MULE1.2 Axial flux test motor/bicycle specific

[spinningmagnets]

I can't speak for anyone who's willing to pay a lot for someone else to produce for the perfect E-bike AF motor for sale (mass production with jigs, etc), but...

I am most interested in a design that is as close to "wonderful" as possible, but still simple enough that the average garage enthusiast could build it from generic components with a jigsaw, drill, vice, etc. I am happy with the Agni specs and real-world performance, but I haven't the first idea about how it works, or why its better than the others.

It would be wonderful if Cedric posted advice about this AF motor spec. Though a few would begin building them, its my experience that the majority of users would rather buy a well-built unit rather than make their own. An E-bike "Agni-Jr" would be great. Trust me the Chinese have already reverse-engineered the Agni. You can't stop it, you can only establish yourself as providing a consistently high-quality product thats reliable (a weak point for Chinese products).

Wow! two people posted while I was typing...

 

[liveforphysics]

They all ready make Agni's that are undersized for an E-bike.

The problem still is that his designs spin the copper... It's not too bad when you're a huge bulky low RPM motor, but it becomes a serious performance limiter for any sort of small motor performance.

 

[Miles]

Just working out possible core techniques. This is supposed an elliptical coil (I cheated)

 

[paultrafalgar]

Thanks for that LiveForPhysics, Thud. I'll shut up now!

 

[liveforphysics]

Just working out possible core techniques. This is supposed an elliptical coil (I cheated)

I'm liking what I'm seeing.

Moose and I talked a bit about this a few days ago, and he suggested picking up a big roll of the copper ribbon used by stain glass folks. I suggested cutting strips to the needed length, and laying a pass of thin kapton tape along one side, then rolling it up on a bobbin made with 1 side removable. To make a coil like this, start from a nearly flat bobbin, wind it up, and perhaps do some post winding squishing as needed to match the shapes of the cutouts to hold the coils. This would be loads better copper fill% than you're going to get with anything involving enameled circular wire.

The tape can handle 1000F ad infinitum. The copper itself is pretty temp safe as well. With good coil mechanical support from the stator framework, we could get a pretty damn high delta-T going between the coil temp and the air passing over it. This means huge continuous power handling ability.

 

[liveforphysics]

Thanks for that LiveForPhysics, Thud. I'll shut up now!

Just doing my part

You asked a valid question, and I'm sure that's been on the fingertips of 90% of the sane readers of this forum. It's good you asked it and got it out in the open.

 

[Thud]

Paul,
No offence taken & hopefully none returned, its all good from from this side & actualy We need a "devils advocate" to keep things grounded. please feel free to chime in any time.

 

[liveforphysics]

http://www.anythinginstainedglass.com/metals/foil.html

This is just the first placed I found, but it seems they have thickness and width that would be appropriate for our needs. I've got a mega stash of kapton tape in various widths all ready.

The problem I see always comes back to what to do with that pesky center tap... If it just crosses over all the coils on it's way out, it's going to be inductively cooked, as well as lower efficiency. I know there has to be some clever solution to this, I just haven't quite thought of it yet... Maybe if I were to start to try an explain a non-solution in words, a real solution would jump into Jonathan's head, and then we would have something.

Ok, bedtime for me.

 

[rhitee05]

You guys have been busy while I was sleeping. Trying to catch up...

If you have back-to-back coils on a twin rotor design, it's best for the cores for each coil pair to be common (axially contiguous)? Right?

You're talking about a multi-stage configuration, like 3 rotors/2 stators? I've seen a few variations on the design, but in most of them the flux flow is continuous from one end to the other, so you'd want all the stator poles lined up for this. You could offset them slightly from each other, which should have the same effect as a skewed rotor for reducing cogging torque in a radial-flux motor.

One other cool thing about the multi-stage designs, not sure if this has already been bought up, but you can switch them to get variable Kv. If you had a 4-stator design, you can connect it 4p1s, 2p2s, or 1p4s and get a 4:1 range of Kv (factor of two difference between series and parallel connection). If you really want you could toss in delta-wye on top of that...

