Background: I'm working on my second hobby small generator design. Particularly difficult to source are (relatively) low speed permanent magnet motors that can be run in reverse as a generator with little or no gearing. Thus if I'm designing and winding my own generator, that's the niche I'm trying to fill. I lack the engineering skills to parametrically design a generator (ie. design based on a desired output voltage, torque and rotor speed) so if my wire size or turn count need adjustment I'll just reproduce the stator iteratively.

My first design used a laminate steel rotor core taken from a DC hobby motor, had 8 concentrated coils of 100 turns and 8 rare earth magnet poles. To avoid using a commutator I rebuilt it inside out so the old rotor core becomes the stator and the outside spins. Spinning it with a lathe I was able to get about an amp at 5 volts and 1000 rpm, which pleased me for the size of the machine, but it has an awful, strong cogging torque at low speed. I have extras of the core but the only way to reduce cogging on an 8 slot core is to use a 2 pole arrangement, so I don't believe I'll wind another.

Research on cogging torque revealed the common solutions of multiphasing and of using a number of poles different from the number of coils, for example 3 phases, 3 coils and 4 poles or 3 phases, 6 coils and 8 poles

To resolve this issue, and also due to means of manufacture available to me I've decided for my next design to work on a mostly air core(ABS/PLA/acrylic) axial flux generator with a few possible coil to pole arrangements.

I'm trying to keep my coil count relatively low for a first design and planning to use rare earth disc magnets. From what I've been reading axial flux is seen by many as a nonideal choice, but it has advantages. If necessary I can iterate to designs with higher pole and coil counts to increase electrical frequency or magnet velocity or decrease magnet weight or further reduce cogging.

-radius of device can be increased easily, especially in a non iron core machine, increasing the velocity of the magnets at a given RPM

-ease of producing dimensionally accurate interchangeable parts via 3d printing and or acrylic molding

-ease of adjusting and producing a thin air gap

-does not necessarily require coils to overlap, making air core stators more feasible

-Ease of part production means I can compensate for the lack of an iron core by having the stator plate sandwiched between two rotor discs with rare earth magnets on them. If it helps field strength on the front of the magnets, the rotor plates the magnets are mounted to could be of iron as I think they're far enough from the alternating fields that I don't think eddy currents would require the plates be silicon steel. I have access to a small lathe, so some dimensionally accurate iron parts are feasible.

This way I should be able to build a first iteration based on reasonable guesses and materials at hand, and build a MPPT circuit for my next project, and experiment with getting better output power/voltage at lower RPM by winding additional cores. Modelling and 3d printing coil winding jigs of whatever dimensions is relatively easy, but I plan to start experimenting with coils of roughly 4mm x 5mm cross section and probably 22 guage wire.

Roughly equally difficult to build would be

3 phase 6 coil 8 pole

5 phase 5 coil 6 pole

7 phase 7 coil 8 pole or 10 pole

Thus far I'm looking at a 3 Phase 6 coil design with 8 poles and I'm evaluating this physical design: (Grid is in mm if you care. I think the proportions and phase relationships are what is relevant though.)

enter image description here

In red and blue are the magnets and in yellow are jigs to wind coils (a second half is added to the jig and the coils are wound around the pegs you can see. the outline of the yellow parts extending from the pegs illustrates the thickness of the coils. I'm hoping you can see the shape of the coils well enough from the jig that I don't need to 3d model a coil just for this diagram. My logic in using these coil dimensions was that if I don't want overlapping coils, each coil must fall within a 60 degree pie shape. This prevents the center lines of the radial sections of the coil from being directly on the 60 degree center lines as they would overlap with neighbouring coils. I modelled a similar diagram to that seen here and realised I could make the coils 45 degrees wide but still centered on 60 degree lines so that the legs of a coil would pass centerlines of N and S poles simultaneously rather than simply taking up all of the space available as you would if you were trying to align 6 coils with 6 poles. The result is seen in this illustration. As you can see the wire length saved is considerable.

As far as I understand, going from single phase to three phase while still keeping a 1:1 coil to pole ratio results in a reduction of cogging torque by splitting the countertorque moment into 3 countertorque moments, thus all other factors being equal, a 3 phase motor would have three times as many "bumps" and the countertorque of each bump would be 1/3 or less. Going to a larger number of phases increases the benefits, with diminishing returns.

