Axial Flux Motor Gung-ho Design/Build

17600 Views 48 Replies 9 Participants Last post by  Coulomb
Hi all,

A have a few quick questions that don’t need a great deal of explanation; unless you really want to because I’m all ears.

I signed up to these forums a week ago and have since spent countless hours reading and learning about EV theory and related technologies. Given my Aerospace background and general love of everything tech I can’t help but tinker.

So I am seriously considering building a 3-phase axial flux motor for use in a potential future EV (also a scratch build).

I have found countless threads and papers all over the net. As I got about ¾ of the way through a mathematical proof earlier today I thought “what’s the point in designing to a desired performance goal? Why not just build the most powerful motor you can within a certain size?” This is easy really. Fit as many poles/teeth/windings as possible within your desired rotor diameter using the strongest magnets you can afford and the thickest wire you can wind.

Simple. But I need counter points! Some I can think of off the bat are:

1. It’s all good having a motor that can output ‘x’ amount of torque and spin at ‘y’ rpm, but totally useless if you can’t give it enough power to reach these numbers.
2. Hence, is it worth just building the most bad-ass motor you can within certain size restraints if you will have an under powered battery system?
3. It will still run with an under powered power source, but more power will be required to get an over-powered motor spinning than a motor specifically designed for a particular power system.
Anyone care to support or dispel simply building the motor this way (as powerful as possible limited only by size....FYI I'm going for an 11" rotor here)?

Peace.

- Matt
1 - 20 of 49 Posts
Hi Modern

This is easy really. Fit as many poles/teeth/windings as possible within your desired rotor diameter using the strongest magnets you can afford and the thickest wire you can wind.

You need to think about cooling as well - as far as I can tell you can easily find a motor that can deliver oodles of power - the issue is for how long!

Permanent magnets have tight temperature limits - get too hot and all the magnetism runs away

Design a cooling system - then build the motor around it -
Yeah that's been in the back of my mind constantly. The coils are what heats up (not the magnets themselves) so that makes it easier because the coils are obviously stationary.

The basic design is the same irrespective of how I cool the stator. So all I need to do is design some kind of housing for the stator that keeps it as thin as possible - the smaller the air gap between rotors the greater the flux density which = more pow-ah.

If I fry a few hundred dollars worth of magnets though I'll be mega pissed lol.
Hi Modern
The coils are what heats up (not the magnets themselves)

Not sure that is true - you will be using the coils to change the magnetic fields the magnets live in - I think you will be dumping heat from eddy current effects into your magnets

If you have a 300 Kw motor that is 90% efficient 30 Kw of heat is being wasted in the motor -
10 off 3 kw fires

that will raise 100kg of steel 90C in 1 minute

However a single stator AF Motor producing 300kW is highly unlikely in any size that would fit in a car. I will most likely be going for a 3 stator design (around 75kW per stator) in series or parallel.

Further more I can do no more than cool the rotors with air seen as they will be spinning at a few thousand rpm, and bathing them in a tank of coolant will cause way too much drag -especially as they will a high number of magnets attached to them, that even when flush mounted will add to the drag dramatically.

I will place a impeller ring between the rotors (and above the stator) in order to suck air in and over the coils/surface of the magnets.

So basically cooling will come down to simply not using too much power per stator! I hope.
You should also choose a controller to match before making the motor. The amp and voltage ratings shouldn't be a problem, especially since you could power each stator with it's own controller if necessary, and change between y/d and series/parallell. The electric rpm limitation however is a more relevant figure. A high amount of motor poles combined with a slow controller may limit you to dissapointing outputs.
So this is where I am yet to do some serious research...

What exactly do you mean by electric rpm? I was under the impression your voltage was tied to the rpm of the motor, or are you inferring the PWM frequency capability of the controller?

You are correct though - I'm looking at around 10 pole pairs per motor so if I go with 3 motors that's 30 poles... separate controllers may definitely be required.
The electric rpm is connected to the maximum AC frequency the controller is able to produce. Indirectly, this is affected by the pwm frequency, but it is not the same thing.

