DIY Toroidal Axial Flux PM

Electric Motors and Controllers
Post Reply
User avatar
HalbachHero   100 W

100 W
Posts: 209
Joined: Apr 05 2021 8:35pm

DIY Toroidal Axial Flux PM

Post by HalbachHero » Apr 13 2021 11:45am

Hello ES community,
I have lurked this site here and there, but really got hooked reading APL's DIY axial flux bike motor build, and it inspired me to make my own post.
viewtopic.php?f=30&t=97860

It's a hell of a thread and I learned a lot and about motors and the community while reading through it all. It's clear that there are people out there in this world with a similar passion to mine, and luckily it sounds like you guys are all way smarter than I. So I welcome the discussion and ideas. While I proceed down a similar journey to APL's

Let me start with some background. I am a software engineer, with background in computer technology. Physics and math were always an interest of mine, but I was not proficient in either. I find that I learn best by doing and getting my hands dirty. I also am fascinated by magnets and their potential application throughout advanced technologies. Another interest of mine is renewable energy, which is where the fascination with the motors piqued.

I love the idea of combining renewable technologies, and in a northern US climate, solar is only so viable. have read a bunch of articles and papers on turbines, and the axial flux diy-er community. Which is what led me to try to make a light weight generator that could be cheaply manufactured, and the neat thing about these generators, is that they are motors too! Therefore I decided a side project of mine would to to electrify my Honda PCX 125.

Now that seems like a lofty goal. It's a 8.5KW ICE engine. Obviously it wouldn't need that kind of power if it was fully electric, and I will need to start smaller, but I felt that it was attainable with enough effort. Especially if a hub motor was a reality, and I could put one on each wheel.

I have been working at this project for a bit over a year, with slow-ish progress with life getting in the way through out. When I had started, I was living in an apartment with very little space, and I didn't have any real tools. I had comprised a very rudimentary rig using K'nex, superglue, and some time. After a while I felt like I had a decent grasp on how this whole induction thing worked, so I bought a 3d printer and started learning a 3d modeling. My thought was that it was the most versatile tool and great for rapid prototyping. I made a few coil cores and toroidal stators, but it wasn't long and my wife didn't like all the room I was taking up soooo, I bought a house.

I would like to dump some photos with out making this initial post too long, but I have so much I would like to talk about from litz wire winding machine that I created to the benefits of FOC type ESCs and how they may work with a coreless toroidal form factor.

Anyway, look forward to more posts!

My YouTube Channel
https://www.youtube.com/channel/UCQk0Ck ... QvpGxj4kxw

the Genesis a CD with a bunch of magnets superglued to it. Super safe!
resized-image-Promo (1).jpeg
resized-image-Promo (1).jpeg (69.2 KiB) Viewed 5439 times

Made it a bit nicer with some aluminum sheets and amore coils. Also a 3d printed rotor that a buddy of mine made before I had my printer.
resized-image-Promo (3).jpeg
resized-image-Promo (3).jpeg (79.07 KiB) Viewed 5439 times
resized-image-Promo (4).jpeg
resized-image-Promo (4).jpeg (79.47 KiB) Viewed 5439 times
resized-image-Promo (2).jpeg
resized-image-Promo (2).jpeg (85.64 KiB) Viewed 5439 times
That's when it got a bit more real:
resized-image-Promo (8).png
resized-image-Promo (8).png (133.68 KiB) Viewed 5415 times
Until I felt good about it:
resized-image-Promo (5).jpeg
resized-image-Promo (5).jpeg (53.54 KiB) Viewed 5439 times
But that was really just a heater. So I tried again. this one spun! I was on to something.
resized-image-Promo (6).jpeg
resized-image-Promo (6).jpeg (41.4 KiB) Viewed 5439 times
resized-image-Promo (7).jpeg
resized-image-Promo (7).jpeg (33.47 KiB) Viewed 5439 times
so I took my learnings and made a bigger "better" one.
resized-image-Promo (9).jpeg
resized-image-Promo (9).jpeg (82.56 KiB) Viewed 5439 times
resized-image-Promo (10).jpeg
resized-image-Promo (10).jpeg (91.72 KiB) Viewed 5439 times
And then it blew up. Wish I recorded that it was pretty wild. I never found some of the chunks of the rotor. They seem to have disintegrated. this I believe was due to only one rotor half being on. (for a test), but what I didn't think about is that the KV was at least double that of a dual rotor, also the two halves helped support each other, but with one on it is much more susceptible to vibration. it was a bad idea, lets just leave it at that.
resized-image-Promo (11).jpeg
resized-image-Promo (11).jpeg (69 KiB) Viewed 5439 times
I then made a nicer version of that same one that I don't seem to have photos of, but I have a video of it here (details in description of video):


And since that one worked so nicely. I decided to give it more turns and see if I could drop the KV.
resized-image-Promo (12).jpeg
resized-image-Promo (12).jpeg (65.44 KiB) Viewed 5439 times
resized-image-Promo (13).jpeg
resized-image-Promo (13).jpeg (72.9 KiB) Viewed 5439 times
resized-image-Promo (14).jpeg
resized-image-Promo (14).jpeg (65.21 KiB) Viewed 5439 times
resized-image-Promo (16).jpeg
resized-image-Promo (16).jpeg (57.78 KiB) Viewed 5439 times
resized-image-Promo (17).jpeg
resized-image-Promo (17).jpeg (33.47 KiB) Viewed 5439 times
To be fair, that last photo shows layer issues with the print. I was experiencing some odd material issues with the printer. It took a bit to get to the bottom of it, and then I broke my stator due to tolerances changing and that's where were at now. Still have not had the chance to run the newest version at high RPM yet. Also, I have made modifications to the stator/hub so that it will no longer be two pieces epoxied together, but something that can be disassembled. Also I beefed up the supports so that hopefully they don't break as easy, but in the end it's only plastic, so it will only take so much abuse.

