"It's possible to do some or all of this in software. Tesla has an ordinary open differential but will apply the brake on an overspeeding wheel. Tesla's all wheel drive system has a separate motor for the front and rear wheels. Power distribution between the two during acceleration is mostly equal, but once speed stabilizes, the rear powertrain takes most of the load. Off-road slip handling doesn't seem to be well documented, but Tesla cars aren't intended for off-roading."

I don't think about this a lot, but I assume that all of the interesting ways to distribute power from a single engine to 1-4 wheels are, all of them, inferior to having a dedicated motor for each.

That is, I assume that no matter how fancy you made your differentials, two motors is better than one and four motors (one for each wheel) is better than two (all else being equal).

Is that correct ? Or is there some scenario(s) wherein mechanical distribution of torque (with differentials) is superior to (again, all else being equal) a dedicated motor on that wheel ?

...

If I try to answer my own question, all I can come up with is:

1. If you have a motor for each wheel, the max output on that wheel is the max output of that motor, and theoretically, you can distribute more than 1/4 of the single engine to that wheel with differentials, so ... maybe that's a very big deal ? Do 4WD vehicles often send 70-80-90% of output to one wheel ?

2. Sending power to a specific wheel via the path of least resistance takes zero time - it's instantaneous - whereas deciding what to do with each of the four wheels (in software, presumably) might have a lag ... although that sort of breaks my "all else being equal" tag, above ...

> If you have a motor for each wheel, the max output on that wheel is the max output of that motor, and theoretically, you can distribute more than 1/4 of the single engine to that wheel with differentials, so ... maybe that's a very big deal ? Do 4WD vehicles often send 70-80-90% of output to one wheel ?

The situation that leaps to mind would be an especially muddy, low-speed, off-road terrain where the differential is keeping three wheels from spinning. The neat thing about that situation is that, in spite of having at most 1/4 of the vehicle's power available, electric motors are still probably a big win, even if you're towing something.

That's because with electric motors you've got almost all the motor's torque available at low RPM. With a gasoline or diesel you've got to get RPM up a bit before getting the power you need. Getting the vehicle moving and balancing RPM and wheelspin and everything is a goofy exercise that would be seem to be made a lot easier by just being able to just gradually bring up the throttle, which you can't always do with a ICE motor.

And more to the point, 1/4 of the total available torque is still more than you're likely to get with even a good diesel and a good automatic transmission that deals with everything gracefully at low-speed, low-RPM. Maybe I'm wrong about that last part: all the off-roading I did was in a primitive Jeep with a manual transmission.

> I don't think about this a lot, but I assume that all of the interesting ways to distribute power from a single engine to 1-4 wheels are, all of them, inferior to having a dedicated motor for each.

While off-roading it's not uncommon to completely loose traction on two wheels (search for cross-axle articulation for a way to mitigate this). I've also lost traction on three wheels during steep hill climbs, rutted, muddy traverses, and on sand.

In situations like that locked differentials are vastly preferred as you can transfer the majority of the car's torque to 1-2 wheels. A motor for each wheel (or each set of wheels) would not be preferred as 1/4 or 1/2 of the vehicle's torque is often not enough to maintain forward momentum.

The challenge is that the axles of many stock vehicles are not designed to withstand all that torque, so breaking your vehicle's axle when all the torque is transferred to one wheel is a real risk.

One more thing to think about, on this line of questioning, is efficiency in "normal" driving situations.

If you're driving in a straight line, with good traction on all wheels, will a single larger motor be more efficient than two smaller ones on each wheel? For many people, I think that's 99% of their driving, so it makes sense to optimize. Especially on electric vehicles, where increasing battery capacity is much harder than just adding an extra gas can.

Maybe in that case the complexity of AWD/4WD/Traction control makes sense in that case.

One motor per wheel, in the wheel, is an old idea. It's been used most notably on LeTourneau heavy equipment. The usual problem is too much unspring weight, not a problem for giant earthmovers but bad for fast cars.

Michelin was pushing it for cars, with their "Active Wheel" concept, from about 2003 to 2012.[1] That seems to have disappeared. Siemens has demoed a motor-in-wheel unit, and Volvo and Nissan have fooled around with this. Protean, in Shanghai, is trying to sell their wheel motor. There are others. Nobody has shipped production cars yet, though.

[1] https://news.ycombinator.com/item?id=383607

"One motor per wheel, in the wheel, is an old idea. It's been used most notably on LeTourneau heavy equipment. The usual problem is too much unspring weight, not a problem for giant earthmovers but bad for fast cars."

hmmm ... I wasn't thinking about motors in the wheel, although I am familiar with that concept.

I was thinking about a more pedestrian motor per wheel configuration wherein the motor is just behind the spring ... and is thus, sprung weight ... is there even space for that ?

> Tesla has an ordinary open differential but will apply the brake on an overspeeding wheel.

Is that different than the traction control used by other automakers?

> Tesla's all wheel drive system has a separate motor for the front and rear wheels.

That is very cool. And it sounds much more complex - I wonder if the complexity is worth the benefit.

Separate motors for front and rear wheels is far less complex than adding more moving parts.

For AWD, you need a driveshaft and a differential. The differential has a lot of parts, there are a lot of bearings that sap power, and it adds a lot of weight.

Using separate electric motors for front and rear instead increases the available power, enables unlimited torque ratios from all-forward to all-rear, and has many fewer moving parts.

Thanks. But isn't coordinating the motors complex? If the power delivered to the wheels is out of sync for a moment, I would imagine it could be catastrophic, and the system has to deal with variables of turns, traction, acceleration/braking, etc.

Why would it be?

It's not like trying to coordinate rowers in crew; you have full, absolute, instantaneously-responding control over the motors, as well as information on exactly how fast they're going, how much force they're applying, how fast each individual wheel is turning, what the steering angle is, etc. And then what's controlling it is a computer, not a human.

Stability control, which already exists in every car, deals with all of those variables and is able to correct for driving with judicious braking (or release of the brake) on individual wheels, even when some or all may be slipping or locked up, the car going sideways, on varied surfaces.

Adding more things for a computer to control makes it easier, if anything.

Keeping the front and rear motors in sync is a cake walk compared to controlling the brakes for traction control.

"If the power delivered to the wheels is out of sync for a moment, I would imagine it could be catastrophic" The control software is probably updating every 3-15 milliseconds, so even if the software had a massive hiccup, it would have to last for multiple frames for it to actually get translated to the pavement. The drivetrains of cars are actually pretty mushy. Everything's mounted on bushings, there's backlash in the gears, and tires are rubber. All of this adds together to buffer out any kind of spikes. Much like a capacitor and resistor can be used to buffer electricity. So if, say, the front wheel was 5% underpowered for a frame, then all that would happen is the gears in the front differential would go slack for a few milliseconds. If on the next frame, the power was corrected, the differential would go back to being taut. If it was just 5% underpowered, you likely wouldn't even feel this. If the front motor totally locked up for one frame, you'd still probably only feel a little jitter.