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

[CollinK]

Originally Posted by Ragnarok Warrior

Hey, I read all 65 pages on this thread last nite and there had been no replies as of late. Now that I see that this thread is still active, can someone (who is actually working on this preferably) give a quick summary of what has been done to date? The best I can tell is that only the controller is working (well beta at least). So no power stage yet?

Well, certainly both Paul and I have built controllers complete with power stage and everything. Both of us have spun a motor with said controller as well. But, I don't have anything I can really release and say "this works. you can build it like this too."

So, quick summary of what has been done to date:
* Firmware which implements PID loop for throttle and V/Hz control of a three phase AC motor
* PCB of an IGBT halfbridge driver
* Prototype controller (control + power stage) put together - A bit of the PCB has been laid out by me but not yet released
* On my end several blown up IGBTs

I haven't worked on the project in a while because I've had other things to do. However, eventually I'll get back to it.

[nlc]

Hi collinK, which IGBT have you choosed for your test ?

[CollinK]

Originally Posted by nlc

Hi collinK, which IGBT have you choosed for your test ?

I have purchased Fuji 2mbi600u2e-060 IGBTs for use in my final design. These are 600A, 600V half bridge modules.

Currently, though, I'm testing with Fuji 2mbi200n-060 which are 200A, 600V modules. I've blown a couple of them up because I was dumb and left too much input inductance. Also, my gate resistors were stupidly small for the 200A modules (I think I was using 3 ohm resistors.)

I have since purchased a big film cap for the input and have minimized loop length as much as possible. Also, I have put 9.1ohm resistors in for the gate resistors. I hope that all of this makes things work a bit better but I've had no time to test it yet.

[nlc]

Originally Posted by CollinK

I have purchased Fuji 2mbi600u2e-060 IGBTs for use in my final design. These are 600A, 600V half bridge modules.

Hummm, great half bridge module !! Where have you buy it ? They are expensive ?

[CollinK]

Originally Posted by nlc

Hummm, great half bridge module !! Where have you buy it ? They are expensive ?

I bought them on Ebay. I believe they were $150 a piece so the total cost for the power IGBTs is $450.

[nlc]

Ok thank for the information !

From my side I continue the development of my controller FOC PMSM controller when I found some free time (not easy ). My last test was today, the second motor short circuited to have maximal brake torque on the first motor and thus very low speed (42 RPM), current consign 35A and.......sensorless !!

http://www.youtube.com/watch?v=Ew7L4-MHG0U

On the scope this is the current in one phase.

[MPaulHolmes]

Papa!!! HAHAHA.

Awesome! I've been studying sensorless, and it's not easy. But if you get everything just right, it can keep the d and q axis within about +/- 1 degree from 90. I also finally figured out the space vector modulation. Now I think I can do all the software without having to look up any formulas or anything. Just do what makes sense mathematically from start to finish. Finally! The driver section is basically done. It's going to drive from -8v to 15v or so. I've found some TO-247 plus igbts I think I want to use too. They're only $6 each. Rated for like 210 amps each, but I'll only run them at maybe 75 amps peak per device.

The whole thing should be just maybe $400, but good to maybe 50kW I'm hoping.

[nlc]

Originally Posted by MPaulHolmes

Papa!!! HAHAHA.

Originally Posted by MPaulHolmes

Awesome! I've been studying sensorless, and it's not easy. But if you get everything just right, it can keep the d and q axis within about +/- 1 degree from 90.

I have not looked in detail the source code, but the ST algorithm seems to be really good. I looked in detail the current curve on phase 1 and the corresponding hall transition, it's really perfect, on all torque request, low or very high
I know the algorithm just use a state observer principle, it use the vbus measured voltage, alpha/beta measured currents, and alpha/beta phases voltages (not measured but with PWM value associated with vbus voltage).
It also needs the phase resistance and inductance, but I have not seen big difference in sensorless performance with differents values but same motor.

Originally Posted by MPaulHolmes

I also finally figured out the space vector modulation. Now I think I can do all the software without having to look up any formulas or anything. Just do what makes sense mathematically from start to finish. Finally!

I also understood the FOC, but don't want for know to try by myself

Originally Posted by MPaulHolmes

The driver section is basically done. It's going to drive from -8v to 15v or so.

-8V is good, and even mandatory for voltage > 150V, to avoid autoconduction.

Originally Posted by MPaulHolmes

I've found some TO-247 plus igbts I think I want to use too. They're only $6 each. Rated for like 210 amps each, but I'll only run them at maybe 75 amps peak per device.

Interesting, I looked some week ago on all manufacturer but I haven't find such an igbt !! You don't want to give its reference ?

Originally Posted by MPaulHolmes

The whole thing should be just maybe $400, but good to maybe 50kW I'm hoping.

The perfect AC controller, high perf, high reliability, and low price

[MPaulHolmes]

OK, maybe $450. They are $6.51 each in quantity 1. But 229 amps!

paul

APT100GN60B2G Microsemi Power IGBT Transistor - Richardson

[nlc]

Really really interesting igbt !!!!!!!!