APL's DIY axial-flux motor

For no backing iron entertainment purposes only:
try searching: "Halbach Array"
 
Thanks KD5ZXG, for bringing that up, good call. :thumb: I thought I knew about Halbach tech, but found out that I
didn't really have a grip on it.

https://www.google.com/search?hl=en&tbm=isch&source=hp&biw=1680&bih=1012&ei=MMRdXIv2FYjKswWY0oGYAQ&q=halbach+array&oq=hal&gs_l=img.1.0.35i39l2j0l8.3136.3674..9011...0.0..0.109.317.0j3......1....1..gws-wiz-img.....0.4RAOaWBre5A#imgrc=_

I didn't know that one side is stronger than the other, and 2 to 3 times stronger than a standard magnet array.





Since the magnets are placed right next to each other, and complete the circle, they act the same as a pice of
back iron would,.. but without it.
Also, since they have 2-3 times the strength,..that means you can use magnets 2-3 times as thin as standard
magnet's? Possibly making that set up lighter than the standard back iron arrangement.
Now thats cool! Magnets and iron are a really big part of the motor weight.

The main problem with a halbach set up in motors, is that the magnets need to be a very specific size, shape,
and thickness. This puts the whole thing squarely into the corporate domain. :(

But as a devious DIY guy, I was thinking of a way to cut up existing neo's into shapes. Youtube has a few crude vid's,
nothing I would recommend, but shows that it 'can' be done. I have a buddy that has a diamond blade, water cooled,
tile cutter, that might do the trick though. When I get some time, I'll give it a try.

Even cruder yet would be to crush some magnets up, and re-bond them on top of another magnet,(separated by plastic)
to realign the fields. Then you could make any shape you wanted. Hey,..if it jerks, it works, right?

The only other way I can think of, is to source some pre made magnets,..arrange them in a halbach pattern, and then
design the rest of the motor from that, using whatever pole count you wind up with.
 
Just thought of a different way,.. use hundreds of small square magnets arranged in the pattern you need,
they can be bought in different orientations, so making a thin Haibach array would be easier. Not perfect, but
possible.

Also found this interesting site, that sells magnets, and also has a magnet gauss calculator for air gaps. Check
out the overstock section,..they have a 3" x 3" x 1/8" neo. for sale that would be a nice size for cutting into
trapezoidal shapes.
Although not Haibach, since you need different orientations for that.

https://www.kjmagnetics.com/blog.asp?p=gap-calculator
 
APL said:
I didn't know that one side is stronger than the other, and 2 to 3 times stronger than a standard magnet array.


Halbach arangement.jpg

One thing I recently learned about halbach arrays is that while they do shift the majority of the field to one side of the magnets, they don't create an even field at all. The field strength above the individual magnets in the array varies greatly but in a predictable pattern.
This video was very informative...
https://youtu.be/ep2PAMrzwEU
 
Cutting magnets is pretty challenging. The material is very hard and you have to keep the temperature below the magnets rating.
It also makes a lot of sparks and the dust is flammable.
I’ve done it a few times but it should be avoided if possible.
 
Also, the neo stuff tends to corrode easily, so you should coat it with something immediatly after cutting. AFAIK tha'ts why they coat them with nickel.
 
Thank's,.. I have noticed that the nickel tends to blister off cheaper magnets, and they corrode quite easily.
Never knew why though.

Just wondering if cutting was an option, since I keep running into magnet size problems, every time I turn around.

As far as halbach field strength goes, I guess it's closer to 2x normal, rather than 3? The two magnets on either side
of the main poles are giving up half their force to the pole magnet, giving it almost twice the strength.

Not that we need twice the strength in motors, but to me, that means that I can get by with half the thickness,
or weight. However twice the magnets 'is' twice the weight,.. except there is no need for back iron, because the
magnetic circuit is completed between all the magnets. Unless I'm missing something?

And I did notice that the fields of all the poles are not exactly even,..kind of a brain boggler, perhaps it's because
they are not connected in a complete circle yet?
More research to be done.

Anyway, a halbach system is is an interesting possible alternative, and may even lead to a lighter PM ring,.. something
that can be pursued on this motor, since the rotors will allow for it. But it would be difficult. :(
 
You're missing the effect of the backiron to increase the magnetic field strength in the air gap.

As stated above: roughly 2x magnet material needed for same field even though you have a Halbach array.

Without the halbach it's even less field strength than that..
 
APL said:
As far as halbach field strength goes, I guess it's closer to 2x normal, rather than 3? The two magnets on either side
of the main poles are giving up half their force to the pole magnet, giving it almost twice the strength.

