More flux, please!

Posted on February 23, 2008


Njay showed me a very cool movie of a Japanese in-wheel motor built for a solar race car. It contains a feature totally uncommon in electric motors: it regulates the magnetic flux through variable geometry.

Check it out:

Fig.1 – “TIGA” Variable Air Gap Permanent Magnet Synchronous Motor of the GOGO81-WSC solar racer.
Copyright “tmise86“, Ashiya University, Japan.

It’s a very cool way to use an essential (and very old) technique called “field weakening”: the reduction of the magnetic flux in the motor. Why would anyone want to reduce the flux??…. because, if you reduce the flux at maximum speed, you also reduce the back-EMF, causing an increase in effective current and motor speed. So it’s like using an overdrive gear after the normal gearbox: you get more speed out of the motor, but less torque (because there’s actually less magnetic flux and therefore force). I tried to find a good explanation for this complex phenomenon on the web, but couldn’t… if you find any, let me know.

Anyway, this video triggered a neural response in me: I can no longer ignore the fact that I’ve been playing around with a lot less magnetic poles than the other boys who have developed in-wheel motors.

If you pay attention to the video, it shows a stator that is full of coil windings, side-by-side, with practically no space between them. And I’m still playing around with the LRK model, which has a lot of space between coils, as well as “dead” stator teeth that have no coils… I’ve also been noticing that my designs are taking very large currents and large numbers of coil turns to produce the high torque I need. The source of my problems seems to be the low number of stator teeth, that not only implies less force-exerting poles but also forces the saturation of the magnetic parts and high iron and copper losses.

I need more flux, but this design can’t cope with it. It’s time to make some changes!…

In a general manner, the more stator teeth the better. But I still want to retain the good things about the LRK model: the low cogging torque and the fact that it works well as a 3-phase motor. So I just multiplied the number of magnetic poles and stator teeth by 2 and by 3 and ran the “magnet scan” script again for each variant. As expected, I got twice the torque for twice the active teeth. And three times the torque for three times the active teeth. 😉

Here are the pics of the generated models. I’m showing you the same model, using the same exact magnets (3x3x3mm) for a fair comparison:


Fig. 2 – LRK 12 teeth, 137Nm.


Fig. 3 – LRK 24 teeth, 411Nm.


Fig. 4 – LRK 36 teeth, 745Nm

However, my scripts try to find the best match of magnet size for each model, and this means the maximum torque computations will yield different magnet choices, depending on the other geometric constraints. At 80 Amps per coil (square wave), 50 turns per coil, with the radial flux Halbach model, the winners were:

  • Original LRK / 12 stator teeth: 240 Nm; 28V; 2,24 kW; (1.232 magnets (D=10,W=10,H=10)mm);
  • New LRK / 24 teeth version: 522 Nm; 56V; 4,48 kW; (4.928 magnets (D=5,W=5,H=5)mm);
  • New LRK / 36 teeth version: 745 Nm; 82V; 6,56 kW; (14.196 magnets (D=3,W=3,H=3)mm).

We can see 2 trends here; as the number of stator teeth increases, the size of the magnets and the radius of the motor decrease. This has to do with my mathematical efforts to make a sequence of homogeneous magnets fit inside the two rotors. But the present model can still be improved, because it was not designed for many teeth, so close together… I’ll be trying that next.

Another thing that pops out is the incredibly enormous number of magnets necessary for these models… hardly the kind of economic decision I’d like to make. I have to improve that too.

Keep in mind that, because I locked the current per coil at a fixed value, the electric power consumed is proportional to the number of coils. But the higher number of coils also gives me higher electric freedom: I can choose to wire the coils in parallel, in series, or a mix of the two. That way I can also pick if the whole system will be high-voltage or high-current. The above simulations where done with all coils in series, hence the high voltage.

High-voltage is better for efficiency because it implicitly means lower current for the same power, and that means lower copper losses everywhere: in the motor, in the vehicle’s wires, and in the controller’s power electronics. And thinner wires also mean less weight. This is the approach that today’s hybrids on the market follow. However, there is always the problem of human safety when dealing with voltages above 50V, and a car is a bad environment for dangerous components… ahhh, but that’s a whole other story, best left for another day… 😉

Posted in: Motorfemmulator