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

[e*clipse]

Unfortunately, I can't find the book right now. One thing I do remember about it is that it was aimed at fairly low voltage ( 144?? ) DC systems. There was nothing about AC systems in it. I do remember the logic for isolating the high(er) voltage was exactly like you describe for safety reasons.

Regarding the motor, the motor I was referring to is the Toyota MGR. I guess you would call it a "3 connection" motor. Seriously; there are 3 power connections, not 6 or 12. The center of the wye connection is buried - right now I can't find it. There are no options for wye or delta connections or how to connect ground. I guess they assume anyone working on this isn't as knowledgable as an industrial person like yourself, so the fewer options, the better...

Originally Posted by thingstodo

Did the book give you any guidance on HOW to isolate the motor, controller and battery pack from ground, or how to detect when that isolation was broken?

[thingstodo]

Originally Posted by e*clipse

Seriously; there are 3 power connections, not 6 or 12. The center of the wye connection is buried - right now I can't find it.

OK. Are you still searching for the star point? It sounds like that's how it's wired.

What changes to Paul's software are required to drive the MGR instead of an induction motor? Anything hardware related?

[MPaulHolmes]

Software-wise, the differences are actually very minor. The main difference is computing the rotor flux angle. In the pmsm case, it's the rotor angle, so you just need to read the dang angle! In the acim case, you have to make a running guess at the rotor flux angle, using some formula that makes no sense to me. haha. In fact, it's the only formula from start to finish that I didn't derive myself. That bothers me, as it's hard to look for bugs when you are doing things that don't make sense. I haven't tried to make sense of it in several months, and I haven't tried that hard, so maybe I should give it another go. One interesting fact about the formula is, you cannot use it when the motor isn't turning, so FOC is useless at zero rpm. You need to use some sort of thing like V/HZ to get it turning, and then switch over to FOC. Otherwise, that formula divides by zero!!! NOO!!!

The PMSM motor also uses a resolver. If the motor is turning at 6000rpm, you'd have to check the angle at about 6.4KHz in order to get a resolution of 5 degrees (64 ticks per revolution). I think that's doable.

[e*clipse]

Ok, this may be interesting to folks:
On this Toyota MGR, there is NO conductivity between any of the phases and the case and/or the ground point on the case.

The resistance between any of the phases (measured with a Fluke multimeter) is 0.3 Ohms.

To be honest, I'm just looking for that star point as a matter of curiosity... I really don't want to re-wire this motor.

So if you - and I guess Toyota - think it's safe to build this way without a "fourth wire" to the star point (or center of the wye) then that should be good.

- E*clipse

Originally Posted by thingstodo

OK. Are you still searching for the star point? It sounds like that's how it's wired.

What changes to Paul's software are required to drive the MGR instead of an induction motor? Anything hardware related?

[e*clipse]

That's great news, especially from the perspective of "getting the motor to spin."
One thing about these Toyota motors are that they are "buried permanent magnet" motors. This results in the motors behaving a bit like switched reluctance motors, especially at higher speeds. This allows them to overcome some of the back-emf and run at higher speeds. There's some great information about this in the Oak Ridge National Labs tests on the various Toyota hybrids. Here are two good examples:

http://web.ornl.gov/~webworks/cppr/y2001/rpt/121119.pdf
http://info.ornl.gov/sites/publicati...s/Pub26762.pdf

Note that these tests also show some cool stuff about the boost converter and the inverter.

Regarding the resolver, I've decided that it would be best for this project if I run with the encoder position sensor. Analog Devices makes an IC that emulates an encoder with resolver input: AD2S1210

So, while I would like the "cleaner" input directly from the resolver, it will be possible to get this bad boy running with an adapter.

- E*clipse

Originally Posted by MPaulHolmes

Software-wise, the differences are actually very minor. The main difference is computing the rotor flux angle. In the pmsm case, it's the rotor angle, so you just need to read the dang angle! In the acim case, you have to make a running guess at the rotor flux angle, using some formula that makes no sense to me. haha. In fact, it's the only formula from start to finish that I didn't derive myself. That bothers me, as it's hard to look for bugs when you are doing things that don't make sense. I haven't tried to make sense of it in several months, and I haven't tried that hard, so maybe I should give it another go. One interesting fact about the formula is, you cannot use it when the motor isn't turning, so FOC is useless at zero rpm. You need to use some sort of thing like V/HZ to get it turning, and then switch over to FOC. Otherwise, that formula divides by zero!!! NOO!!!