The Curtis AC controllers for an example will only handle 300Hz, meaning 18 000 electric rpm. Since you have 10 pole pairs, this controller would limit you to only 1800 motor rpm. The standard kelly controllers can handle 40 000 electric rpm, still only 4000 motor rpm. Kelly has upgrades to deliver 70 000 and 100 000 electric rpm, which should suffice. But kelly isn't exactly the finest brand, the quality is considered quite low.

The pole pairs need not be added with multiple rotors, 10 pairs each at four rotors (three stators surrounded by rotors) still adds up to 10 pole pairs if properly synchronized.

The rpm also depends on voltage, but since you are making the motor yourself you will be able to choose the voltage dependecy freely. You will wind as many turns as necessary to get the decided max rpm at the max available voltage. This is another reason to choose controller first, so the max available voltage is known.
See less See more
I’ve been reading on the theory and design of motor controllers this morning, so your reply does make sense to me lol.

I’ll have to look around for an appropriate controller, because at this stage I believe building a controller, while not beyond my technical skill or knowledge, is a pain I don’t really want to deal with.

I see your point on the rotors and their magnet pairs. So long as the rotors are all aligned perfectly, the one controller can be used to control the AC-phase in all 3 stators.

Some work to be done just to do that though.

EDIT: Question – this is a bit embarrassing but what numbers am I looking for to determine the electric rpm? The only numbers I can see are frequency of operations and at 16.6kHz this is someway off 40,000…

Another embarrassing question; the power from the batteries runs through the controller yes? So if you have 3 motors you are limited to powering each motor by the total power the controller can handle divided by 3. So if the controller can handle 144v and 900A, the most I can feed to each motor at any one time is 48V and 300A or 14.4kW. Kinda lame…The other option is a controller per motor. Provided the rotors are all aligned correctly this would still work fine…

I could also have my understanding of how it’d work totally wrong, so please correct me if this is the case. Bear in mind I started this a week ago and have no prior knowledge on the subject….and I’m trying to build from scratch!
See less See more
Another embarrassing question; the power from the batteries runs through the controller yes? So if you have 3 motors you are limited to powering each motor by the total power the controller can handle divided by 3. So if the controller can handle 144v and 900A, the most I can feed to each motor at any one time is 48V and 300A or 14.4kW. Kinda lame…The other option is a controller per motor. Provided the rotors are all aligned correctly this would still work fine…
The question is, are you connecting the motors in series, or in parallel.

If they are connected in series, then using the controller listed each motor would see 48V and 900A.
If they are connected in parallel, then each motor would see 144V and 300A.

Some advanced controllers also have series-parallel switching, so they start the motors in Series mode, to get the most out of the high amperage, then when the back EMF starts decreasing the acceleration, the controller switches to Parallel mode, to increase speed higher at the cost of decreased acceleration.
If they are connected in series, then using the controller listed each motor would see 48V and 900A. If they are connected in parallel, then each motor would see 144V and 300A.

I’m assuming in these examples only the one controller is being used for all three motors? Is one controller per motor an issue other than increased complexity cost and battery drain? lol “other than” – I think that’s everything that could change!

I’m assuming in these examples only the one controller is being used for all three motors? Is one controller per motor an issue other than increased complexity cost and battery drain? lol “other than” – I think that’s everything that could change!
I think the main complexity would be keeping all the motors in perfect sync. If any of the motors is out of sync, you'll get "Transmission wind-up", which may break your connecting shafts at best or trash the motor entirely at worst.

Other than that, however, you would get full voltage and full amperage to all three motors.

Of course, another option would be to have 2 of the motors on the rear wheels, one per wheel, and the last motor on a differential to the front. It would give you 4-wheel drive, a 33/66 power split (Best for handling without the typical 4WD bogging down and torque-steer), and would have no problems at all using separate controllers per motor, though if they were linked in series it would act as a limited-slip differential, and if they were connected in parallel, it would act as an open differential.
The Curtis AC controllers for an example will only handle 300Hz ...
These things seem to want to run at high frequencies. That becomes possible with an ironless design.

