[jddcircuit]

What does anyone think about this motor.

http://www.youtube.com/watch?v=2olyxIhN_HE

The VIN on the transaxle tells me it is from a 2001 prius hybrid. I got the motors spinning last week using a couple of motor control boards and a small inverter. The transaxle has two large brushless DC motors inside. The whole unit has transmission fluid inside dripping through the stator and rotors of each motor and the housing has water cooling jackets for each electric motor. MG1 and MG2 are connected by a planetary gear with the planet carrier shaft protuding where the hybrid ICE would connect. MG2 connects to a differential with a 4.11 to 1 gear ratio.

In the video I am driving MG2 which is the larger of the two motors and MG1 is held stationary. My test setup in the video is only using 24 VDC limited at 3Amps. The output of the differential spins at 1.6 rev/sec with my 24V supply. I calculate this would equal 6.8 mph with a 24inch tall tire. So I would expect to be able to go 68 mph top speed if I had a 240 volt battery pack since my understanding of brushless dc motor speed to voltage relationship is linear.

In the video you can hear some grumbling of the gears. I'm not 100% sure I put it back together perfectly and there isn't any fluid in it right now. Or maybe that is why it was scrapped in the first place.

My wife just texted me and said a large box just arrived at the house. I would assume that this is the High Voltage prius inverter that I ordered from a wrecked 2002 prius. My goal is to use this transaxle and the prius HV inverter with my own control unit design for my electric volkswagen project.

I assume that if this motor worked for a prius then it would work well for my ev. Not sure.

Has anyone used this motor before or have they built their own controller for the prius inverter?

Thanks
Jeff

 

[Coulomb]

In the video I am driving MG2 which is the larger of the two motors and MG1 is held stationary. My test setup in the video is only using 24 VDC limited at 3Amps. The output of the differential spins at 1.6 rev/sec with my 24V supply.

So that's about 96 RPM. So MG2 would be doing about 96 * 4.11 ~= 400 RPM.

I calculate this would equal 6.8 mph with a 24inch tall tire.

If it's 24" diameter, that would be 24*pi = 75" = 6.3' = 6.3/5280 miles. That's per second, so multiply by 3600 for miles per hour; I get 4.3 mph.

[Edit 2: forgot the 1.6 rps; 4.3 * 1.6 = 6.88 mph. Sorry!]

On a stock Prius (Gen II) with 15" wheels (not tyre diameter), they get about 68 mph at MG2 = 4000 RPM. So maybe I'm wrong above; I'm not good with imperial measurements.

[Edit 1: the above assumed a final ration of 3.905 and a rolling radius of 11.1" (i.e. 22.2" effective tyre diameter). I figure that these two differences would almost cancel out though.]

So I would expect to be able to go 68 mph top speed if I had a 240 volt battery pack since my understanding of brushless dc motor speed to voltage relationship is linear.

The relationship is linear, unless you use flux weakening. My understanding is that the Prius inverters do perform flux weakening, so they can exceed the speed implied by the DC bus voltage by increasing the frequency and weakening the flux, hence reducing the back EMF, so the motor doesn't regen back to the speed predicted by fixed field strength.

I assume that if this motor worked for a prius then it would work well for my ev. Not sure.

As designed, the Prius transaxle isn't perfect for EV work. However, you have a 33 kW motor plus another capable of about 10-15 kW (nobody seems to be sure about the power of MG1 in the various Prius models), so if you can combine these sensibly, you should be able to power a light EV from that. A Gen II transaxle with 50 + 15(? maybe 30) kW motors would provide better performance especially with larger host vehicles.

The main limitation is the speed of MG1. It seems to be capable of 10,000 RPM (Gen II), but seems to be limited to 6500 RPM if the ICE isn't spinning. Presumably, you won't be using the standard control software, so you should be able to let MG1 go to 10,000 RPM (make sure it has lubrication!).

 

[jddcircuit]

So that's about 96 RPM. So MG2 would be doing about 96 * 4.11 ~= 400 RPM.

