Homemade Axial flux PM motor

[enrengineering]

Hello ES members,
I've been into electric vehicles for a while, and after reading APL's and HalbachHero's threads I got some inspiration and decided to start tinkering about building an axial flux motor (and found out that i'm not the only one that likes this stuff ).
I'm currently studying mechanical engineering and I have a lot of interest on electric vehicles, I've built two electric go karts so far and going strong, also, I spend all my summers working on a electric motor maintenance and repair shop and I learned a lot working there.
My plans are to make a small motor (like the one you see on the pics, the rotor has 150mm outside diameter), make sure it works the way it should and keep improving from there (and maybe scaling it up). I would love to build a small buggy/crosskart with my own electric motor.
I've been working on it for a few days, first I searched for a supplier for the magnets and picked a reasonable size to start designing, I went with 24 magnets of 10x10x5mm (N42 grade) and 24 magnets of 10x5x3mm (N45 grade), all of them arranged in a Halbach pattern (24 poles). Once I had that I made a rough design of the 3D printed piece that would allocate all the magnets distributed evenly and printed it with polycarbonate carbon fiber (5.4 mm of total thickness), here's the part:

And all the magnets mounted (almost at the limit of being completely pulled away):

When I got the magnets all mounted I had to design the backplate, I made a drawing in AutoCAD and machined on my MPCNC out of aluminum and threaded all the necessary holes to mount the countersunk screws, also I made the center piece out of stainless steel that bolts the aluminum plate to the axle. Notice that the magnets are pushed against the aluminum plate, the thickness of the bottom part of the 3d printed piece is 0.4 mm, I did some tests with half of that but wasn't strong enough to hold firmly the magnets.


Once I had the rotating assembly done, I 3D printed the stator (36 slots) out of PETG, and embedded two 6002-2RS ball bearings onto it.


The air gap between the rotor and stator is about 0.6 mm.
That's all the progress so far, I'm currently waiting for the copper wire to start winding the stator, maybe the most practical way is to wind the coils in series but I still need to think about it.
All comments and suggestions will be greatly appreciated, I will try to post more about this project as I'm making progress with the machining and winding process.
PS: I got some suggestions from the workers of the electric motor shop about the windings, but it will be my first time winding a real motor, so maybe I will practice a bit before the real deal.
Thanks for reading.
Henry.

 

[NoSmoke]

Interesting project. A couple of questions if I could:

How will the motor be mounted? Metal brackets fastened to the outside of the stator disk? Will the PETG be strong enough in this situation?

The two shaft bearings seem awfully close together. Will they be strong enough (along with the PETG they are mounted in) to keep the rotor from skewing into the stator?

What are the advantages of this type of motor over more conventional designs?

Thanks and please let us know how it goes...

 

[APL]

Awesome to see another build in progress! Welcome to the grind.

Looks like your off to a good start,.. I like the 3D 'covered' magnets idea, .6mm gap is nice. It will be interesting to see what you
come up with for windings, I guess it depends on what you plan to do with the motor and what controller is used.

Anyway, for a first time build she's looking sweet,.. all we can do is build one and improve from there, unless your a master of the FEMM and formulas.

Looking forward to your progress. :thumb:

 

[Barncat]

You young guys have CAD, mini CNC, and printing tools and skills that didn't exist when us older dudes grew up, so that's cool.

This will probably work OK as a bench prototype, but I'd redesign the axle to backplate connection. That single axial SHCS screw is transmitting all the torque, and only the limited surface area of the axle end interface with the aluminum plate is responsible for maintaining rotor perpendicularity and air gap- which if I'm seeing this right is governed by that thin plastic spacer slid over the axle. Please correct me if wrong.

I'd print a wider beefier spacer that fully contacts the inner bearing race. And perhaps in future turn a steel hub with a wider outward facing step bored for a light .001-.002" press fit with the axle which could be TIG welded from the back after insertion. That high strength assembly could then be reverse chucked in a lathe to check for runout and even trued a bit if necessary.

Hope the wire winding goes well.

 

[enrengineering]

Anyway, for a first time build she's looking sweet,.. all we can do is build one and improve from there, unless your a master of the FEMM and formulas.

