Ultimate lightweight wheel-motor concept

[Miles]

Yes, I would have assumed a curved face to offer better overall performance. Have you tested a bread loaf shape? It could offer a tighter average airgap than flat without the high detent and ripple typically gotten with simple airgap radius faces. Depends on magnet coverage and stator shape, so running a sim would tell the tale.

A quick way I have found to estimate breadloaf shape is to peak the middle at min airgap, and radius the top so the edges are at max gap found on other design iterations. From there, the leading/ trailing edges can be curved or chamfered to further soften magnetic transition from pole to pole. It's basically the inverse of a flat top, but with tighter coupling and more magnetic material.

I got Stephan to modify the breadloaf template in Emetor, so I could experiment with this very thing. Previously, you could only use the airgap radius as the magnet radius. Now, you can set the magnet radius independently. For the design I was working on, reducing the magnet radius actually made torque ripple worse.... Of course, it probably is more likely to be the other way round, generally. Anyway, there was no penalty for using flat magnets, in that case. You can achieve the same fundamental airgap flux density by adjusting other factors.

Ref: http://www.emetor.com/blog/post/release-2013-10-18-more-flexible-template-breadloaf-magnets/

 

[johnrobholmes]

Since all my experience lies with low pole small diameter rotors, I'm off the mark with my assumptions here! I should know better than to generalize so much, motors dont scale well. Carry on!


If you post up a drawing of magnet I can submit it with our next batch for cost estimates.

 

[johnrobholmes]

Just saw 4x10x15mm on one of your posts. I could get a quote on a range of grades. 700 pc should be enough for ten motors with a few spares.

If you are going to make them key in, I'll wait till the shape is determined.

 

[speedmd]

Interested in the simulation results of flat faced vs breadloaf. It may work out to not matter much if at all.

On the magnet manufacturing, are they compression molded in a permanent mold and then sintered? Would be good to know exactly how they make them so we can better work within mold limitations (mold cost /cycle issues) when designing final/ ultimate mag. shapes. Would be good to know also if molded, if the supplier has in house mold making abilities to better consult what exactly is possible shape wise if we wish to go for any undercuts in the mag sidewalls.

 

[Miles]

Since all my experience lies with low pole small diameter rotors, I'm off the mark with my assumptions here!

I just measured the model. The difference in airgap distance at the edge of the magnets from that at the middle of the magnets is 0.06mm

 

[Miles]

If you are going to make them key in,.........

You mean like here? http://www.endless-sphere.com/forums/viewtopic.php?p=860752#p860752

We could inset them almost completely, if there's an interest in field weakening....

 

[johnrobholmes]

There is a certain amount of shrinkage during sintering, so unless there is a large negative draft angle there are no special considerations on shape to be made. Magnets that are pressed still have to be ground to final tolerance, I would assume a slightly tapered magnet to use a basic square sided mold.


The break even would be about 100 motors, so a run of 200 motors is the point where I would consider molding as worthwhile. I can get mold life cycle estimates at the same time.

 

[johnrobholmes]

If you are going to make them key in,.........

You mean like here? http://www.endless-sphere.com/forums/viewtopic.php?p=860752#p860752

We could inset them almost completely, if there's an interest in field weakening....

Like that, and with a wider base if epoxy alone isn't suspected to be enough. Keeping the mags square would mean only one type, and from an assembly and inventory perspective I like it simple.

 

[Miles]

What's the price differential between EDM cut and moulded?

 

[Miles]

Keeping the mags square would mean only one type, and from an assembly and inventory perspective I like it simple.

Me too

 

[johnrobholmes]

What's the price differential between EDM cut and moulded?

I'll get the breakdown for it asap.

 

[speedmd]

As far as the shape is concerned, if molded, one could open mold from the sides rather than top/ bottom direction on this narrow shape. Draft can form the dome in the bread loaf face. Parting line would run length wise centered on the top and bottom faces. Interested in draft angles required for the material so I can draw up some concepts. What are some of the ideal mag dimensions .

 

[johnrobholmes]

it looks like miles has determined a flat magnet to be ideal for torque rippl. 0.06mm of airgap difference is really low for a flat mag, so there isn't much material gain or airgap tightening to achieve either.

 

[speedmd]

I am mainly looking to eliminate most of mag machining and reduce assembly costs by eliminating the bonding process.

 

[johnrobholmes]

I don't see an easy way to completely get rid of epoxy for the magnet affixing. Getting a press fit tolerance on regular price magnets is really tough if not impossible. With enough clearance to pass the mag through without force, there would be rattling and fracturing during use. There would also need to be a "lock ring" to hold the mag in transversely, which would increase labor and parts count considerably.


