APL's DIY axial-flux motor

Electric Motors and Controllers
User avatar
APL   10 kW

10 kW
Posts: 579
Joined: Aug 06 2018 8:44pm

Re: APL's DIY axial-flux motor

Post by APL » Jan 30 2020 3:26pm

Probably not a lot of difference, especially at slow speeds. Fact is, just about anything will 'work'. But It's good to know
about this stuff, to hone in on getting a better efficiency. At any rate, I feel a lot more comfortable about it.

I was going to mention the other side of the motor geometry equation, but ran out of time.
That would be the coil size and placement in a motor,.. my opinion, from what I've read so far anyway.

The coil width, or pitch, would be the same as with the magnets,.. all the slot to slot widths around the diameter of the
motor, added up, to make the 180 degrees.

But theres also 'Coil Span', which is all the coil widths added together,.. minus one.
Engineers have determined that you can diminish harmonics by adjusting coil span. A coil span of 144 deg. will neutralize
5th order harmonics, and a coil span of 154.3 deg. will take care of 7th order harmonics. So by averaging the two together
we get 150 degrees to take care of both.
Most all modern motors are made with 150 deg. coil spans.

That means that on an 18 coil motor, I need to take 17 of them and adjust their size until collectively, they have a 150
degree distance around the stator end to end. Then space all 18 of 'that size' evenly, to get them to fit in the 180 degree
diameter. After the winding size and fill has been determined, the core size would be whatever fits in the center.

I sure hope I have all that right, although it's kind of unclear as to exactly where on the coil the measurements are taken,
(outside, middle, or inside), I guess I'm assuming outside, or center of slot.

I also read something about the tooth overhang, or brim, and that it's used to even out the cogging fluctuations between
the cores, which will induce eddy currents in the magnets and back iron, as they pass over open slots.
It's an attempt to even out the slot gaps magnetically. I'm assuming that that the brim size is half the slot gap, minus
2-3mm so that wire can be placed in the slot, and to avoid magnetic fringing. The radius, or taper, under the brim is to
further even out the transition. :?:
Just my take.

I'll have to re-measure this motors coils, to see if I'm any where near any of this. Not that it matters too much, it's
a little to late in the build to change anything very much, and I'm not that concerned about harmonics.
Like fetchter says, based on observation, were in the ball park. But it's good to know for the next build. :)

User avatar
fechter   100 GW

100 GW
Posts: 14648
Joined: Dec 31 2006 3:23pm
Location: California Bay Area, USA

Re: APL's DIY axial-flux motor

Post by fechter » Jan 30 2020 4:27pm

If the magnets are touching, there will be fringing near the joint so the effective magnet width is slightly less. To optimize things, I suppose you don't want the extra weight of magnet doing nothing, so you should have a gap about equal to the magnet thickness. This may make mounting the magnets a bit more challenging. You could 3D print an alignment fixture to get the magnets all evenly spaced. And use very slow setting glue as things can easily get out of control when handling magnets.
"One test is worth a thousand opinions"

User avatar
APL   10 kW

10 kW
Posts: 579
Joined: Aug 06 2018 8:44pm

Re: APL's DIY axial-flux motor

Post by APL » Jan 30 2020 11:04pm

Yea, I've been giving that a lot of thought.. spacing and bonding the magnets. I had planed to print a spacer, and was
hoping for some space between the magnets to make that work. It will be over 8", so I need a large bed printer, but
that's not a problem.

The problem is bonding the spacer along with the magnets, which I don't want to do. I'm thinking of using cling wrap
in between them, but don't know how well that's going to work.
JB-Weld has a super slow setting epoxy, like 4 hours, that's also their highest bond rating, that I'm thinking on using.
Rated at 500 degrees.

I tried some very expensive glue last time, but wasn't all that impressed. I don't think it takes that much as long as the
prep is done right, and the surfaces are sanded well. Especially with large area magnets, and low RPM.

