This DIY guide by Simone Rambaldi is intended for people willing to design their own outrunner motors with rectangular magnets. However, most of the hints are also valid as general suggestions for the design of every type of electrical motor.
This guide is intended to help people willing to make their own motor for bikes, motorbikes, cars, boats or planes in the power range between 1 and 100 kW. The target audience are people with low-to-medium skills, such as rc hobbyists with for example experience from rewinding motors, trying for the first time to design a motor from scratch. This hint guide will give them a starting point.
By following these hints you can build a more or less "working" motor, that can then be further improved by fine tuning your design. This process is illustrated on the example of a "working but not so good" motor in Figure 1a) and a motor that "I think can be considered as good" in Figure 1b).
Fig. 1 Example of outrunner designs with a) 90.3% and b) 97.2% efficiency.
Commercial mainstream controllers have a peak working frequency of about 300-400 Hz, but waveforms can be really bad close to the maximum frequency. Plan to stay below 80% of the maximum controller frequency. Accordingly, choose a pole count such that your required RPM stays below the maximum controller frequency.
A very common and cheap iron lamination is M400-35A, but it is suitable only up to 60-90 Hz. For higher frequencies you need a better iron, which can be hard to find. M270-35A is better and still available mainstream but is suitable only for frequencies up to 120 Hz. Thinner iron laminations can allow higher frequencies, but if you get your iron laminations laser-cutted, the price will go up because thinner iron means more cutting time for the same stack length.
Outrunners (Outer-rotor motors) have a very high power density, but are more difficult to cool. The peak power will be high but nice continuous S1 power can be hard to reach. When thinking about the motor think also how heat can get away from the copper and iron.
The Neodimium N series are affordable magnets, with the N35 to N45 grades are good options and are available from many online stores in different sizes. The weak point is that they can be destroyed (demagnetized) by temperature, so keep your rotor cool.
Magnetic coverage should be around 66% for a sine wave motor, choose your magnet size to cover approximately 66% of your pole pitch.
The rotor yoke height should be at least 1.5 times as thick as the magnet thickness. More is better but does also increase the mechanical inertia and weight of the rotor. A too thin rotor may brake or bend under load.
The size of the tooth top (tooth width plus twice the tooth tips) should be similar to the magnet size. Tooth tip angles between 30° and 60° are common. Slot opening widths between 1 and 3 mm are reasonable values (depending on the copper wire diameter), less then 1 mm may damage the copper wire during the winding. Thin tooth tips can be bent and damaged easily.
An airgap of 1 mm is a reasonable starting value.
Motor proportions with a stack length similar to the airgap diameter are a good starting point.
Iron saturation should be below approximately 2 T (depending on the used iron lamination). If there are parts above this value in your simulation result, try to add material to reduce the value. The rotor can have higher values, around 2 T. The stator needs to be lower between approximately 1.6 and 1.8 T. If the iron saturates you loose efficiency and power. A higher working frequency may require even lower values in the stator.
Try to increase the tooth width, increase the tooth tip angle, increase the slot opening depth, increase the rotor outer diameter, or decrease the shaft size. On the contrary, try to remove all the iron not doing useful work. If you see some parts with very low flux densities, you can remove them and save weight and cost.
The back-EMF should be as sinusoidal as possible, without any harmonics. Try to change the magnet strength (Nxx) to see if localized saturation causes problems. Change the tooth tip size and angle and use a graphical program to overlap the graph result to see if you are improving. Always have a pure sinusoidal waveform as reference. Harmonic cause the motor to heat up, make noise, and loose power.
When doing simulation under load, a current density of 6 A/mm2 is reasonable. Choose the option "PMAC" for a sinusoidal supply voltage and current.
Keep this general goals in mind:
Good luck with your motor design and thanks to Simone for this DIY guide.