Prius 3 MG2 Wye/Delta and controller

[vehiculeselectriques]

Hello,

i'm working on solutions for driving this motor at reasonnable voltage instead of the 650VAC it needs.
As i see a shared connexion here seems to me it's possible to wire it in Delta for 375V (650V/1.73) at double current instead of actual Wye !



am i right ?


Then problem is finding a controller which can drive these internal permanent magnet motors (saliency factor)

I suppose a VFD with auto mode would be irelevant finding the right driving specification ?

thanks,
Phil

[coulomb]

As i see a shared connexion here seems to me it's possible to wire it in Delta for 375V (650V/1.73) at double current instead of actual Wye !
am i right ?

Wow. That sure looks to be the case. Except that the current won't double, it will go to 1.73x (for the same ratings). I assume that it is already running at close to its peak power, but there may be some capability spare.

I assume that's MG1 from a NHW20 Prius... is it? [ Edit: oops, from the title, I assume it must be a ZWV30; the NHW20 ran at up to 500 VDC. ]

Then problem is finding a controller which can drive these internal permanent magnet motors (saliency factor)

Well, it's not going to need a 375 VRMS sine wave, but something like a 375 V p-p trapezoid wave; I'm hazy on these IPM motors. It was being driven from an up-to-650 V DC bus (assuming a Prius NHW20 or similar), so I think that wiring in delta means that you would get the same performance from 375 VDC if you rewire it in delta. I assume that the star point didn't connect to anything outside the motor.

So you will need a motor controller that can run at about 375 VDC and knows about IPM motors. The Tritium Wavesculptor 200 would do nicely, though it would probably be somewhat overkill for the job. Since its 300 A capability is likely way more than needed, I wonder if you could drive MG1 and MG2 (with locked rotors, carefully aligned) in parallel? It sounds more than a little dodgy to me, but maybe it would work. But get advice from an expert before attempting that. At the very least, you'd want to spin them together and make sure that the back EMF was within say 5% over a wide speed range. I vaguely recall that they have quite different characteristics, so that's likely not possible.

Maybe a Prius inverter, with the outputs designed for MG1 and MG2 connected in parallel? You'd bypass the boost converter, of course. [ Edit: no, the motor constants are vastly different, at least on the NHW20 model: 1.33 VRMS/Hz (MG2) and 0.46 VRMS/Hz (MG1) (from Evaluation of 2004 Toyota Prius
Hybrid Electric Drive System ]

I suppose a VFD with auto mode would be irrelevant finding the right driving specification ?

I believe that most VFDs don't know about IPM motors, but I could be wrong there.

[coulomb]

This would appear to be the equivalent photo for a NHW20 MG2 motor:



I can't be certain, especially with something that could possibly be a star point at about 11 o'clock in the photo, but I'd say that this motor is connected in delta, with no star point accessible.

It's interesting that they would change from delta to star configuration when going from 500 VDC to 650 VDC (and from memory, 50 kW to 60 kW peak).

[vehiculeselectriques]

Thanks for your answer

yes mine is prius 3 model MG2 so ZWV30 ref number.

Seems to me the NHW20 MG2 motor you are showing also have Wye shared connexion though there is little more work accessing to it.

ZWV30 motor should be a 650V Wye 375V Delta motor (with 1.73 more current sorry for approcimation

and NHW20 motor should be a 500V Wye 289V Delta motor ?

[coulomb]

Seems to me the NHW20 MG2 motor you are showing also have Wye shared connexion though there is little more work accessing to it.

My initial thought was what connection. But then I looked again:



I was initially looking at the spot marked 1, and was quite dubious. But now I see you are likely referring to the area marked 2; that does look like a star point. Neat!

and NHW20 motor should be a 500V Wye 289V Delta motor ?

Yes, I believe that wiring in delta and powering from 298 VDC means that each winding will see the same voltage and current (assuming the controller can dish out the extra 73% of current) as the original star/wye motor would see with a 500 VDC power source.

Actually, I'm making a big assumption here, that the motor behaves like a 3-phase load on a sine-wave AC system.

However, many brushless DC motors use "6-step commutation", where only two windings are on at once. If you are not PWMing the outputs, then with star you would get two windings on at once, but with delta, only one. So if the motor has crude commutation like that, switching from star to delta would not be valid, and the motor would run extremely roughly, if at all.

