APL's DIY axial-flux motor

[fechter]

I'd guess the Magnax motor has a solid disk in the middle of the stator so it looks like two of the cores in the picture above it back to back. The middle part just needs to be strong enough mechanically.

The question about optimum copper/iron ratio is going to be hard to answer. A MFEA model is probably the best way to optimize it. I generally just look at successful design and copy it. You want a little tooth overhang but the amount is apparently not that critical as I've seen both extremes in good motors. My rule of thumb is the space between teeth should be about 2x or more the distance from stator to magnet (air gap). Too much copper and not enough core will result in saturation. Too little copper will result in lower power density. A coreless motor can't saturate but at some ridiculously high current you might demagnetize the magnets.

 

[APL]

I know, it gets really technical, there's plenty of formulas for it, that will get you very close, and the MFEA model's will hone
it even closer.

I'm probably asking too much, for something more general, for earlier in the game. The 2mm or so gap between teeth is a
good start though. And the kind of thing I'm after. Tooth overhang seems to be all over the board.
Maybe something to do with copper fill that I'm not getting. Needs more research.

Seems like core size and mass are somewhat a product of design too. Given the same 9" size diameter, a 6S/8P motor will
have very large cores compared to an 18S/16P motor.

Then again, the 18 slots, when divided by three, will be close to the 6 slot motor in mass. So nothing really changes there.
More,.. or less cores, does not change core mass for same size motors.
(given the same width)

So the only thing that can change to a greater degree is the copper? Still, you can only fill so much before tooth overhang,
and core thickness become an issue, and saturation.

How do we know when a core is oversaturated,.. by doing a single core experiment? It seems to me that the thing to shoot
for in a bike motor is the absolute minimum amount of steel, so you'd want a core's steel to be just a 'little' more than 1T,
or the PM's strength, before it saturates.

 

[APL]

Maybe thats a good start for a general formula.

Given an axial motors diameter, and slot count, the core faces being be spaced no less than twice the air gap,
the rest of the core mass is then determined by the minimum amount of steel, for core saturation, at just over the
PM magnets maximum Tesla strength.

And copper fill is determined by turns for speed, gauge for current, and current just less than it takes for saturation.

 

[fechter]

With a linear hall sensor (like the kind in your throttle) and a voltmeter, you can see if a core is saturating at a given current by plotting flux vs. current and looking for a bend in the graph. A fancy gauss meter will tell you the same thing but will also let you know what the level of flux is at saturation.

I wouldn't overthink it at this point. The real math is incredibly complex and hard to get right. Doing "what if" models in MFEA is probably a more practical approach.

 

[APL]

Sorry for being lazy, and not doing the research, but I guess I'm not getting the Bsat rating completely yet.

Bsat's on the chart for different mix's, rated in Gauss, 18,400G being 1.84 Teslas, needs to relate to something for me.
Is it for a partial, a cubic mm, cubic centimeter, etc.? Or the point at which a substance saturate's regardless of size?
Or maybe I'm missing it altogether.

Bsat second column, left.


Micrometals; https://www.micrometals.com/materials/pc

 

[fechter]

Bsat is a material property and independent of size/shape. The type 40 stuff looks pretty good. 1.8T is about the same as typical silicon steel laminated core.

 

[APL]

I'm looking into info. on Bsat and saturation properties at the moment,.. might be awhile, but I'll get it.

I found this video on Youtube of using steel shot for a core. Mildly interesting. LONG in the tooth, so move through it.
Unfortunately, he gives no performance tests, or information on how it compares to anything else.
So it's fairly useless, but the concept is still valid. I think powdered iron is better than steel though, and the amount of
air between the shot seems extreme. But there it is,.. worth a mention at best.

It does, however, conform to any shape, and theres probably much smaller shot, and metal choices out there.
'Steel' is an alloy, and 'iron' is different.