As far as I understand, separate from the reduction in cogging torque from having a larger number of phases, the method of having a number of poles not evenly divisible by the number of coils is intended to desynchronize further the moments of countertorque produced by the generated current.

If I look at a design where the legs of the windings are closer to a full 60 degrees apart, not only would copper losses be greater due to a larger coil per turn, but also would de synchronize the emf waves produced in the two halves of the coil as they would pass the strongest field points separately from one another, so the induced emf is the sum of two sine waves of equal frequency but out of phase and always smaller in amplitude than it would be if they were in phase. Having them out of phase would also result in an output emf (and thus current) shifted in phase from either, separating the moment the coil produces its strongest field from either of the moments that it passes through the strongest field. This would appear to be an advantage that my 45 degree coil design may forego, although I've been unable to figure out thus far if inductive lag would produce that desynchronisation of permanent magnet and coil field moments, and I'm also not 100% certain it is desirable.

I've also noted that because I'm using 6 coils and 3 phases, the two coils of a phase are exactly opposite eachother. 6 poles will produce cogging, 8 are recommended. With any multiple of 4 magnets, the same phase coils are acted on in an identical way by the magnets, due to the symmetry of the magnet arrangement, thus the coils of the phase are in phase with each other.

If I was using 9 coils in phases of three coils in series separated 120 degrees from eachother mechanically, with either 10 or the preferred 12 magnet poles, I believe there may be an additional anti-cogging benefit to be had in that the three coils of a phase would be slightly out of phase with eachother, but still share current and again I guess I may be looking at the same potential benefit as I outlined with the two halves of a single coil being out of phase with eachother.

My question is, by using a smaller 45 degree coil to match the timing of the magnets while still using 6 coils and 8 poles, am I giving up the benefit of using a number of poles not evenly divisible by the number of coils?

I analysed what would happen if I started with phase A aligned with a magnet pair and rotated the magnets in steps of 15 degrees at a time. I found that every 15 degrees of rotation a phase will peak, and a full electrical cycle is 90 degrees. As far as I understand, this should produce a torque bump each, so 360/15=24 torque bumps per mechanical rotation and 6 per electrical rotation, which makes sense. A significant improvement over the first generator I built, but what I'd really like to achieve is the smooth countertorque effect you get when you spin a modern 3 phase brushless RC motor as a generator.


3 Answers 3


My question is, by using a smaller 45 degree coil to match the timing of the magnets while still using 6 coils and 8 poles, am I giving up the benefit of using a number of poles not evenly divisible by the number of coils?

In a coreless generator, cogging torque is a result of the changing flux density as the magnets get closer to and further away from the coils. This generates a voltage waveform that causes the load to draw varying current and torque. The amplitude of the torque ripple is dependent on the exact dimensions of the coils and magnets, so it's hard to say whether changing the coil dimensions will make it better or worse.

Using an odd ratio of coils to poles helps because it spreads out the torque peaks, but again the exact effect depends on dimensions of the components. You could try to calculate the performance of various designs using a 3D FEM program, but it might be easier to just make different coils and test them.

