Simple axial flux Torus motor

[hillma]

I'm thinking about building a simple axial flux non slotted torus machine (single interior stator with two external rotors) with the idea, if the experiment works, of moving onto a twin stator three rotor design to power my e-bike. I like many others are limited to basic tools but unlimited enthusiasm!

After being motivated by Lebowski's (and others) work on coreless designs I thought a simple torus might be a good way of generating heaps of torque and reducing the amount of reduction required to drive the rear wheel, it will be mid mounted btw.

Basic outline of the machine is:

Rotor
ID = 60mm
OD = 150mm
Thickness = 5mm
Material = Steel
Magnets per rotor = 12 x 30mmx20mmx5mm N50 (total of 24) Magnets arranged NN so that flux is driven circumventionally (is that even a real word?) around the stator to the adjacent magnet on the rotor rather than NS where the flux path passes through the stator to magnet on the opposite rotor.

Stator
ID = 84mm
OD = 150mm (ratio of ID/OD = 0.58)
Thickness = 10mm
Material = spiral wound electrical steel

Coils
Total of 36 surface wound coils, 12 per phase
Width = 7mm
Length per turn = 90mm ish (including end turns)
No of turns per coil = 10
Wire = 1.4 mm

I'm hoping to construct this with basic tools and would welcome any comments or observations before I start out. I haven't set myself any parameters for the performance yet as this is mostly a learning exercise at the moment. But if it could eventually manage 1kw and spin at sub 1000rpm efficiently I'll be a happy camper

My main concern at the moment is if the stator is thick enough to not saturate with all the flux being driven though it, (I'm physically limited to 10mm with an eye on the triple rotor design). If this is the case I may need to switch to the NS design where stator thickness is not really an issue but does impact the torque and efficiency.

 

[Miles]

circumventionally (is that even a real word?)...........My main concern at the moment is if the stator is thick enough to not saturate with all the flux being driven though it, (I'm physically limited to 10mm with an eye on the triple rotor design). If this is the case I may need to switch to the NS design where stator thickness is not really an issue but does impact the torque and efficiency.

Hi hillma,

"Circumferentially" is....

You can check the flux density in the stator core using FEMM.

You already have the magnets?

 

[hillma]

Hi Miles

Many thanks for your response - I'll have a go at modelling the stator in FEMM which I've just found, much to learn . When I have something that's representative of the stator I'll post up the results for any feedback.

I've not been able to find much reference on ES to the Torus topology - would you consider this to be a suitable approach to lower rpm higher torque motor (compared to a coreless AFPM) that would be easy to assemble for the DIYer?

I've not yet bought the magnets but based the design around some cheapies I found on ebay that matched the dimensions I was after.

Thanks again Miles

Cheers Mark

 

[Miles]

I've not been able to find much reference on ES to the Torus topology - would you consider this to be a suitable approach to lower rpm higher torque motor (compared to a coreless AFPM) that would be easy to assemble for the DIYer?

Anything is better than coreless, if torque density is your priority....

It's a promising direction. Of all the cored axial flux topologies, it's the simplest to construct. Looking forward to following your build.

In case anyone is wondering what we are talking about, I've attached a pdf that might help......

 

[hillma]

Hi Miles,

First attempt at modelling in FEMM. I've replaced N50 with N42 magnets which have kept levels below saturation...... just!.

I've not yet included the coils as part of the circuit and a guess this will have a impact on results, need to work on this later today.

Any comments and thoughts always appreciated.

Cheers

Mark

 

[Miles]

Hi Mark,

Well done. Take the magnets down to N40 and you'll be about right, I'd say.

I guess that's a section near the periphery, as it's quite a long yoke run, between magnets. Using segment arc magnets would make better use of the available area but need a thicker core and rotor back iron, of course...

 

[IanFiTheDwarf]

you need to simulate a longer section to ge an acurate saturation level. more than half the flux is traveling in one direction where in reality it would be exactly half for a continuous ring. idealy you need to simulate an infinate length to represent a ring.

 

[Miles]

Very true. I should've noticed that, it's rather fundamental.... Still waking up....

 

[Miles]

I modeled the rotor to get the proportions:

 

[Miles]

As this is sub 1000 rpm, perhaps you should consider increasing the pole count? Less waste of potentially active area. Less yoke material (core and back iron thickness) required. Shorter end turns.

 

[hillma]

you need to simulate a longer section to ge an acurate saturation level. more than half the flux is traveling in one direction where in reality it would be exactly half for a continuous ring. idealy you need to simulate an infinate length to represent a ring.

Hi Ian, thanks for your reply. I've re-run FEMM with additional magnets (changed to N40 - thanks Miles) as suggested and the results show a more even distribution of flux to the adjoining magnets and avoiding saturation as per my previous effort 8)

Cheers

Mark

 

[hillma]

As this is sub 1000 rpm, perhaps you should consider increasing the pole count? Less waste of potentially active area. Less yoke material (core and back iron thickness) required. Shorter end turns.

Hi Miles, many thanks for posting up the great CAD drawings (I'm still at the pen and paper stage!), that's exactly what I have in mind . For my first attempt I might stick to the 12 magnet layout for ease of construction and treat it as a learning exercise though the 24 magnet approach does maximise the active area and will result in a much improved motor, maybe that's V2?

For the 12 magnet rotor the stator will have 36 coils, 7mm wide and spaced at 6.5mm at the periphery (they will practically touch at the inner circumference of the stator). The motor will be an inrunner and to hold the stator and provide a thermal path to the can I was planning on a concentric 10mm thick aluminium ring attached to the stator (with 36 teeth on the inside, wide enough to allow for the end turns). Each of the 36 contact areas should be 10mm x 6.5mm(ish) which I hope will be sufficient transfer any heat generated and provide sufficient stability to the stator.

