APL's DIY axial-flux motor

Single disc stator, yea.. I guess we'll find out, but it seems to be pretty stiff. I clamped some to a table, with a 3"
overhang, and put 30 pounds of weight on it, just to see how much it can take. It didn't bend very much at all,
maybe a millimeter or two, so I think we're good.

It is a spring though, and there might be harmonics that can happen. But I think it's WAY stronger than some of the
single rotor - double stator designs I've seen,.. where the rotor is mostly cut out for magnets. So there must be
a fairly good magnetic balance going on there between the coils, or those thin rotors wouldn't work very good.

Axial center rotor .png

In fact, given that there is a balance, I have to wonder if maybe the stator plate could be 3D printed out of ABS.
Perhaps when this build is up and running, I can go back and try it. It would be interesting to see what happens.

Anyway, I got some cores 3D printed, so now I can wrap some wire on a few and see how that goes. I made the tooth
lip 4mm thick, which may be a bit much, but this SMC stuff is pretty weak, and I don't like the idea of breaking cores.
The cores do have a radius on the tooth lip.. although the printer didn't seem to produce it very well on the high build
setting.

View attachment 1

I need to tweak the cad core drawing a little already.. and I can change the thickness easy enough if needed.
We'll see how it looks once the wire is on it.
 
I looked at a few different sizes of screws for the fastener set up, and decided on the 4mm size. But when I went
to find a 13mm long standoff post in that size, I was out of luck. I did find them in SAE 8/32", so I went ahead with
that. (same as 4mm)
A thin threaded tube of aluminum, and two small stainless screws. Hopefully they won't be affected by eddy
currents too much.

Core Fasteners.jpg


Then I got out the wire and wrapped 10 turns around one core, (two layers of 5 turns),.. it turned out just about right,
and I think everything looks pretty good there. I'm hoping that I can use the plastic cores to wrap the wire, and then
transfer the coils to the steel versions.

View attachment 1

All in all, things are looking good,.. the cores are mounted very sturdy, with no movement at all, and the wire is
fitting just right. This new version is going to weigh a few pounds more than the last motor, but it will also have
over twice the tooth face area.
 
That's good progress, can't wait to see it finished!
What kind of power/torque figures do you expect to get from this motor if everything goes well ?
 
I wish I could do the math, or could run the motor calculator well enough to answer that. All I can say is that I expect
it's going to be a vast improvement over the last version. Although it's running more steel in the cores, which means
more core hysteresis, and twice the amount of copper, which means more losses, so there could be some disappointment
yet.

But the 18/16 design is a sure performer, and I can always take windings off. There will be three cores and windings
less than the first motor, so that will figure in too. Having full trapezoid tooth shapes, and SMC material, should make
the biggest improvement.

So I guess we'll just have to wait and see, The windings are able to handle up to 2000 watts, and as long as the cooling
fans do the job, it should be fine. Once it gets spinning I can take some Kv readings and get an idea on efficiencies.
The true test for me, is putting it in the bike and riding it down the road,.. thats when I can tell how it's really working.

It will be interesting to see the difference in weight between the new core sets and the old laminated stacks. Up till now,
I've been doing pretty good with saving weight.
 

Attachments

  • Core comparison..jpg
    Core comparison..jpg
    10.2 KB · Views: 3,059
Slowly moving forward on this project. I was able to bother some friends again, and had this spacer ring printed.
I'm going to have to pull the trigger on a 3D printer sooner than I thought,.. these things are just handy as hell.

The ring allows me to scribe the slot placement for the cores, so I can mill them on the manual mill. The cad program
automatically spaces all the slots perfectly, so I can be sure that they will be exactly in the right spot.

Scribed Carier..jpg

I'm not sure if this stuff can be laser cut, but it can probably be water jet cut or CNC'd, in which case, this part of the
process would be unnecessary.
 
Laser cutting is possible but not recommended due to toxic fumes from the plastic. CNC is probably the best bet.
 
APL you made some nice progress on that new motor. Everything looks fantastic and kudos for your skills and your will to learn something new, like the thing with the CAD program and choose of materials to make it as good as possible. :thumb:

I have not much to add, only that i still have the meaning that you should not have the endturns within the magnetic circuit as they do not contribute to torque on a PM motor, and the other thing is that i would try to find a different solution to put the cores in place without decreasing the thickness.
If you decrease the cross section, you probably get less peak torque out of it because the flux density in those areas is higher as within the rest of the cores and so this part will go elarier into saturation which then lowers kT.
 