I've been spending a lot of time looking for magnets, and trying to figure out how to construct them, (on the cheap),
by arranging various sections. By choosing the right sizes, I can get the cost down quite a bit. Otherwise, magnets for
these things can run into the hundreds $$.

Having a trapezoid shape is what really makes it difficult. Right now I'm considering using two 3/4" squares on top,
and one 1" square on bottom, and doing a little shaping. It's by far the cheapest way so far. About $90. for all 32 PM's,
at 1/8" thick, and N52.

Another out of the box idea, is to make PM shoes, or caps, out of SMC material, and mill out pockets in the back for
multipule magnets, or one big square magnet. This way you wouldn't have to buy trapezoid shapes, which are extremely
difficult to find in the size and shape you need.

The cap would need to be very thin though, on the order of 1-2mm. and I'm not so sure that its possible to do that
with SCM. Once everything is bonded to the back iron, it would probably be strong enough.

This is an earlier concept.. with just a thin top. Now I'm thinking a fully covered cap, that fit's all the way around the PM.
SMC PM cap.png
SMC PM cap.png (57.98 KiB) Viewed 1652 times
I also read somewhere that a trapezoid PM will tend to saturate at the large end, and this cap might reduce that a bit
with slightly less T in the top center,.. and the back iron FEMM tends to show a 'part' in the flux there as well.
Just rolling the idea around at the moment.

major   10 kW

10 kW
Posts: 799
Joined: Mar 10 2013 10:12am
Location: NW Ohio, USA

Re: APL's DIY axial-flux motor

Post by major » Jan 31 2020 12:12am

Not ready for prime time... 3D printing magnets. Just saw this today.
https://phys-org.cdn.ampproject.org/v/s ... gnets.html
Maybe some day soon.

major

User avatar
fechter   100 GW

100 GW
Posts: 14648
Joined: Dec 31 2006 3:23pm
Location: California Bay Area, USA

Re: APL's DIY axial-flux motor

Post by fechter » Jan 31 2020 11:55am

"Tiling" magnets together to make a big one out of several smaller ones can work but is extremely challenging. The magnet sections will be repelling each other strongly and also have a strong tendency to flip over and stick to the other pieces. Being on a back iron helps a lot but you still need to force the sections together against their repulsion which takes a lot of force. All this while the glue is starting to get hard.

I've done it on a couple of motor builds. It can be done, but it's just not as easy as it might look. A spacer that keeps things in position will be very helpful.

Machining magnets is probably even harder. Any heat produced from friction can demagnetize the cut and you have a bunch of magnetic dust produced that is hard to clean off. Stuff is also quite flammable as a dust.
"One test is worth a thousand opinions"

User avatar
APL   10 kW

10 kW
Posts: 579
Joined: Aug 06 2018 8:44pm

Re: APL's DIY axial-flux motor

Post by APL » Jan 31 2020 2:21pm

I've tried cutting magnets many times, but to no avail, they always shatter into fragments. But they can be 'shaped' fairly
easy with a diamond disc on a Dremel tool, or a belt sander with a fine sandpaper belt. You have to go slow, to avoid heat.
It needs to be done outside of course, with a mask, as you've said, the stuff is toxic, and a fire hazard.

It will take some planning, and experiments, to get the segment bonding to go smoothly. Perhaps do all the north first,
and then all the south.
I had thought about using many really small magnets, 1/4" x 1/4" in a pixel style shape, But that would be a mess.

Ideally, length-wise segments would be the best, but so far, that has lead to much more expense. This arrangement helps
reduce eddy currents in the magnets as well, which is a nice plus. I'll keep searching though, perhaps I'll find something yet.
laminated trapezoid rotor magnets.jpg
laminated trapezoid rotor magnets.jpg (6.14 KiB) Viewed 1614 times
I guess I'm not sure how important it is to have a 'straight line' shape on the side of the pole magnet. I've seen some
motors with segmented magnets that have a 'stepped' shape to the sides.