The PMSM motor also uses a resolver. If the motor is turning at 6000rpm, you'd have to check the angle at about 6.4KHz in order to get a resolution of 5 degrees (64 ticks per revolution). I think that's doable.

[thingstodo]

Originally Posted by e*clipse

On this Toyota MGR, there is NO conductivity between any of the phases and the case and/or the ground point on the case.

That's Good!

The resistance between any of the phases (measured with a Fluke multimeter) is 0.3 Ohms.

What accuracy does the Fluke have? Can id display 0.31 or 0.315 ohms?

I guess most of the voltage drop would be from the inductance. What carrier frequency is Toyota using to drive the motor? Is it much different from the 8 Khz that Paul is using?

To be honest, I'm just looking for that star point as a matter of curiosity... I really don't want to re-wire this motor.

Agreed. It took me a while but I managed to read through your DIYelectriccar thread ... it's pretty heavy, detailed stuff ... so I MIGHT have skimmed a few sections Is that why I don't remember the thread discussing the carrier frequency?

So if you - and I guess Toyota - think it's safe to build this way without a "fourth wire" to the star point (or center of the wye) then that should be good.

I definitely think it's safe. But I'm NOT AN EXPERT! And I will try *REALLY* hard to remind everyone of that!

[thingstodo]

Originally Posted by e*clipse

One thing about these Toyota motors are that they are "buried permanent magnet" motors. This results in the motors behaving a bit like switched reluctance motors, especially at higher speeds. This allows them to overcome some of the back-emf and run at higher speeds. There's some great information about this in the Oak Ridge National Labs tests on the various Toyota hybrids. Here are two good examples:

http://web.ornl.gov/~webworks/cppr/y2001/rpt/121119.pdf
http://info.ornl.gov/sites/publicati...s/Pub26762.pdf

Oak Ridge National Labs seems to know no definition for 'overkill'.

*WOW*

I saw 5 Khz and 10 Khz carrier frequencies on the boost converter. Did I miss the carrier for the controller?

I agree that this would make an AWESOME drivetrain for an EV. And I like the simple interface - 3 wires in, motion out. Whatever is bolted to this drivetrain will move as commanded.

[MPaulHolmes]

I can only get 1 loop around the lem with #4 or #6 welding cable, so I guess I have to be happy with a "lem 150". Although 8 gauge would work.

[e*clipse]

What accuracy does the Fluke have? Can id display 0.31 or 0.315 ohms?

The Multimeter is a Fluke 179. If I measure a 1K resistor, I'll get a result like 0.997kOhms. A very low resistance like the motor windings is simply 0.3 Ohms.

Regarding carrier frequency, one issue is simply audible noise. One way to avoid an audible whine is to move the carrier frequency above the human hearing frequency range, like 18 kHz or 20 kHz. (probably lower for those of us who went to too many rock concerts...)

Paul had a great discussion about a "moving" carrier frequency on the EV tech list. If done correctly, this can result in a much less annoying variable hiss - type sound. I would definitely say Paul is a much more qualified person to answer those issues.

Also, Oak Ridge Labs has a lot more interesting tests than the ones I posted. It's definitly worth it to see what they've been up to.

[e*clipse]

Paul,

Those LEM current sensors all have a rectangular window to shove the wires through. A big wire, like a #4 welding wire's generally round shape would be a restriction. I haven't done the calculation, but there might be space for two square things of a given cross section, just not two round things.
Two possibilities:
1) Squish the welding wire into a more square or rectangular cross-section, so you might fit two, for example.
2) Use multiple smaller gauge wires in parallel, so the window can be more efficiently utilized.

Perhaps if this is too much work, we could try those inductive current sensors?

- E*clipse

Originally Posted by MPaulHolmes

I can only get 1 loop around the lem with #4 or #6 welding cable, so I guess I have to be happy with a "lem 150". Although 8 gauge would work.