Dual rotor axial flux motor design

[fechter]

I'd solder the connections after pressing them in. Using high melting point solder might be good there too (but it's even better if these parts don't get hot enough to worry about that).

On your rotors, it might be good if there was a small lip on the OD to help keep the magnets from flinging off at high speeds. I don't trust glue that much. Also, if the rotor hub was mild steel, it could help the back iron or replace the back iron.

[flathill]

How about the following idea for connecting up the coils?

- Before assembly, crimp the ends of each coil strip around a gold-plated tube (bullet connector?).

- Use gold-plated copper wire to form the bridging links.

- Wire is a push fit into the ID of the tube (end could be tapered).

How about some poki-poki style

Also know as a joint lapped core
http://www.mitsubishielectric.com/compa ... 12_tr3.pdf

Not exactly applicable but it gets you thinking

[Miles]

On your rotors, it might be good if there was a small lip on the OD to help keep the magnets from flinging off at high speeds. I don't trust glue that much. Also, if the rotor hub was mild steel, it could help the back iron or replace the back iron.

Yes, I'm intending to put a lip on, possibly by milling shallow pockets for each magnet, which would also give the spacing.

Your second point has prompted me to try something that we talked about before. With this, all the internal height of the motor could be used for active material.

Increasing the height of the cores to match would lower Rm whilst also increasing heat dissipation but would mean higher iron losses. Thoughts?

[fechter]

I'm missing something.

In the pic above, the gap needs to be shortened. I guess you're suggesting making the cores longer to fill the space. It would be better to move the rotors inward to shorten the gap. Adding iron behind the magnets can't hurt.

[Miles]

Yes. It was/is work in progress:

[Miles]

I guess you're suggesting making the cores longer to fill the space. It would be better to move the rotors inward to shorten the gap.

Why would it be better?

[Miles]

Ok. Here's everything updated to match:

[bigmoose]

Miles, might I suggest that you throw a lip on the OD of the rotor to restrain the magnets with respect to centripetal forces and to align them when they are epoxied on.

[Miles]

Thanks Dave.

Richard has prompted me on that, too

On your rotors, it might be good if there was a small lip on the OD to help keep the magnets from flinging off at high speeds. I don't trust glue that much. Also, if the rotor hub was mild steel, it could help the back iron or replace the back iron.

Yes, I'm intending to put a lip on, possibly by milling shallow pockets for each magnet, which would also give the spacing.

[Miles]

What do you think of this solution?

[fechter]

OK, nice. I like the magnet wells.

My point earlier was just in response to the drawing with a huge gap. I'm not sure if there is some advantage to having the core shorter. As long as it isn't saturating, it should be good.

[bigmoose]

Miles, my apologies, I missed Richards recommendation. I wouldn't have repeated it if I had seen it.

Here is a FEA that I did of the Mars Axial motor a year or two ago. The pix in that thread got hammered during the migration. Here is a pix of that rotor (not yours) at what I believe is it's max RPM of 6750. What is of interest in this design is the 0.008 in deflection in the lip area. It gives an idea of how things might move around at speed.

6750rpm500lb-DISPLACEMENTmagnets.jpg (54.52 KiB) Viewed 369 times

Stress plot for same condition:

6750rpm500lb-STRESSmagnets.jpg (57.82 KiB) Viewed 369 times

[Miles]

That is interesting. Thanks Dave.

I had a quick scoot around and came across this
http://acuteaero.com/2010/06/16/breakin ... ess-motor/
Anyway, it has some good pics of the rotor...

[Miles]

I'm reviewing the option to use a wedge shaped core section, matching the heads........

[Miles]

Advantages:
- More iron
- More complete utilisation of available volume
- Greater exposed area of coil on the outside for heat dissipation
- More area for coil pair link to be on inactive copper
- Fewer coil turns (better heat transfer and fill factor)
- Smaller proportion of flux path is circumferential (better for G.O. steel)
- Complete control over core module morphology

Disadvantages:
- Greater proportion of coil length is non-contributory (only marginally extra active copper)
- Requires segment shaped magnets to take advantage of the extra iron
- Likely to give a more "blocky" bEMF
- All machining needs to be done after assembly of lamination block

Anything else?

