Hot-rodding the Toyota MGR

[freebeard]

(freebeard told me the name of those, and now I forgot)

I think it's willful. But I'm Okay with that.

As long as the trade dress Open Revolt is attached to the result. That would look so good on my Baja.

[e*clipse]

Jiminy Xmas!!!! WoW Paul - you're awesome!!!!

How fast was the motor turning at 375V?

Why did it intermittantly depower when it got near the top speed? I thought it would just top out smoothly like a DC motor when the BEMF matched the input voltage. That's strange...

Just a couple mechanicy things, to answer some questions about orientation and mounting.

- Freebeard: I would recommend NOT spot-facing the backs of those mounting holes. That will most likely result in a stress-riser, leading to fatigue failure. Aluminum is nototious for fatigue falures; I personally wouldn't push my luck by removing material in a highly stressed place.

- Paul: those square thingeees are mounting flanges for halfshafts. For independant suspension drivetrains (front or back) the halfshafts need to be free to change angle without changing the output speed.

This may seem pretty strange, but a universal joint causes the output speed to vary throughout a revolution. This is why they're not used (anymore) on halfshafts, especially on FWD cars. Here's a description of the issue:
https://en.wikipedia.org/wiki/Universal_joint

Fortunately, there are a few good solutions, in constant velocity joints:
https://en.wikipedia.org/wiki/Constant-velocity_joint
Most common are rzeppa joints and tripod joints. Tripod joints don't allow a lot of articulation, but allow in-out movement for suspension articulation. They are frequently found inboard and FWD cars. For my setup, I'm going to use the Rzeppa joint. There are 3 variants made by the VW automotive group (this includes Porsche and Audi) They are very robust and commonly available (relatively inexpensive).

Regarding orientation - I would HIGHLY reccomend orienting the MGR like Toyota intended. The reason is the oil system, and Astro pretty much summed it up. The oil system flings oil into some "containers" that hold the oil and slowly let it dribble out to the gears, bearings, and motor. The oil is used to cool the motor. Any other orientation will pretty much defeat the lubrication/cooling system because the oil won't make it into those containers.

Anyway - AWESOME job getting the MGR running I am estatic!!

- E*clipse

[MPaulHolmes]

I had set the PI loop to a slow, conservative setting (about half of what it could be). Also, I had made a change to the "field weakening" where I allowed IdRef and IqRef to be chopped way down as soon as it it the voltage ceiling. I had been messing around with it, and then I forgot to change it back. So, it was the equivalent of a super bouncy ball when hitting the voltage boundary rather than a mushy not-so-bouncy ball bumping into the voltage boundary. I'm gong to make it more mushy. haha. That effect didn't show its ugly head in the induction case. My guess is because there's a load on this motor, due to the gears spinning, whereas there was no load for the AC motor, so it didn't bother to slow down when its super bouncy ball did the bouncing (to put it technically. haha)

I had made a mistake when I was cutting and pasting stuff all over the place in the code, so the RPM kept saying 53,000. haha. I found out that I had forgotten to save oldPosition, so it was just garbage. I spun it by hand, and now RPM works. I'll test it again tonight if I get the chance, to see the top speed, and change the boundary mush factor. I"m going to coat the voltage boundary in cream of wheat. Then I'll mix in a little butter, and also some sugar on the top, so it looks really shiny, and then I'll eat the sugar off the boundary, leaving the cream of wheat, ensuring that it is very mushy when things bump into it.

[MPaulHolmes]

It's a big improvement now! First I changed how you get the angle offset, so now it's very accurate. Next, revved it up at 120v and the peak mechanical RPM of the motor post inside was 1560RPM clockwise and 1507 counterclockwise.

At 376v, it was 4898RPM clockwise, and 4762 counterclockwise.

I made the bounces a lot smaller when you crash into the voltage boundary, so you can't notice it at all. IT just revs as fast as it can and is stable right at max RPM.

[e*clipse]

Great news!

I just plugged those #'s into my simulation spreadsheet to see how things would work out.
The gear ratio 6.86:1 and the tire diameter is 24.9"
Just as a quickee check, it looks like 4898 rpm would be good for about 52mph. 1560 rpm would go about 17mph.

Were you doing any field weakening or making use of the reluctance torque? I'm wondering because according to the ORNL test, they got 5000rpm at 225VDC (bus voltage) - it's on page 75 of the 2010 Prius test.