I would love to print these out of metal in the future, but I need to prove the design first. Also, there may be an issue with swapping an air core for a ferrous core and expecting only improvements.

I welcome your comments and suggestions. I will post more here soon going into more detail on my manufacturing process. Hope you enjoy the pics!
Last edited by HalbachHero on Apr 13 2021 1:25pm, edited 1 time in total.

mxlemming   10 kW

10 kW
Posts: 868
Joined: Jul 17 2020 7:56am

Re: DIY Toroidal Axial Flux PM

Post by mxlemming » Apr 13 2021 12:11pm

This is brilliant. I love it.

How much scope is there for flattening the stator while keeping the same amount of copper? If you can get the magnets close enough to have the flux linking between them i think the mag field gets stronger and better aligned.

Wondering how much torque you can get out of it... Whether it's mountable on an ebikes.

If you've read APL's thread, try Lebowski axial flux, he did coreless as well, and it appeared it worked fairly well.

I love that this packages so neatly.

The obvious hassle of adding iron to the core is that suddenly your magnets want to stick to the core and everything has to be much much stronger to resist the magnets pull.
Hold my beer while I divide by zero :flame:

User avatar
HalbachHero   100 W

100 W
Posts: 209
Joined: Apr 05 2021 8:35pm

Re: DIY Toroidal Axial Flux PM

Post by HalbachHero » Apr 13 2021 12:29pm

Thank you for your interest. I think it packages nicely as well, and it's fairly simple to print 5 pieces in total. The version in the video is pretty thin as there are only 4 turns per slot, but the more turns the thicker it needs to be due to the wires crossing over, therefore the newer version is about 25mm thick. The difference in the pull of the magnets from rotor to rotor is significant. I can tell that its much weaker now that its further apart.

This is the conundrum I have been facing, and seems like the tradeoff, either get stronger, heavier, bigger magnets or make the stator thinner. I cannot find a clever way of adding turns but keeping it thin without adding significantly to end turns and ultimately the diameter of everything.

I worry that adding iron to the core has a greater impact than cogging and tolerance issues though. I fear that the form factor will not work the same as currently induction is direct to the coil where as with iron, my understanding is that its to guide the flux into a path where it can be optimally inducted. My concern given the stator shape is that the flux would be pulled into multiple phases at once.

Thoughts?

User avatar
APL   100 kW

100 kW
Posts: 1063
Joined: Aug 06 2018 8:44pm

Re: DIY Toroidal Axial Flux PM

Post by APL » Apr 13 2021 1:23pm

Super awesome thread HalbachHero! My hat goes off to your tenacity and ingenuity,.. which is what it takes to do this
DIY stuff. Lots of sleepless nights, he, he... :wink:

Now I can see what your up to, great pics, and I can see that the trail of tears has been long and hard. I love the 3D
printing aspect of you're build, and we haven't really explored that too much here on the ES. I can see a lot of potential
there, and it can only get better with the advent of newer materials coming all the time. :thumb:

From what I've seen so far on core-less axial's, is the gaps between magnets seem to hoover around 13mm or so,
(memory), but it's dependent on strength like you say, and adding steel will only get us more axial problems, with the
'pay off' ratio still being reletivly unknown. Ive always wondered about the hybrid aspect, half coreless, half cored.

A thinner stator comes down to materials strength, and bigger magnets add weight and expense which are unnecessary if
the stator is thinner. Lots of conundrums with motors,.. bigger diameter amplifies the torque, but also amplifies the
strength problems.

Well, thats what were here for, to thread the needle and find a solution. I see that your using lots of Litz type wire in your
experiments, which is very nice stuff, but have you thought about single strand as a way to maybe get a thinner stator?
How much current/voltage are you using these days?

User avatar
HalbachHero   100 W

100 W
Posts: 209
Joined: Apr 05 2021 8:35pm

Re: DIY Toroidal Axial Flux PM

Post by HalbachHero » Apr 13 2021 1:53pm

Thanks APL,

The rotors I have constructed are using halbach arrays.
resized-image-Promo (15).jpeg
resized-image-Promo (15).jpeg (63.6 KiB) Viewed 5411 times
In that photo you can see that there are thin magnets placed on their side to prevent some flux leakage, while minimizing extra weight. However, their is still a decent amount of flux leaking through the back of the rotor. I am considering including a iron backing with a halbach. though that might not be the best power to weight choice.

Weirder is that the axial facing magnets are N52 1/2" x 1/4" x 1/8" and the ones on their side are N50 1/2" x 1/8" 1/16". which probably has something to do with flux leakage.

The litz wire is homebrew. I have created a winding machine with a combination of K'nex and 3D printed parts. Here's a video of it in action.


I use 20 strands of 32 gauge wire. I recently bought a solder pot. Absolute must buy for anyone using litz wire, but don't buy to 200C stuff (Polyamideimide). Its ruthless to get the enamel off.

I buy all my Magnet wire from Remington.
https://www.remingtonindustries.com/magnet-wire/

I bought a 5lb spool for $75 US or so. 25,000ft, gives me a bit over 1,000ft of litz. @ 20 strands. and I use less than 1m of it to make a phase, so 9ft per motor. ~100 motors given some waste. That's a lot cheaper than what I was finding. But with my luck Remington started selling litz so, now I make it for the fun.