However, it seems that at these higher frequencies you need to consider using litz wire (a sort of wire woven from many thin strands of wire all electrically insulated from each other). There are few places in the world any more where such wire can be sourced. The idea of course is to limit the effect of the skin effect; even at 50 Hz, there is a fair bit more current at the surface of a thick conductor than at the middle.

Litz wire used to be common in intermediate and high frequency radio frequency coils in radios and TVs.

I suppose you could ignore the skin effect, and use ordinary thick round or rectangular cross section wire, and put up with slightly lower efficiency (due to the higher effective resistance).
you need to consider using litz wire
If I was rich and able to afford it I would be using Litz wire, a Halbach array and N52 magnets. Unfortunately I don't have the cash lying around to invest that sort of coin

At this stage I am using 1mm (18AWG) copper for my conductor, and will be air cooling it (split across 3 stators it can be run cool enough) however I will be leaving room in my design for liquid cooling just in case I want to push the system harder.

Still early days though - hell, I'm still trying to figure out what size rotor and hence how many magnets I'll be using.
litz wire is probably not needed, and can actually lower your efficiency. The fill factor decreases due to increased "insulation thickness to conducting area ratio" and decreased order, when you decrease individual strand thickness. Twisted litz, which is quite common, is even worse and should never be used. Just make your own bundle of, for an example, 0.3mm wires. Depending on the motor design it may actually be better.
I'm going to use the thickest wire I can reasonably work with - less resistance and greater heat capacity.

I do have something I need to know a little bit more about before I can continue with my design...

Does it matter if the controller can output more current/voltage than the motor can handle? Inversely does it matter if the motor can handle more than the controller can supply?

So if I just go ahead and build an “as powerful as possible” motor and hook it up to whatever controller I can afford, other than giving the motor too much juice and blowing it up, are there any other reasons to avoid this route?

I am of course assuming motor controllers give you a decent level of control over the current and voltage draw.

The way I see it I can build my motor, buy the controller and make sure the temperature of the stator core is closely monitored. If it gets too hot then I know I need to back off. Through trial and error I could deduce the motors continuous and peak power ratings.

It’s obviously the wrong way to go about things, but if there is no overly hazardous reason to do this then I’m going to plough on.
See less See more
Does it matter if the controller can output more current/voltage than the motor can handle?

Motors often don't have a very hard limit; they can usually take what the controller can dish out until they get too hot. If this happens routinely, then it's a shame about the cost of the silicon that rarely gets used, that's all.

Inversely does it matter if the motor can handle more than the controller can supply?
It means you could be frustrated by the lack of performance, knowing that if the controller could only push out a bit more, the motor is ready to take it.

So as far as I can see, there is no huge disaster if the controller and motor are not matched well.

One thing though - controllers need a certain minimum of inductance. If there is not enough inductance, that's really bad news; the silicon can't protect itself fast enough. That's not usually a problem with iron cored motors; they have a factor of ten or a hundred more inductance than the minimum the controller needs.

But if your design is ironless, you may actually need external inductances. This is the case for the Ultramotive Carbon motors, for example, which are axial flux and ironless.
One thing though - controllers need a certain minimum of inductance. If there is not enough inductance, that's really bad news; the silicon can't protect itself fast enough. That's not usually a problem with iron cored motors; they have a factor of ten or a hundred more inductance than the minimum the controller needs.

But if your design is ironless, you may actually need external inductances. This is the case for the Ultramotive Carbon motors, for example, which are axial flux and ironless.
My design is core-less, but the magnets are attached to magnetic steel to focus the field properly. Or is this exactly what you are talking about?
My design is core-less, but the magnets are attached to magnetic steel to focus the field properly. Or is this exactly what you are talking about?
I'm not a motor design guru, but it seems to me that steel in the rotor won't increase the inductance in the stator much, if at all. So you may need to measure the inductance once you've made your creation.
Most axial flux motors place an iron backing plate behind the magnets to complete the magnetic circuit. This seems to work fine for every motor I've managed to lay my eyes on over the last few days.

Nothing quite as powerful as what I am attempting to make, but considering this one has nothing but wood backing (a poor design choice) and manages a continuous power rating of over 10kW (according to the designer it has a peak of around 18-20kW) I think the inductance may not be too much of an issue.