Youtube steel shot video; https://www.youtube.com/watch?v=kpHEyzvc4dY

Ive got one of those big #40 mix blocks coming in the mail, so we'll give it a going over. Hope it doesn't brake the mailbox.
I have a Gauss meter app on my phone, that wturber turned me on too,.. that should help with that.
I saw a guy on Youtube using a 'Dremel' with a diamond tip, doing some fairly intricate shapes in ferrite, that might help
with the finer shapes.

 

[fechter]

The #40 stuff will be interesting. Most ferrite stuff I've messed with in the past was super hard and super brittle, like glass. I don't think any of it was powdered iron like the #40 though. For DIY powdered iron I'm not sure how you do the phosphate coating for insulation. Pressing it into a mold with a lot of pressure would compact the air spaces more but you'd have to have somewhere for excess binder (epoxy?) to squeeze out. It won't be as good as the #40 or laminations in terms of saturation.

 

[larsb]

Sorry for being lazy, and not doing the research, but I guess I'm not getting the Bsat rating completely yet.

Bsat's on the chart for different mix's, rated in Gauss, 18,400G being 1.84 Teslas, needs to relate to something for me.
Is it for a partial, a cubic mm, cubic centimeter, etc.? Or the point at which a substance saturate's regardless of size?
Or maybe I'm missing it altogether.

Gauss is a density unit, maxwell per area. google is your friend

 

[InterestedinEVs]

I just purchased two of the #40 blocks (guy is in Texas!) and will try machining on them. He also has over 200 pieces left. The supply probably won't go away anytime soon unless he sells the lot. They are cheap enough to buy many to keep on hand, though.

 

[APL]

Good to know how many he has,.. if the material pans out, I'll have to pick up some more, I'm sure he's dying to get
rid of it. I doubt weather anybody else has a use for such a large piece.

From what I've been reading, most ferriet's are just not up for the task of a motor core. Most are in the .3T to .5T range.
But powdered iron mixes can achieve close to silicon steel. The differences, are in the frequency's.

A simple comparison chart that seems to show SMC materials where we need it to be.



Also found this interesting article on a Composite, powdered core - lamination hybrid axial, thats full of imfomation.
I don't think I can view the full document without joining the site. But it's worth the look.
A very interesting motor, complete with test results.



From: https://www.semanticscholar.org/paper/Axial-Flux-Motors-Using-Compacted-Insulated-Iron-Jack-Mecrow/5ef72ec69d71b583f9f22a1e85f6f0aa8a59553c

 

[larsb]

These might be of interest, even found a doctors thesis on powder core motors but it was too large to upload

View attachment 10.1.1.847.9796.pdfView attachment 10.1.1.882.1738.pdfView attachment 55_Experience-ATOMET-SMC-55.pdfView attachment 2View attachment PowderMet_2013_Chicago.pdfView attachment designOfAxialFluxPMM.pdf

 

[APL]

Thanks larsb, that's a lot of reading! The last one didn't want to download for me, so I don't know what thats about.
From what I've read so far, it confirms the SMC direction we've been going, and I see lots of motors being made out
if it, now that I've been looking. It can even be used for back iron and tooth caps.

I found some machining info. on it,.. evidently there are some offerings made 'specifically' for it, but most will be
machinable.

Heres a simple way to get axial teeth, if you have a 'ring', or toroid of it.
Slot the side all the way to the end, for the tooth top, and then cut the tops for the tooth overhang's.
If you slot just the middles, then there would be a double sided bar, or two tooth tops, for a motor like mine.







My bar came in the mail this morning,.. heavy as hell, and enough 2x4's in the box as fillers, to feed a wood stove.
If I get some time tomorrow, I'll try hacksawing a piece off and milling a few slots in the block,.. to see how it goes.

If that guy in Texas has over 200 of these, at 21lbs each, then he's sitting on over two tons of it!

 

[APL]

I've been trying to find out the weight differences between SMC and laminations, but to no avail. I'll have to compare them
here, when I cut some pieces. I'm sure SMC is lighter, but don't know by how much.