I am using some Lipo batteries I have. I am still keeping voltage at 16V until I can drive the Kv down, otherwise, I worry about a rapid unscheduled disassembly of the rotors. I have tried to keep it around 4k RPM to limit stresses. Not sure why that number, just felt right.

Current peaked at 32A, and I don't understand how it can go higher. I simply can't comprehend more current going through those thin wires. But I have read on here and other motor spec sheets hundreds of amps being possible.

I based the choice of wire on an ampacity chart that I derived from looking at cross-sectional area of wire and their ampacity by gauge. I thought with 20 strands of 32AWG I could handle 10A, and bursts of more, but it seems that its still a bit of a mystery to me.
Screenshot 2021-04-13 145130.png
Screenshot 2021-04-13 145130.png (34.5 KiB) Viewed 5411 times

mxlemming   10 kW

10 kW
Posts: 868
Joined: Jul 17 2020 7:56am

Re: DIY Toroidal Axial Flux PM

Post by mxlemming » Apr 13 2021 4:06pm

To have the halbach array work properly your would probably need shaped magnets. The gaps will leak a lot of flux. See the FEA done in this thread: viewtopic.php?f=30&t=46476&hilit=Axial+flux

Wire can take a lot more current than those charts say. That's based on very conservative values for wiring up machines etc where voltage drops are important, for your motor the heating effect is the only real limit so you can probably pass far more current.

I'd just flatten the ends of as much as possible. I'm not even sure half the plastic is needed in your last stator.
Last edited by mxlemming on Apr 15 2021 5:38pm, edited 1 time in total.
Hold my beer while I divide by zero :flame:

mxlemming   10 kW

10 kW
Posts: 868
Joined: Jul 17 2020 7:56am

Re: DIY Toroidal Axial Flux PM

Post by mxlemming » Apr 13 2021 4:09pm

Possibly your halbach magnets should be much wider, try to keep the gap between the rotor less than the gaps between the magnets?

Love the knex winding machine.
Hold my beer while I divide by zero :flame:

User avatar
HalbachHero   100 W

100 W
Posts: 209
Joined: Apr 05 2021 8:35pm

Re: DIY Toroidal Axial Flux PM

Post by HalbachHero » Apr 13 2021 4:26pm

I want to also go into a little more detail of what I have tried so far. Initially I was using PLA just to really simply test things. Not only was I new to making motors, but also 3d printing. So I had no real idea of what tolerances were possible especially with the printer I bought, a hobbyist level printer. I was also new to 3D modeling, so the ramp up process was slow as there was just a lot to take in.
Anyway, eventually I made some of the stators that were in the previous photos, I tried using angled stator legs and straight, and I tried angling the magnets and straight magnets. While angling the magnets I think makes for complicated flux paths, it also makes it so you cant fit as many magnets and not have overlap as they pass over each phase. Therefore, I just went with straight.

This worked well, so I experimented with iron backing. I understood enough to know that I was going to direct the flux but I would have to give up structural integrity, and would produce some eddy currents. This actually worked quite well in my "Mk1" motor. it had horribly tolerances, but I just went for it anyway because I was excited, and it got hot enough that it fused itself together because the plastic melted.

So I sort of abandoned the iron backing and tried the halbach array and immediately had success, and could feel the difference on the sides of the rotors, so I knew it was doing something, and while I never calculated the weight savings (other parameters changed since then too), I imagine it was reasonable given the rest of it was mostly plastic and copper.

Once I had the pretty much the tightest possible arrangement of magnets given the 12 pole pair count I wanted (24 poles, 48 individual magnets due to halbach), I only made some changes to support a ring in the middle instead of standoffs attached to one rotor and a groove for the ring to sit in on each rotor half. This helps seat the ring. There are still 2mm standoffs at 5 points where its "keyed" just to have a solid way to transfer rotational energy without shifting.

Once I was happy with the rotors, I moved to the stator and tried to optimize that. The format I went with was 3 Phase 12 "Teeth"
per phase so 36 in total. I practiced winding on stators with small amounts of purchased litz wire since it was expensive. and found that the amount of twist in the wire is important to prevent it from getting loose as you go. This proved to be exhausting on the hands to do more than one phase at a time (at first it took like 20 minutes per phase). Holding the wire tightly and having to twist it over and over sucked.

Thus spawned the litz wire machine. I figured I could just spin a bunch of spools. of course, it was a bit more complex. and I ended up with a few iterations of that before I have what I have now, but now it nicely twists 20 strands into a small 3d printed spool. Which is small enough to fit through the center of the stator, and has enough wire for at least 1 phase. ( you can fit 2 but its tight, and theres no real point that I can see). Great, now I am drowning in litz

I tried twisting strands of litz together, 2 and 4 strands of 20/32 on different attempts, The downside to this is having to cut the litz to length, twist it up then solder the ends together. That proved to be a pain in the butt (before I had my solder pot). After that I tried just doing laps. I would do one phase at a time. This is what I have found to be the most pain free. The Mk1 had two strands of litz wired simultaneously. (Ill add it goes on neater this way, but the soldering after it messier, and a pain (without that solder pot, see what im getting at?)). The Mk2 had 4 laps of a single litz wire strand. and so did Mk3.

Mk2 and 3 were great super thin, the whole motor is like 13mm thick, and weighs 300g or something close. Mk2 exploded because im dumb and spun it with one rotor half with the same voltage, and it didn't last long. Like only a few couple seconds long. Mk3 was a replica, and survived a few stress tests. in the "hardest" stress test I ran it at 100W load for 10 minutes continuously and it survived, the temperature on the coils was 38C after. Albeit, it has a propeller on it that may have cooled it a bit, but eventually on another test, it got too warm, and a magnet was pulled out of place, and the stator started to drag on it.

