The motor, in ASCII art
A fellow tinkerer has contacted me showing his own project of in-wheel motor.
His design is of Axial Flux with double rotor, and he shows me a series of details and shares a few enhancements over my original axial flux single rotor design.
I found this event quite gratifying; it’s the kind of confirmation I like to see that the smartest people on the planet are always the first to openly share what they know and have, in hope that their contribution will feed progress. Everybody wins. 
But meta-remarks aside, I deeply enjoyed the simplicity and elegance with which he showed me his design: using ASCII art. It was very clean and explicit. And it also made me think that it’s been a long time since I posted the design I’m working on (and since people are still not sure if I’m working on an Axial or Radial Flux motor, I think it’s time to make things clearer).
So, here it goes. My model is a Radial Flux motor, with double Rotor. And it is designed to fit outside the brake drum, with the rotor fixed between the wheel hub and the brake drum, and the stator fixed to the suspension arm. This is a view in cut, along the axle of the wheel. On the left is the outside of the wheel, and on the right the inside (suspension).
r r
rrrrrrrrrrrrrrrrrrrrrrrrrrr
r
r RRRRRRRRRRRRRR
rr R MMMMMMMMM
rr R
rr R CCSSSSSSSSSCC
rr R SSSSSSSSSSSSSSS
rr R CCSSSSSSSSSCC S
rr R S
rr R MMMMMMMMM S
rr RRRRRRRRRRRRRRRR S
rr R S
rrrR ddddddddddd s S
rotor rrrRdddd dd s S
bolts --> rrrRdddd dd sssSsss
here rrrRdddddddddddd sssSsss <--- stator bolts here
rrrR ooo ooo sssSsss
rrrR aaaaaaaaaaaaaaaaaa
(Exterior) rrrR aaaaaaaaaaaaaaaaaa (Interior)
rrrR ooo ooo sssSsss
rrrRdddddddddddd sssSsss
rrrRdddd dd sssSsss
rrrRdddd dd s S
rrrR ddddddddddd s S
rr R S
rr RRRRRRRRRRRRRRRR S
rr R MMMMMMMMM S
rr R S
rr R CCSSSSSSSSSCC S
rr R SSSSSSSSSSSSSSS
rr R CCSSSSSSSSSCC
rr R
rr R MMMMMMMMM
r RRRRRRRRRRRRRR
r
rrrrrrrrrrrrrrrrrrrrrrrrrrr
r r
Already existing components in the car (small cap letters):
"r" - light-alloy wheel rim;
"d" - brake drum;
"o" - bearings;
"a" - axle;
"s" - suspension arm;
New components to add so to build a motor into the wheel:
"R" - rotor (stainless steel for disc plate, soft iron for magnetic core);
"M" - neodymium-iron-boron magnets;
"C" - coil windings of stator;
"S" - stator (stainless steel for disc plate, soft iron for magnetic core);
There it is, in all its non-scaled glory. The red parts group is the rotor I’m building, and the blue parts group is the stator.
This fellow tinkerer, named John Bass, sent me his design explaining that the “1 ton force problem” was resolved by using a double rotor. I had thought of that too, at the time, but that disposition creates another problem: assembly. Such a motor (axial flux, double rotor) is impossible to separate into 2 parts (rotor and stator) without undoing the rotor itself. This is something I didn’t want for my design; I wanted something easy to assemble and disassemble into the wheel, something that integrates 100% into the wheel and car without modifications and also allows for easy maintenance of the brakes, just the way they are now. That’s why I dropped the axial flux in favour of the radial flux: this way I can use double rotor layers and still easily remove the rotor to allow access to the brake drum.
Another difference between his design and mine is that I put the motor outside the brake, whereas he put the motor on the inside and the brake (disc) on the outside. I prefer my design, because this way I can get more torque from the same electric power (the mechanical arm – the radius – is longer).
Anyway, I’m still running the batch simulations… I already found 2 good candidates, 1 Halbach variant with 36 poles and very thin core capable of 950Nm, and another normal variant with 60 poles and thick iron core capable of 1700Nm, both at 100A per phase. Now, there are a lot of approximations incorporated into these simulations, so a better analysis will be done later. But now I just want to find out the best geometries, and this is good enough.
July 7, 2008 at 11:01 am
Have you been running the simulations non-stop since your 1st post? I can give some CPU time if you want
July 7, 2008 at 1:31 pm
No, I’ve been on vacation!!!