If it's 24" diameter, that would be 24*pi = 75" = 6.3' = 6.3/5280 miles. That's per second, so multiply by 3600 for miles per hour; I get 4.3 mph.

I think you forgot to multiply by 1.6. You are correct that 4.3 mph is for 1 rev per sec. I was getting 1.6 revs per second at 24 volts

 

[Coulomb]

I think you forgot to multiply by 1.6.

Oops! Yes, I did. When the 1.6x is factored in, the result is 6.88 mph.

Sorry!

 

[jddcircuit]

As designed, the Prius transaxle isn't perfect for EV work. However, you have a 33 kW motor plus another capable of about 10-15 kW (nobody seems to be sure about the power of MG1 in the various Prius models), so if you can combine these sensibly, you should be able to power a light EV from that. A Gen II transaxle with 50 + 15(? maybe 30) kW motors would provide better performance especially with larger host vehicles.

What exactly does it mean that MG2 is rated at 33kW. I would think that I am mostly limited by the amount of continuous current through the motor before it overheats. How much current do you think it can handle? If I stay within the maximum current I would think that I could increase voltage to acheive higher wattage?

 

[zaxxon]

Has anyone used this motor before or have they built their own controller for the prius inverter?

I am using a Microchip development board to drive a Gen II Prius Inverter and Ford motor. Have driven car a block or so, but still have control and or current limit issues.

You were looking for pinout info. In the Gen II (2004 to 2008 ?) , the inverter's control pin out is shown on the inverter control board. Not sure same for a 2002 Inverter.

 

[Coulomb]

What exactly does it mean that MG2 is rated at 33kW. I would think that I am mostly limited by the amount of continuous current through the motor before it overheats. How much current do you think it can handle?

I would think its peak current would be close to 33,000/273 = 121 A (input to a 100% efficient inverter; actual phase current is possibly half that if 2 of the three windings is energised at any one time.)

I guess that the continuous power rating would be less than that, limited by the cooling system. But I'm only guessing.

If I stay within the maximum current I would think that I could increase voltage to achieve higher power?

Well, yes, as long as the wire insulation doesn't break down, and the IGBTs don't avalanche. My guess would be that you could apply a fair bit more voltage without the wire insulation breaking down, but the voltage limits of the IGBTs will be pretty firmly fixed. The higher you go, the greater the chance that some random glitch of voltage will add to the bus voltage and break down an IGBT, which will then turn into a dead short.

You can probably get an idea of the breakdown voltage of the IGBTs if you could get a part number. However, they could be proprietary devices.

A common voltage for IGBTs is 600 V, meaning you can safely use bus voltages up to about 450 V. But these may not be common IGBTs.

 

[jddcircuit]

I am using a Microchip development board to drive a Gen II Prius Inverter and Ford motor. Have driven car a block or so, but still have control and or current limit issues.

You were looking for pinout info. In the Gen II (2004 to 2008 ?) , the inverter's control pin out is shown on the inverter control board. Not sure same for a 2002 Inverter.

I partially dissambled the inverter last night to see what was inside. It is a 2002 so I guess that means gen 1. I don't remember seeing any pin description labeling but the thing appears very modular and I hope to be able to figure out the pieces that I need. If you have any source or suggestions on where to find pinout or diagnostic documentation It might help me reverse engineer the thing.

I bet there are some similarities between the gen 1 and gen 2 versions that might help.

Thanks
Jeff

 

[Coulomb]

If you have any source or suggestions on where to find pinout or diagnostic documentation It might help me reverse engineer the thing.

There are wiring diagrams available; I bought a CD-Rom with hundreds of diagrams for about AU$10, from Ebay.com.au, similar to this one:

http://cgi.ebay.com.au/TOYOTA-PRIUS...lothing_Merchandise_Media&hash=item230bfec46a

Example page attached. This particular vendor posts only to Australia; I'm sure you can find a similar vendor that posts to the USA, or use the Toyota download facility (available only in the USA; grizzle) to get a copy of an appropriate PDF. The PDF this image came from is from this file:
D:\Toyota_Prius_01_03_WSM\2001-03 prius\Wiring_Diagrams.pdf
(in case that helps).