And that brings us to now. I then decided to go bigger with the stator. I wanted 10 laps, or turns. so I made it a bit thicker to 15mm, after a many small tweaks and reprints I got something I liked. So I wound it up and put it on the rotor halves. but there was an issue. I didn't compensate for the height change in the stator in regard to the hub. Damn... its already epoxied together. No worries, I modified the rotor design and added a stand off on the shaft collar. for the stator to sit on. This was actually a blessing in disguise, because it also solved another issue, which was a very weak shaft collar on the rotor, due to plastic and what-not. I added some braces and it seems to have helped. Also fixing an issue with the printer resolved some layer adhesion issues I was facing.
resized-image-Promo (19).jpeg
resized-image-Promo (19).jpeg (42.47 KiB) Viewed 5394 times
Great now it fit, but theres another issue with the printer. Because it was printing like crap, I messed up all the settings, so when I fixed it (New bowden tube, new hot end, new nozzle, replaced corroded wiring, etc). I was printing with different tolerances. I was over extruding. So nothing fit. Well I'm impatient so I thought Ill just smack it with this hammer till it goes in. Its tight but cant be that tight (I've had to hammer in the stator some iterations). Anyway, it was too tight, and everything broke. So. I needed to beef up the stator now.

The stator was held to the hub by 12 little 2mm nubs. this was working (sorta) for a while. But they did break or print badly quite a few times. So they had to go. I ended up going with a new design that I have yet to play with much, but it should receive up to 12 M2 screws. which will go into threaded heat insets in the stator hub. this way I can replace those two parts independently, and its a bit thicker, so hopefully it can take a bit more abuse.
resized-image-Promo (18).jpeg
resized-image-Promo (18).jpeg (40.56 KiB) Viewed 5394 times
I think I need to make more litz wire, but I have enough for at least 2 phases, so I can see how winding goes with the new form factor. But I'm also waiting on some M2 screws and heat set inserts to show up in the mail, so that I can test fit the stator to the new hub design.

Let me know what you all think!

mxlemming   10 kW

10 kW
Posts: 868
Joined: Jul 17 2020 7:56am

Re: DIY Toroidal Axial Flux PM

Post by mxlemming » Apr 13 2021 5:36pm

I would
1) consider doing an inrunner, then the screws could be on the outside where side is less constrained
2) make the wire guides as thin as possible, then you can make them lower for the same wire fill.
3) if the ends start to get large, you can always cut a channel in the rotor?

What are you modeling in? CAD wise? Your shapes are moderately advanced for any of the open source CAD packages.
Hold my beer while I divide by zero :flame:

Thecoco974   10 W

10 W
Posts: 86
Joined: Dec 07 2016 7:01am

Re: DIY Toroidal Axial Flux PM

Post by Thecoco974 » Apr 13 2021 5:59pm

Your corelless statosr are looking really good ! :shock:
Happy to see someone else inspired by Ben katz axial flux build, I've been following APL's build also for quite some time and find it really inpiring.
Since you presented yourself I might do the same : I did mechanical engineering studies and started working as an electrotechnical engineer 5 month ago (I was just a lot in electromechanical stuff at the time last year reading all the posts in the Motor Technology section).

Since I'm also working on coreless axial design I will be happy to share my understanding and advancement on your thread if possible or start my own thread if needed too.

I've build a really small test motor some month ago to validate my findings, here are some pictures of it :
mini-axial2.jpg
mini-axial2.jpg (174.42 KiB) Viewed 5386 times
mini-axial.jpg
mini-axial.jpg (188.12 KiB) Viewed 5386 times

The esc add trouble starting it from stand still because of weak BEMF I suppose (the air gap had to be widen because of rubing problems) but once at speed it work pretty smoothly (no load testing at least).

This build was exactly a 1:5 scale of the one I was designing at the time.

Here is the 3D printed winding support i'm planing to use for the full scale build :
stator_winding_guide.jpg
stator_winding_guide.jpg (167.55 KiB) Viewed 5386 times
Sorry for showing my stuff and not adding to the discussion..

I've been going a different route than yours for the rotor and not planning of using an hallback array for two reason :
- The need of stronger and expensive magnets for counteracting the flux leakage and still have decent airgap flux
- And since I'm not building an airplane motor the heavy back-iron is a non issue for me.

Also if your goal is a lower Kv, keeping the stator thickness under control is key. I've found a good video series on youtube from a guy building an axialflux generator explaining just how you can find an optimum with FEM analysis and a Figure Of Merit calculus pointing to it (here is the link : https://www.youtube.com/watch?v=53GIT3c0M3M). The flux in the airgap is decreasing pretty fast as you open it up and make for a waste of copper going past the optimal point.

I've done just that using my cheap N37 magnets and found for my particular application an 11mm gap as an optimum (need to find my spreadsheet back it's been a while). As mxlemming stated it as been found that going with an airgap about two times the thickness of the magnets is a good ballpark and I just find something close to that with my FOM calculus (i'm using 40x20x5mm magnets).

You're winding your stators with litz wire but as you said the crossover is making difficult to achieve a good copper fill in a thin stator so why not winding it straight ? It will be harder but permit a higher density overall and a thinner stator.

Another way of making it thinner is to squeeze and laminate it once the winding is done like Ben Katz did. But it won't be as good looking as the one you just build for sure.

I think a FEMM analysis of your hallback array will point you in the right direction for optimizing it.

