Miles' 90mm inrunner build thread

[Miles]

Slinky type construction goes well with high pole counts... It might be possible to get away without cutting the head of the tooth (use the edge of the strip)?

 

[speedmd]

On the rotor/magnet side a straight edge would work great the way your planning on attaching the magnets. As long as the strip is not too wide it should be easily formed into the slinky. You would need a aluminum hub for it. Not sure how that impacts the losses. This could double as a internal fan if machined with slight helix spokes. Extrusions are also a possibility. With a good shrink fit it would have excellent heat transfer also. Wonder if these slinky type cores are available in standard sizes for prototyping.

 

[Miles]

Wonder if these slinky type cores are available in standard sizes for prototyping.

This is your size: http://www.gasgoo.com/showroom/autowb/auto-products/1178656.html Samples free 8)

 

[speedmd]

Great, free sample. 1000 pc minimums unfortunately.

Looks to be 1.0mm? Close to your 90 mm size also. looks like 36 teeth, at least what is pictured. May be good for testing possibly. If you can get them to stamp teeth into the thinner stuff in small quantities, that would be a super solution to our ebike needs on the stator side.

This one is even closer with .5mm thickness. http://www.gasgoo.com/showroom/autowb/auto-products/1165937.html

 

[Miles]

After the final tweaking these are the simulation results (3000rpm; AC sinusoidal source).

No load current 0.38 amps. Torque ripple at no-load 106.8%.

[pre]Amps Torque Eta T. Ripple Power Out Heat Out[/pre]
[pre]1A 0.12Nm 61.91% 42.81% 23.0W 14.14W

2A 0.24Nm 80.79% 21.76% 60.0W 14.28W

3A 0.35Nm 87.00% 14.57% 97.0W 14.47W

4A 0.47Nm 90.05% 10.94% 134.0W 14.8W

6A 0.71Nm 92.99% 7.26% 207.9W 15.7W

8A 0.94Nm 94.35% 5.58% 281.3W 16.9W

12A 1.41Nm 95.46% 3.67% 428.4W 20.3W

20A 2.33Nm 95.81% 2.36% 718.0W 31.5W

30A 3.45Nm 95.28% 1.58% 1070W 53W

40A 4.53Nm 94.41% 1.71% 1407W 84W

50A 5.56Nm 93.41% 1.97% 1730W 122W

60A 6.53Nm 92.31% 2.00% 2034W 170W

70A 7.44Nm 91.15% 1.80% 2320W 226W

80A 8.29Nm 89.94% 1.59% 2589W 289W

90A 9.09Nm 88.68% 1.41% 2840W 362W

100A 9.84Nm 87.38% 1.55% 3074W 445W

120A 11.19Nm 84.69% 1.82% 3498W 633W

140A 12.37Nm 81.88% 2.17% 3866W 855W

160A 13.38Nm 79.00% 2.51% 4183W 1112W

180A 14.26Nm 76.07% 2.84% 4459W 1402W

200A 15.03Nm 73.13% 3.28% 4701W 1728W[/pre]

 

[Miles]

More data added to above table.

 

[liveforphysics]

>90% efficiency from 4amps to 60amps is fantastic my friend.

>94% from 8amps to 40amps is also quite amazing. That's a 500% torque range while still >94% efficiency.

Electric motors are phenomenally efficient things when done correctly.

 

[Miles]

Let's see how close I can get to the simulation.......

 

[Honk]

Hi Miles.

Can I ask you a somewhat technological question?

Your motor looks promising according to simulations!
But what differs in your design from professional manufacturers? What makes it better?
Is it copper density, lamination choice, magnetic field strength or perhaps a combination of all?

 

[Miles]

Hi Honk,

Good question.

I used a copper fill factor of 0.6. That's definitely at the top end of the scale and I might struggle to achieve that but it's not unrealistic.

Laminations used in the sim. were Sura 0.2mm silicon steel, for operation at 500Hz that's reasonable.

Field strength is limited by the compromise struck for the ratio of space allocated between the copper and iron circuits.

I think I've got close to the optimal airgap diameter for this stator OD. and pole/slot combination.

All I can say is that I think I've made the right compromises at every stage.

I see the simulation as showing the maximum I can achieve from this design. If I get anywhere near it, I'll be happy.

 

[Honk]

Seems you have used mostly good materials and a well optimised design solution.
It could be that a commercial design like this is simply not economical to produce.

 

[crossbreak]

really good figures! I like the choices you made. I think a lot about the newer 0.33lams MAC, which is 32N 36P. Two of these would make a great single reduction setup. Are you still planing on going two stage with the (ok still simulated) figures you got? Single speed would be OK for most people who go offroad i guess. just spend more $$$ on the motor and save the $$$ and kilograms of the 2-speed gearbox. Just a thought... maybe we should make a spreadsheet for simulating a driving cycle. Sadly my Matlab/Simulink approaches didn't really yield any fruit.. just very artificial figures that won't tell much

 

[toolman2]

Seems you have used mostly good materials and a well optimised design solution.
It could be that a commercial design like this is simply not economical to produce.

its probly not AS economical to produce as a ~$10 build cost chinese motor (that sells for say $100), but to US this motor could be more economical than any other because it doesn't need a rewind, quality bearing changeover, hall slots, cooling, and a gearbox..

this happens all the time honk, a switched on dude on his own designing something that is waaay ahead of the commercial offerings, for some reason people think its the big manufacturers who are going to have the best designs for things, but somehow millions of really crap designs are sprouting up everywhere.

im left wondering how it is that a motor like the ca120 (and the rest) can go into mass production seemingly without anyone even spending 20 minuets checking it with FEMM or similar, -that to me is not economical. :wink:

btw miles, joby seem to now be using .13mm lams, pm me if you want more info.