I haven't seen a component level circuit of the inverter; that of course would be better, but knowing that a signal is supposed to do is also important (and not always obvious from a component level circuit diagram).

Of course, another challenge is figuring out what codes like "GSNG" actually mean. You can guess because there is also a "MSNG" from MG2 that the first letter means "G" for generator and "M" for motor (older Prius terminology for MG1 and MG2 respectively). Maybe there is a list of codes on the disc somewhere that I haven't found yet.

 

[zaxxon]

I partially dissambled the inverter last night to see what was inside. It is a 2002 so I guess that means gen 1. I don't remember seeing any pin description labeling but the thing appears very modular and I hope to be able to figure out the pieces that I need. If you have any source or suggestions on where to find pinout or diagnostic documentation It might help me reverse engineer the thing.

I bet there are some similarities between the gen 1 and gen 2 versions that might help.

Thanks
Jeff

Check out this link:
http://www.techno-fandom.org/~hobbit/cars/ginv/i3elec.html

 

[jddcircuit]

Check out this link:
http://www.techno-fandom.org/~hobbit/cars/ginv/i3elec.html

Zaxxon

Thanks for the link. I think it will help me. I see his waveform for a single phase of the inverterj in his electrical analysis portion. I am still studying it but it is not clear to me yet how I am going to implement my 6 step BLDC commutation pattern since it requires me a third state. I need to be able to turn on high side, low side, and no side. During each step one of the 3 phase legs is floating in my scheme. I would assume that this would require two control lines per phase and not just one. I haven't had much time yet to figure it out but if this makes any sense to you please share.

Thanks
Jeff

 

[zaxxon]

Zaxxon

Thanks for the link. I think it will help me. I see his waveform for a single phase of the inverterj in his electrical analysis portion. I am still studying it but it is not clear to me yet how I am going to implement my 6 step BLDC commutation pattern since it requires me a third state. I need to be able to turn on high side, low side, and no side. During each step one of the 3 phase legs is floating in my scheme. I would assume that this would require two control lines per phase and not just one. I haven't had much time yet to figure it out but if this makes any sense to you please share.

Thanks
Jeff

First I am no expert, but if you use space vector control it seems you do not need independent control of each of the six transistors.

See: http://en.wikipedia.org/wiki/Space_vector_modulation

All PWM phase states can be controlled by three inputs with states of "on" or "off". Since you never want to turn the Top and Bottom transistor of a Phase output on at the same time (it would short out the battery) you can use one control line per Phase. When "high" it turns on the Top (high side) transistor and when its "low" it turns on the Bottom (low side) transistor for example. You can set each of the states needed with this three wire configuration.

 

[jddcircuit]

First I am no expert, but if you use space vector control it seems you do not need independent control of each of the six transistors.

See: http://en.wikipedia.org/wiki/Space_vector_modulation

All PWM phase states can be controlled by three inputs with states of "on" or "off". Since you never want to turn the Top and Bottom transistor of a Phase output on at the same time (it would short out the battery) you can use one control line per Phase. When "high" it turns on the Top (high side) transistor and when its "low" it turns on the Bottom (low side) transistor for example. You can set each of the states needed with this three wire configuration.

Zaxxon,
Thanks for the link. I think that will help me. I can modify my control to use this type of modulation and then be compatible with the prius inverter. It looks like my reference vector will need to shift 15 electrical degrees when I do make the change. Should not be too difficult.

Jeff

 

[Coulomb]

... I am going to implement my 6 step BLDC commutation pattern since it requires me a third state. I need to be able to turn on high side, low side, and no side.

I'm not sure that that is possible or wise. After turning on any transistor, the opposite transistor is needed as a sort of replacement for the catch diode in a DC controller. Without the equivalent of a catch diode, there would be nowhere for the energy stored in the inductance of the windings to go, and the resultant high voltage spikes would likely kill everything in short order.

