Toyota MGR Rear Traction Transaxle

[UStas]

I think MGR is perfect only as Motor+Reductor.
If you want to keep existing reductor, just find standalone motor and connect it to propellershaft or directly to reductor if space enaugh.
If you want to keep only motor, you can remove reductor and keep motor housing, but you must made a new shaft and seal a bearings and think about lubrication and cooling.
Here is sample of similar modification, but here general idea was a two output shaft for 4WD -

[ZooKeeper]

Options I'm considering are:
1 - Hack the casing apart, take a shaft straight off the motor then find a way to keep it cool and stop it pissing oil everywhere.

While "anything" is possible, marrying Al to FE is an engineering task for threaded fasteners and alignment is critical; so not a viable option.

2 - Lose the casing and make a new one for just the motor

Same as above, plus the resolver will not appreciate.

3 - Rewind the motor for silly speed and hope the gears can take it

Since you would need DOUBLE or more, that is not advised, yet possible.

4 - Accept defeat, use this as a test motor and swap for something more suitable when I've got the rest of the build worked out

If the subject can be adapted to IRS, that would make things much easier in terms of fabrication and modification.

[Damp_Cuttlefish]

Thanks for the replies!

While "anything" is possible, marrying Al to FE is an
engineering task for threaded fasteners and alignment is critical; so not a viable option.

I was hoping to avoid most of that.
I still haven't actually got my hands on the motor, so I'm probably missing something obvious.
Looking at these pictures Kevin posted earlier it seems like you can run the motor on its original bearings without any of the rest in place.

gearboxsideofstatorsection_zps136725da.jpg (24.88 KiB) Viewed 7667 times

My hope is that I can replace the now exposed inside bearing with a sealed variety and have it do double duty as oil seal.
I'd then tap the oil channels between motor and gearbox and either block or use them as an oil feed/drain with an external pump.

Independent suspension would definitely be the optimal route, but would be a fairly major operation. I think if it came to that I'd give up and buy a new motor.

Here is sample of similar modification, but here general idea was a two output shaft for 4WD -

That certainly looks promising. I hadn't considered taking power off the encoder side, but perhaps that would be the easier route.

[celeron55]

I've gotten this to spin using a Leaf gen2 inverter and FOC firmware. My plan is to test how much torque it can exactly create, given 500 amps.

So far it's spinning at 180rpm (= 24.2rpm at axle) given 14.4 volts which seems what one would expect.

[celeron55]

About 5000rpm @ 275V when feeding it from a 10A 230VAC mains circuit. Iq reaches -20A which isn't much. I'm trying to use some kind of parameters that won't immediately blow the fuse. Next I'll start setting up the torque test.

EDIT: Attached the parameter file

[mdrobnak]

Nice!

[celeron55]

Rate my chances of staying alive operating this standstill torque testing device...

EDIT: Just a 12V supply and a laptop away from SCIENCE

[ZooKeeper]

Depends on how good your helmet is -

[Bratitude]

just send it

[celeron55]

So far I'm not dead, and the jig stayed intact, which is already better than expected.

I made some measurements increasing the current in steps, but then got oscillation at higher current and my scale went all wonky. I need to log some stuff and see if udc goes below udcmin or something. Maybe I'll also get a cheap digital scale, those probably stand more abuse than these fiddly mechanical things.

Anyway, this is what I got so far. The current is simply throtcur*potnom. I haven't checked whether the inverter is reading correct current values compared to a clamp meter so you should assume it's probably wrong. On the other hand, I'm assuming the force measurements and calculated torques are somewhat accurate. Probably a good idea for me to check Iq and Id also, the inverter shouldn't be using much Id in this case.

You can extrapolate from these values that to make the published 900Nm torque, it would take 391A. And I need to go above that in order to learn what I want to know.

[celeron55]

Okay that way of measuring current isn't reliable and I'm already facing my input current limit at the 450Nm tipping point in that graph. So up to that point it's (probably) accurate, from there the actual current is lower than what I'm requesting. I decided to use the phase currents from now on and made new measurements.

I'm feeding the inverter using a 500W isolation transformer and it's already pulling more than 1000W at the maximum here. Need to set up another power source (EDIT: about 830W actually). This obviously means though that to make 551Nm at the axle, the MGR and inverter will be heating up at about 1kW. I'm going to guess most will be at the MGR as 190A is pretty much nothing to the Leaf inverter I'm using here.

Overall this looks better from the standpoint of driving this using a 350A Prius power stage, as extrapolating from here gives about 1000Nm at 350A.