User avatar
APL   10 kW

10 kW
Posts: 579
Joined: Aug 06 2018 8:44pm

Re: APL's DIY axial-flux motor

Post by APL » Jan 31 2020 2:34pm

The project is not moving along very fast lately, lot's of appointments, and things being up in the air decision-wise.
But I managed to get the weight reducing holes in the rotors, and compare the weight with the last rotors.
(minus the magnets)

Going with more steel and less aluminum has been somewhat of a success, in that I've lost about 100 grams per rotor.
This will be lost again when the larger magnets are used, and possibly more back iron added on the outside,.. but at
least for now, it's a better design, and makes for a thinner motor. :thumb:
V1 & V2 Rotor comparison..jpg
V1 & V2 Rotor comparison..jpg (52.96 KiB) Viewed 1613 times

User avatar
fechter   100 GW

100 GW
Posts: 14648
Joined: Dec 31 2006 3:23pm
Location: California Bay Area, USA

Re: APL's DIY axial-flux motor

Post by fechter » Jan 31 2020 3:35pm

APL wrote:
Jan 31 2020 2:21pm
I guess I'm not sure how important it is to have a 'straight line' shape on the side of the pole magnet. I've seen some
motors with segmented magnets that have a 'stepped' shape to the sides.
The part of the magnet next to the adjacent magnet isn't doing much so the shape won't be critical. As pointed out before, if the magnets are touching, the area around the joint will be fringing and not contributing flux to the pole. MFEA model would tell for sure.
"One test is worth a thousand opinions"

User avatar
stan.distortion   100 W

100 W
Posts: 140
Joined: Jun 12 2019 12:53pm

Re: APL's DIY axial-flux motor

Post by stan.distortion » Feb 02 2020 8:19am

For 3d printing a guide for magnet placement, ABS parts can be chemically welded with acetone (nail varnish remover but check brand first, some are useless). That allows large parts to be easily built up from smaller parts, accurate keying is all that's really needed. Superglue works well on PLA, certainly well enough for a magnet guide but it's a bond rather than a weld, with acetone and ABS the join is equal strength to the base material.

Not sure what's the best route with CAD exports, the obvious route is different CAD software but after seeing how well you've got on with that one it's worth the $500 imo. Depends how good the export is though, it could simply be an STL to STEP conversion (inaccurate) with a rip-off price tag, there are plenty of free tools for that.

User avatar
madin88   1 GW

1 GW
Posts: 3060
Joined: May 27 2013 2:02am
Location: Austria

Re: APL's DIY axial-flux motor

Post by madin88 » Feb 02 2020 2:30pm

APL wrote:
Jan 30 2020 3:26pm
But theres also 'Coil Span', which is all the coil widths added together,.. minus one.
Engineers have determined that you can diminish harmonics by adjusting coil span. A coil span of 144 deg. will neutralize
5th order harmonics, and a coil span of 154.3 deg. will take care of 7th order harmonics. So by averaging the two together
we get 150 degrees to take care of both.
Most all modern motors are made with 150 deg. coil spans.

That means that on an 18 coil motor, I need to take 17 of them and adjust their size until collectively, they have a 150
degree distance around the stator end to end. Then space all 18 of 'that size' evenly, to get them to fit in the 180 degree
diameter. After the winding size and fill has been determined, the core size would be whatever fits in the center.
It all sounds very interesting, especially the thing with harmonics which should make the difference between cheap and advanced motor designes.
Also, for an optimized design, the BEMF waveform should be at least of similar importance.
Aside from the slot/pole count of your motor, magnet spacing, magnet shape and coil spacing have the biggest influence on it.

For cogging torque there are these skateboard hubmotors (there is a thread here in ES) which have special assembly of stator teeth to get a very small gap.
These are known to have the lowest cogging torque and another benefit is the higher winding factor.
If you are familiar with the bavaria simulator, then you probably know about the "slot opening width" input field if you go in advanced settings. If you enter a number, like around 2mm as it's often the case, the winding factor will decrease.