Maybe the "parasitic" diodes (actually real in the case of the Prius IGBTs, I think) of the IGBTs would effectively take over, but they may not be designed to take full current, or at least not for a large portion of a PWM cycle.

 

[jddcircuit]

I'm not sure that that is possible or wise. After turning on any transistor, the opposite transistor is needed as a sort of replacement for the catch diode in a DC controller. Without the equivalent of a catch diode, there would be nowhere for the energy stored in the inductance of the windings to go, and the resultant high voltage spikes would likely kill everything in short order.

Maybe the "parasitic" diodes (actually real in the case of the Prius IGBTs, I think) of the IGBTs would effectively take over, but they may not be designed to take full current, or at least not for a large portion of a PWM cycle.

You are probably correct. Having a transistor on is probably more efficient then relying on the freewheeling diode.
I was a little stuck with how to interface with the High Voltage Prius inverter since it appears to only use 3 control lines (one for each phase) instead of all six (one for each transistor). The low voltage inverter board that I was using for prototyping independently controls all six transistors. The demo software that I am modifying uses a modulation scheme that relies on one of the three phases of the motor to be floating (both high and low side off) at any given time. It needed this floating phase in order to measure the zero crossing of the back emf for synchronization with the magnetic field rotation. Since the Prius motor has a resolver for angular position feedback I don't need to use the back emf of the motor to sync my commutation. Zaxxon enlightened me to another modulation scheme that does not produce a floating phase and that should be compatible with the high voltage inverter from the Prius. If my understanding is correct then I will need to adjust for this magnetic field change by shifting by 15 electrical degrees in my software (I think). Another advantage I just realized is that since the inverter only requires 3 control lines I can now drive both MG1 and MG2 using the six outputs that my motor control board has. Hopefully I will have time soon to test this out. I am now working hard on the mechanical adapters needed to fit my motor into the 66 Volkswagen I just bought. This is another engineering challenge in itself.

Thanks
Jeff

 

[peggus]

If you're using an IGBT you're relying on the anti-parallel diode whether you're turning it on or not, and yes you have to have an anti-parallel diode or the magic smoke escapes. For mosfets they are intrinsic so the mosfet will be protected, but it is more efficient to turn on the mosfet than use the diode.

Six step sensor less control relies on one phase being un-powered to detect the zero crossing, but you want to use a position sensor anyway since sensor less control requires the motor to spin to work, leads to a slight problem at standstill.....

One phase will still be un-powered for some time, it is inherent in all voltage fed inverters. Sounds like the prius inverter has this deadtime built in and therefore fixed. Should be no problem in your application.

 

[Coulomb]

The demo software that I am modifying uses a modulation scheme that relies on one of the three phases of the motor to be floating (both high and low side off) at any given time. It needed this floating phase in order to measure the zero crossing of the back emf for synchronization with the magnetic field rotation.

Ah, of course. They must wait for the average current (not just PWM instantaneous voltage) to go to zero before they disconnect the other transistor. Maybe this is something that is usually done only on smaller motors.

 

[Batterypoweredtoad]

Why not just use the Highlander rear motor/diff setup? Seems easier.

 

[Arlo]

Interested in how this goes for you. IM not sure i understand how the thee wires control the 6 igbts and on top of that if you need you can find igbts with higher ratings to suite you needs. I think the prius motors can take a lot more amperage for a short burst then they are rated for.

 

[jddcircuit]

Interested in how this goes for you. IM not sure i understand how the thee wires control the 6 igbts and on top of that if you need you can find igbts with higher ratings to suite you needs. I think the prius motors can take a lot more amperage for a short burst then they are rated for.

It is going well thanks. The 3 half bridges also have a shared enable signal. Once enabled then all low side xtrs are on. Pulling one of the phase control lines low will switch the low side off and high side on. You don't have any control over deadband.

I don't think more amps are going to be necessary. The motor spec shows in excess of 300Nm of torque for the larger motor. I now have both motors connected so plenty of torque at the rated currents.

I am doing some testing very soon to better characterize the motor. These interior permanent magnets have some special considerations.

Jeff