A rotor with alternating magnets and slots to jig them will take very minimal labor and epoxy to finish with a trained worker. If I were setting up an assembly line for it, I would use a steel plate to set all magnets on. Degrease them all at once, sandblast all at once, degrease again, then mix epoxy and start wetting out the magnets and stator. Placement won't take more than 3 minutes with a fast hand. Slap em all on, then edge align them all at the same time with a flat surface. I would estimate at most 15 minutes for one serial rotor process and 100 cents of high dollar epoxy. With three workers on a serial line it could be cut down even further.


Winding and impregnating the coils will be much more cost intensive. Just getting a worker trained well enough to do it right with speed will be tough. One hour winding for the average hand would be doing well. A very skilled winder that didn't bullshit could probably do it in 30 or less if copper fill wasn't too high. Resin setting the coils with brushed on application would take a few more minutes. Vac chamber application would take a few minutes to pull in, 15 minutes to drain, then 4hr oven cure for most industrial coil resins.

 

[johnrobholmes]

my math for winding:

4 seconds per turn, 6 turns per slot, 72 slots= 29 minutes.



I could wind faster than this, but there will be mistakes and setup time to consider. untrained people will have a hard time hitting the slot consistently with a bundle of wires in hand. Using fish paper for insulation can help guide the wire in.

 

[speedmd]

Depending on manufacturing methods of the magnets, this is what I was thinking could be possible for a magnet shape with a straight open /close mold going from the sides.

http://freepdfhosting.com/94eca19689.pdf

You would most likely still need to bond them if there is play with a non sagging type epoxy and it would benefit heat transfer as well. Bonding would not be as critical and mechanical failures of the bond would not be catastrophic with the mags anchored by the dovetails. Fit Tolerances, yes, a bigger issue.

 

[speedmd]

Some more interesting info I found on magnet manufacturing. Plating has other benefits. Interesting issues with pressing / grain orientation. Large shrinkage percentages in sintering also.

https://www.kjmagnetics.com/blog.asp?p=how-neodymium-magnets-are-made
http://www.arnoldmagnetics.com/Magnet_Manufacturing_Process.aspx

 

[liveforphysics]

my math for winding:

4 seconds per turn, 6 turns per slot, 72 slots= 29 minutes.

Some motors I'm happy to get one turn per 1 minute. All depends how difficult to manage your wire is and how flat you need to pack it down to lay to get the copper fill you need.



I could wind faster than this, but there will be mistakes and setup time to consider. untrained people will have a hard time hitting the slot consistently with a bundle of wires in hand. Using fish paper for insulation can help guide the wire in.

 

[speedmd]

Thanks johnrobholmes for the time estimates and the process step break down. I see myself taking much much longer doing any of this the first few times. Would love to see a video of you winding at this rate once we get things stated.

 

[johnrobholmes]

Some motors I'm happy to get one turn per 1 minute. All depends how difficult to manage your wire is and how flat you need to pack it down to lay to get the copper fill you need.

Tis very true. I have some 12 slot 22mm diameter cores here that take me 2 hours to wind, even with a power winder helping me. Very high wire fill and high turn, so there is no room for a misplaced wire on the base layers.


Glad to help where I can, when I have some spare time to think about this fun motor project.

 

[johnrobholmes]

Example of the more difficult cores to fill

 

[speedmd]

Great work there johnrobholmes. You have my vote to be our official wiring process layout specialist for the prototypes.

On the magnets, they do also extrude shapes which may be a suitable option here if we go with some sort of sidewall undercuts. They may be a bit wide and flat to get away with much of this and we may need to discuss with your magnet folks what may be possible / reasonable low cost solutions. Grinding/ Machining is most likely the lowest cost option to give good fits.

@Miles, You had it drawn with twin magnets. Been a while now, and I am assuming that was to use standard size blocks. Advise if I am under the wrong impression here and if a one part magnet in something like the 15x10x4mm size would not be what we would want ideally.

 

[Miles]

@Miles, You had it drawn with twin magnets. Been a while now, and I am assuming that was to use standard size blocks. Advise if I am under the wrong impression here and if a one part magnet in something like the 15x10x4mm size would not be what we would want ideally.

If you mean the 68p one, there was no circumferential or axial segmentation. One 10mm wide, 15mm long magnet per pole.

 

[speedmd]

Thanks Miles

I have not had a chance to draw up what you last simulated for rotor core profile shape. Would like to take a stab at it with modifications to possibly dove tail into a custom magnet shape. If you have some updated core dimensions you would recommend, I will try to give it a go when time allows to see what core material is left to use in this tight spot.