Magnets, wire and steel for the rotors seems fairly easy to find, biggest headache so far is trying to get my hands on a suitable source for the tape wound stator, the best solution seems to be looking at wound transformer cores unless there's a better alternative?

Cheers

Mark

 

[Miles]

I was planning on a concentric 10mm thick aluminium ring attached to the stator ....

Bear in mind that anything conductive, close to the path of the magnets, is going to get eddy currents induced in it. Luckily, you're operating at less than 100 Hz

 

[hillma]

Luckily, you're operating at less than 100 Hz

Thanks Miles, I hadn't considered the aluminium being a factor....you mentioned the operating frequency of 100 Hz, at the risk of a dumb question how do you arrive at that number and at what value would the frequency become an issue?

Cheers

Mark

 

[Miles]

The fundamental frequency is rpm / 60 * no. of pole pairs.

In your case 1000 / 60 * 6 = 100 Hz

Eddy currents increase as the square of the fundamental frequency and as the square of the lamination thickness. 0.5mm laminations will be fine for 100Hz. Because you are using a slotless design, the wires are exposed to the flux, too.

 

[major]

Stator
ID = 84mm
OD = 150mm (ratio of ID/OD = 0.58)
Thickness = 10mm
Material = spiral wound electrical steel

0.5mm laminations will be fine for 100Hz.

That would be a 12 meter long strip. I'm guessing that won't be cheap.

I knew some guys who built an axial generator. Pretty much on the cheap. I was just an observer as we shared some common shop space. But after they found out the aluminum stator plate was cookin' hot, they asked me why. So anyway, had them wind up a toroid out of steel banding. Guess what? It stayed cool and the voltage went way up.

So, I know that steel banding strap for packages isn't electrical grade by any means, but it might be an inexpensive way to construct a first go-around prototype. But I don't know; maybe transformer core winders use narrow slit electrical steel all the time and you can get what you want.

 

[Miles]

Refill for loggers tape?

 

[major]

 

[hillma]

Thanks Major :lol: :lol: :lol: :lol: !! Though I think the large clip on the end might interfere with the end windings ?

I have been searching around and have come across these guys who also happen to be on my side of the pond, they seem happy to deal in small quantities and I'm waiting to hear back from them

http://www.airlinktransformers.com/toroidal_cores/ros174/

Do you reckon that these might be up to the job?

Cheers and thanks again to all for their help

Mark

 

[h0tr0d]

Subscribed

 

[hillma]

About time for a bit of an update

After re-reading tons of online information and much scratching of head I've settled on the following revised spec. So far magnets, wire, steel and acrylic have all been ordered and should start turning up this week ! (Unfortunately as Jersey is a smallish island we are not able to pop into a hardware store and find what we need in stock - ebay is your friend!)

Rotor
140mm od, 50mm id by 6mm mild steel
Magnets
16 (per rotor) 20mmW1 x10mmW2 x25mmL x6.5mmH wedge magnets N45 (22.5 degrees) magnets originally created a 100mm od x 50mm id circle.

Magnets will be spaced out by 5mm which will push the outer diameter to approx. 130mm. Matching holes will be cut in 4mm acrylic sheet to retain the magnets in place and this will all be epoxied to the steel rotor.

Do you think that the epoxy will be strong enough or should I bolt through the steel and acrylic plates, if anyone has advice on what epoxy to use that would be great too?

Cant wait to get stuck into this project once the bits start turning up.

Cheers

Mark

 

[hillma]

Now that I've settled on the wedge magnets I've discovered an unexpected consequence of reducing the gap between magnets from 10mm to 5mm. In the attached FEMM image the bottom rotor shows every other pole saturating the rotor iron which is very odd, if I open the gap back to 10mm the flux distribution becomes even again. To try and combat this I've introduced additional 5mmx5mmx25mm magnets between the wedge magnets orientated like a halbach which seems to help out with the saturation of the rotor (shown as the top rotor for comparison) and evens up the flux distribution in the air gap and stator.

Should this uneven distribution be expected by closing the gap between magnets or is there something else messing up the analysis and does inserting the additional magnets to reduce saturation in the rotor provide a solution or am I heading off in the wrong direction (again). Any help or comments would be greatly appreciated as its got me beat at the moment.

Cheers

Mark

 

[hillma]

With the attachment !! :lol: :lol: :lol:

 

[IanFiTheDwarf]

This comes from having nothing for the end magnets to interact with on one side, so all of their field is going in one direction. I presume that as the gap between the magnets decreases they interact more with each other. As you can’t simulate a circle in this plane or an infinite length try adjusting the size of the end magnets until the flux lines from the next magnet split evenly left and right.

You have talked about your spiral wound laminate core but how are you going to make the stator teeth?
Are you planning on three complete sets of stator with two steel backed rotors for the three phases or are you going to use both sides of the magnets that are between two stators without iron backing and only have iron on the end rotors?

 

[Lebowski]

Do you think that the epoxy will be strong enough or should I bolt through the steel and acrylic plates, if anyone has advice on what epoxy to use that would be great too?

Cant wait to get stuck into this project once the bits start turning up.

Cheers

Mark

I bolted through

http://endless-sphere.com/forums/viewtopic.php?f=30&t=30061&p=434496#p434496

(picture ....308...)

the centrifugal forces are quite large and it gets very messy when the magnets are not kept in place...

P.S. I used stainless steel screws so as not to mess up the magnetic field
P.P.S. can you imagine, I did 9000km already on that motor !