The past week I installed the software and messed around a bit with FEMM, but I don't know how to use it.

Metal brackets fastened to the outside of the stator disk? Will the PETG be strong enough in this situation?

The two shaft bearings seem awfully close together. Will they be strong enough (along with the PETG they are mounted in) to keep the rotor from skewing into the stator?

The motor will be mounted on a "L" type metal bracket that I still need to build, the bearings are a very tight fit on the petg, no problems so far by spinning the rotor by hand, I will need to make some changes if the rotor scratches the stator, or simply redesign the whole thing, I hope this first prototype works and start making more "serious" versions.

the aluminum plate is responsible for maintaining rotor perpendicularity and air gap- which if I'm seeing this right is governed by that thin plastic spacer slid over the axle. Please correct me if wrong.

Correct, I forgot to mention that the stainless plate with the six holes will be welded to the shaft, once the welding is done I will chuck it on the lathe and machine the shaft fo make it perpendicular to the plate. I joined the two together with a bolt because I still don't have (yet) the stainless round bar that will be the axle. Here's a pic of how it will be:

And then using the same center hole on the aluminum plate drill with a hole saw to make clearance for the weld.

 

[Barncat]

I'd still recommend drilling/boring through the stainless hub, inserting the axle, put a deep chamfer on the joint and weld from the back side to get fairly accurate alignment. Then chuck the axle in a lathe and true the inner face of the hub. Turning the axle will not true the assembly. And you're probably using stainless because it's less magnetic?

And there's your mounting plate

 

[enrengineering]

And you're probably using stainless because it's less magnetic?

Correct, the stainless that I’m using isn’t that magnetic. Last year I converted a car alternator into a brushless motor by replacing the rotor (that has the electromagnet) with a custom 3d printed one with neodymium magnets, I had the axle machined out of a round stock and now the axle is an entire magnet, the axle runs through the rotor but doesn’t make any contact with the magnets.
Back to the axial flux motor: I did all the windings, 7 turns per pole and then connected to a cheap speed controller from aliexpress (12V) and does an 1/8 of a turn only and stops for 3 seconds, next I tried an amazon 1500w e-bike controller and the same, except this time all the stator started became warm to the touch (I was pulling around 2 amps @ 48V from the power supply) and the last attemp was using a vesc from the alternator project, I changed the settings and all the motor poles and it vibrates. I also tried using a homemade low voltage three phase generator and the rotor vibrates like the vesc.
I’m going to I make new designs reusing the same magnets and make the windings bigger, only 6 coils. Also, how many magnets I should install? Or doesn’t matter how many are?
I will upload a video of the testing and to show the behavior of the motor.
Here’s the motor:

I placed two rolls of tape under the stator and tested it vertically, I held the stator vertically and did the same. Those spots of hot glue are to hold the ends of each winding.

 

[Barncat]

I'm just a machinist/welder type so am unable to advise you in motor electronics, and the controller stuff is an order of magnitude above that. There are other guys on site who can assist you. But you're doing nice tidy work and I'm sure will get results before long.

You must have already tried this but manually give the rotor a good spin right before you apply current. Starting a motor from zero rpm is electronically difficult.

Oh, btw, you must have some sort of locking set screw collar on the axle outboard of the bearings to keep the rotor from flying away if spun up?

 

[nicobie]

Looking good. What brand of filament and printer are you printing with?

 

[enrengineering]

You must have already tried this but manually give the rotor a good spin right before you apply current. Starting a motor from zero rpm is electronically difficult.

Oh, btw, you must have some sort of locking set screw collar on the axle outboard of the bearings to keep the rotor from flying away if spun up?

Yup, I tried giving a reasonable full send and no luck at all [emoji23]
I have a 3d printed small collar on the axle just like the spacer between the rotor and one of the bearings.

 

[enrengineering]

Looking good. What brand of filament and printer are you printing with?

Thanks! I’m using PETG for the stator and Polycarbonate carbon fiber for the ring that holds magnets. I use these filaments with my CR-10, I got a hardened steel nozzle and swaped the stock bowden tube with a capricorn one. Everything else is stock and working like a charm. I also use TPU (SUNLU brand) with retractions and no problems at all.