You can see there's a lot of Id being applied, I'm not sure if it's a good idea at a standstill situation but given it's there this test does give a more realistic view on what the torque will be at speed also.

EDIT: Re-uploaded the graph with a small fix

[Jack Bauer]

Thats excellent work. Reckon it will move an E39 quite nicely:)

[thornogson]

Excellent pioneering stuff there , just like science but with more fear !

Should move this along ok in the spring:

viewtopic.php?f=11&t=1165#p21776

[celeron55]

Skipped the isolation transformer, now I'm limited by my 10A fuses and 10A mains circuits. Still, got some more measuring done.

My goal is to get to the point where the graph is no more linear. So far it's straight as a nail. At the highest measuring point about 2.3kW seems to be used.

EDIT: I'm thinking I might need to double the torque and current to get to the non-linear region. To achieve that I probably need to quadruple the power, that is about 9kW and the Leaf inverter wants that at under 500V. I'll need to rig up a HV battery.

[ZooKeeper]

So if I understand your set-up, you are feeding a max of 2300w (230v @ 10A into the inverter) and able to achieve 700NM equivalent wheel Tq?? I simply must be missing something

If so, that is indeed VERY promising and it matches well with my guesstimates based on the limited testing I have done thus far.

[celeron55]

We have positive HV battery arrivement. 80kW @ 360V should be more than enough to either destroy everything or give me the results I want.

(Somehow this 1980s garage only needs 750W of heating to stay at 0°C while it's -27°C outside. Also, -27°C is probably the ideal condition to run an MGR powered vehicle.)

[Jack Bauer]

Don't die on us! I need to rip off more of your software:)

[celeron55]

I increased idcmax up to 40 and this is the result.

I think there's like 35A going into the inverter at maybe 335V at the maximum point which would be around 12kW. At the end the MGR did feel slightly warm to my extremely cold hands, meaning maybe its temperature increased from 0°C to 20°C or something, so the total of maybe 60 seconds of maybe an average of 5kW definitely were turning into heat.

So this proves the MGR will output 1200Nm at the axle, given a total phase current of around 650A.

The "bang for amp" did decrease after 800Nm, but the graph isn't getting flat. I think I need to crank up idcmax even more. Obviously there's no way it can produce these torques for more than a couple of minutes without literally melting, but sometimes that's all the time you need.

EDIT: Graphed rough power values

[tom91]

Does the motor not contain a simple thermistor you can monitor to determine rise in temperature?

[celeron55]

My board doesn't have the correct pull-up resistor and resistor divider for that sensor so the default firmware doesn't show the correct temperature. It is connected though, I guess I should log the bogus value whatever it is.

(Openinverter feature request: A simple way to roughly calibrate a temperature sensor no matter what it is without modifying the firmware or PCB. Personally I'm fine with a +-15°C accuracy)

[rstevens81]

Don't die on us! I need to rip off more of your software:)

I'll second that I have spent most of the christmas holidays (on& off) studying your charger code, at the very least I have learnt a huge amount on how own on the atmega 328p (Arduino) works

Thanks for your work, the potential for your work is huge!
(Your making me think about a 2nd conversion before I have started my first )

[ZooKeeper]

Does the motor not contain a simple thermistor you can monitor to determine rise in temperature?

Yes, one can use the oft-shared standard Toyota resistance to temp chart to *fairly accurately* measure mtemp. The sensor is within the windings so fairly accurate.

[celeron55]

The firmware expects a 10 kohm pull-down for the Toyota temperature sensor instead of the 510 ohm default one, is that correct?

This doesn't seem to be documented anywhere, not even in the code. What's certain is that my 510 ohm pull-down doesn't give correct readings when the Toyota motor temperature sensor is selected.

[MattsAwesomeStuff]

Damien - Someone on the Isle of Man is giving away their MGR. I supposed that you might want it, for... purposes. I spoke up and suggested you might make use of it, and what you might do with it that would benefit the community. Just throwing it out there.

https://www.diyelectriccar.com/threads/ ... st-1067899

[johu]

The firmware expects a 10 kohm pull-down for the Toyota temperature sensor instead of the 510 ohm default one, is that correct?

This doesn't seem to be documented anywhere, not even in the code. What's certain is that my 510 ohm pull-down doesn't give correct readings when the Toyota motor temperature sensor is selected.

Indeed that didn't fit the range, see here: viewtopic.php?f=14&t=1087&p=17817&hilit ... ure#p17817

You will find that kind of stuff also in misc/temp_sensors.ods