As for magnet coverage the rule of thumb from fechter of leaving a gap similar to the thickness of the magnets is a good advice. It makes alot of sense.
Of course, thats it if you wanna optimize weight or BEMF waveform, because in terms of torque with 100% it should not be worse and most hubmotors i know have 100% and also those 18N/20P bafang gear drive motors have 100% :wink:

Regarding the magnet grade, i think that it would be a bad idea to use N52 because thats 1) one of the most expensive grades, and 2) they cannot withstand much abuse in terms of temperature.
It would be more advisable to use something like N42SH or even UH grade.

If you think about having custom magnets made then check out this magnet supplier:
viewtopic.php?f=30&t=91144&start=125#p1465783

larsb   1 MW

1 MW
Posts: 1641
Joined: Dec 10 2014 5:12am
Location: Gothenburg, Sweden

Re: APL's DIY axial-flux motor

Post by larsb » Feb 03 2020 1:14am

Yes, agree, a slightly warm n52 is weaker than a lower grade magnet. The n52 grade is only good if you never overload your motor.
Ride on!

User avatar
APL   10 kW

10 kW
Posts: 579
Joined: Aug 06 2018 8:44pm

Re: APL's DIY axial-flux motor

Post by APL » Feb 03 2020 3:04pm

Haven't heard that about N52 magnets. A little late now though, since I just sent off for some. Never fails! :(

Well, they 'say' N52, but who's going to know if their not. I don't trust any advertising anymore, at least, on the Bay.
I'm sure they will be fine.. same as the last cheap magnets I bought.. and they worked all right. Cheap is the thing
when it comes to experimental's. (with my budget) I don't think I can afford custom's for this motor, but it's good to
have the link. Things might change, if I make new rotors, and others may need the link. (I also like the square wire
link, that Larsb posted, and is worth repeating here; https://www.wires.co.uk/acatalog/rt_ec_wire.html)
The last motor didn't get all that hot, and this one is a better design, with better cooling, so I'm hoping to squeak by.

For some reason the 60x10x3mm magnets were cheaper than the smaller ones in quantity. I can get one pole magnet
by cutting two 60's in half. I found out that by scoring them with a Dremel disc, I can snap them accurately.
I plan on making a metal fixture to hold them in the pole shape, and sand the edges.
$60. for all 32 PM's is a good deal, so I'm willing to put a lot of work into it.

Might be just a lot of hope though,.. we'll see how that goes. At least I'll get the segmented design I wanted, like
In this photo.
They have two more segments on the inside radius, making for a 'long' core. I wonder how much more power/torque is
to be gained versus the extra weight, copper, and small diameter?
Neodymium-Magnet-Stator-Assembly.jpg
Neodymium-Magnet-Stator-Assembly.jpg (42.82 KiB) Viewed 1488 times
Stan.distortion, thanks for the PLA/ABS bonding info. good to know. I've often thought about doing smaller sections, and
bonding them into a large disc, because most printers only have a 150mm table. When I buy one, I'll get a 230mm, so
I can do 8" disc's for motor stuff.
I've been watching Onshape tutorials lately, and fairly certain I'm going that route.. not all that different than what I'm
using, and it exports in all solids. Lot's more tools and options, but like most things, you only use few of them the most.
Last edited by APL on Feb 03 2020 3:20pm, edited 1 time in total.

User avatar
stan.distortion   100 W

100 W
Posts: 140
Joined: Jun 12 2019 12:53pm

Re: APL's DIY axial-flux motor

Post by stan.distortion » Feb 03 2020 3:19pm

Good stuff, you've certainly got the hang of the CAD software you're using and the first one is the painful learning curve, you've got your feet under you now so others are usually pretty easy to get a handle on. Had to laugh when you mentioned the block for filing the magnets level, that's a perfect job for a 3d printer but it takes soooo damn long to remember to put that 2 and 2 together when you have one sitting there :) I'm pretty sure folks already mentioned corrosion on magnets when the coating is stripped off them, I have some here that have been in dampish conditions for about 5 years with bits of coating pulled off and they're only lightly corroded but it would enough to break them free if only glued on a face.

