Dual rotor axial flux motor design

Miles

100 TW
Joined
Mar 16, 2007
Messages
11,031
Location
London UK
I've finished my gearbox designs and have started thinking about motors again....

I've decided a dual rotor axial flux would be more interesting to do than the single rotor one that I modelled before.

Anybody think the following approach, for the stator, is worth pursuing?

It's the same basic topology as the Helmut Schiller and Oxford YASA motors.
 

Attachments

  • AF-core-design.jpg
    AF-core-design.jpg
    65 KB · Views: 5,724
Yes, I think that's quite reasonable. Symmetry is useful and probably makes the mechanical design easier in terms of dealing with the magnetic attractive forces. You must have something serious in mind with the copper strips in those windings. :)
 
Looks like the coils will be easy to cool. :D
 
Miles said:
I've always wanted to try out the contiguous double-spiral strip idea that we discussed before

I can tell you've been thinking about this... you even have the laminations drawn in the right direction. :)

I'm curious what you have planned for the mechanical design (if you've thought that far ahead). If you use big magnets to match those coils, some serious thrust bearings and lateral stiffness of the rotor are in order!
 
rhitee05 said:
I can tell you've been thinking about this... you even have the laminations drawn in the right direction. :)
:mrgreen:

rhitee05 said:
I'm curious what you have planned for the mechanical design (if you've thought that far ahead). If you use big magnets to match those coils, some serious thrust bearings and lateral stiffness of the rotor are in order!
The idea was to brace the two rotors against each other as close to the inside of the core as possible so, the axial forces will be neutralised with respect to the bearings and things will be as rigid as possible around the rim...
 
pelle242 said:
Any plans on how to get the stators made? Laminations or iron powder?
Definitely laminations. The idea is to have the minimum of iron so that it's close to saturation at max. continuous torque. No dedicated flux return path, minimum cross-section and thinnest laminations = low parasitic torque :D
 
Miles said:
The idea was to brace the two rotors against each other as close to the inside of the core as possible so, the axial forces will be neutralised with respect to the bearings and things will be as rigid as possible around the rim...

That sounds like a pretty good plan to me (not that I'd question you anyhow). A bracing structure of some kind between the rotors would absorb the axial forces and then you can just use regular non-thrust bearings. If you want it, I can probably help you try to estimate those axial forces when you get closer to the design phase, unless you just want to over-engineer the snot out of it! :twisted:

Miles said:
Definitely laminations. The idea is to have the minimum of iron so that it's close to saturation at max. continuous torque. No dedicated flux return path, minimum cross-section and thinnest laminations = low parasitic torque

Another plus here - each core is just a stack of rectangular lams, so that should make it easier/cheaper to source materials rather than needing custom-stamped parts. Judging by the above, I assume you're planning to use proper electrical steel laminations like you'd find in a "real" commercial motor.

Miles said:
There's not much of a conduction path through the core, though....

Not sure how effective this would be, but one suggestion. You could space the two coils apart slightly and clamp a thin-ish aluminum plate or something around the center of each core to help conduct some heat away. There would be some eddy losses, but most of the magnetic fields should be pretty strongly confined to the cores so it probably wouldn't be excessive. I have no idea if that would provide enough cooling/extra power potential to offset the losses. Perhaps a better idea would be to design the bracing structure for the rotors like a squirrel-cage fan and just pump lots of air between the coils?
 
Miles said:
pelle242 said:
Any plans on how to get the stators made? Laminations or iron powder?
Definitely laminations. The idea is to have the minimum of iron so that it's close to saturation at max. continuous torque. No dedicated flux return path, minimum cross-section and thinnest laminations = low parasitic torque :D

i ve seen designs like that , fine efficiency . Finally you started this , i think aial flux design is underrated today . i bought axel borg book - diy - nice chapters there - you should look it up . Laters Axel design was based on dual rotors with mags i think
 
rhitee05 said:
Perhaps a better idea would be to design the bracing structure for the rotors like a squirrel-cage fan and just pump lots of air between the coils?
Yes, it would definitely be worth experimenting with that - especially in combination with opening up the case ends and sucking air in through the rotors, using some sort of integral axial fan arrangement.
 
rhitee05 said:
If you want it, I can probably help you try to estimate those axial forces when you get closer to the design phase, unless you just want to over-engineer the snot out of it! :twisted:
That would be a great help, Eric. I'd also appreciate advice on calculating the right balance between the iron and copper volumes.