I spent 10 days driving around England, and at night I used a borrowed laptop to tweak my scripts while doing a little simulation.
I had finished the simulation of the Halbach model by that post, and now I was doing the normal model, but it needed some manual work before it could be 100% automated.
I synchronized with the sourceforge repo, and this eventually saved my work, because the laptop burned out the hard drive on friday, losing all the output…
Although I don’t think it was my usage; there where already some file corruptions when I started using it. 
I had about 70 variants computed over 4 days in that machine, but it’s ok, I’ve already recovered that much work since last night by running the simulations in my home machine. Hehe… multicore rocks!!
Thanks for the offer, though. But I may actually make use of it someday… beware!!!
July 7, 2008 at 6:45 pm
1700Nm times 2 motors = supercar territory… nice!
My usual A-class math:
0-100 in 4,26s
0-200 in 9,72s
0-300 in 16,18s
(almost EliminatorIII territory http://www.wurts.net/ElectrifiedMotorSports/)
And a world record of 534km/h in a hybrid or electric!
If this doesn’t go well here, go sell in the states!http://www.nedra.com/
July 7, 2008 at 6:53 pm
ouch … brake drums will be dumping 500+C heat across your rotor. Both your magnets and adhesives will need to take up to maybe 500-600C heat. Worst case is coming to a panic stop from 90-110+km/hr and not moving for a while … about 1/8 of that energy is going to heat the upper rotor segments as the cast iron drum is going to be cherry red at a panicstop.
High temp magnets are not cheap, nor very powerful. Check the operating temp and currie temps
July 7, 2008 at 8:17 pm
Yes, I’m aware of that, John.
In fact, that’s been worrying me since the start, because Neo magnets are normally graded to work at a max temperature of 80 C. Some go as far as 120 C.
The Curie temperatures are higher than that (300 C, I think), but they start losing their strength around 80.
So yes, temperature management is a big challenge.
That’s why I’m planning to have water cooling on the stator.
Of course, that doesn’t really help the rotor…
But H0tR0d already gave me the crazy idea of spraying the rotor with water in case of a “panic stop”, and I may just do that – with a built-in sprayer and automatic temperature monitoring!!!
July 7, 2008 at 8:53 pm
The biggest problem is that you have put the rotor right in the middle of
drum’s cooling air. This will reduce the airflow around the drum by closing
up the space between the drum and the inner wheel diameters, which is a
carefully designed in cooling fan/chamber. Plus, your rotor plate will need
serious holes to allow air in from the outside of the wheel.
Net of this, is magnets on the rotor will always run very hot in stop and go
traffic … first the drums heat them stopping, then the waste energy from
accelleration will continue the heating.
Any liquid around cherry hot cast iron is going to crack/warp something big
time. Probably your magnet rotor too.
This is the reason I’ve been looking at axial side-by-side with disc brakes.
Both independently get cool air from opposite sides of the wheel, and can be
heat baffled from each other.
July 7, 2008 at 9:05 pm
since the inside of your drum vents directly to the stator plate and into the gap between your rotor magnets,
how are you going to keep the cast iron metalic dust out of the rotor magnets?
July 7, 2008 at 9:39 pm
John: both excellent points.
I’m seriously considering a 100% regenerative braking system, which would avoid the overheating problem – the drums would then only be needed as parking brakes (no heat) or in case of electric system failure (hopefully extremely rare).
I know, I know… nobody has 100% regen… the Prius gives up regen’ing at above 60km/h (or something like it), and most of the DIY EVs don’t even use it.
However, I’m dead serious in investigating this. My first cost estimation of supercaps was very disappointing, but I need to do it again with more detail. I believe it would be a great development and worth the extra investment.
As for the brake dust, you’re again absolutely right: it is a big problem if you use them frequently. And even if the usage is extremely reduced by regen’ing, there is always some dust production. I was thinking of making the internal rotor slightly conical (and longer towards the stator than in the picture), in such a way that the dust would slide out towards the wheel rim and out the rim holes. Of course the rotor plate would have to have holes too. Think its feasible?
July 7, 2008 at 10:01 pm
I’ve changed the ASCII pic to reflect the above (longer internal rotor).
July 7, 2008 at 10:57 pm
Regen mode is not the only way to break, you know?…
July 7, 2008 at 11:07 pm
Yeah, I know… But my philosophy is: if you’re gonna use electricity to brake the car, you might as well make the power flow in the right direction!!!…
To charge the batteries!