[Miles]

.....

[Kingfish]

The AF design I have been pursuing uses wedge-shaped coils in the belief that it is more beneficial for the layout, especially if the magnets are wedge-shaped as well.

On the connections between coils, it is apparent how the outer layers are wired though not obvious how the inner are connected with the flatness of the inside end-turns: How is this accomplished?

+1 on the shoulder to retain the magnets

What is your plan for active cooling?

Nice work & good progress, KF

[Miles]

Thanks KF!

On the connections between coils, it is apparent how the outer layers are wired though not obvious how the inner are connected with the flatness of the inside end-turns: How is this accomplished?

I haven't modelled that part here, yet. It's something that came out of Thud's epic "MULE" thread: viewtopic.php?p=214408#p214408



Sorry about the inage - I can't find the original.....


Here's a 3D PDF:
[Click on the image in the PDF to activate the model. ]

[Miles]

What is your plan for active cooling?

The original plan was to have a centrifugal fan doubling as the brace separating the two rotors. Now, I'm not sure that will be necessary....? It would be a continuous addition to the parasitic losses. The passive cooling from air flowing through the stator, around the large areas of exposed copper, should be pretty effective.

[Miles]

Ok. Turning attention to the coils

The best place to have the internal connection would be on the endturn on the outside of the motor.

This would mean that there would be 2 half turns (one for each coil). So, the number of turns will change to odd numbers.

At the moment, we have 6t coil modules (2 x 3t). I was thinking that with the extra iron (assuming......) we could drop to 4t (2 x 2t) but 5t (2 x 2.5t) might actually be closer.....

[Kingfish]

Double-coil/internal link: That is devilishly brilliant and certainly solves a large issue! The only issue I can foresee is that one side of the coil will have more pull due to the double-back unless the begin & end terminations are on opposite sides of the coil – so that you have the same count of conductors on each side. Make sense? The physicality though may not be a problem because now we are distributing linked connections on both sides which may prove to be shorter and thus potentially reduce overall resistance. Now I want to go model it…

To make the cross-over, if using flat wire, fold the conductor at a 45° angle to create the 90° bend, and then repeat again. I’m an old dog PCB designer and we used to hand-tape boards before the invention of CAD, therefore the concept of folding tape is not foreign idea, and it might work here. My plan though is to use Litz wire (ironless core) thus the challenge will be more difficult.

Good on you, KF

[Miles]

The only issue I can foresee is that one side of the coil will have more pull due to the double-back unless the begin & end terminations are on opposite sides of the coil – so that you have the same count of conductors on each side. Make sense?

Yes. In fact I've just been contemplating this very point, as I'm starting to model it There will be an asymmetry, but there will still be the same number of conductors on each side, in total. Right? As the force acts from the core and not the coil, I assumed the asymmetry wouldn't be a problem.

I could make the link on the endturn on the inside of the motor, which would give whole turns, but it would be messier...

[Kingfish]

If the coil is tapered as you recently suggested, the end-turns do not contribute to the motor other than affecting losses, so that’s a good place to make the links to other coils.

I redrew the concept and determined that even-numbered turns are best if the begin-end terminations are to be on the same side (presuming inside the magnet radius). To visualize from top-down, if we had an ovoid shape with the narrow end at the bottom pointing towards the axle, we begin winding from the outside towards the center from 6:30 clockwise, round and round until we reach 6:00, fold and flip to the other side, continuing clockwise round and round – matching the same number of turns until we reach 5:30. The direction of current remains the same and we have an equal balance of conductors on all but the inside end-turn. That sounds reasonable for your design. I will have to recalculate for mine cos I did not plan for double-windings of late, and allowed instead for an odd number of turns per coil. However – it definitely resolves a cumbersome design problem

Very pleased, KF

[Miles]

Ok. This is an accurate model of a 2 x 2.5t coil...

[markobetti]

Hal are you watching this ? LOOKY what Miles is doing ....We should bring him to Croatia ..Would make Sundays e-bike meetings much more interesting
Miles ,great progress