Let me share a bit of what i have learned about mangetic circuits:

Let's assume you going to use N42 grade magnets which have a remanence or magnetic field strenght of around 1.3 Tesla.
If you now place two of such mangets side by side, one N and one S, with 100% steel return (sufficient back iron), each magnet will eaxctly provide the 1.3Tesla on the surface.
With insufficient steel return (back iron too thin as often done on cheap motors), the field strength will be less than 1.3 Tesla.

What you want to know now is how many % of this 1.3T will reach the stator teeth, as this will highly effect kV and torque of the motor.
This can be calculated with following formula (and as you will see now the THICKNESS of the magnets comes into play):

-> field strength on stator = initial field strength x (thickness / (thickness + airgap width))
Let's assume the magnets in your motor have a thickness of 4mm and the airgap is 1mm and you put those vlaues into the formula:

1.3T x (4mm / (4mm+1mm) = 1.04T or 80% of the initial field strength

If you would reduce the airgap width, you also could reduce the thickness of the magnets to achieve the same:

1.3T x (3mm / (3mm+0,7mm) = 81% of initial field strenght

So 4mm mangets with 1mm airgap do the same as 3mm magnets with 0,7mm airgap.

I don't know how this could be calculated with an Halbach array, but from my readings and as larsb mentioned it will be worse, or with other words you need more magnets and/or with very high field strength ($$$) to achieve the same as you could achieve with a standard manget array with back iron.
The other "unknown issue" is magnet coverage. With Halbach you will always have 100% coverage because otherwise it will not work, but with standard magnet array you can simply adjust this just by leaving gaps between N and S poles. The coverage will highly effect BEMF waveform and torque.
 
I thought that the video provided by Archer321 explained things fairly well. Theres tons of info out there and I've
personally only scratched the surface,.. I'd have to do the usual week long studys before I could argue the point on
anything Haibach very strongly.

At the moment I'm not even really clear on what it is that I'm arguing about. But it was my understanding that back
iron only serves to complete the magnetic circuit, and allows the PM magnets to share each others fields, somewhat,
thus making the flux in the air gap stronger,..'relative' to single disconected magnets. (no back iron)

The distance between the magnets on the back iron, and-or the thickness of the iron, and magnets, will obviously
change the gap flux as well.

The halbach array is different, in the fact that extra magnets are added, solely to increase the main pole fields,
and since 'they' complete the magnetic circuit by connecting all the poles, just as the back iron would normally do,
there is no need for back iron in that system.

Obviously there is an advantage to halbach, it's both very expensive, and physically hard to do, and motor companies
are going way out of their way to do it. Even so, there are literally thousands of motors with the halbach set up.
 
You know, there's never any "need" for backiron. You can always use more magnet material or copper wire :D
It's that it's an effective and cheap way of getting high field strength in the right place.

Backiron together with halbach is also good for the field strength, off course you need less magnets in this case.

Madin stated it well, the worst drawback for a diy guy are the close tolerances needed for a good halbach (cost is actually not that huge for the extra magnet material)
 
Wires, strands, and turn counts.

Well, this is a subject that has been talked about many times over the past decade on the ES, and I'm not going to
step too hard on that quicksand. There are a few good posts on the subject, and I'll put the links below.

https://endless-sphere.com/forums/viewtopic.php?t=85960
And some great research by Doctorbass; https://endless-sphere.com/forums/viewtopic.php?f=16&t=19301


As I understand, first, you need to start with the controller, and what it can handle. Then it's up to tooth turns, which
determine the speed of the motor, given the same voltage. Theres also the size of the wire, or strands, that determine
the current capability, and overall resistance of the windings, and phase wires.

To complicate things, theres the number of motor slots. or windings, of different motors, that will change resistance's
and phase frequency's. (take two aspirin now)

Most of these motors use multi strand wire, because a solid strand wire is very stiff, and hard to wind within the radial
motors slots.10 strands of 24 awg wired in parallel, is the same as one strand of 14 awg wire, but much more supple to
work with. Same weight too.

One way you can convert multi strands to solid strand is to use the amount of ohms per foot,.. or thousand feet. (chart)
24 awg wire is 25 ohms per thousand feet, so ten of those strands together will be 2.5 ohms.
(decimal goes backwards with ohms).
Looking at a wire chart we can see that 14 awg wire is 2.5 ohms per thousand feet. 10x24 = one 14 awg.


Ohms per 1000ft chart.jpg


There are endless combinations of strand gauges and counts that can be employed, but some will result in better copper
fill, because of the wasted space in between the round wires changes. If wire was made in a square cross section, you
could possibly achieve 100% fill, but it would be difficult to keep all of the squares perfectly oriented with each other.