 

[Miles]

Just working out possible core techniques. This is supposed an elliptical coil (I cheated)

I'm liking what I'm seeing.

Moose and I talked a bit about this a few days ago, and he suggested picking up a big roll of the copper ribbon used by stain glass folks. I suggested cutting strips to the needed length, and laying a pass of thin kapton tape along one side, then rolling it up on a bobbin made with 1 side removable. To make a coil like this, start from a nearly flat bobbin, wind it up, and perhaps do some post winding squishing as needed to match the shapes of the cutouts to hold the coils. This would be loads better copper fill% than you're going to get with anything involving enameled circular wire.

The tape can handle 1000F ad infinitum. The copper itself is pretty temp safe as well. With good coil mechanical support from the stator framework, we could get a pretty damn high delta-T going between the coil temp and the air passing over it. This means huge continuous power handling ability.

This was for the core. As you say, the copper would be wound the same way, on top. Is this ok for the core structure?

 

[Thud]

I knew I saw Copper tape somewhere, I have a buddy who does staind glass work, I recall a roll on a shelf in his workshop.
Time to start networking.

Miles, you are working on Cores (not coils) I know Luke Likes the coreless design for several reasons.

lets sumerize this little exchange-
Just to keep us all on the same page.
We are using cores in the 1st test motor.
lets settle on a stator tooth count. I would like to get some better models drawn for evaluation.
determim a start point magnet aray. I am not opposed to attempting a halbauch,I just need a little help figuring it with readily available magnets. Who has the best Flux path simulator? Help please.
what else is there-I know i am forgetting something.

 

[rhitee05]

I think the flat copper windings are a winner. Definitely a good way to get high fill and massive current capacity. Make elliptical coils with the inside height about the same as your magnet width. Stack them double-layer CSIRO-style and you've got yourself a very nice stator.

 

[Miles]

You're talking about a multi-stage configuration, like 3 rotors/2 stators? I've seen a few variations on the design, but in most of them the flux flow is continuous from one end to the other, so you'd want all the stator poles lined up for this. You could offset them slightly from each other, which should have the same effect as a skewed rotor for reducing cogging torque in a radial-flux motor.

No, just 2 rotors 1 stator. Well, to start with..

I'm confusing myself, now...... With a coreless design using flat strip copper coils, back-to-back pairs made sense - it enabled the connections to be made between the pairs. With an iron core, it's not so important that the coils end close to the magnets, so the connection could be made at the end.

 

[Jeremy Harris]

The problem I see always comes back to what to do with that pesky center tap... If it just crosses over all the coils on it's way out, it's going to be inductively cooked, as well as lower efficiency. I know there has to be some clever solution to this, I just haven't quite thought of it yet...

How about two copper strip coils laid one on top of the other, wound in opposite directions, for each full coil? This would allow the centres of each to be soldered up, with the true coil connections coming out from the outside edges.

The downside is that the coil ends up being wider, so the air gap goes up.

An alternative is to look at bringing the centre out by twisting it and feeding it to the centre. This might work OK, as the very ends of the coil don't contribute much.

A final idea for now is to think about drilling a hole through the coil at the centre (assuming oval coils as Miles has drawn) and feeding a round wire through to the centre coil connection. It would increase the resistance of the coil very slightly, but might be a workable answer to the problem.

Jeremy

 

[rhitee05]

No, just 2 rotors 1 stator.

I'm confusing myself, now. With a coreless design using flat strip coils, back-to-back pairs made sense - it enabled the connections to be made between the pairs. With an iron core, it's not so important that the coils end close to the magnets, so the connection could be made at the end.

