-wheeling) diode, then you are seeing the motor voltage, but the waveform is more indicative of a low PWM than the 80% or so that is indicated by the gate drive.

I'd really need to see a schematic to provide any more than an assumtive guess. I can't figure out the modifications from the video. But it would help to try a resistive load so inductive effects will not be present.

I had to look up the VLA502. It should have desaturation detection built-in, and its own DC-DC converter for the gate drive. This app note shows the details: http://www.pwrx.com/pwrx/app/vla502_applnote.pdf
 
#34 ·
The first thing i would do here is go back to basics. Drive one igbt and one cap. No snubbers. Also look at the 15v supply to the gate driver. Does it contain a lot of ripple? Is it dipping on load ? That 15v rail must be stiff as a board. Then start with a gate resistor that is too high. Say 33R and drive a pure resistive load and watch the C-E voltage on the scope. The miller plateaus should be quite obvious. Then connect up the starter. You should get no turn off kick with that high a gate resistor. If it does kick then the free wheel diode is not recovering in time. Look at the datasheet for the igb and make sure the the reverse recovery time for the fwd is faster than the turn off (toff) of the igbt.

Now , assuming its working ok at this point the igbt should be getting warm as its not switching fast enough due to the high value gate resistor. the trick is to get the gate resistor low enough to ensure the device switches as fast as possible but not so fast as to exceed the recovery time of the fwd. The vla502 driver switches on with +15v but off with -8v so you may need to use two gate resistors and a fast diode to shape the turn on and turn off to get best results. Only when you get this worked out on one igbt should you proceed to parallel up the others.
 
#39 · (Edited)
OK, I found a simple schematic for an EV, and I was very surprised to find that, indeed, the battery (+) is connected to the motor (+), and the other (-) goes to the drive, which modulates the DC to the negative (B-). Well, I suppose this is OK, but it's not what I expected - or what I would like to have in an EV, but I'm not going to use a brushed DC motor anyway. I suppose this design may be an extension of simple low power PWM motor controllers where the switching element was a BJT or MOSFET which is on the low side, and the motor is the load on the collector or drain. And a simple commutating diode across the motor completes the circuit. Here's what I found:
http://www.evconvert.com/eve/ev-schematic



I don't know why it was so hard to find. It seems like a schematic would be the first thing to have when making an EV. Well, at least now I have one, and the only thing that needs to be added is the actual IGBTs and the driver circuits. The Open Revolt website has some schematics, but they are for their MOSFET version. It does not actually show the MOSFETs - or the freewheeling diodes - so I'd have to dig deeper to see how they are connected to the motor. I guess it really doesn't matter, electrically, but I just like the idea that the terminals on my motor will be at the same potential as the frame when it is not running. So, OK, I learned something! :D

PS: And here are more schematics and other resources:
http://evhelp.com/Wiring_Diagrams.htm
http://evdrives.com/
http://waynesev.com/ev/ev_wiring_dia.html
http://visforvoltage.org/topics/schematic
http://www.evworks.com.au/tech/?section=circuits
 
#40 ·
OK, I found a simple schematic for an EV, and I was very surprised to find that, indeed, the battery (+) is connected to the motor (+), and the other (-) goes to the drive, which modulates the DC to the negative (B-). Well, I suppose this is OK, but it's not what I expected - or what I would like to have in an EV, but I'm not going to use a brushed DC motor anyway.
<snip>
I guess it really doesn't matter, electrically, but I just like the idea that the terminals on my motor will be at the same potential as the frame when it is not running. So, OK, I learned something! :D
<snip>
http://www.evworks.com.au/tech/?section=circuits
The traction (144V) pack should be completely isolated from the chassis at all points - neither the positive or negative end is tied to the chassis of the vehicle or the motor frame. So it doesn't matter whether the B- is tied directly to M- or B+ tied directly to M+ - either way, there should not be a current path between the chassis and the motor terminals. The only thing that can happen here is that commutator and brush dust will accumulate in the motor and this may form a low-grade conductive path to the frame; but in that case it still doesn't matter which end is permanently connected to the traction pack.
 
#43 ·
Yes, I see now. I just ASSumed that the B- was at chassis potential as it is for ICE cars. If it is isolated and floating, then the IGBT drive must also be isolated and floating, and it does not matter if it is on the top rail or the bottom. The gate reference can be connected to either E1 or E2, and the isolated PWM drive will still provide the same +15V and -8V as specified for IGBTs.

How is reverse handled in such DC drives? A full H-bridge would take care of this nicely, and it could also short out both motor terminals for dynamic braking (although for series wound motors I don't think this would apply). And an H-bridge requires four IGBTs and drivers for each, so complexity is increased.

Thanks for teaching me this fact about EVs. I still don't know why the waveforms are screwed up and the IGBTs are blowing. Hopefully the OP can find out why and fix it!
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top