Since this motor has straight cores, I can go either way with strand or solid wire, although since I've already bought the
24 awg wire, thats the way I'll go.
But I think the solid strand looks better, and comes out to exactly 12 turns. (I didn't have enameled wire for the photo)


Strand vs solid.jpg


Since I don;t know enough about designing motors, I'm trying to mimic the basic constructions of the Crystalyte motor
and controller. That's why I'm using the 10/6 tooth winding, but since both the N and S poles are on one core now, they
each will have 12 turns of 10 strands.

Plus, I have less poles now, so there will be less resistance overall, and I need to add a few turns to each pole to
compensate. (42 vs 53 poles)
This will make the motor turn a little slower, but having less poles should make it turn a little faster.
(a lot of finger crossing here)
At any rate, I have gearing, so speed doesn't really mater, I'm more concerned in the resistance of the final phase wires.
 
Wire is made in square and rectangular shape. I've wound with rectangular 2.5x1.25mm, it was stiff but still ok to work with. I got close to 65% fill if i remember correctly which is at the top end of the winding scale. The parallell wire winds get crossed and lose fill so easily, they are only better for an easy winding job..

You'd probably want to test wind a coil and spin motor up as a generator and confirm what kV you can get, otherwise you might do all the winding to realise that you got the wrong winding scheme or turn count. Good luck!
 
Neodymium magnets are 14 parts iron to three parts other stuff.
2 parts neodymium. 1 part boron. So much for eliminating iron...

Samarium Cobalt: 5 parts Cobalt to one part Samarium.
Extremely brittle, but won't corrode, and takes heat better.
Definitely not saying Cobalt makes a better motor than iron.

49%Cobalt 49%Iron 2%Vanadium isn't permanent, but carries
the highest flux of anything you might use for "back iron."
 
I've also read here on the ES somewhere of using coated strip, or foil copper for a winding, which is a really neat idea.
I have a hard time wrapping my head around exactly how to do that though. Close as i can come, is to take a pice
of 14 awg and beat it into a foil 10-20mm wide, and then measure the thickness. Ha! not likely.. :roll:

Otherwise theres the ohms per foot thing. But I'm sure there is a mathematical way to do it, which would make the
most sense. 14 awg is .0675" diam. so you'd just need to unroll it?

Anyway, it might result the best copper fill, and would be an easy wind on a straight core. I have no idea on how it
would actually perform though.

So many experiments could be tried here. :thumb:

larsb, I'll have to look up square wire, first I've heard of it. Any idea of a source? Just curious, as I don't think it
would improve this particular motor much, but it would be cool.
 
Also, I not worried about rewinding the coils, because it's so easy. Each one only takes about one minute to wind,
And if I put more winds on rather than less, its an easy job to just remove a wind or two from each one.
Adding winds is a different story.
 
Well, yea, sure,..I did look it up. but most of it is overseas. thought maybe you had a US source already.
Never mind, it's just a reference thing.

But I did find some interesting wire, such as square wire with an internal hole,(cooling?), and some square hobbyist
gold, and silver filled,(coated) wire, and even some square 'solid' silver wire, that would make the ultimate 'bling motor'. :)
Although expensive as hell. And would have to be enameled.

I like to dream.


View attachment 1


Gold filled square copper wire.jpg
 
Electrical conductivity of gold isn't great - about 1/3rd less than copper. Of the common metals only silver is better than copper, but only slightly and harder to work with.
 
APL said:
Well, yea, sure,..I did look it up. but most of it is overseas. thought maybe you had a US source already.

sorry for the pun, it's only that i am in sweden.. only time i buy from the US is when i really must since the air freight and taxes are huge..
 
Rectangular Compacted Litz.

Skin effect may not matter for reasonable motor frequencies.
But proximity effect still matters, as does breaking up eddies.
https://en.wikipedia.org/wiki/Proximity_effect_(electromagnetism)
 
Whoops! Sorry larsb, I forgot to look at where your from! :oops:

As a crazy American I assume that everybody is from here.. Ha! My apologies, I should be more careful on the ES.

It's fantastic to be able to talk to people all over the wold! :thumb:
 
Thanks KD5ZXG,.. I have seen the square or rec. Liz wire. Awesome looking stuff, and good to know about.
But very expensive I imagine, hard to solder, and this particular motor is such a low frequency that it wouldn't
be very advantageous. Still, thanks for the info! Keep it coming. :)
 
Back
Top