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

[MPaulHolmes]

Hi Etc! I'm ALMOST done with the 1000amp kits. Then I'll get back to it. There's not much more to do. Those igbt driver boards will go a long way to driving really big igbts. I still need to test the mosfet based AC controller, which is completely assembled and I think will work just fine. It can do 300 amps per phase. It's basically just software that needs to be finished. I need probably 2-3 uninterrupted weeks.

[sawickm]

Originally Posted by etc

Will there be an open source inverter before they are cheaply commercially available?

Not open source but, the cost for a 100Kw AC Controller now! is about the same as a highend DC Zilla or Soliton ...... MES TIM600 ($3700).

-Mark

[ecorsa]

Originally Posted by bennelson

My friend Tom built a 300 volt AC Dodge Neon.
He used a stock AC drive from some decommissioned factory.

The problem is that those drives are not designed with a car in mind. There are a lot of "little things" that had to be taken into account to make the drive really work for a car application.

He's an electrical engineer, and enjoyed the challenge. For the average guy, it would be a lot of work to make all the little tweaks for it to all work.

With an open source build, everything can be designed in right from the start.

would it be a lot easier to program and modify an off the shelf industrial drive than to design and build one from scratch. has anyone tried this?

[drbobwoolery]

The members of the Australian Electric Vehicle association have several cars using this setup. Search out their mail list.

[Mesuge]

Yep, there have been quite some successfull implementations of this approach around the world in recent years. Some limitations apply though. Firstly, the packaging and power, you are limited lets say by aprox. 40-60kW max peak output, otherwise these units are just huge. Also they are designed to be aircooled, usually incorporate useless (for EV) addons, wich are quite heavy, it's a stationary box (not designed for automotive vibrations). So, you will have to repackage them into something more compact, add a chill plate on water circuit etc. Not easy. Last but not least, the price is defined mostly by 2nd hand ebay prices/surplus and availability, currently it's quite overpriced and the prefered brands/models are hard to find, i.e. the firesale has taken already place around 2007-2009.

Given the fact there are now several companies offering "mid voltage" AC/BLDC drivetrain systems <450VDC, the route of industrial VFD is not that much appealing anymore. In terms of high voltage 700-900VDC system, which give you the best efficiency possible, it's still interesting, but even this area is being covered by diyers (in progress).

For the "low voltage" domain you got Curtis, and hopefully this P&S AC open source controller in the near future. So, the field of AC/FOC is pretty much covered. Yes, more consolidation, and documented practical experience is always needed, so the more betasters means faster adoption and debugging..

[pedjag]

don't forget Tumanako the kiwi opensource controller...

SourceForge.net: tumanako

Parts are a bit expensive I think..

[tesla500]

Hi Guys,

I just found this project recently, really nice progress. I wish I'd found this project earlier, I've actually been doing similar work, unfortunately in complete isolation, for the last while. I had the inverter working nicely about a year ago, but sort of lost interest. Seeing there's lots of interest in such a project has rekindled my interest now. I've got a video showing what I've done so far:

Oops, I can't post a link. Go to youtube and search for "AC Drive System for EV", it's the first result.

We should compare notes and see if the work I've done could help on your project. I've done quite a bit of software work with the Microchip app note code, specifically with respect to setting the magnetizing current for optimal use of the power stage. I also have an autoranging encoder algorithm that uses a 50 pulse encoder that can easily be made DIY, as opposed to high count ones that are difficult to make.

I've also had a good go at getting the sensorless version of the code to work, with very limited success, unfortunately. I think I'll stick with sensored for now.

On the hardware front, I've done a lot of work on determining how far the IGBTs can be pushed safely, and it turns out that as a very rough estimate, the RMS output current you can run on most modules is about 1/2 of their DC rating. Eg, if you have a 600A module, you can get about 300A RMS (If you can keep the heatsink at about 50C). Also, the maximum inverter power throughput can be estimated as Pout = DC link voltage * RMS output current * 1.17. When you go through all the math, with the inverter at max duty cycle and unity power factor, the ratio of DC link current to RMS output current is exactly SQRT(2) * 5/6.

I've got lots of other little tidbits, probably too many to list. Ask away if you have any questions or want more details.

I think the next step for my project is to "productionize" it with everything on a PCB, and a proper power stage. The present power stage is cheap (about $300 for the IGBTs brand new for a 300A RMS output), but is too difficult to make, and has too many points of failure with so many parts. I'm currently compiling a spreadsheet of the various IGBT options, I'll post it when I have it filled in more. At the moment there's some Powerex and Microsemi modules that look pretty good.

Regards,
David

[MPaulHolmes]

DAVID!!!!!! Holy cow!!! that's awesome! I've been distracted by the DC controller lately. That's so great! I think I have a lot to learn from you! I'm so glad you wrote on here!!!!! I have a mosfet based low voltage power stage that is good to around 150v peak, but I haven't tested it yet. I've just been so dang blasted busy. The igbt power stage I did first worked OK, but the 0 to 15v driving wasn't working very well. Very long turn off times, so I made a driver board specifrically for igbts. basically a copy of the bg2a from powerex, but I haven't populated the board yet. I had a pmsm motor that I was trying to get working with the field oriented control, but I need time, dang it! haha. I can't wait to go watch your youtube video. Thank youso much for posting on here!!!

[mrbigh]

I had seen Tesla500's work in another short-cut place. Very interesting things to learn to.
Would be great if both of you can finish the current project anytime soon.
Keep up the good work!!!!

[tesla500]

Thanks Paul!

Did you have a general list of IO features for your DC controller? Since a controller is more than just a power stage, I'm trying to figure out what IO (switches, LED drivers, etc) to put on the board.

My current list is:
Pot inputs

• Throttle
• Regen
• Auxiliary

Inputs

• Throttle enable
  
• Regen enable
  
• Valet switch
  
• Forward
  
• Reverse
  
• Motor Encoder
  
• Motor Temperature Sensor (What should this be? Thermistors can't handle Class H motor winding temps)

Cruise control:

• On
  
• Off
  
• Set speed
  
• Encoder 1
  
• Encoder 2

Digital Outputs

• Analog gauge driver (speedometer/tachometer)
  
• D/N/R LEDs
  
• Thermal Derating indicator LED

Contactor drive

• Main
  
• Precharge (should this be internal?)

Fan drive (variable speed, 5A max)

• Inverter
  
• Motor

Regards,
David