July 7, 2008 at 11:59 pm
No one said you would have to pick just one method
July 7, 2008 at 11:59 pm
Seems to me you have eliminate the drums, in favor of disk brakes, unless you only intend to use the parking brake.
The heat is manageable but the dust… Sure we can have holes, but anything with iron won’t go out.
And there’s another rather BIG problem: the magnetic force of neo magnets will pull the brake shoes against the drums constantly braking the car!
July 8, 2008 at 12:00 am
Concerning supercaps (for the regen part), are they cheaper than Li-Ion?
July 8, 2008 at 12:02 am
I think they won’t h0tr0d, because the magnets are enclosed by iron, which will divert (most) the magnetic forces from everywhere else.
July 8, 2008 at 10:58 am
I can’t remember if supercaps are cheaper or more expensive than Lithium, but I do remember the blocking point was the energy density: supercaps probably won’t replace a chemical battery because they would be too large for any useful capacity. But I’ll be working on that…
About the brake dust and the magnets: if the motor is well built, it will have very little magnetic loss (flux that goes outside the iron cores). Then again, these magnets are extremely powerful and I am trying to save weight on the iron cores, so anything is possible. I guess prototype testing will be the phase to clarify these points. Still, the geometry will help keep the dust away from the magnets: there is an air gap of about 2 cm between the brake drum surface and the interior rotor surface, and the rotor wall can be skewed conical like I said before. I hope this way the brake dust will slide out of the wheel, and if not, it accumulates on the rotor outer surface, not inside the magnetic air gap.
For the same reasons, I don’t think the brake pads will be attracted by the magnets. The escaped flux should be far too weak to cause any attraction on them, and as NJay said, they are shielded behind an iron drum which is at least partially ferromagnetic.
The friction heat should not exist if braking is purely regenerative. NJay, are you proposing a combined braking system (electric regen + friction brake)? That sounds good, but it would be a bitch to fine-tune… and the necessary servo-system would be too complex for my taste – too many complicated parts waiting to break down. No, I don’t think so… KISS.
July 8, 2008 at 8:54 pm
We’ll see my Friend, we’ll see
…
July 9, 2008 at 9:19 pm
I can understand the appeal of using all regen braking. But do you think you can really get the power densities required out of diy wheel motors. Even Volvo used disk breaks on their concept using PML designed wheel motors.
If you are only using your wheel motors in the rear, rather than on all four wheels, why not use the breaks on your front wheels for higher speed stops. This type of combined breaking system wouldn’t require much mechanical complexity since the abs system probably already uses speed sensors, etc. The work could be done by reprogramming the computer (which I figure you will probably have to do anyways with all the hybrid enhancements you are making).
July 15, 2008 at 10:16 am
Will: you’re absolutely right. The scope of the conversation (the rear-wheel motors) made me forget about the front wheels.
It seems I cannot escape the smart integration between the hydraulic brakes (front axle) and the regen brakes (rear axle). However, I am tempted to throw out the whole hydraulic system and put motors in the front wheels as well (without any brakes over there). My car has a very complex braking system, with independent speed sensors and brake controls at each wheel (ABS, ASR, ESP, blablabla). Integrating that with the electric motors may be too big a hassle; I’d be better off replacing the whole thing with reversible electric machines. I’m still thinking about it, though…
Why do you think that’s easy? I’d like to hear your thoughts.
“But do you think you can really get the power densities required out of diy wheel motors?”
Yes, I do. First, I’m aiming for an overall power that is lower than Volvo’s. Second, I don’t think the problem is in the DIY motors, but in the electrical storage system – it will be impossible to harness the full power of an emergency breaking without supercaps – and even then… And throwing it out via power resistors is not an option for me. But if my car weighs as much as the Volvo, then it has to have the same braking power even if it doesn’t have the same accelerating power. Tough cookie, isn’t it? What do you suggest?
July 15, 2008 at 10:29 am
Oh, about reprogramming the computer: that’s not my intention, because I don’t have the tools or expertise. My intention is to isolate and complement the original computer with another one of my own.
In the parallel version, the DIY computer has to pay attention to the inputs and outputs of the original computer, and compensate the required power (accel. and braking). It cannot overdo it, or the original computer will go crazy. But because the brake and accelerator pedals have a big “dumb spot” at the beginning of their course, I can use that to make the electric motors shine at low power levels. At higher power values (pedals depressed more deeply) the original computer and engine/brakes do most of the job.