Oh, I think I see. I think you're better off doing a staggered winding, lap-winding style. Saw a nice picture of this somewhere, but can't find it now. You end up with two layers of coils. Start with phase A in layer 1, then phase B is in layer 2 overlapped halfway over phase A, then phase C in layer 1 next to phase A with a 50% overlap to phase B, etc. This gives lower Kv and higher copper density for a given coil size and turn count.

Did I guess right this time?

 

[Miles]

The problem I see always comes back to what to do with that pesky center tap... If it just crosses over all the coils on it's way out, it's going to be inductively cooked, as well as lower efficiency. I know there has to be some clever solution to this, I just haven't quite thought of it yet...

That was the reason for back-to-back pairs on the coreless design - so that the connection could be made in the middle, away from the magnets.

 

[rhitee05]

I don't think bringing out the connection to the center of the coil should be that problematic. You have to cross the coil somewhere, just do it at the lowest point of the coil so you cross a minimum amount in the shortest distance.

 

[Miles]

This was the earlier illustration I did for Thud. It shows the back-to-back coil pairs:

 

[rhitee05]

I'm speculating a bit, but I think it'd be easier to get a low Kv by using a large number of poles rather than a few poles with a whole bunch of turns. That's the approach CSIRO and Launchpoint seemed to take. If you overlap the coils, you get more copper fill in the stator and less wasted volume.

I know that's just a conceptual sketch, but I want to point out again that the coils need to be larger in size than the magnets. You really want the inside height of the coil to be the same as the height of the magnets, so the coil can capture the flux of the magnet thru it's full number of turns.

 

[Miles]

I don't think bringing out the connection to the center of the coil should be that problematic. You have to cross the coil somewhere, just do it at the lowest point of the coil so you cross a minimum amount in the shortest distance.

But for a coreless twin rotor that means the airgap will increase to accomodate the centre tap - unless you have back-to-back coils. Or am I making too much of this?

 

[Miles]

I know that's just a conceptual sketch, but I want to point out again that the coils need to be larger in size than the magnets. You really want the inside height of the coil to be the same as the height of the magnets, so the coil can capture the flux of the magnet thru it's full number of turns.

Yes, the illustration was done last week. We've come a long way since then I was just using it to illustrate the solution to the centre tap problem... Imagine the same idea with ellipsoid coils, with the end turn beyond the magnets.

 

[rhitee05]

But for a coreless twin rotor that means the airgap will increase to accomodate the centre tap - unless you have back-to-back coils. Or am I making too much of this?

No, that's correct, it would be an issue for a single-layer stator. If you did a two-layer design with overlapped coils, the center tap of one coil could come out in the other layer between the two adjacent coils. Since the center of one coil would be lined up with the edges of the other two, that could work out quite nicely. Does that make sense without a picture? I'm not a CAD master like you.

 

[liveforphysics]

I'm speculating a bit, but I think it'd be easier to get a low Kv by using a large number of poles rather than a few poles with a whole bunch of turns. That's the approach CSIRO and Launchpoint seemed to take. If you overlap the coils, you get more copper fill in the stator and less wasted volume.

I like the overlapping coils method as well, but It doesn't seem to be compatible with cores, and it looks like we're going with cores on this one, which is probably a good thing for this first motor.

I know that's just a conceptual sketch, but I want to point out again that the coils need to be larger in size than the magnets. You really want the inside height of the coil to be the same as the height of the magnets, so the coil can capture the flux of the magnet thru it's full number of turns.

This would be true only when using core's right? You're saying you want the core OD (coil ID) to match the magnet size. Everything for coreless seems to imply to make the magnets match or better, overlap the coil on the ID and OD.

 

[Miles]

Does that make sense without a picture? I'm not a CAD master like you.

That makes perfect sense. What I modelled was 2 layers with the coils aligned - which means a wasteful separation between them to make the centre taps. What you propose is 2 layers with the coils staggered, which is a better solution.

 

[liveforphysics]

rhitee05 = The man.

Ok... going to sleep for reals now... flu is fading away, but I still gotta work another 12hr shift tomorrow.