IPM motors are somewhat between PM motors and BLDC motors, so I don't know where that leaves us. Suddenly, I'm not as confident that the star-delta rewiring will work. I think we need someone more familiar with IPM motors to chime in at this point.

[BigMouse]

Actually, I'm making a big assumption here, that the motor behaves like a 3-phase load on a sine-wave AC system.

However, many brushless DC motors use "6-step commutation", where only two windings are on at once. If you are not PWMing the outputs, then with star you would get two windings on at once, but with delta, only one. So if the motor has crude commutation like that, switching from star to delta would not be valid, and the motor would run extremely roughly, if at all.

IPM motors are somewhat between PM motors and BLDC motors, so I don't know where that leaves us. Suddenly, I'm not as confident that the star-delta rewiring will work. I think we need someone more familiar with IPM motors to chime in at this point.

I'm far from familiar with IPM motors, so please consider what I write below to be assumption and guesses.

Is it safe to say that the waveform that a PM motor would be happy with is the same as the waveform it produces at its output when spun by an external source?

For example, if one of these motors is re-wound to delta, the motor is spun, and the waveform observed at the terminals is sinusoidal, then the motor would happily run with a sinusoidal input. Likewise, if the output is trapezoidal, then it would want a trapezoidal input.

The fact that the rotor position is sense using hall effect sensors does suggest that it's a "6-step" motor, but if the absolute rotor position was known through the installation of an encoder/resolver, shouldn't it be possible to run the motor with any FOC controller capable of the voltage and current?

I've been fantasizing about using a Prius transaxle as an EV motor for a while now. Even better if I can put it in a car and have it just look like a transmission floating there with no engine attached, then drive away in it. Boggle people's minds

Really though, the Prius transmission, from MG2 through to the diff and output shafts, is a very nicely packaged unit with a (I think around) 8:1 ratio. It's like a mini version of the "holy grail" Borg Warner + Siemens motor. An enterprising entrepreneur would design a stand-alond cast aluminium casing into which the MG2, reduction gear, and differential from wrecked Prius transaxles could be swapped. I'm picturing a Subaru Impreza with one of these in the front, another in the back, and 120kW AWD.

</dream>

Still, 298vdc is still very high for a motor on a 8:1 fixed gear reduction (7000rpm @ 100kph). Does anyone know how many poles these motors have?

[vehiculeselectriques]

I planned looking at his waveform as generator for sure as i agree with you it's the best id type method actually.

What i found is this MG2 is a internal permanent magnet motor, 8 poles, 48 slots...
Each phase of the 2010 Prius motor stator windings consist of
twelve – 20 AWG wires with 8 coils (poles) in series, with each coil having 11 turns.

from motor data is here


about 20kW continuous and 60kW peak
200Nm max
13000rpm max
with 650VDC bus so i presume 460VAC RMS ?(i'm not sure about the real RMS voltage rating on this motor)
Thermistor is 3900-3950K 47kohm (source)
Resolver 2X-VR (Source)

For sure IPM are really new curiosity beast for EV community though with such impressive performances numbers these motors seems a very interesting solution for hobbyists.

I find lots of studies about IPM motors and their driving specifications but actually no controller available is said to be made for such ...

time to find IPM motor/driver specialist [investigation mode ON]



Phil

[PlanB]

Looks to be similar to that Lexus RX400H electric transaxle (viewtopic.php?title=rx400h-transaxle&t=2435) living under my bench. One day motors will have the VFD built into them & life will be easier for us lot.

[Tritium_James]

The waveform when spun as a generator only shows you the PM component of the motor. IPMs are a hybrid of PM and Switched Reluctance motors, and there's no way to characterise the SR component of the design without dyno testing, ie it's not possible to 'autotune' IPM motors.

[coulomb]

... there's no way to characterise the SR component of the design without dyno testing, ie it's not possible to 'autotune' IPM motors.

Ah. So what happens if someone puts a random IPM motor onto say a Wavesculptor 200, and just selects IPM as the motor type (I can't remember if that's even possible; it's been a while since I played with the motor configuration software.)

Are you likely to get poor efficiency?

Motor likely to overheat? (I guess this would be a consequence of poor efficiency).

High torque ripple (way more than if configured correctly)?

Edit: is this what the torque verses angle diagrams are about in the ORNL reports? They had to throw together a motor controller, since they didn't want to use the Prius ECUs for some reason. So the angle of maximum torque would possibly be one (perhaps the main? only?) parameter you can't auto-tune for?