In the full series version, the original computer becomes “blind” to most of the car, and it manages only the engine, being directly controlled by the DIY computer. The engine’s accelerator input is an output from the DIY computer into the original computer. The hydraulic system being gone, the ABS and other control strategies are implemented via the DIY computer and electric motors. Braking and accelerating are fully controlled by the DIY computer.
How’s that for a plan?
July 15, 2008 at 6:04 pm
First off let me say I usually know how to spell “braking,” at least when I’m not posting at around 1am. Perhaps easy wasn’t the correct word. But is there anything worth doing that really is easy.
While cars share similar abs characteristics I admit I do not know that much about your car’s abs system. From what you said though, it sounds like a 4 channel 4 sensor setup. When I said “reprogram” I think I had more in mind what you are planning. If you can find the circuit board diagram of your abs system, (You can find just about any car on http://www.alldatadiy.com), you could use your DIY computer to duplicate the signals from pedal and wheel sensors and wire directly to the inputs. (Admittedly, finding that signal info might be difficult). The control would be a straightforward logic circuit using the criteria of a sudden stop to trip the signal telling your front brakes to stop. Just an idea, it allows you to keep
July 15, 2008 at 6:15 pm
also remember that your motor design, at least according to an older post, was based on 2 motors powering the car, if you add 2 more motors you would probably have to redesign the lot of them, figure out controlling your motors at different speeds according to the degree of turn, and possibly beef up your power steering to account for the new weight. All of these things are not impossible, but they do set your motor design back a few posts.
I am designing an electric car with rear wheel motors right now myself, so I am paying very close attention to your blog, I’m interested in your design, and am very anxious for motor construction to begin. Good luck.
July 16, 2008 at 11:23 am
Will: thanks for the support and the “alldatadiy” tip.
There’s only a “small” problem with that thinking: it’s not enough to just “fool” the original computer with “poisoned” inputs (as if that wasn’t complicated already). I will have to disable the hydraulics to the rear brakes to allow full regen braking, and I don’t know how much trouble that will bring me into the front brakes and ABS system…
As to the motor design, it is still in a state of flow, so I haven’t made any compromises yet that cannot be easily redone. So, presently, it doesn’t make a difference (for the rear motors) if I’ll also use front motors or not. The only impact I see is the choice of power rating. Ok, making a motor for “X” kW or “2*X” kW is not exactly the same, but that should be it in terms of differences for the rear motors.
If and when I set my mind to the front motors, it will be a new project entirely. The present motor design is completely dependent on my rear axle geometry and should not be adaptable to the front axle without major re-engineering. We will cross that bridge when we get to it. I can always build the car with a pair of rear motors that simply covers the basic needs, and later on add a pair of front motors. Then I may need to electronically restrict the total acceleration power, but the higher regenerating power is welcome.
As to the intelligent control of the motor speeds while in curve, that’s a necessary requirement no matter what you do. From the moment you put in more than one motor, you have that problem to solve. Like you said, it will be necessary to calculate the different wheel speeds according to the steering angle, but this will have to be done whether you have 2 or 4 motorized wheels, so we really have to deal with it anyway. This is the only place I envy the guys with DC motors: they can connect them in series, put in a voltage across both motors, and they will share the load and the speed “automagically”. This is called the “electric differential”, and with AC motors it has to be done by electronic control (software).
But as you also said, “is there anything worth doing that is really easy?”
I wish good luck to your project as well, and if you have some way of keeping us posted (a blog?), please publicize it – I also want to follow your progress!
October 29, 2008 at 11:59 am
I am searching for a complete motor to test it by beginning of january .
Do you think that it is possible to have drawings ( Solidworks or equivalent by that time ?
I find the project very interesting and pratical works could be possible .
Best way is to contact me directly by mail
October 29, 2008 at 2:47 pm
Pierre:
Right now I can’t get you anything more concrete than a 2-D CAD file.
To get such a file, you have to:
- download my FEMM lua scripts;
- change them to your needs (put your own parameters and criteria);
- run them inside FEMM;
- save the output model that you want as a DXF CAD file.
I don’t think I will have anything better than this by January – I have a lot of work to do until then.
Meanwhile, be advised that my models DO NOT YET HAVE THERMAL MANAGEMENT considerations, which means that the windings are not dimensioned for real-life usage.
What you do with my models is exclusively your responsibility.