I also found out that a person should shorten the middle part of a core of SMC, in order to optimize steel loss, which
has the benefit of reducing copper length, and steel at the same time. So a SMC coil and core will be much lighter than a
laminated one. Good news for bike motors!

Notice the center of the tooth..

Article;https://www.sintex.com/en/about-sintex/cases/item/stator-design-med-smc

 

[APL]

I was able to do some work on the iron block today. I put it in the bandsaw , and it cut like butter, which is a big relief.
It cut the whole 3" x 3" piece in half in about two minutes!

Same way on the mill,.. fly's through it with ease. I did have some trouble on the ends with some chipping, but I'm not
using a very new cutter, and don't know what size will work best, or at what speed. I'm sure theres a learning curve.
Seems to want to chip when leaving the piece on the end, so I stopped half way, and came at it from the other end,
and had no trouble. So there are some tricks to be learned yet.
It might be a good idea to grind or sand the finer shapes.

Anyway, looking at it through a magnifying glass, you can see that it seems somewhat porous, which I assume is binder.
And when I put an ohm meter on the surface, it would reed anywhere from 25 to 425 ohms, so it's not a dead short either.



Next I made a piece the same size as the lamination cores I have in the motor, and weighed them on the scale. There
wasn't any big difference between them, I was surprised to see,.. the lamination core was 91.4 G and the SMC was 89.4.
Since the laminated core has tape and fiberglass on it, they are pretty close to the same weight.

 

[fechter]

That's great. Seems like you could make almost any shape with the right cutter bit.

 

[APL]

Yea, I think it's a GO so far. As long as you don't make tooth overhang to thin, and can mount the cores without
too much stress, it's looking pretty good.
Any shape can be made as long as it's not too thin, or long, and some fiberglass reinforcement might help with that.
It would be interesting to be able to try other SMC offerings, to see if some have more strength.

I read somewhere that there is still a certain amount of surface 'smear' from cutters, that can be addressed with etching
if needed, (possibly sanding, or sandblasting?), but one great benefit over laminations is that it doesn't lose any values
from machining. And has a uniform 3D low hysteresis direction, so it doesn't matter how you machine a tooth out of it,
it's all the same.

Next step is some magnetic test of some sort I suppose. For a start I should just wind each core with the same piece
of wire, and take a Gauss measurement from the same distance. using the same power supply.

 

[Lebowski]

Still... your magnets make a certain flux. With the coil current flux rises to 1.41 times this. For this flux, keeping saturation in mind, you need a certain metal area. The smallest circumvence (wire length) your get for a circular area..

 

[APL]

Yea, a simple magnetic comparison isn't going to be much use. What I really need to know is how many Gauss the PM
magnets I'm using are. I can put a spacer on the end of my Gauss meter, and take a comparative reading from them.

Then I can take the same reading at the same distance from the coil and core, at a certain volt and amp rating.
If I can find out where the coil and core saturate, then I can adjust the amount of steel to where it saturates at around
the same Gauss as the rotors PM's. (Or as you say 1.41 times coil flux,.. not sure what you mean,.. you want the core to
saturate at above the PM T?)

Still rolling it around in my head. But, I don't have the equipment to do the volt/amp tests at the moment anyways.
I'm thinking maybe a cheap PWM motor controller from Ebay,..(10v - 50v, 100A), and a shunt type 100A amp meter, with
a car battery or two, should have the ability to drive a 10 or 20 wind coil/core into saturation, and not burn up..

It would be nice to know what a single coil is using for volts and amps, 'in the motor', at 500 watts from the battery at
full throttle. Just to know about what range of volt-amps to be working with.

 

[APL]

Heres an entertainment break. You've probably seen it before, but it's always a kick to see how people struggled so,
with these things, way back when,.. just like were doing now!