[vehiculeselectriques]

i'm suposing waveform will be sinusoidal as it's just a particular PMAC...

For IPM driving parameter there is lot's of information available but my problem is finding harware !
I mean > a controller said to be IPM capable.
Which significate it give access to parameters for adapting the control algorythm to the saliency specifications of the rotor or to have an autotune step by step mode using measure and calculations from torque/efficiency etc. i don't see why it can't be possible, difficult yes for sure as it must be a lot of hardware high specs design and work but impossible

There is driving data everywhere on the net

more driving data

interesting data

And the prius motor saliency is pretty well caracterised ("dynoed" from all perspectives) on ORNL prius reports.
these are just few of them


these are just few of them

Seems to me it should possible to charasteris by small steps without dyno using all this data available...
so from my customer perspective problem is at first finding a controller which have the IPM motor settings available.
   
In data i have readed about Prius IPM motor, saliency is said to be responsable of about 10% total torque for low speeds (<2000rpm) and about 30% for high speeds (>6000rpm)
so i suppose driving at the nominal Voltage an IPM motor as if it was a PMAC motor will make that become torque losses ?

We need an IPM driver specialist here, the CNC world use them more and more so this man/woman should exist !
Phil

[Tritium_James]

Our controller can run an IPM motor just fine (in fact, mathematically provable to be near-optimal) but there's no way to set it up without dyno testing.

The torque vs angle data from ORNL's dyno shows the optimal angle *at one speed only*. There is not enough data there to figure out the parameters over the entire envelope of operation. In reality you need this data for all speeds.

Actually, the "must have dyno data" is not 100% correct, it is theoretically possible to do without, but selecting the correct operating point (d/q currents/angle) requires solving a fourth-order polynomial in real-time at several hundred Hz = not possible in reality.

[vehiculeselectriques]

thanks for your "mathematical clarification" James,

once more my dreams and reallity are different ... ok ok you take the point

i must add my respectfull apologises as i have readed your last manual
here and motor configuration include IPM motor...
so there is at least one EV hobbyist available IPM motor controller
Does it provide field weakening for high rpm and maximum torque per Amp tracking functions ?

But i don't understand how industrial Yaskawa A1000 can drive IPM with autotuning functions ? or it's commercial tong ?

ok the Delta voltage suits the tritium capabilities for max usable motor power but i'm still stuck with complete dyno data availibilty about this motor... is there a solution/procedure to make the dyno machine/caracterisation myself ?

[vehiculeselectriques]

Hi every EV motor dreamers, seems this interesting discussion is fading which is "sad" so reading your very informative comments

Our controller can run an IPM motor just fine (in fact, mathematically provable to be near-optimal) but there's no way to set it up without dyno testing.

The torque vs angle data from ORNL's dyno shows the optimal angle *at one speed only*. There is not enough data there to figure out the parameters over the entire envelope of operation. In reality you need this data for all speeds.

Actually, the "must have dyno data" is not 100% correct, it is theoretically possible to do without, but selecting the correct operating point (d/q currents/angle) requires solving a fourth-order polynomial in real-time at several hundred Hz = not possible in reality.

James could you please answer on this interesting question:

...
Ah. So what happens if someone puts a random IPM motor onto say a Wavesculptor 200, and just selects IPM as the motor type (I can't remember if that's even possible; it's been a while since I played with the motor configuration software.)

Are you likely to get poor efficiency?

Motor likely to overheat? (I guess this would be a consequence of poor efficiency).

High torque ripple (way more than if configured correctly)?

and what if we use "The torque vs angle data from ORNL's dyno shows the optimal angle *at one speed only*" as the same for all speeds ?

would this give bad commutation ? or a correct one ? good one ?
a starting point for tweaking at other speeds ?

have a nice day,
Phil

[vehiculeselectriques]

Hi,

someone around ?

[vehiculeselectriques]

[acmotor]

Just coming into this topic late...
Not helping with the technicalities but wondering...

Do Nissan Leaf and Mitsubishi iMiEV run pure PMAC while Tesla runs pure IM ? Toyota is alone in running IPM ?
What is the upside of IPM ?
If the PMAC and IM are getting 95%+ efficiency then how can an IPM get 10 to 30% better torque at various revs ? There isn't the maths to do that. OK that may be in power to weight or volume maybe.
Just wondering where the IPM is heading ?

[BigMouse]

I'm pretty sure Honda and GM (I assume Ford too) are also using IPM, though don't quote me on that.