 

[nicobie]

I was curious about which brand of CF you are using. I've tried Sainsmart CF/nylon and it had crappy layer adhesion. At the time I didn't have a filament dryer but the filament was new so should have been dry. The CF/PETG I got from Filaments.ca printed nicely with my highly modded Ender v2, but it was just PETG so couldn't handle much temperature. Since then I've gotten a roll of Matterhackers NylonX and the new Sunlu S2 dryer but haven't had a chance or project to try them out on.

Have you found that an enclosure is needed in order to print your PC/CF? The CF/PETG printed fine for me without one.

 

[enrengineering]

Have you found that an enclosure is needed in order to print your PC/CF? The CF/PETG printed fine for me without one.

For the CF and PETG I use a drying box from eSUN that I bought on amazon for 50€, I start it 15 minutes before printing with the filament inside and works like a charm, before buying it I didn't use the drying box and also printed fine. I bought it to keep the CF dry because the 1kg spool cost me around 50€. The brand of PC/CF that I'm using is PRILINE and the PETG is ERYONE.

 

[enrengineering]

Here's the video of trying to start up the motor:
https://youtu.be/1uNAQXWePfE

 

[stan.distortion]

Here's the video of trying to start up the motor:
https://youtu.be/1uNAQXWePfE

You're certain all your polls are wound/wired in the correct direction? I've messed that up a few times before, with generators so the only symptom was rough running and low output plus but the symptoms would be similar to the vid as a motor, trying to drive against its self.

 

[enrengineering]

You're certain all your polls are wound/wired in the correct direction? I've messed that up a few times before, with generators so the only symptom was rough running and low output plus but the symptoms would be similar to the vid as a motor, trying to drive against its self.

That made a difference, I connected the wires from the star connection and vice versa and worked for less than a minute!
I used the small bldc controller, tried with the ebike controller and nothing. Here's the video: https://youtu.be/0HW6PDHwUcE
The small controller gets hot really fast and shuts down, I made sure that each phase wire was connected in order to match the controller direction (which has a jumper that you can move to select CW or CCW). To start the motor I gave it some speed with my hand and kept adjusting the speed pot of the board with the other hand until I felt it started going itself. I think maybe that the problem I have is that there's too much mass to start itself.

 

[stan.distortion]

Cool! Could be worth going over each poll one by one with a magnet and a light dc current to the winding just to make absolutely certain all polarities are correct. An oscilloscope would be a really useful tool too, maybe even essential when you get into the fine tuning. I've got a Hantek PC based one here that was about 40 euro new, right at the cheap and cheerful end of the market and I wasn't sure if it was worth getting even then but it's turned out really useful, would have been well worth spending a few of hundred on a new higher end model or something really good used.

 

[APL]

It works! :thumb: That's the main thing. Excellent workmanship, and I haven't seen to many prints that nice!

Like Stan says, checking each coil for the correct N-S orientation is always good, and has stumped me and many others on the initial motor run. Not a very fun to do though, (and don't use a compass because it might re-magnetize the needle and make it useless.)

But, it looks like your motor is working fine, just a little under powered at the moment. Your correct in that you have a lot of mass in the rotor. It's an air core, so power usually comes from speed, or lots of amps which makes thing get hot fast of course.

Speed comes from volts and coil count, you might try a little more V's.

The controller might be a little small for the task, do you know how many watts your using? Also, you don't have that many turns, so some controllers might not be getting the right back EMF signal they need. It might need more input volts to get a higher speed to produce a better signal too.

I assume it's wired like this Bavaria scheme?


I suspect that the coils just don't have the power to be useful at the moment. Usually air core motors have two rotors to squeeze the coils, and the thickness of the coils, & the distance between the rotor magnets is primary for power. (Usually around 10mm.)

If you replaced the cores with steel, and added back iron, this motor would fly!

Changing the coil count to six is not going to help, it will just make the motor run faster and start harder, 36 slots is more useful and more controller friendly.

I think the main problem is that the coils are one sided and long, so the magnets are not getting very much flux yet.
Keep up the good work, as I'm sure you know by now, success and failure go hand in hand with DIY motor building.

 

[APL]

A coils strongest magnetic flux is at is center, so if I have a coil 20mm long, then the most usable flux is 10mm from each end at the center. The ends of a coil have hardly any flux in comparison.