- E*clipse

Originally Posted by MPaulHolmes

It's a big improvement now! First I changed how you get the angle offset, so now it's very accurate. Next, revved it up at 120v and the peak mechanical RPM of the motor post inside was 1560RPM clockwise and 1507 counterclockwise.

At 376v, it was 4898RPM clockwise, and 4762 counterclockwise.

I made the bounces a lot smaller when you crash into the voltage boundary, so you can't notice it at all. IT just revs as fast as it can and is stable right at max RPM.

[MPaulHolmes]

Ya, that was with the Id = 0, Iq = throttle. I still haven't run the tests for the saliency (and maybe oil is where it shouldn't be?). HOpefully tomorrow I"ll do that.

[freebeard]

- [f]reebeard: I would recommend NOT spot-facing the backs of those mounting holes. That will most likely result in a stress-riser, leading to fatigue failure.

I ordered the mounting bolts and they are at the Toyota parts counter for me to pick up. I also got a $67.87 price quote to replace the rear mount rubber puck, mine is sort of beat.

My example is up-river so I can't look at it right now. Are those front mounts threaded in the aluminum? Because the parts diagram doesn't show a nut on a through-bolt.

Do you have a solution for the VW Group CV joint on the Toyota axle flange?

[e*clipse]

I'm trying to figure out a "better for DIY'ers" method for connecting these things to a car.

Because there is that rubber puck in the back, there really needs to be some similar equivalent in the front. If the back can flex, then the front will be under stress unless they can too. It may be possible to just ditch the rubber pucks, period. However, there will be more NVH issues.

Yes, the mounting bolts thread direcly into the aluminum. I would be much happier if there was some standard nut&bolt arrangement. You can't fit a nut on the other side of that flange because of the gussetts. However, it may be possible to modify the nut to fit.

Regarding the drive flange - it appears Toyota had some custom tripod joint that bolts to the output flange. I'm trying to make a more general purpose set-up, which is one of the reasons for the VW Rzeppa choice. The type II joints, designed for the Porsche 924 and VW busses, looks appropriate for torque ratings, etc. If you want serious overkill, there is the Porsche 930 joint. Within a particular range, the axle splines are common and its fairly easy to find halfshafts for dune buggies that work. If you need a custom length, there are several companies that make these that don't cost a fortune.

So, I think a fairly reasonable solution is an adapter that can provide a 4 bolt Toyota bolt pattern to a 6 bolt Type II VW. I haven't worked out the details yet, but it is something I could build. The tolerances, metal requirements, etc are things most machinists could pull off.

Since a good part of the machining cost is set-up, maybe there would be enough interest for a group buy?

- E*clipse

Originally Posted by freebeard

I ordered the mounting bolts and they are at the Toyota parts counter for me to pick up. I also got a $67.87 price quote to replace the rear mount rubber puck, mine is sort of beat.

My example is up-river so I can't look at it right now. Are those front mounts threaded in the aluminum? Because the parts diagram doesn't show a nut on a through-bolt.

Do you have a solution for the VW Group CV joint on the Toyota axle flange?

[e*clipse]

Originally Posted by MPaulHolmes

Ya, that was with the Id = 0, Iq = throttle. I still haven't run the tests for the saliency (and maybe oil is where it shouldn't be?). HOpefully tomorrow I"ll do that.

I'm quite happy to hear that - it corroborates a guess I had about this.

Some years ago, a fellow on another forum did a conversion using the MGR, with something like a 375V battery pack. He used a commercial BLDC controller and the car was a Saturn, if I remember correctly. I was stoked to find someone who wanted to do the same thing I wanted to do. Unfortunately, he gave up on the project because he could only get the car to go 50mph. My guess was that there was some problem with the controller, but I didn't understand enough about what was really going on.

Now the problem is clear - the controller probably didn't allow an Id Iq adjustment, and maybe field weakening wasn't done correctly. Since it was a commercial product (rather than open source) the code was inaccessible.

So - - Open source is the best way to approach problems, and Open Revolt Rocks!!

E*clipse

[freebeard]

I got the bolts as samples, the next step is finding shorter ones in the same grade. The marking on the recessed head is two dots 120° apart. I guess I'm reconciled to a small box that fits the mounts like a hood made from 1/4" plate. Then that bolts to the stock transmission mount. In the back maybe I can cut a section out of that subframe I have no other use for. If I get luck I can just rotate it 90° and drill some holes for the bolts in the end of the frame horns.

If you need four and I need two, that's a half dozen. Would a run of ten be no more expensive?