User avatar
APL   10 kW

10 kW
Posts: 579
Joined: Aug 06 2018 8:44pm

Re: APL's DIY axial-flux motor

Post by APL » Feb 03 2020 3:30pm

We need more magnets like this! 13.5 T!!! Might take a bit of power though... :lol: https://home.cern/news/news/engineering ... n-record-0
smc-magnet_1.jpg
smc-magnet_1.jpg (155.87 KiB) Viewed 1477 times

User avatar
fechter   100 GW

100 GW
Posts: 14648
Joined: Dec 31 2006 3:23pm
Location: California Bay Area, USA

Re: APL's DIY axial-flux motor

Post by fechter » Feb 04 2020 8:48am

APL wrote:
Feb 03 2020 3:30pm
We need more magnets like this! 13.5 T!!! Might take a bit of power though... :lol: https://home.cern/news/news/engineering
But it's superconducting so needs to be cooled to liquid helium temperature. At least that would solve motor heating issues. I'm looking forward to practical graphene conductors.
"One test is worth a thousand opinions"

User avatar
APL   10 kW

10 kW
Posts: 579
Joined: Aug 06 2018 8:44pm

Re: APL's DIY axial-flux motor

Post by APL » Feb 04 2020 1:27pm

OK, that one was pretty big, heres a little smaller one, an 8' x 2" neo. thats just plain scary.
A lot of you have probably seen brainiacc75's video's, but if not, then check it out.

https://www.youtube.com/watch?v=yM4Xe2c0B8M

Graphene is a really promising material that is making it's way into the market more and more, I don't quite understand
it yet, but can't wait for it to wind up in our motors some where.

https://roboticsandautomationnews.com/2 ... ial/14725/

I think I read on Wikipedia that plasma was the most conductive thing we have, at 100 percent, but of course,.. a little
hard to make coils out of yet. :)

User avatar
amberwolf   100 GW

100 GW
Posts: 29129
Joined: Aug 17 2009 6:43am
Location: Phoenix, AZ, USA, Earth, Sol, Local Bubble, Orion Arm, Milky Way, Local Group
Contact:

Re: APL's DIY axial-flux motor

Post by amberwolf » Feb 05 2020 1:27am

not gonna have to worry about motor overheating once we build them using plasma, are we? ;)

ErgonomicMMO   1 µW

1 µW
Posts: 1
Joined: Feb 05 2020 8:45am

Re: APL's DIY axial-flux motor

Post by ErgonomicMMO » Feb 05 2020 8:50am

Last edited by ErgonomicMMO on Mar 09 2020 7:23am, edited 1 time in total.

User avatar
fechter   100 GW

100 GW
Posts: 14648
Joined: Dec 31 2006 3:23pm
Location: California Bay Area, USA

Re: APL's DIY axial-flux motor

Post by fechter » Feb 05 2020 9:00am

Big magnets like that are very scary. I have a bunch of much smaller ones (2" dia x 1/2" thick) that are scary enough. Those can pinch really hard (ask me how I know). I've also played around with things in a MRI magnet which is 3T. Chunks of copper fall in slow motion. It will demagnetize magnets if you force them to repel the field.
ErgonomicMMO wrote:
Feb 05 2020 8:50am
Where Can I buy 350mm magnets?
I haven't ever seen any that big that are one piece.
"One test is worth a thousand opinions"

User avatar
APL   10 kW

10 kW
Posts: 579
Joined: Aug 06 2018 8:44pm

Re: APL's DIY axial-flux motor

Post by APL » Feb 05 2020 2:21pm

I think they made the 'Flux Capacitor' out of plasma.... :confused:

User avatar
APL   10 kW

10 kW
Posts: 579
Joined: Aug 06 2018 8:44pm

Re: APL's DIY axial-flux motor

Post by APL » Feb 06 2020 2:34pm

A little more about segmented pole magnets, this add for custom made magnets talks about it a little bit, and mentions
something about a specific glue used to insulate, and bond them together. I've read about the glue before, but haven't
found references to what brand it is, or whats 'specific' about it. Just curious.
Segmentation-Neodymium-Lamination-Motor-Magnet.jpg
Segmentation-Neodymium-Lamination-Motor-Magnet.jpg (24.14 KiB) Viewed 1315 times
https://www.hsmagnets.com/product/segme ... or-magnet/
(click on the 'more')

User avatar
fechter   100 GW

100 GW
Posts: 14648
Joined: Dec 31 2006 3:23pm
Location: California Bay Area, USA

Re: APL's DIY axial-flux motor

Post by fechter » Feb 06 2020 3:12pm

I would imagine it's a high temp epoxy. To insulate segments you'd need to coat the faces with a very thin layer of the epoxy stuff first and let it cure, then glue the segments together. Most people don't worry about eddy currents in the magnets, but there will definitely be some.
"One test is worth a thousand opinions"

User avatar
APL   10 kW

10 kW
Posts: 579
Joined: Aug 06 2018 8:44pm

Re: APL's DIY axial-flux motor

Post by APL » Feb 06 2020 8:15pm

Good to know though,.. this motor isn't refined enough to worry about it yet.
I had thought about using 1/8" x 1/8" x 1-1/2" long magnets, and bonding them all together like the photo above,
for a home brew version, but they are not common, so the price is high, and it would be hard to align all the pieces.
Something for the future maybe.
Perhaps this company has decent prices for their customs,.. could drop them a line too.

User avatar
APL   10 kW

10 kW
Posts: 579
Joined: Aug 06 2018 8:44pm

Re: APL's DIY axial-flux motor

Post by APL » Feb 06 2020 8:32pm

I managed to get all the spacers made. I left a few millimeters on the ends, for later,.. once the motor is together I'll
have to sneak up on the air gaps by milling the spacer length.

Theres a 'step' on each side of the top of the spacer, for the fan segments to be attached to.
24 - 3mm screws to hold the fan segments on the spacers, is not making me happy, but at least they don't all have to
come out, when the motor is taken apart. I can leave them in two or three sections.
The fan sections will cover the holes in the rotors, and seal them up for air flow.
Spacers V2..jpg
Spacers V2..jpg (47.85 KiB) Viewed 1284 times
Then I tried snapping a few of the 60mm magnets, and lined them up to see how much will have to be ground off.
Worked out pretty good, and I won't have to do much grinding. (I marked the sides with a sharpie, to show whats being
removed.)
First cut - PM's V2.jpg
First cut - PM's V2.jpg (86.61 KiB) Viewed 1284 times
Only 62 magnets to go. :)

User avatar
fechter   100 GW

100 GW
Posts: 14648
Joined: Dec 31 2006 3:23pm
Location: California Bay Area, USA

Re: APL's DIY axial-flux motor

Post by fechter » Feb 06 2020 10:21pm

I found it handy to mark the faces of the magnets so you know which side is N vs. S. When tiling pieces together, they will strongly repel each other while gluing. You need a fixture of some sort to hold them in place while the glue hardens. I used some little C-clamp and blocks of wood on one project. I made wooden wedges that could be tightened once all the magnets were on.

It will also be hard to place all the magnets in one pass. You might want to do all the N facing ones first then do the S facing. The back iron helps keep them down but the glue lowers the friction so they want to slide around. Small gaps between the pieces shouldn't be a big problem.

I used a dremel with a cutoff wheel to score one side then try to snap them like a piece of glass. Sometimes worked OK, sometimes the magnets broke not along the line.
"One test is worth a thousand opinions"

Post Reply