Donut Lab's next-generation in-wheel motors | Hacker News Reader

Donut Lab's next-generation in-wheel motors

donutlab.com

70 pointsklemola39 comments

I spent a number of years designing outer-runner direct drive permanent magnet motors for industrial equipment and I've got some questions.

Cheifly, bearings. They're not shown in any of the oh-wow images, but these will likely be the most expensive component of each motor. Big bearings are expensive, and to accept the loading of normal wheel operation, these will have to be pretty beefy. That's not even discussing operational life and maintenance.

After you've stuffed a pair of angular contact roller bearings into this "wheel", you're going to want to keep salt water and road grime from entering those bearings, so what do you use as a seal? Whatever you use is going to be big, expensive, and suck up huge amounts of power due to the large contact surface.

Finally, once you've got big ass bearings and big ass seals, how do you have enough room to put a decent amount of copper in there? Power in these things always amounts to maximizing the amount of copper in the space, and I just don't see room for it.

Does "suck up huge amounts of power" for a seal imply "generate lots of heat"?

If so, is that heat another issue or is it a "don't care" because the heat is over a large enough surface?

It absolutely matters. Every Watt of energy that doesn't become torque at a rotational velocity is just heat.

Contact seals work by contact and friction, friction generates heat proportional to linear velocity and linear velocity goes up proportionally with radius.

The motors I designed were intended for food production washdown areas, and if I were designing large motors for use in road environments, I would use a lot of similar methods, including high quality contact seals.

Teflon seals would probably have the required capabilities, but they will get destroyed by dust and grit. Nitrile seals would do it too with the detraction of a huge power loss at the seal. I wouldn't trust a plain labyrinth seal to do the job.

> Every Watt of energy that doesn't become torque at a rotational velocity is just heat.

Do you have any rough numbers to put on this? If there was 1000w of electrical power going to a wheel like this, what kind of heat loss are we talking? 5%, 10%, 30%?

Depends on the seal manufacturer. Those numbers are usually provided through their engineering data system. Their data will be fit to a particular tolerance for seal race surface finish, which will be influenced by the reality of manufacturing.

Right, but are we talking closer to 5% or 50%?

Allow me to embrace my engineer nature and hedge my bets. A contact seal of that size is usually specced for a power transmitting shaft. An (assumed) 20" shaft is going to transmit a huge amount of power. So much that the seal losses will be negligible. Those same losses would exist on any shaft/seal combo of that size, but would take a greater fraction of total motor power, given the size and power constraints given by an internal motor design.

This is an offshoot of a motorcycle manufacturer. They have this product on the market already, so there must be something you missed?

https://arstechnica.com/cars/2024/07/sci-fi-looks-high-end-p...

I cannot speak to their actual means and methods. Maybe they've figured it out. But my experience designing and manufacturing similar products informs my skepticism.

I also look back in history and see many "revolutionary" technologies in the automotive/transportation space that didn't turn out as the inventors hoped. A veritable graveyard of "good on paper" ideas that failed due to the harsh realities of environment, maintenance, safety and manufacturing.

As I said in another comment, I want them to succeed. I am familiar with the great efficiency that a direct drive motor can offer. They can be great motors in the right application. But my evaluation of what I can glean through the marketing is that this particular product will end up in the graveyard.

The way I understand hubless wheel designs (powered or not) is that you don't build them as one big bearing, wasting huge amounts of load bearing capacity in all parts of the rotation that aren't the ground contact point. I assume that the moving part is the rim is designed as a rail, with tiny trucks (as in the rail car component) riding on it that are fixed to the non-moving part. You'd have a high density of strong trucks near the floor, some at the three and nine o'clock positions for braking and acceleration force and perhaps some flimsy guiding on top. Those trucks would not necessarily require more sealing (outside their own small bearings) than the rail/wheel contact in railroads need sealing. And moving that rail a little hubward, behind a lip that extends rimward would already get you strong centripetal forces driving out all ingress in contact with the moving part, and adding some overpressure (that you might need for cooling anyways) would help even more. I think it could all remain contactless on that first, whole-wheel level, at least if you don't design for routine wading.

Plus there's the unsprung mass. At least traditionally part of suspension performance is reducing the unsprung mass as much as possible because mass and spring rate inversely correlate.