History of the electric motor (early days); https://www.eti.kit.edu/english/1376.php

 

[fechter]

Here's a motor you can make with just a magnet, battery and a drywall screw:
https://www.evilmadscientist.com/2006/how-to-make-the-simplest-electric-motor/comment-page-1/
Not suitable for bikes, but a fun demonstration.

Don't get hung up too much on the saturation thing. If you have decent sized cores it won't be an issue until you run extreme amounts of power. Just being able to make a core with a tooth that overhangs on all sides will be pretty special.

One thing I didn't see (or notice) was if there was a thermal conductivity spec for the #40 stuff. I imagine it's not great, but I'd be interested to compare it to silicon steel or other materials like copper.

 

[KD5ZXG]

I grossly misunderstand how ferrite works, and could easily be in complete error: Trouble with saturating ferrite is that it may become either hard permanent or lossy supressor ferrite. Another reason to be cautious of it in a high torque motor.

Only manganese oxide (in the presence of randomly oriented iron oxide) is the soft magnet. iron oxide is a hard flippin' high loss magnet in parallel. Manganese is all done well below threshold of the iron oxide, so hard iron need never become involved. But keep pushing, and chance creating permanent magnetic domains.

Ordered domains won't help manganese act low loss. Manganese isn't even magnetic by itself. Wants disorganized iron nearby to gain a clue. "Demagnetizing" rituals can rebalance domains, but can't restore small scale randomness that really mattered. Can't anneal ferrite back to Curie temp after its been assembled, epoxied, or got copper wound on it.

If we are talking back-ferrite without eddies, permanence isn't a problem, might even be a feature. There we need no manganese.

Anyone know why ferrite also contains zinc oxide? Not a trick question. I really don't know what the zinc does, and its always bugged me. Mechanical? Thermal? Magnetostrictive? What gives here???

 

[fechter]

The #40 stuff the brick is made from is not what I would call ferrite. It is powdered iron, a bunch of ground up silicon steel with a super thin coating of phosphate to insulate the particles, and stuck together with something like epoxy. The datasheet shows it is not quite as good as typical silicon steel laminations but you can make up for this by the ability to fabricate more optimal pole shapes.

There are dozens of types of ferrite. Most are made for high frequency radio stuff. Magnetically hard ferrites are used to make permanent magnets.

 

[damirsky]

The #40 stuff the brick is made from is not what I would call ferrite. It is powdered iron, a bunch of ground up silicon steel with a super thin coating of phosphate to insulate the particles, and stuck together with something like epoxy. The datasheet shows it is not quite as good as typical silicon steel laminations but you can make up for this by the ability to fabricate more optimal pole shapes.

it could be not even a steel but a pure iron, i guess..

 

[APL]

Love the screw and magnet video, someday I still want to try a large homopolar disc motor, or several disc's,..so simple!
It would probably run on about 3.7v, so all the batteries would be in parallel, and would need a heavy duty controller.
Welder type stuff.

I went back to the web site to see if there was any specs on chemical make up, but of course it's proprietary. But I did
see some temperature specs, on aging. I also see that they have some 'high temp' offerings, that specify silicon steel
content, with Bsat's still in the range. Nothing specifically about thermal conductivity though.

Micrometals; https://www.micrometals.com/temperature-considerations

Micrometals;https://www.micrometals.com/materials/200c

We might be taking it for granted that this 'is' Micrometals #40 and not something else, as I didn't see any markings on the
core, or print on the wrapper, nor did I see it in the company product line up,.. but I'm sure they do a lot of custom
products not advertised, and have been doing so for a long time. Anyway, I'm sure it will be fine,.. going to use it anyway.
I haven't noticed any rust yet, but I'm sure thats coming up next, so I better put some grease, or primer on it.

The main reason I'm so hot on saturation, is so I can make the motor as light as possible, steel is the heaviest part, and I'd
like to use the least amount possible. Plus I'd like to find a simple way to find out these parameters at home, with simple
equipment. But otherwise your right, ball park sizes will work. The cores in the motor I have now are probably twice as
much steel as needed, and it works.