Hot-rodding the Toyota MGR

[P-hack]

Originally Posted by e*clipse

But you don't get something for nothing, and the current will have to be increased if the voltage is decreased. This extra current will result in more I^2R heating, demanding more heat rejection just to keep the same continuous power output.

Just to clarify, the motor windings will see the same current, if you have 100 amps going through two fields in series, or 200 amps split between two 100 amp fields in parallel, sameo-sameo as far as the motor is concerned. The supply will see higher currents, but not really within the motor itself, since you were focused on the motor heat issue. Same is true for proper rewinding for lower or higher voltage, in that the power stays the same.

[e*clipse]

Just to clarify, the motor windings will see the same current, if you have 100 amps going through two fields in series, or 200 amps split between two 100 amp fields in parallel, sameo-sameo as far as the motor is concerned. The supply will see higher currents, but not really within the motor itself, since you were focused on the motor heat issue. Same is true for proper rewinding for lower or higher voltage, in that the power stays the same.

Ok, that makes sense, since it all comes down to current flowing through a coil.

Could you clarify this issue?
This may be more of a control issue, but I've heard that the circulating currents in a delta wound motors can be an issue. It's pretty clear how things work in a wye motor, but how can imbalances be controlled in a delta motor? For example, the impedance through every coil may be the same, but the momentary BEMF could be different, resulting in circulating currents that don't provide any torque.

Thanks,
E*clipse

[P-hack]

Originally Posted by e*clipse

Could you clarify this issue?

Probably not, it is a bit nebulous. Admittedly I have focused on induction, and in DIY they are almost exclusively delta (i.e. HPEVS, reconfigures), pretty much make it reasonably symmetrical and you are golden.

BLDC might be an issue, pmac perhaps less, dunno, reluctance maybe less so. Lots of conflicting information out there. Minimize 3rd harmonics

here is the ES take on magnet rotor delta
https://endless-sphere.com/forums/vi...4&hilit=+astro

"Astro's share poles and overlap.


That tends to be a problem when delta connected. The non-overlapping configuration is less likely to have circulating currents in delta"

one other thing to note is that the phases in the picture *look* like they are different lengths, i.e. one phase is at the root of the slot, another in the middle, another towards the rotor. But there are a number of delta pm motors in use.

[freebeard]

The 25kW continuous power rating of the MGR is more than adequate for a small runabout like freebeard wants to build.

Thanks for the mention. It's been what? a year? since I got that sub-frame missing the bracket. I'm looking to get out of the air-cooled scene to finance an electric tadpole, but the last pieces to go will be the '58 Beetle and the Lexus MGR.

The project he mentions replaces a 36-hp prime mover with the 68hp/50kW device—which should be adequate for a single-speed drive train. I'm curious about optimizing high-torque, low-rpm operation (rock-crawling) and I think the solution here is oversizing the inverter components. The motor cooling would be a combination of oil cooling and a heat emissivity coating with close-fitting fan shroud.

[P-hack]

here is another ES post, I've read a fair bit of liveforphysics stuff, he *seems* to understand it pretty well.
https://endless-sphere.com/forums/vi...=25136#p362983

Recirculating currents (sometimes called winding circulating current, or a few other names) are due to the inherent imbalance of induced potential inside the conductor loop created by the winding termination. This creates an inherent voltage imbalance between one side of the loop to the other, and since they share termination points (hence making the loop), they have internal closed system current flow inside this loop that changes every time the motor has a change in magnet-tooth position. Depending on the motor wind/design, this current can change from being so extremely mild it would be almost impossible to notice in a practical application (like under <1% efficiency) to something in an extreme effect that makes 10-15% difference in efficiency (but you would pretty much have to optimize a motor around maximizing circulating current losses to get it that bad.)

[e*clipse]

Originally Posted by P-hack

Probably not, it is a bit nebulous. Admittedly I have focused on induction, and in DIY they are almost exclusively delta (i.e. HPEVS, reconfigures), pretty much make it reasonably symmetrical and you are golden.

BLDC might be an issue, pmac perhaps less, dunno, reluctance maybe less so. Lots of conflicting information out there. Minimize 3rd harmonics

here is the ES take on magnet rotor delta
https://endless-sphere.com/forums/vi...4&hilit=+astro

"Astro's share poles and overlap.


That tends to be a problem when delta connected. The non-overlapping configuration is less likely to have circulating currents in delta"

one other thing to note is that the phases in the picture *look* like they are different lengths, i.e. one phase is at the root of the slot, another in the middle, another towards the rotor. But there are a number of delta pm motors in use.

Thanks for the links & info.
It appears from the various links that one can expect about 3% losses due to the circulating currents. If it's a trapazoidal controller, things could be much worse due to the 3rd harmonics issue.

So, there's basically an engineering compromise decision at hand - whether to accept the 3% losses in exchange for a lower bus voltage. 3% doesn't sound like much, but it is 1.5kW more losses at full power. To put this in perspective, my 2kW induction stove boils water pretty rapidly - like a "professional" gas stove.

With the MGR, the wires aren't brought out so setting it up for delta would involve messing with the internal windings. Even though I've completely disassembled one of my MGR's I haven't found the midpoint of the wye. I'm not saying it's impossible; just there's more work involved.

Because of similar extra costs and losses I decided to drop the booster idea. I figure I'll be just as dead if I touch 650VDC or 325VDC - but I guess that's a personal choice. If anyone is considering doing this please please please do it with super-paranoid safety precautions.

- E*clipse

[P-hack]

well, if delta and boost are out, short of rewind the next logical step is seeing exactly how many amps it can handle at low rpm (and how long), and playing with the gearing to get the wheel rpm back up.

got a thermal imaging camera and a dyno?

edit, maximizing cooling would be helpful too of course.

edit2, can you skip a reduction stage in your gear holder thingy?

https://www.youtube.com/watch?v=h6-G8Zn0CrE

[e*clipse]

That's a real cool video of the thermal image/ dyno test!

I'd love to get my paws on a thermal imaging camera - for a whole bunch of reasons.

Maybe it would be possible to do some dyno testing like thingstodo is doing using another motor as the load.

I do know it's not possible to remove the midshaft part of the gearing. While the gears look close, they don't fit.

- E*clipse

[dremd]

Subscribed.

[MPaulHolmes]

I've been trying to get it running for the last couple days. I've got the resolver to encoder working, and the PI loop tuned. I've never seen behavior like that before when tuning the PI loop. At 48v it would take around 0.01 sec to converge. Every other motor I've ever tried would converge in 0.001 sec to 0.002 sec. When I tried 120v it would converge in 0.003 sec. Also, I got it to run "smoothly for a fraction of a revolution, and then it sloshes back and forth (but very smoothly). I'm becoming convinced that it's like 50% synchronous and 50% reluctance. I don't think you can just ignore the reluctance component, even for casual testing to get it to spin. What seems to be happening is, it rocks back and forth between where the reluctance component is near zero and where the synchronous component is near zero. So, what I"m going to do is get the reluctance correction code running on the Leaf motor (since that seems to be like 90% synchronous 10% reluctance), and then switch back to the mgr.