At first I was like "this is crazy, those are going to be crazy expensive, they'll never make it" and then I realized we're not the customers, at least not willingly

https://www.donutdefence.com/

The Oruga Unitrack [1] is listed as one of their case-studies, it looks awesome and I would really like to see it move.

I found [2] on YouTube, but it doesn't seem to contain any actual video of the vehicle (and the voice-over says "unit-rack" rather than "uni-track" which I didn't love).

1: https://oruga.eu/#unitrack

2: https://www.youtube.com/watch?v=clL4sDXxvWQ

Looks like they're targeting "civilian" products as well https://www.vergemotorcycles.com/

Maybe they thought of this, but I didn't see any mitigation for the unsprung mass problem that these motors bring.

If the weights are listed on the are to be believed, this shouldn't be an issue. It says 40 kg for a 21" wheel. A quick Google is telling me that a regular old steel wheel from a truck of the same size is around 50 to 80 lbs (23-36 kg).

That said, I was already thinking 630 kw per wheel was a pretty incredible claim before I realized these are apparently not much heavier than a non-mptorized wheel. These have to be some marketing department numbers or something. 630 kw is roughly 850 horsepower.

There's a whole lot of marketing speak, finger waggling, and wishful thinking in what I am seeing here.

Copper is heavy. So is silicon steel. So are high strength magnets.

I love the idea, but this one's going to have a real world bite in the ass once they get it out of the lab.

Well, you can already buy one: https://www.vergemotorcycles.com/

Well that is interesting.

I still have my doubts, but I wish them luck. I've always wanted to get Kenada's bike.

> If the weights are listed on the are to be believed, this shouldn't be an issue. It says 40 kg for a 21" wheel. A quick Google is telling me that a regular old steel wheel from a truck of the same size is around 50 to 80 lbs (23-36 kg).

I'm skeptical; They say 40kg, but I think that's just for the motor, not the entire wheel. I working off the (maybe incorrect)[1] assumption that the 40kg doesn't include the steel/mag rim and the tyre.

As far as a 21" standard mag wheel, the tyre alone is around 12kg, so quite believable that a standard wheel with tyre would weigh maybe 30kg. However this still means that putting a tyre on a 40kg wheel is going to take it to +52kg.

[1] Just the minimum metal and rubber needed for a 21" wheel to maintain its shape and structure should be around 30kg. Maybe this motor is structurally round already, so doesn't need any rim to reinforce it?

A regular wheel is about 12kg bringing wheel+tire to ~25kg

At these weights, I suspect the vehicle won't handle as well as a vehicle with the motor offboard.

I think it still needs the wheel, and all the related parts in the middle like an upright/steering knuckle, which probably cancels out any mass of the motor that is acting as a structural part of a wheel. Other things like brakes would still be required. These should allow for regenerative braking normally the braking torque of a motor is nowhere near the acceleration torque, plus they don't mention braking function anywhere so it's safe to assume there's no special performance there. In all, I would assume for now that a 40kg motor would add approximately 40kg to the overall wheel mass, possibly even more. It would be interesting to see if there's a concept for an inboard version of these motors and compare to others.

They mention in their CES interview that putting two of these in the rear axle + a lighter version in the front can give you a 2500hp+ car. That is the other breakthrough, and matches the comparison chart they have on the website against other tech.

Disk brakes add another 4-5kg, maybe their wheel hub doesn't need one? At these power levels, regen is probably stronger than the largest brake disc you could fit.

You'll also have at least 150kg less on the car (motor/differential). Hard to tell what handling would be like until someone builds a test car.

I'm thinking about the motorcycle motor. If they're really lighter than competing motors, I don't see the downside of using them like a traditional motor, taking the weight savings, putting it inside a sealed cavity, and coupling it to a driveshaft. Simplifies some of the other problems with sealing and bearings mentioned elsewhere, avoids unsprung mass.

More on their design can be found in their patents, https://patents.google.com/?assignee=verge+motorcycles&oq=as... Although that may not be their most recent work due to how long patents take to get published.

NB reminder that your employers legal policy may be to never look at patents.

There's a concept called "unsprung mass", which basically destroys handling of all vehicles from race cars to trucks. Basically the greater the unsprung mass, the harder it is to damp the input into the suspension because of the inertia of the moving suspension components themselves. An ideal suspension has zero undamped mass, and all input to the suspension is a direct result of contact with the surface the vehicle is traveling on.

