Paul and Sabrina's Cheap 3 Phase Inverter (AC Controller) with Field Oriented Control

[CollinK]

Heh, it's interesting that you found this thread. I have had your website bookmarked for a long time. Yeah, I've now blown up the controller twice. I think this makes my average about 60% since I've successfully tested with no blow ups like three times and blown it up twice. Luckily this last time was just a transistor, gate resistor, and a few diodes. I'm using $25 surplus 200A IGBTs right now so that explosions are less sad.

Here's my thoughts on my last blow up (which I unfortunately did not film at all.... stupid!):

I ordered a 50A 800V three phase rectifier off of ebay and used it connected to 240V single phase as my power source. Since I had precharge capacitors installed they were able to smooth the voltage to about 339V at rest. Unfortunately I never got to testing how much voltage fluctuation I was getting. However, I was only testing a 2HP motor unloaded so I doubt that the amperage was too high. I did have an oscilloscope running at the time. I put a probe on one of the three phase outputs to the motor while testing. I could see really large spikes on the output. They looked very short, perhaps like 1us. Here is where I wish I really wish I had video because all I can go on now are vague recollections after the fact. I'm not sure if the spikes were because A. I somehow managed to use IGBTs with no freewheel diodes. I doubt this because any reasonable half bridge should have them but I'll check B. using a rectifier would have caused the input voltage to be very unstiff and maybe that caused some issues C. I had way too long of a connection between the capacitor bank and the bus bars. I knew that was naughty but was hoping to get away with it. One thing is sure though, I had way too much inductance on the input side of the controller. D: Maybe my leads from driver to IGBT are too long. I mean, the wires are like 2.5" or something but maybe that's still too much. Or maybe something was getting induced on the lines. I might be able to shorten the leads even more.

Regrettably I didn't test for many of my guesses above. It blew up before I could. Well, it's fixed again and I'll keep at it.

Originally Posted by etischer

Your first spin video is just like mine =) I blew up my IGBTs too =)
<<URL removed>>

Now, two years later, I'm running closed loop vector control, and have racked up almost 30k miles!

It's great to see other people take on a such a project. I wish you luck.
Here's a link to my inverter project: <<URL removed>>

It's my first post, and I'm not allowed to post URLs yet =( You can google etischer and find my electric vehicle webpage.


Cheers
-Eric.

[etischer]

My best guess is:

C. I had way too long of a connection between the capacitor bank and the bus bars. I knew that was naughty but was hoping to get away with it. One thing is sure though, I had way too much inductance on the input side of the controller.

This could also be why you were seeing large voltage spikes on the scope.

[CollinK]

Yes, that's my guess too. I've since shortened the links and I think I'll try to make them even shorter. I know that the less distance and inductance between capacitors and bus bars the better. Chalk this one up for knowing better and trying it anyway...

[JasonG]

Are your leads cable?
May I suggest using flat bar from the caps to the IGBTs, I know its a pain to fabricate but it will kill the inductance.

BTW etischer, I haven't kept up. Didn't know the AC was up and running. Nice !
Since your post count isn't up yet http://etischer.com/awdev/

[CollinK]

I was using bars across the IGBTs. The batteries where connected through a contactor to the busbar. The capacitors, though, were connected to the bus bar with 2 or 4 gauge wire. It was way too long too. What a recipe for disaster! I've since bought a 66uF 800V film cap that can handle 100Arms. It is now bolted directly to the bus bars. My calculations suggest that this should result in a total of about 15V ripple at 100Arms AC ripple current. However, I also got some 1500uF 450V elcos. I might try to get bars long enough to put one or two of those on the bar too. That will make the bar sort of long though. I suppose the real question now is, what sort of ripple might the busbars experience? If I run a motor up to 250Arms then what equations should I use for the AC ripple at the caps? From what I can gather it seems that the AC ripple of a three phase system should be approximately half of the Arms to the motor. This would give me about 125Arms at the caps at peak output. Additionally, I might want to clamp the ripple voltage to less than 15v so I'll need more caps because of that too. So, I think I'll need a bit more than just my 66uf film cap but maybe not until I start really loading the motor. I'm reading that elcos in the system were mostly used in VFDs because they input rectified mains voltage and thus have very ripply voltage input. A battery is very stiff and constant so it seems like tons of capacitance is no longer needed. Instead, the caps in that case are just to deal with the ripple current and to decouple the IGBTs from the batteries.

I have a ton of PDFs about this whole situation but I certainly wouldn't reject suggestions either directly or for other PDFs or books I might want to read.

[mrbigh]

And here is my approach to the IGBTs and Capacitor mounting

[madmike8]

That's pretty impressive looking... I like it!

[dave koller]

mrbigh

Nice!!!

[etischer]

Make sure your bus bars don't rub through the vinyl insulation on your capacitors! You might also verify the vinyl insulation is rated for your bus voltage.

[adamj12b]

Looks great MrBigh!

How did you end up bending those curves on the bus bars?

-Adam