So for air core, you want 'short' 'fat' coils to get the magnets closer to the flux power. But then the coils are too wide, and you cant
put too many of them very close together. That's why most air cores look like Halbachhero's, sinusoidal flat wound motor, to get the magnets as close to each wires individual Lorentz force as possible.

The reason we use steel,( laminations), is to draw the coils central flux out to the ends of the coil, and be more useful. The problem with that is, now the steel becomes a problem, as it has hysteresis and cogging.

You probably know all this, sorry to ramble on....

 

[Lebowski]

My opinion on the purpose of steel cores for coils is different. Motors and generators are the same thing, just the power flow is different (electrical to mechanical or other way around). Anyway, for a generator it is easy to see that focussing the magnetic field from the permanent magnets tthrough a smaller area means the wire length for X coils around the field is less, so less resistance (less losses). The iron is the way of focussing the field.

It would be cool if focussing magnetic field could be done without iron, like with some sort of lense that acts on magnetic field lines.

 

[enrengineering]

I assume it's wired like this Bavaria scheme?
36 24 SP count.png

So for air core, you want 'short' 'fat' coils to get the magnets closer to the flux power. But then the coils are too wide, and you cant
put too many of them very close together. That's why most air cores look like Halbachhero's, sinusoidal flat wound motor, to get the magnets as close to each wires individual Lorentz force as possible.

The reason we use steel,( laminations), is to draw the coils central flux out to the ends of the coil, and be more useful. The problem with that is, now the steel becomes a problem, as it has hysteresis and cogging.

You probably know all this, sorry to ramble on....
[/quote]
Not ramble at all, I didn't know this! Thank you so much!

 

[enrengineering]

I assume it's wired like this Bavaria scheme?
36 24 SP count.png
Yes, actually the same

So for air core, you want 'short' 'fat' coils to get the magnets closer to the flux power. But then the coils are too wide, and you cant
put too many of them very close together. That's why most air cores look like Halbachhero's, sinusoidal flat wound motor, to get the magnets as close to each wires individual Lorentz force as possible.

The reason we use steel,( laminations), is to draw the coils central flux out to the ends of the coil, and be more useful. The problem with that is, now the steel becomes a problem, as it has hysteresis and cogging.

You probably know all this, sorry to ramble on....

Not ramble at all, I didn't know this! Thank you so much!
[/quote]

 

[enrengineering]

I assume it's wired like this Bavaria scheme?

36 24 SP count.png
Yes, actually the same

So for air core, you want 'short' 'fat' coils to get the magnets closer to the flux power. But then the coils are too wide, and you cant
put too many of them very close together. That's why most air cores look like Halbachhero's, sinusoidal flat wound motor, to get the magnets as close to each wires individual Lorentz force as possible.

The reason we use steel,( laminations), is to draw the coils central flux out to the ends of the coil, and be more useful. The problem with that is, now the steel becomes a problem, as it has hysteresis and cogging.

You probably know all this, sorry to ramble on....

No ramble at all, I didn't know this! Thank you so much!

 

[enrengineering]

Sorry for the entries I did before, the quote thing pissed me

 

[APL]

Yea, I don't use the quote thing unless I have to,.. you can also edit or reduce your previous entries by using the pencil/edit box at the upper right on your post. Can't reduce it down to less than one word though it seems.

Anyway, steel concentrates the coil flux, reducing the need for more windings, agreed. I'm no expert for sure, in fact the closer you look at it the more complicated it gets, an electron moving force through a wire and causing magnetic spin? How does that work,.. and what about north and south never being separate no matter how many times you divide them, mono poles are theory but not observed. Plank waves and positrons, it's all too much for me.

Luckily we really only need to deal with the basics, try, measure and observe.

A motor needs to have a basic "magnetic circuit", or loop, in order to work well though. Looks like a box, or two staples facing each other. One side has the two magnets N & S and a piece of back iron connecting them. and the other side has the two coils with back iron. Air gap in between.

Or, in this case, two more magnets and back iron complete the circuit. (second rotor)



Your motor at present seems to have half the circuit, but still works. In order to get power/torque you'll need to close the other side of the magnetic circuit somehow, I think.