Edit (just saw your explanation post :oops:) :
Why do you think rotor back iron add eddy current ? To my understanding eddy current are created when a conductive material is exposed to variating magnetic flux, in the case of the rotor back iron magnet flux is fixed and coil field is also fixed in the rotor frame, am i missing something ?

User avatar
HalbachHero   100 W

100 W
Posts: 209
Joined: Apr 05 2021 8:35pm

Re: DIY Toroidal Axial Flux PM

Post by HalbachHero » Apr 13 2021 6:35pm

mxlemming,

Great thoughts. Here are just some counter thoughts as these have crossed my mind already.
1) consider doing an inrunner, then the screws could be on the outside where side is less constrained
I could do an in runner. However, I had thought that I could swap the ring out for a something that included mounting to have a direct drive attached directly to the motor. But the in runner concept is looking more and more attractive as I deal with very small parts on the inner diameter.
2) make the wire guides as thin as possible, then you can make them lower for the same wire fill.
currently the guides coincide with the magnet width quite nicely, and as soon as I make the grooves flat, the subsequent laps of wire are difficult to hold into position. With the current architecture, I can pull the wire as hard as I want and it just tightens up, I haven't broken a stator once that way yet, which is a nice feature for getting things nice and tight.
3) if the ends start to get large, you can always cut a channel in the rotor?
This I had not thought of, but I'm not sure of it's feasibility, as the overhang is already quite flexible on one rotor, and making that any bigger is asking for clipping under load. I am dealing with millimeter tolerances pretty much everywhere at this point.



Thecoco974,

Thanks for sharing. It's awesome to know there are others like myself out in the world, and I'm certainly getting some Ben Katz vibes from your motor there. Looks very nice, and seems to spin up and down nicely. Looks like a pretty high Kv assuming that was an RPM readout in your video. What is the diameter of your full size version. It looks very close to mine (120mm). Do you know what you are designing for yet? I'd love to see where it ends up. I would encourage you to post about your endeavor. It's fun.

Also, that video you shared. I'm the third comment from the top haha. I watched that video several times, and modeled my motor quite closely after his.

Can I also assume the airgap you mention is rotor to rotor, and not rotor to stator?

Also can you elaborate on what you mean by winding it straight. I am using litz specifically for its benefits in reducing eddy currents at high frequencies, and with a 36 slot stator, I thought that was enough to merit the use of litz. but also there is the added benefit of being pretty easy to work with.

Thanks for the thoughts!

User avatar
HalbachHero   100 W

100 W
Posts: 209
Joined: Apr 05 2021 8:35pm

Re: DIY Toroidal Axial Flux PM

Post by HalbachHero » Apr 13 2021 6:36pm

Thecoco974,
Why do you think rotor back iron add eddy current ? To my understanding eddy current are created when a conductive material is exposed to variating magnetic flux, in the case of the rotor back iron magnet flux is fixed and coil field is also fixed in the rotor frame, am i missing something ?
Perhaps its minor, and/or a complete misunderstanding but does back EMF not find its way into the back iron?

User avatar
HalbachHero   100 W

100 W
Posts: 209
Joined: Apr 05 2021 8:35pm

Re: DIY Toroidal Axial Flux PM

Post by HalbachHero » Apr 13 2021 6:43pm

Also mxlemming,
What are you modeling in? CAD wise? Your shapes are moderately advanced for any of the open source CAD packages
I am using Fusion 360

Thecoco974   10 W

10 W
Posts: 86
Joined: Dec 07 2016 7:01am

Re: DIY Toroidal Axial Flux PM

Post by Thecoco974 » Apr 14 2021 7:26am

HalbachHero wrote:
Apr 13 2021 6:35pm
currently the guides coincide with the magnet width quite nicely, and as soon as I make the grooves flat, the subsequent laps of wire are difficult to hold into position. With the current architecture, I can pull the wire as hard as I want and it just tightens up, I haven't broken a stator once that way yet, which is a nice feature for getting things nice and tight.
I see why you're doing that, winding is quite a pain, but i agree with mxlemming on this, spreading the winding over the full electric revolution decrease KV because all turns will not contibute the same (I think thay might contribute sinusoidally) but the flux increase will more than make up for. Should be interesting to model it.
HalbachHero wrote:
Apr 13 2021 6:35pm
Looks like a pretty high Kv assuming that was an RPM readout in your video. What is the diameter of your full size version. It looks very close to mine (120mm). Do you know what you are designing for yet? I'd love to see where it ends up. I would encourage you to post about your endeavor. It's fun.
Yes it is rpm. The KV is really high because of the thin and weak magnets and the increase airgap I got. Full scale stator to magnet clearance was supposed to be 1mm, scaling 4 times down mean 0.25 mm, not really dooable with my craftings skills. Also the full scale one is design to spin fast at 8-9Krpm at 55V, scalling down the diameter means also dividing by 4 the BEMF (linear with magnet speed) and multiplying by 4 the KV also.
I still need to check if my sims results where correct though ...
The full scale version have changed since then, I'm now looking for a three stator inrunner with a 167.5mm magnet end to magnet end rotor diameter (so since going inrunner the motor diameter will be 195mm). I'm really ambitious about it but i'm looking for 5kw continuous and 10kW for 1min. I might start a thread about it in the next few days.
HalbachHero wrote:
Apr 13 2021 6:35pm
Also, that video you shared. I'm the third comment from the top haha. I watched that video several times, and modeled my motor quite closely after his.
Just saw your comment :) I too watch this video (the entire series actually quite e few time :lol: )
So have you done some FEM simulation of you're motor design ? I'm interested by the flux in the airgap you're achieving with the hallback array.
HalbachHero wrote:
Apr 13 2021 6:35pm
Can I also assume the airgap you mention is rotor to rotor, and not rotor to stator?
Yes of course rotor to rotor ! being aircored everything betwen magnets is airgap for me.
HalbachHero wrote:
Apr 13 2021 6:35pm
Also can you elaborate on what you mean by winding it straight. I am using litz specifically for its benefits in reducing eddy currents at high frequencies, and with a 36 slot stator, I thought that was enough to merit the use of litz. but also there is the added benefit of being pretty easy to work with.
Winding the litzwire without twisting it first, I know it's not litwire anymore but for our use case (reducing eddy current on the winding) the twisting isn't needed. winding it straight will also marginally reduce the turn length and resistance. I think we only need the multiple strand enamebled small diameter copper wire part of the litz wire, not the fancy twisting (for my build i'm planning 700 strands of 0.1mm parralleled per turn).