Looks good! Agreed with CNC probably being the best. Laser won't do it well and waterjet will likely delaminate it. Perfect part for a router with a vacuum table.
 
I managed to machine the stator plate without messing it up,.. takes a lot of concentration on a manual machine.
Found out that you have to go real slow when milling through the bottom side, as it tends to delaminate when it
pushes through. (using a large end mill anyway)



Decided to go with flat head screws to hold the cores in place,.. there is a lot more area on the tapered head to
push on the SMC material. The cap screws only had a few mm's to push on, between the threaded post and the head.
I was afraid they could just push that through, and then the cores would be free to self destruct.

Coleasterling has graciously agreed to make the cores for this beast,.. I thank you profusely sir, :thumb: .. and as soon as I can
get him a decent rendering, things can begin moving forward on that.

Madin88, the slots for attachment are a compromise, and perhaps not the best for magnetic transfer between cores,
a better design would be a trapezoid shape, with only a millimeter or two for a ledge. But the bigger holes will make
the plate weaker too, so there needs to be a compromise, or design with less cores. I wanted a slot, so that I could
easily do it here.

I have given it a lot of thought though, and I'm still working on it. No easy solution. Perhaps on V3.
I also feel that the cores on each side are somewhat independent,.. the coils are centered on each one, and the connection
between them is almost unnecessary, except to make the magnetic circuit solid.

I'm not so sure the flux density between them is higher, it seems like it should be the same as the face,.. being equal
distance from the center of the coils.
If I stack five magnets on end,.. the face of the last magnet is the same, not five times stronger. :?:
It is a conundrum though, and I'll have to give it some more thought.
 
Looks amazing so far. :thumb:

You want to recess the screw heads somewhat to avoid eddy currents in the screws. A few mm will make a big difference. Gluing the cores to the stator plate might not be a bad idea too, but would make it hard to rework.
 
I keep saying it, but looks great! You're chugging right along!

Might be a dumb question, but would Ti screws help significantly with the eddy current issue?
 
I can recess them a few mm's. And yea, I think Ti screws would be better. If they were 4mm, then it would be easy,
because thats common bike hardware. But I went with 8/32" because of the standoffs. I'll look around..maybe they are
available too. Otherwise, maybe I can look a little harder for some new metric standoff's.

Stainless isn't too bad though,.. as far as I know it's pretty neutral, but I'll see what I can find. Theres always nylon..
not really motor-worthy though.

What I need is a chart, that shows all the different metals, and their magnetic susceptibility, to get a better grip on this
kind of thing.
 
For the screws, as long as they are non-magnetic, most of the flux will avoid them and take the easier path through the core material. What counts most is their electrical conductivity as we want to avoid eddy currents. Titanium would be the best. Otherwise I would use 316 (or any 300 series) stainless which is easier to find and less expensive. Don't use 18-8 stainless as it's magnetic.
 
8-32's are actually pretty available! Little spendy for the size, but I think still reasonable for the project. Standoffs, however, I'm not sure. Don't think I've ever seen that size of ti standoff.
 
The screw head is the most important part. The standoffs can be anything non-magnetic. Very little flux will be passing through the standoffs.
 
Maybe you could consider using screws made of FR4 material (the kind of plastic used for electronic boards).
They will resist the heat cycles and this is a very tough plastic with very high insulation characteristics.
I think it might work, but can't be really sure, there are lots of forces involved and these forces are not always in the axis of the screw.

Otherwise, as someone suggested earlier, the easiest way to get entirely rid of the conductivity issue is probably to epoxy the cores to the board, or, even better, to each other.
Gluying them to the board would have the advantage of being easy to do, but the inconvenience of making them very hard to remove without destroying the board.
Gluying them to each other would have the advantage of being easier to remove (you just break one core and the opposite one falls off), but the inconvenience of being a bit more difficult to install in the first place: the process would be something like that:
-Protect the sides of your slots so that the glue won't stick to them.
-Protect a screw so that the glue won't stick to it
-Put glue on both cores
-Fit the cores in the slot
-Put the screw in place and tighten both cores firmly in place
-Wait for the glue to cure and then remove the screw
-Pour some epoxy in the screw hole.

