Sunday, September 20, 2009

What's Next: Epic Axial Motor

Nothing good can start this way...

It was only a matter of time.

The axial-gap motor has been haunting my dreams for some time now. At least since I found $4 wedge magnets (see above for proper shipping and handling procedures). There has to be an easy way to make motors with these! But there are some problems to solve:

Most importantly, how do you make the stator? On the one hand, motors are forgiving and will let you get away with almost any ridiculous configuration of magnets, iron, and coils that you want. Think about it. Nothing inherent in the geometry creates power. You put in voltage and current, you get out torque and speed, and you lose some heat.

That said, if you're going to build a motor in-house you probably want your power in the form of more torque, lest your unbalanced mess of a rotor disintegrate at 20,000RPM. The key to getting torque is to "link as much flux" as possible. In other words, get as many turns of wire under as many magnet poles as you can. And then pass an absurd amount of current through them. But you just can't put as much of it where you'd want it (right under the magnets), so a good, high-torque motor will use steel to help funnel and carry magnetic flux from right near the magnets to the windings. It follows, then, that it would be nice to get as much steel surface area as possible near the magnets, leaving just tiny gaps for wires to fit through. (Take a look at most real motors, axial or radial. At the air gap, it's mostly steel adjacent to magnets.)

The steel is laminated to minimize the circular eddy currents that result from changing magnetic flux. This presents a big problem for axial motors, though. They have features in all three dimensions, so each lamination would have to be a different shape. Expensive to buy. Hard to make. This is about where my though process stopped....

Until I met the YASA motor. This Oxford creation solves so many practical problems that it almost makes you forget about the "trivial benefit" of higher power density resulting from an inherently minimalistic stator design. Originally, the EVT TTXGP crew was going to partner with these guys to produce a motor for the racing motorcycle. But now that partnership looks to be impossible and I'm left wondering: Can we make a version of this cheaply, quickly, and efficiently? I believe the answer is yes:

Just my take on it...

1) The stator problem is solved. Well, Oxford solved it by making the individual stator segments out of iron power core. But pretend for a second that the goal was to make a motor that didn't want to return to a powerdery state when dropped on the ground. If the pole count is high enough, each segment can be made from a stack of progressive H-shaped laminations:

I probably overuse powerpoint.

If you are running on a tight budget, you can even make every lamination the same shape, and then just (gently) grind down the flanges later.

2) The next obvious practical benefit not really highlighted in the technical paper: You can wind each segment independently. No feeding of tiny wires through tiny slots. No finger blisters. No trying to clamp the stator without destroying the existing windings. Once they're all wound, you can just bolt them to the hub and make the interconnections. Speaking of windings, with these magnets and 16 poles, the flux rate at any significant RPM is going to be quite high. I'm not really interested in making a high voltage motor, so the number of windings will be pretty low and the current pretty high. Which begs the question: why use wire at all? How about copper sheet!

Copper sheet conductor.

If each segment only needs less than a handful of windings, why not wrap some sheet around them concentrically? And the insulation? How about Kapton tape! Now you've got a motor that has a pretty good fill factor due to the rectangular conductors, a way to hold them together, and a temperature rating that has got to be higher than that of magnet wire enamel.

3) Another problem with axial motors is high thrust loading due to the fact that the lowest energy state of the motor is a collapsed, stationary pancake, hopefully without finger remnants inside of it. The bearings take a lot of axial load and the rotor especially has a significant bending moment acting on it at the radius of the magnets. This design provides a nice clean way to deal with this: connect the two disks together with a rotating outer can that can go into compression (not pictured in the paper or my little drawing, but I guess it's implied). Good bearings are still a must, but this will make life easier.

4) The segments are more-or-less independent of each other. Flux goes in one end, comes out the other. There is no steel path in between. The practical implication of this is that you can test the motor without a full set of segments (less development investment). How? Build the rotor. Wind one segment. Spin the motor up externally. Measure the back EMF. By virtue of geometry, superposition, and power conservation this can give you most of the important specifications about the motor....I think.

So it begins. We've got laminations inbound and a significant amount of design and testing to do. Could be a big project. Whether it winds up on a motorcycle or not, I don't know. But I am a believer in component-driven design. That is, if it's too cool not to have (like an 8" neodymium wedge magnet set), buy it now and figure out how to use it later...

Now what?

Check back for more axial motor updates and the inevitable design and fabrication of a controller for it when all the off-the-shelf options fail yet again.


  1. you mentioned the YASA motor used a powered iron core. That's not exactly true. It looks like they used Somaloy, which is an exotic material. It is powered metal with high tech coatings on the microscopic particles to reduce resistance. It has to be pressed at ~400MPA and then sintered.

  2. Thank you to the anonymous post above, the SMC material is getting better and better from a losses and cost point of view too.

    Interesting discussion Shane, I'd like to know if you have an updates on your take on the yasa axial flux machine? We're always interested in improvements and cost reduction opportunities.

  3. What are the voltages and amperages, that is the wattage or KW of the 120 NM and the 500 NM motors?

  4. "Check back for more axial motor updates" Do you have any updates on the axial motor? I was considering to build a motor on my own but finally gave up a few years ago. Very difficult if you have too less technical background, but I am impressed.

    Chris R.
    Webmaster, Learning Linux

  5. Oh yeah, there have been plenty of updates since this post. It chanced to a coreless motor and then I discovered why winding it with flat wire was not as good an idea as I originally thought:

    Right now it's sitting under a bench, but some day I will return to the axial motor...