User avatar
HalbachHero   100 W

100 W
Posts: 209
Joined: Apr 05 2021 8:35pm

Re: DIY Toroidal Axial Flux PM

Post by HalbachHero » Apr 14 2021 7:48am

I'm really ambitious about it but i'm looking for 5kw continuous and 10kW for 1min
That's the beauty of these axial flux motors, lots of power in a small package. I'd love to see how it turns out.
So have you done some FEM simulation of you're motor design ? I'm interested by the flux in the airgap you're achieving with the hallback array.
No I haven't I haven't spent the time to learn that yet. Though I feel it could be really helpful with understanding the complex flux lines, it's something I will have to put some time into. But, likewise. I am very curious what this sort of halbach array looks like in FEMM



The twisted litz wire is certainly easier to manage and turn, but I get what you two are saying now. I will already have some twisted litz, but I have just about run out. For the next spools I wind, I'll not twist it up.

I'll make two stators and compare how many turns I can fit on the two given the same stator and assuming I can fit more turns. I'll modify the stator design to be much thinner with thinner wire guides.

I suspect I will need to find a creative way to hold the wires flat in their place while I wind each phase. Maybe I can make a jig.

User avatar
HalbachHero   100 W

100 W
Posts: 209
Joined: Apr 05 2021 8:35pm

Re: DIY Toroidal Axial Flux PM

Post by HalbachHero » Apr 14 2021 10:55am

Okay, well I think you guys were right. I decided to just try the untwisted litz just to see if I could fit more. I have to say its a nicer looking coil in the end. Much neater, and thin, and clean on the outer diameter of everything. But it was a pain in the butt to wind. I'm going to need to make a jig to help with that.

But dang doesn't that look sexy?
resized-image-Promo (20).jpeg
resized-image-Promo (20).jpeg (47.33 KiB) Viewed 5342 times
resized-image-Promo (21).jpeg
resized-image-Promo (21).jpeg (68.28 KiB) Viewed 5342 times
The issues I see now are, since I beefed up the supports in the middle they kind of get in the way of winding the phase that touches it. Also, keeping the wire in its slot while winding is a bit more of a challenge. Thinking of making a modification to the stator design to facilitate that. But also the wire is going to overlap the other phases. I make the whole stator shorter, that will be less of an issue but still a bit of a problem. Might need to find a way to wind all 3 phases at once....

The photos make it clear too that there is more room between those teeth. I stopped at 10 turns which is what my last one was with the twisted litz. I think I can fit a few more. I will keep playing with that and see what's possible.

shirk   10 mW

10 mW
Posts: 25
Joined: Oct 11 2013 4:26pm

Re: DIY Toroidal Axial Flux PM

Post by shirk » Apr 14 2021 12:03pm

Thanks for sharing your builds here.

I have been looking at axial style motors for a ebike build that's churring away in my head and this is all amazing info.

A questions regarding size/weight to power. The motor I want needs to put out about 300w nominal and I'd like to pair it with a cycloidal drive for it's reduction. Battery would be 52V. Do you a rough idea on the size / weight of this style axial motor that would hit this 300w 5k RPM 52V spec?

User avatar
HalbachHero   100 W

100 W
Posts: 209
Joined: Apr 05 2021 8:35pm

Re: DIY Toroidal Axial Flux PM

Post by HalbachHero » Apr 14 2021 12:41pm

Shirk,

It's difficult to say without some real thought. But a 52V @ 5k RPM will give you a Kv of ~96. 300W @ 52V is only 6A without losses. This means you could get away with pretty small conductors, and likely small magnets in a pretty small form factor. But the power to weight ratio is really achieved through very small tolerances and minimization of losses. The ability to fabricate one is also really important. You can only make things so small, and so close together. Another important thing to consider is materials. Composites and some metals like titanium are great for strength to weight, but are difficult to source or just expensive. Cheaper materials come at the price of weight and size. Everything is a balance.

Also, specifically axial flux motors benefit from having a large diameter, as they can produce exponentially more torque the further away from the center of the hub. This can mean big power. 300W is not a LOT of power, and there are many very lightweight brushless outrunners that can achieve 300W and are very cheap and widely available. I feel that axial flux really shows its worth in high power applications.

Now if you're talking about a 3Kw motor, that may be a better application for a geared down axial flux hub motor
Last edited by HalbachHero on Apr 15 2021 6:56am, edited 2 times in total.