  • \$\begingroup\$ Thank you for your response! The effect of reluctance cogging due to iron cores seems to be simpler and more well documented/easy to look up, and I was indeed hoping that using an air core generator would allow me to focus on the electrical cogging, which is a very strong effect in my first design. I've spent a lot of time thinking about it and the phase relationship of the peak torque points still feels like something that can be evaluated theoretically. I'm working on figments remembered from an industry video showed us in third year electrical school(electrician not engineer) \$\endgroup\$
    – K H
    Commented Jan 17, 2021 at 2:19
  • \$\begingroup\$ A lot of the industry videos they showed us were chosen to give us a bit of a window into the engineering side, so this exact topic was examined but above my skill level, making the advanced parts very difficult to remember after 10 years. Would it be possible to confirm or deny that having the two sides of a coil out of phase with eachother would produce two torque bumps per coil instead of one? If I can figure this out it appears the other drawbacks are longer coil per turn, increased magnet cost/weight and lower power at a given rpm and advantages are decreased demagnetization over time. \$\endgroup\$
    – K H
    Commented Jan 17, 2021 at 2:35
  • \$\begingroup\$ I've been considering busting out FEMM, which I've used before, to evaluate effect of using square rather than arc shaped magnets in a radial flux halbach array. If I remember correctly from last time I used it, 3D modelling is done with a very expensive product. The radial flux generator we're looking at is beyond my skill to model in 2d due to the necessity of looking at a side view of 3d circular magnet. It could still be modelled as a linear array, but I understand those relationships, inasfar as the sum of two induced voltage waveforms out of phase would be smaller than that in phase, \$\endgroup\$
    – K H
    Commented Jan 17, 2021 at 3:08
  • 1
    \$\begingroup\$ Here is an article that might be of interest:- researchgate.net/profile/Junaid_Ikram2/publication/… \$\endgroup\$ Commented Jan 17, 2021 at 6:50
  • 1
    \$\begingroup\$ That article also gives you some idea of what torque variation to expect in a 'conventional' design. I agree that circular magnets could have a similar torque smoothing effect. How hard would it be to build a generator to find out? \$\endgroup\$ Commented Jan 17, 2021 at 8:14

The only way to reduce torque ripple is to choose more than 2 poles. 3 poles is typically used.

  • if poles are replicated in one rotation, you can get an electrical gear effect with lower RPM. Best performance is using Hall sensors or FOC. This is what is done in dryer motors with dozens of poles.

You can also convert an alternator into a BLDC variable speed motor with a suitable controller .

  • \$\begingroup\$ Thanks always for your response! The electrical gear effect cogging you're describing is what I'm attempting to eliminate. My first attempt generator does not cog at all without a load, which I think means I can rule out cogging due to magnetic reluctance and concentrate on the "electrical gear effect" cogging you're describing. As far as I understand, hall sensors are used to determine rotor position and an FOC is more for running a machine as a motor than a generator. Do they still have applications with a permanent magnet machine run as a generator? \$\endgroup\$
    – K H
    Commented Jan 17, 2021 at 1:48
  • 2
    \$\begingroup\$ use 3 phase.... \$\endgroup\$ Commented Jan 17, 2021 at 12:10
  • \$\begingroup\$ I'll definitely use 3 phase 8 pole design for the first iteration and depending on the result I may even move to 5 or 7 phase but I'm trying really hard to understand the countertorque effects of odd pole arrangements as much as I can before I commit to a 45 or 60 degree coil. I like the idea of lower copper losses and higher peak field, so I'm leaning toward trying the 45 degree coil first but also considering making both just to post the results here as low speed hobby generators are a popular project. I've been trying to look up my school notes on countertorque now. \$\endgroup\$
    – K H
    Commented Jan 18, 2021 at 0:03
  • \$\begingroup\$ Hmm your emphasis on three phase caused me to realize that with my single phase motor the torque ripple is probably from 0 to 100% like the output of a single phase rectifier, so switching to the model of 3 phase rectifier output ripple the improvement from single phase to 3 phase should be huge, so that's reassuring. \$\endgroup\$
    – K H
    Commented Jan 18, 2021 at 3:06
  • 1
    \$\begingroup\$ yes use 3 phase !! \$\endgroup\$ Commented Jan 18, 2021 at 3:17

I would like to help, but this is from 2021 so I am not sure if it is still a going project. So I have built several motor/generators. there are some thing that may not be 'wrong' to that you and many people keep doing. Hughes did a wind powered generator, and every one seems to just follow the same design, and this is much like the same thing.Ok for some of my thinking and this will go against all of the engineers. for a start do not try to keep making a 'three phase' generator. next use one more or one less magnets than there are coils. one more seems to be the best when using just 4 coils. next do not rotate the magnet poles, N,S,N,S, that is you are going to use an axial type of generator. also use round rod magnets. there is a lot more if this subject is still l going then I will explain more and why.

  • \$\begingroup\$ As it’s currently written, your answer is unclear. Please edit to add additional details that will help others understand how this addresses the question asked. You can find more information on how to write good answers in the help center. \$\endgroup\$
    – Community Bot
    Commented Feb 26, 2023 at 18:56

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.