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

[thingstodo]

Originally Posted by MPaulHolmes

Here's the golf cart model:

No joy on the golf cart. I can't find an AC model that will drop in.

Taylor Dunn 2372P appears to be kinda old.

[thingstodo]

Originally Posted by e*clipse

Regarding the code debugging, I'm wondering if it would be possible to share that load?

At one point, many years ago, I messed about with PICs. I don't think I have anything loaded on this laptop, though ...

At the very least I can do a code review, read through the comments and see if I can figure out what you are doing. Another set of eyes never hurts!

[MPaulHolmes]

That sounds good to me. The clarke and park transforms and their inverses are easy peasy.
What is more tedious is the space vector modulation part, tuning the Id and Iq PI loops, and the running guess at the rotor flux angle. It seems like everything needs to already be working in order for it to work at all. A motor that just sits there isn't very useful for finding mistakes. I guess I need a strategy for finding bugs without the motor actually turning. Maybe a bunch of data that could be looked at.

[e*clipse]

Originally Posted by thingstodo

No joy on the golf cart. I can't find an AC model that will drop in.

Taylor Dunn 2372P appears to be kinda old.

Were you able to find what the DC motor's shaft diameter is? Maybe it would be possible to get a **close** 3phase motor and fudge things with an adapter plate and some sort of shaft coupler?

I'm willing to make or modify some parts as necessary - I've got a mill & lathe.

[freebeard]

Originally Posted by thingstodo

Quote:
Originally Posted by e*clipse
Regarding the code debugging, I'm wondering if it would be possible to share that load?
At one point, many years ago, I messed about with PICs. I don't think I have anything loaded on this laptop, though ...

At the very least I can do a code review, read through the comments and see if I can figure out what you are doing. Another set of eyes never hurts!

SourceForge - Download, Develop and Publish Free Open Source Software

The help you're wanting is available, all you have to do fully embrace Open Source and put it up on a repository like Source Forge and let the whole world hammer on it. Many eyes, and all that.

In addition to bugs you should show concern for in-car systems security. We don't want people hacking on things that involve mass and velocity.

[MPaulHolmes]

There's a reason I don't work at intertrode making fixes for the y2k bug. The structure is safe. AC and DC code is easily structured the same. You just do a few more transforms. Its been running in vehicles since 2009 without issue, and I learned it from an expert in the industry whose EV won best in show at the L.A. auto show. Everyone has their opinion about how things ought to be done with software. For me, 500 armchair quarterbacks that have never made FOC code is less helpful than having discussions with 1 or 2 people. And the code will not be a secret. If anyone wants to put it where ever they want, it can be mutated until I don't recognize it anymore. Just don't insist that I personally use code that no longer makes any sense to me, that I didnt write.

For me, all the fun is in the act of creation. It's not in having everyone tell me how stupid something is, and try to change everything so everyone is happy.

[thingstodo]

Originally Posted by MPaulHolmes

That sounds good to me. The clarke and park transforms and their inverses are easy peasy.
What is more tedious is the space vector modulation part, tuning the Id and Iq PI loops, and the running guess at the rotor flux angle. It seems like everything needs to already be working in order for it to work at all. A motor that just sits there isn't very useful for finding mistakes.

Id and Iq determination ... The industrial controllers (AC in, AC out, DC bus between) all have a 'wizard' thing that steps you through a 'tune'. You enter the current and the voltage that the motor is rated for, it's rated rpm and frequency. The controller then pushes various pulses through the connected motor and determines about a dozen different parameters that it then uses to set up it's loops. Perhaps these parameters would be useful to you?

We (day job) power up each of our spare controllers once every 2 years on a Preventive Maintenance (PM) schedule. During these PMs I was planning (I have not started as yet) on getting our electricians to connect up a motor. Likely a different motor for each size of controller. After hours, I will run through the wizard setup and grab the parameters. When I see these parameters, they sorta make sense to me .. although I don't understand a lot of what they are for.

I am aware that the parameters are slightly different for EACH motor, even if they have the same part numbers. But if I can match voltages and currents to AC motors being used by all of us, I may be able to come up with something that you can start with. It will still take some 'playing around' to get it perfect ... but I'm all about GOOD ENOUGH.

I guess I need a strategy for finding bugs without the motor actually turning. Maybe a bunch of data that could be looked at.

Motor turning or not. During one of the demo/training sessions I went to (salespeople dress up sales talks as 'training'), the controller had some optional magic loaded that would measure the back EMF closely enough to approximate a linear positioner ... you could turn the shaft by hand and get a readout from the controller regarding angle and number of revolutions. Perhaps the PIC libraries don't do that. Perhaps the libraries are there and it will be easier to find now that you know about it. My point is that it is possible, at least. Powered up with no torque sounds pretty interesting ..

Let me know if you'd like some help on some task that you think you can split out.

[thingstodo]

Originally Posted by e*clipse

Were you able to find what the DC motor's shaft diameter is? Maybe it would be possible to get a **close** 3phase motor and fudge things with an adapter plate and some sort of shaft coupler?

I'm willing to make or modify some parts as necessary - I've got a mill & lathe.

Nope. No information at all that appears to match. The older manuals do not appear to have been digitized and the ones available (at least, the ones I can find) do not match even a little bit.

Perhaps I am not reading the model number correctly. I can't find a Taylor Dunn model, even in their archives, that could match the 2372P. Rumor has it that Taylor Dunn manufactured for third parties that sold them with their logo pasted on the side.

Sorry.

[thingstodo]

Originally Posted by MPaulHolmes

There's a reason I don't work at intertrode making fixes for the y2k bug.

I had to look up the Office Space reference ... memory is going!

The structure is safe. AC and DC code is easily structured the same. You just do a few more transforms. Its been running in vehicles since 2009 without issue, and I learned it from an expert in the industry whose EV won best in show at the L.A. auto show.

Cool!

Everyone has their opinion about how things ought to be done with software. For me, 500 armchair quarterbacks that have never made FOC code is less helpful than having discussions with 1 or 2 people. And the code will not be a secret. If anyone wants to put it where ever they want, it can be mutated until I don't recognize it anymore. Just don't insist that I personally use code that no longer makes any sense to me, that I didnt write.

However the code is maintained and published is fine with me!

[e*clipse]

Bummer. (regarding the golf cart) I guess it's all still possible, it would just take more work - and maybe we're bumping up against that "diminishing returns" barrier.

I guess it's 100% Paul's call.

Another alternative may be much more productive:

Paul has a Superbeetle mentioned in his sidebars. Freebeard is doing a VW-type conversion. Any chance these efforts could be combined?

There's a really cheap MGR on E*bay right now - it's in New York and pretty rusty. I think it's about $250. Otherwise, I've offered one of mine as a tester. Or maybe we could scoure some junk yards for Highlanders - E*bay can't be the only source.

One of the main problems is that it changes the control philosophy significantly. It might be a bit difficult to go back to an induction motor once these are running.

-E*clipse