User avatar
HalbachHero   100 W

100 W
Posts: 209
Joined: Apr 05 2021 8:35pm

Re: DIY Toroidal Axial Flux PM

Post by HalbachHero » Apr 14 2021 2:37pm

I was able to get 12 wraps around the same stator slot without it becoming a huge pain. I worry what its going to take to wind the next version as things get thinner and more fragile. I have started redesigning the stator. The connections on the inner diameter are no longer sufficient, There is now very little material to grab on to, and the copper will want to take that up. Hmmm, looks like an inrunner going forward. :? I really liked the outrunner, it was clean, but so be it if it works better.

I squished everything to 9mm and changed the shape of the slot to better facilitate winding. but I will likely need to make that a bit thicker. I have leaned on the suggestion of 11mm air gap assuming I will be able to have 1mm tolerances between the rotor and stator.
Screenshot 2021-04-14 153602.png
Screenshot 2021-04-14 153602.png (288.26 KiB) Viewed 5313 times
The cool thing is, at 9mm I am back to the thickness of my Mk2 which only had 4 turns. So thats cool (if this works).

I would love to bring the Kv down in a linear fashion given the number of turns, and I think that's only possible if I keep the thickness and airgap the same.

Any suggestions on how to mount this stator on the outside? I do worry the copper is going to be difficult to work around with it this thin.

Thecoco974   10 W

10 W
Posts: 86
Joined: Dec 07 2016 7:01am

Re: DIY Toroidal Axial Flux PM

Post by Thecoco974 » Apr 15 2021 2:25pm

HalbachHero wrote:
Apr 14 2021 7:48am
No I haven't I haven't spent the time to learn that yet. Though I feel it could be really helpful with understanding the complex flux lines, it's something I will have to put some time into. But, likewise. I am very curious what this sort of halbach array looks like in FEMM
I just did some quick sims of the arrangment you use layed flat as others did for axial FEM analysis :

Flux line results :
hallback1_line.PNG
hallback1_line.PNG (41.38 KiB) Viewed 5252 times
here is the flux at the midline :
hallback.PNG
hallback.PNG (11.08 KiB) Viewed 5252 times
I Thought there were too much leakage flux so i did the same without insert magnet :
same_without_hallback.PNG
same_without_hallback.PNG (31.96 KiB) Viewed 5252 times
same_without_hallback2.PNG
same_without_hallback2.PNG (11.37 KiB) Viewed 5252 times
and with back iron for base line evaluation of hallback efficiency at redirecting flux on one side :
same_without_hallback_with backiron.PNG
same_without_hallback_with backiron.PNG (11.93 KiB) Viewed 5252 times
same_without_hallback_with backiron2.PNG
same_without_hallback_with backiron2.PNG (27.55 KiB) Viewed 5252 times
So I think the insert magnet may be to thin and add litle to the flux (if my sims are correct). Plus side if I am you might decrease the KV working on that too.
HalbachHero wrote:
Apr 14 2021 10:55am
But dang doesn't that look sexy?
Holy shit youre fast working on this ! And it look really good for sure :shock:
HalbachHero wrote:
Apr 14 2021 10:55am
But also the wire is going to overlap the other phases. I make the whole stator shorter, that will be less of an issue but still a bit of a problem. Might need to find a way to wind all 3 phases at once....
If you're keeping the wire on its spool can you make one turn by one turn all phase at once like you said ? might be a pain and need a holder of some sort.
HalbachHero wrote:
Apr 14 2021 2:37pm
I was able to get 12 wraps around the same stator slot without it becoming a huge pain.
:thumb:
HalbachHero wrote:
Apr 14 2021 2:37pm
There is now very little material to grab on to, and the copper will want to take that up. Hmmm, looks like an inrunner going forward. I really liked the outrunner, it was clean, but so be it if it works better.
I d'ont know if you're reusing wire for each evolution, but the simpler thing to do (not the cleanest for sure) might be to epoxy a center mounting ring directly to the inner winding like i did in my small scale build.
HalbachHero wrote:
Apr 14 2021 2:37pm
I would love to bring the Kv down in a linear fashion given the number of turns, and I think that's only possible if I keep the thickness and airgap the same.
Yes it's the best way, I think you'll need to sacrifice copper section for KV.
HalbachHero wrote:
Apr 14 2021 2:37pm
Any suggestions on how to mount this stator on the outside? I do worry the copper is going to be difficult to work around with it this thin.
It's more dificult to deal with because of end turn (tha's another reason why i'm folowing Ben Katz on the no end turn winding), but if going outrunner you can make it wider on the outside without making the rotor weaker (since it's already floating) and it might give you more mounting options

User avatar
HalbachHero   100 W

100 W
Posts: 209
Joined: Apr 05 2021 8:35pm

Re: DIY Toroidal Axial Flux PM

Post by HalbachHero » Apr 15 2021 5:06pm

Thecoco974,

Woah! thank you so much for doing that. That is very helpful. Now I am very curious how close it would have to be to prevent flux leaking, and seeing the difference between the back iron and not is cool, though it complicates how to make this with a printer a bit. I also wonder if on the inner diameter since its basically touching, if flux isn't leaked. Also, what airgap did you use in your simulation? was that with the 11mm?