I feel a responsibility to revive the DIY motor challenge... :) :twisted:
 
Any concerns about flux lines cutting through at an angle to the lam stack, as the lam stack isn't curved but flat?
 
liveforphysics said:
Any concerns about flux lines cutting through at an angle to the lam stack, as the lam stack isn't curved but flat?
Good point. I've know idea how detrimental that would be? I guess one could curve the laminations. Maybe even as part of the stamping process?
 
markobetti said:
Finally you started this , i think aial flux design is underrated today . i bought axel borg book - diy - nice chapters there - you should look it up . Laters Axel design was based on dual rotors with mags i think
Thanks Marko. I'll look it up.
 
liveforphysics said:
Any concerns about flux lines cutting through at an angle to the lam stack, as the lam stack isn't curved but flat?

I suspect that this shouldn't be too significant so long as the pole count and radius are sufficiently large, so the angular size of each core is not very big. Looks like maybe 18 teeth/cores? Since Miles knows the dimensions, I'm sure he could figure out how much of a difference there is between the straight lams and the ideal arc. Probably not worth the trouble to do anything about, especially if the lams are made from thin electrical steel.

I was also going to suggest that it would probably end up being worth your while to shape the cores so the center section is narrower and the ends flare out somewhat. That would give you room for more copper in the winding while keeping the tooth ends the same size, which I suspect is a good tradeoff for slightly lower saturation flux threshold. This is a good item for some simulation and optimization.

If you wanted to be really fancy, you could shape the ends such that the flare is wider on the outside edge of the core and narrower on the inner, making the end somewhat more wedge shaped (looking axially). Again, not entirely certain how much benefit this would provide (this is a 3D geometry, so would be difficult to simulate using the freeware) and would make fabrication more difficult.
 
rhitee05 said:
If you wanted to be really fancy, you could shape the ends such that the flare is wider on the outside edge of the core and narrower on the inner, making the end somewhat more wedge shaped (looking axially). Again, not entirely certain how much benefit this would provide (this is a 3D geometry, so would be difficult to simulate using the freeware) and would make fabrication more difficult.
I seem to remember modelling something like that on the single rotor design. Unless there's a significant gain it would be nice to keep all the laminations identical. I should imagine it would be more beneficial if you were using SMC cores?
 
Miles said:
I seem to remember modelling something like that on the single rotor design. Unless there's a significant gain it would be nice to keep all the laminations identical. I should imagine it would be more beneficial if you were using SMC cores?

Come to think of it, you might actually be able to use the square ends as a "feature". I expect this would have the effect of making the BEMF of the motor more sinusoidal, especially if you also used rectangular magnets on the rotors.
 
Thanks Eric. :)

Anyone know how much there is to be gained from using grain oriented steel for this application? It would certainly be possible to take advantage of it with this type of core design.
 
rhitee05 said:
liveforphysics said:
Any concerns about flux lines cutting through at an angle to the lam stack, as the lam stack isn't curved but flat?

I suspect that this shouldn't be too significant so long as the pole count and radius are sufficiently large, so the angular size of each core is not very big. Looks like maybe 18 teeth/cores? Since Miles knows the dimensions, I'm sure he could figure out how much of a difference there is between the straight lams and the ideal arc. Probably not worth the trouble to do anything about, especially if the lams are made from thin electrical steel.

Miles said:
It's only a 90mm motor. The minor diameter across the cores, as modelled, is 62mm. Major diameter 82.

Maximum deviation is 0.4mm across 10mm wide core.
This is what it looks like with a 0.35mm lamination thickness (first two laminations from the inside):
 

Attachments

  • Arc-across-lams.jpg
    Arc-across-lams.jpg
    9.2 KB · Views: 4,404
At worst, a 1.5% increase in "effective" lamination thickness (2.25% hit), on entry and exit..... So, less than 1% increase in eddy current losses from not having curved laminations?
 
rhitee05
Why do the laminations have to be horizontal (as Miles has shown) rather than vertical. The first application of the split flat coil wrap that I ever saw was done by Shane Colton. Is there any benefit in making the coil end plates more 'pie' shaped to maximize the flux face area of the coil? Does this approach not engage the rotor magnets for a slightly longer period of time and result in a more sinusoidal BEMF? I have not yet had time to start my axial design, but I had assumed that a powdered core with intergal flux face plates could be all ground at the same time on a lathe fixture to ensure that the resultant gap between the coil flux end plates and the rotor magnets is more uniform/constant as opposed to using laminations. Miles, have you considered how or what would need to be done to replace any coil if it is damaged in any way? How do you plan to connect the various coils and ultimately bring out the wiring? Is this a single 3-phase or multiple 3-phase groups? Very glad to finally see a discussion like this ongoing. Thanks everyone for answering.
kenkad
 
Back
Top