I would be entirely confident an epoxy like the 3M DP420 could do the job, this thing is crazy tough.
 
APL said:
I also feel that the cores on each side are somewhat independent,.. the coils are centered on each one, and the connection
between them is almost unnecessary, except to make the magnetic circuit solid.

Both cores are part of the same magnetic circuit and if they do not touch, the flux will decrease (similar as it does in the airgap between magnets and stator).
Less flux means less torque (kV goes up and kT goes down for the same number of turns).

I'm not so sure the flux density between them is higher, it seems like it should be the same as the face,.. being equal
distance from the center of the coils.

As a rule of thumb, when using Neodym magnets (1.1 - 1.4 Tesla), the cross section of the cores should be 2/3 of your magnets surface area (twice as much as the back iron), so if your magnet is 60mm², the core should have 40mm².
If it is less, the core will be in saturation which happens at around 1.5T for steel (don't know about SMC though).
The back iron should have at least a thickness of 1/3 of the magnets width.
So if your magnets are 18mm wide, it should be 6mm thick. Trapezoidal shape would mean that you could decrease the thickness from outside to inside.

If I stack five magnets on end,.. the face of the last magnet is the same, not five times stronger. :?:
It is a conundrum though, and I'll have to give it some more thought.

Yes, thats right, if you stack mangets above each other, the flux on the surface will not be stronger, BUT it will remain stronger the further away you get.

example:
lets say you have two magnets, one with 10x2mm (width x thickness) and the other one 10x4mm. They are N42 grade, so ~1.3T flux density.
The airgap in your motor is 1mm wide.

Now the flux which will reach the stator can be calculated as following:

2mm:
1.3T x (2/(2+1)) = 0,87T

4mm:
1.3T x (4/(4+1)) = 1,04T

So roughly, with the 4mm magnets your kT will be around 20% higher.
If you stack two of the 2mm above each other, the result will be similar as one single magnet with 4mm.

What is the max RPM you want this motor to spin at?
 
I looked up the Ti screws and see that they are indeed available, but not cheap. If I only needed a few it wouldn't be
a problem, but I need 36 of them, and that comes out to $60. - $100. Hmm... I'm thinking, we'll try the stainless ones
first, and call the Ti an upgrade option. I've got to many projects that need money at the moment, especially this one.

Maybe I can get the laser thermometer on the screws once it's spinning, and check that reading against a few Ti screws
and see if there's a difference? Or maybe I'm just being cheap, and I'll change my mind.

Dui, I would like to use some sort of bonding agent on the cores, DP 420 seems a bit extreme,.. I was thinking maybe red
Loktite? or just soak the whole thing with red motor enamel. The metal cores should have something to keep them from
rusting anyway.
Whats nice about Loctite is that it comes apart with a heat gun. But I'm not sure how strong it would be in this situation.
Have any other bonding suggestions?

Madin88, thats a lot to ponder, I'll have to get back to you a little later...
 
Thats awesome coleasterling,.. totally doable now. With the discount, it comes out to $30. I'll buy some just before
final assembly. I'm on my fourth pack of stainless screws, because I keep changing my mind.
Well, I can always use screws somewhere, and they're only a few-$ a pack.

Madin88, I see what your saying, lots of good information. :thumb: (motor RPM should be around 400 - 600 under load)
I think the original post was about the contact area between the cores, which I admit, should be a bit larger. It'll take
some experimentation to find a good compromise, and get a sturdy mount, and small enough hole through the carrier.
Making the slot a little longer would help.

I split a core in cad, and put them back to back,..then drew a curve that represents the probable flux path, just to help
visualize. I'm still thinking about how much of the flux is present, or concentrated at the contact point between the cores.

Flux curve.jpg

I'm a bit shocked by your statement that the back iron should have a thickness of 1/3 the PM magnets width though.
I remember hearing that on the last build, but forgot to take it into account on this build with the wider trapezoid's.

The 18 cores will be 30mm at the top, and 20mm at the bottom,.. so average it out to 25mm? The 16 PM magnets will be
about the same size, maybe a little bigger with some overhang, so that works out to be about 8mm thick back iron rotors.
Dang thats heavy! I'm not ready for that,.. I want to use 1/8" - 3mm thick rotors. :(
 
I don't want to make very much of a design change in the middle of a build,..like last time. It leads to big trouble.
But I could make the back iron a 'little' thicker without too much design change. I don't know how much fudge factor
I have with thickness,..as a 'rule of thumb' formula.