Wonder if I can jam more magnets in there if I get rid of the halbach ones.... food for thought

I might be able to turn all three phases at once, but I think its going to prove to be a challenge, without a good groove or twisted wires, its wants to go everywhere. I printed a test stator based on my model change and wound 2 laps. While the trough (that's what I'm gonna call the female side of the wire guide thing.) is nearly the same cross sectional area that the other stator, due to the larger width the wires have more of a chance to run at an angle, which takes up some precious extra space. Will keep going with this test to see how many laps I can fit.
resized-image-Promo (23).jpeg
resized-image-Promo (23).jpeg (277.05 KiB) Viewed 5231 times
Also I made a tool, which I think could be helpful.
resized-image-Promo (22).jpeg
resized-image-Promo (22).jpeg (48.86 KiB) Viewed 5231 times
resized-image-Promo (26).jpeg
resized-image-Promo (26).jpeg (59.23 KiB) Viewed 5231 times
resized-image-Promo (24).jpeg
resized-image-Promo (24).jpeg (38.28 KiB) Viewed 5231 times
resized-image-Promo (25).jpeg
resized-image-Promo (25).jpeg (55.39 KiB) Viewed 5231 times
It's more dificult to deal with because of end turn (tha's another reason why i'm folowing Ben Katz on the no end turn winding), but if going outrunner you can make it wider on the outside without making the rotor weaker (since it's already floating) and it might give you more mounting options
I have a concern with material strength if I make the stator a larger diameter. Since one of the rotors have no support on the inner diameter. The 6mm of plastic likes to flex already. My goal is to keep this as this as possible, so I don't really want to add any more overhang or thickness if I don't have to. If it makes more sense to go in runner. I will, but it won't look at good I don't think

mxlemming   10 kW

10 kW
Posts: 868
Joined: Jul 17 2020 7:56am

Re: DIY Toroidal Axial Flux PM

Post by mxlemming » Apr 15 2021 6:08pm

I think for this you should get higher power density with an inrunner, since the outer shell takes up a lot of volume, being on the outside and all... the middle is generally pretty empty so you have a lot of space for bearings, flanges and whatnot.

Getting tabs to add screws on the outside of the stator should be far far easier than the tabs you previously put in the middle, again because the circumference increases, so there is just generally more space on the outside.

You could use a solid steel disc for your magnet backing, just drill a single hole (exactly...) in the middle. or, if you are really cheaping it/bodging away, I just checked my bike disc brake rotors and found they are made from highly magnetic steel. That's a circle that can sustain a lot of force and RPM. :lol:

This is looking pretty cool! I can really get behind this, since the costs involved are super low, I have a 3d printer and piles of ESCs to energise it with :D I also have a solidworks licence, so I can model away.

Any idea what the BEMF profile of it looks like? You have an oscilloscope?
Thecoco974 wrote:
Apr 15 2021 2:25pm
I just did some quick sims of the arrangment you use layed flat as others did for axial FEM analysis :
same_without_hallback_with backiron2.PNG
What are you doing this FEA with? Is it some open source thing I can play with? Or is it some mega expensive thing you get through work/uni/...?
Hold my beer while I divide by zero :flame:

User avatar
APL   100 kW

100 kW
Posts: 1063
Joined: Aug 06 2018 8:44pm

Re: DIY Toroidal Axial Flux PM

Post by APL » Apr 15 2021 8:35pm

You guys must drink a lot of coffee,.. I have trouble just keeping up with this thread! :shock:

I guess I'm not really sure what your having trouble with, on the wire moving or not staying put, but it's common for
motors to use string to help hold coils together. Theres a fabric fishing line thats supper strong and mega-thin that
wouldn't take any space,.. might work well for that, and it super glues well. Just a thought.

User avatar
amberwolf   100 GW

100 GW
Posts: 33733
Joined: Aug 17 2009 6:43am
Location: Phoenix, AZ, USA, Earth, Sol, Local Bubble, Orion Arm, Milky Way, Local Group
Contact:

Re: DIY Toroidal Axial Flux PM

Post by amberwolf » Apr 15 2021 11:58pm

APL wrote:
Apr 15 2021 8:35pm
it's common for motors to use string to help hold coils together.
A helpful page on techniques:
http://www.aeroelectric.com/articles/ca ... _lace.html
If you found this advice helpful, supporting contributions are accepted here.

Beginning of a New Life


Give a listen to my music!

(More here, including experimental)
.

User avatar
HalbachHero   100 W

100 W
Posts: 209
Joined: Apr 05 2021 8:35pm

Re: DIY Toroidal Axial Flux PM

Post by HalbachHero » Apr 16 2021 8:07am

mxlemming,

Thanks!
I will keep playing with the stator a bit more to see what I can fit, but you're totally right bring the mass to the middle as much as possible. If you can move the mounting points to the middle as well, then you isolate the mass that is on the outside to only what needs to be there.

I am considering making the same style supports I had on the inner diameter since it allows for support on the top and bottom but still allowing for easy winding.

I bought some large diameter washers initially, which ended up in the Mk1, but were working with a larger diameter now. I worry about making my own, because they will have to be very well balanced. But will definitely give it a shot when I get there.

I believe strongly in renewable technology and would love for others to be able to use this as a generator in a wind turbine. I will definitely make the STLs available once the design is solid. (or now if someone wants them)

I do not have an oscilloscope. That would be really helpful to visualize the waveform.

FEMM is a free program for Windows. I think it runs in Wine too
https://www.femm.info/wiki/Download


I guess I'm not really sure what your having trouble with, on the wire moving or not staying put, but it's common for
motors to use string to help hold coils together. Theres a fabric fishing line thats supper strong and mega-thin that
wouldn't take any space,.. might work well for that, and it super glues well. Just a thought.
The issue with this is that I am doing laps around the whole toroid with a single phase. So I wouldn't be able to tie it until I was done all the laps. And it wants to fall apart in the process of winding it. I just have to pull everything super tight, and the thinner this gets the more difficult that becomes. However a change to the stator again may help allow for tying it at the end. But I may rather go with a high heat epoxy to keep everything solid at the end, which would likely make tying things unnecessary. We will have to find out what works best.


Amberwolf,
Thanks for sharing, that is a handy resource. Bookmarked!

Post Reply