There is zero chance this tech will make it into sports cars unless it can beat the weight of a magnesium or AL alloy rim. Even casual vehicles like minivans have rim weight minimized for comfort.

Not an expert :) just watched enough Donut media on youtubes :P

One thing to add, while your wheels and tires are a big chunk of the weight here - you also have the uprights/knuckle and the wheel hub, bearing, brakes, rotor etc etc. All of that adds up quick. Most times if you are looking to reduce this on a car you will have the most gains (or losses haha) switching to an aftermarket wheel. No wonder the manufacturers slapped alloys on their cars too!

“The unsprung mass (colloquially unsprung weight) of a vehicle is the mass of the suspension, wheels or tracks (as applicable), and other components directly connected to them. This contrasts with the sprung mass (or weight) supported by the suspension, which includes the body and other components within or attached to it. Components of the unsprung mass include the wheel axles, wheel bearings, wheel hubs, tires, and a portion of the weight of driveshafts, springs, shock absorbers, and suspension links.

…

The unsprung mass of a typical wheel/tire combination represents a trade-off between the pair's bump-absorbing/road-tracking ability and vibration isolation.“

https://en.m.wikipedia.org/wiki/Unsprung_mass

There's also the inertia problem where it's harder to accelerate mass that's further away from you. The classic spinning ice skater demo. This feels physically disadvantageous. Hopefully I'm wrong though, the more competition in spaces the better.

But here that mass is the driving force: the further away from the center you put your magnets, the less force (the less magnet) you need for a given amount of Newtonmeters. When you move that mass away from the center, you need less mass, changing at the same factor as rotational inertia changes. So no improvement or drawback in terms of rotational inertia, but improvement in terms of total mass (and improvement in terms of unsprung mass when comparing to a wheel motor in the hub).

In a way this is a transfer of semiconductor miniaturization to motors: when your power transistors can switch fast enough, you can replace fewer bigger coils with more smaller coils (that switch more often per rotation) and moving it all rimward gives you more torque per Newton. That ice skater effect? It's on your side.

Pretty powerful stuff. The Model Y Long Range AWD has a pair of motors that deliver Peak Power of 286 kW and Peak Torque of 510 N⋅m. Donut's claims their Automotive (21") motor can 2.2x the Power (630kW) and 8.4x the Torque (4300 Nm)

That's high enough that I have to assume it's for all four wheels... if it's for a single wheel, then an all wheel drive vehicle would have 2520Kw Power + 17,200 Nm Torque, which is 1.6x more than the most powerful production car in the world: the Lotus Evija (1,500 kW).

This page [0] says "Donut Platform empowers hypercars [...] Delivering 1,500kW and 9,000Nm of total wheel torque, with acceleration from 0-100km/h in under 2 seconds"

I don't see how you go from a single 286kW/510Nm hub motor x4 gets you 1500kW/9000Nm instead of 2520Kw/17200Nm. Wonder what the limitations are + what nonsense they are trying to pull in their single motor stats.

[0] https://www.donutlab.com/industries/

I guess I'm not really following. The engine's in the wheel?

Yeah. In fact, the article linked below[1] is more informative than the page I linked. TL;DR: there’s a big reduction to the overall weight of an EV using these, but they do add to the unsprung mass of the vehicle (the weight below suspension, as mentioned in another comment).

[1]: https://www.cnet.com/home/electric-vehicles/this-donut-shape...

In the same space https://yasa.com/ axial flush motors.

Is the control of these things good enough that there's no need for a steering mechanism? Or I should say "mechanical steering". Devices like the Segway steer by changing the wheel speeds just a bit. Could this deliver enough control to steer well enough at highway speeds? I'm guessing it could work in a parking lot.

Have they put one in a car yet? Maybe I’m a simple person, but the first thing I’d do if I was CEO of that company would be putting my ultra-powerful motors in a car and heading to the drag strip.

I see how the one on the motorcycle looks. But why does the car and semi just look normal? I wonder what the bearing configuration looks like.

Looks like one of many "shovels for gold diggers" unitized motors for robotic dogs and humanoids. There's been a lot of such compact all-in-one motor units from China lately.

Reply on news.ycombinator.com