Another idea is to use a separate back iron ring on the outside, and it could be used to fine tune, or sneak up on the
minimum amount of steel, with different thicknesses. It would add thickness towards the top of the trapezoids, which
works out well for trapezoid taper, and minimum weight.

Selective back iron..png

Maybe not the best idea, but at least for now, it keeps this project moving forward. Meanwhile, I've got some searching to
do on back iron thickness. Since it's one of the heaviest pieces of the motor, I don't want to use any more steel than I have to.
 
The best way is to model your motor on FEMM and play with the back iron thickness, it's not a difficult programm too learn.
Once you're model is set you just need to add thickness until all the flux line are in the back iron. I'have done that for my design and came out with a thickness of 4.8mm back iron for 40*20*5 mm N38 magnet, but it's a value for a design with air core (less magnetic flux because of the large airgap) . femm.PNG
If you give me the mesurements of your current design i can try to model it
 
Thanks for the offer Thecoco974, that would be great! And your right, I do need to learn that FEMM program, since
I'm building motors,.. I always thought that it was beyond my level, but if you say it's not to difficult, I'll give it a
look. Is there a decent download link you can suggest?

I'm not sure how fine of a model your making in FEMM, or what you need for measurements, so I included everything.
Not exact, but close enough. I'm not sure about PM thickness yet, or width, I haven't really sourced the parts out yet.
The PM's will be segmented,.. or four 10mm magnets high, probably 4mm thick.

Each core is going to have 10 winds of 14ga wire around it, two layers of 5 winds connected in series with the other side.
It will be driven at around 500 watts, but 1Kw won't be uncommon if I get my wish. (48-56v)

Core measurments..png
Core Measurements.png
Rotor Measurements.png

Let me know if you need anything else, and I'll start looking at some FEMM tutorials on Youtube. :)
 
APL said:
But I could make the back iron a 'little' thicker without too much design change. I don't know how much fudge factor
I have with thickness,..as a 'rule of thumb' formula.
It depends on the magnet grade like N35 or N50 and the fact that steel starts to saturate at around 1.6T, but only a simulation can tell you how thin you can go without worsening the flux density within the entire motor too much.
APL said:
Making the slot a little longer would help.
Thats what i would do.
Another idea is to use a separate back iron ring on the outside, and it could be used to fine tune, or sneak up on the
minimum amount of steel, with different thicknesses. It would add thickness towards the top of the trapezoids, which
works out well for trapezoid taper, and minimum weight.
Good idea :thumb:
APL said:
motor RPM should be around 400 - 600 under load)

Thats low so you need lots of torque which means that you should not make much compromises on above mentioned things.
Imagine that it turns out that you are not happy with the torque at 400-600RPM, then you really cannot do much to improve it.
While instead if torque is excellent but you are not happy with no-load consumption (or drag), then you simply could increase the airgap to reduce the no-load losses and play with kV (flux weakening).

I was wondering why you decided to have the endturns within the magnetic circuit :?:
It is known that the part of the winding which is pointing in spin direction does not contribute to torque on PM motors so the core overhang, and probably from the magnets than too seems to not add much to the performance (but it will increase cost, weight and iron losses).
 
Rpm is low, but it's a mid drive, geared down 3 to 1, so theres less torque than a hub motor, and typically sees around
500 watts max.

Another idea I had for the rotor was to use a one piece tapered shape for minimum weight, and a more organic look.
Still need to know how thick though, and it would be more machining, so maybe not on this motor,.. but we'll see.
Taper matches the trapezoid taper.

Tapered Back Iron Concept.jpg

These cores and windings are pretty much the same as most motors I've seen so far, except for those that don't use any
tooth profile, or overhang at all.
End turns away from steel don't contribute, but end turns next to steel should contribute to the power of the overall core
flux, I would think.
Perhaps the lip on the top and bottom should be deleted, steel moved out to the edge, and the end windings moved out
a little further,.. out of the circuit more.

I have been thinking about it though. (end windings) This is one half baked idea I've been trying to work out,..sort of
a radiator style core. End turns get reduced, and radial turns are doubled. I'm not sure how the magnetics would wind up
being yet, and I'm probably missing something major, but I'm trying.

Radiator Core 1.jpg
 
Back
Top