 

[Miles]

Are you still planing on going two stage with the (ok still simulated) figures you got? Single speed would be OK for most people who go offroad i guess. just spend more $$$ on the motor and save the $$$ and kilograms of the 2-speed gearbox.

Yes. The motor was designed as part of this system. Two reduction stages (including the one to the wheel), with a 2 speed gearbox integrated into the jackshaft. The 2 speed components only add 600 grams to the jackshaft assembly weight.....(less than that if you take into account the fact that the gearbox has integral freewheeling, so a separate freewheel is not required).

 

[Miles]

btw miles, joby seem to now be using .13mm lams, pm me if you want more info.

Thanks for the info. Given that they are designed to operate at around 1000hz, that makes sense... There wouldn't be enough gain for me to justify the extra cost for my motor, though, unless I wanted to run it a lot faster.......

 

[speedmd]

With a slinky style core, you would be using a strip of material half the thickness and not have the massive amount of added scrap. It is a waist using twice the square footage of expensive laminate sheets to stamp out and scrap 80% of the material. It is a matter of getting the design worked out for manufacturability to eliminate the added costs of using the thin lams. You could also take advantage of directional alloys. For a single speed setup, it would be best to gear low and let the motor rev on to the speed the power limit would provide. It would be a game changer for ebikes. Still a bit unsure on best method of holding magnets in place and if added stiffness of the rotor core is possibly needed. I do like the rotor center hub design of the joby. This style would help keep the light laminate core round when hit with the pulsating mag forces. Unfortunately it would add the cost of machining one from solid or eventually making a extrusion if one suitable can not be found. Look forward to what you come up with.

 

[Miles]

Fair point. I really should investigate the practicalities of 'slinky' core construction......

 

[crossbreak]

The 2 speed components only add 600 grams to the jackshaft assembly weight.....(less than that if you take into account the fact that the gearbox has integral freewheeling, so a separate freewheel is not required).

Why even use a freewheel? Indeed, with a dual rotor MAC or BPM single speed, both rotor inertia and the drag would really suck. But with your motor both would be acceptable.

 

[Miles]

I think the drag would still be too much for me....

 

[bearing]

How about making the gearbox able to go into neutral, or the freewheel able to lock?
Then you are able to use regen.

The spiral/helical/slinky process seems like a good idea for high volume production. But not for prototyping, of course. And probably not for low volume production either.

 

[Miles]

The spiral/helical/slinky process seems like a good idea for high volume production. But not for prototyping, of course. And probably not for low volume production either.

Constructing a helical core for my motor would require a strip over 40 meters in length, with 3600 slots in it..... Forming it into an accurate and stable cylindrical seems a pretty daunting task, too......

 

[Miles]

How about making the gearbox able to go into neutral, or the freewheel able to lock?
Then you are able to use regen.

I've been working on this gearbox design for 7 years. I've considered everything.......

 

[speedmd]

The spiral/helical/slinky process seems like a good idea for high volume production. But not for prototyping, of course. And probably not for low volume production either.

Constructing a helical core for my motor would require a strip over 40 meters in length, with 3600 slots in it..... Forming it into an accurate and stable cylindrical seems a pretty daunting task, too......

Happy new year.

Slinky core is not for one offs IMO. Not sure unless you were skilled with shaping ribons of steel you could get there at a reasonable quality level. Certainly once you prove out the design and have lots of 1000 units or more you will see some serious savings going that route.

When looking at some of the advantages, consider the following.

Material waist reduction. Rough calculation is 50-60% minimum.
Stamping from a reel of ribbon, on a continuous basis.
Deburring and forming continuously.
Suface coatings and treatments also continuous.
Directional grain. Worth another few percent?

 

[Miles]

[pre]Comparison of 24t 20p and 24t 22p designs

Identical stators; same magnet pole arc & thickness; same magnet rating; both at 3000rpm (floating voltage).

No load:
Gap flux T. ripple Iron loss
20p 0.969T 106.8% 14.1W
22p 0.957T 17.1% 14.4W


Loaded:
Torque T. ripple Eta
6A:
20p 0.71Nm 7.26% 92.99%
22p 0.72Nm 1.63% 92.96%

20A:
20p 2.33Nm 2.36% 95.81%
22p 2.38Nm 0.97% 95.81%

40A:
20p 4.53Nm 1.70% 94.41%
22p 4.62Nm 0.37% 94.47%[/pre]

 

[bearing]

24t22p is pure and complete love

Is this with double layer windings?
The 1-layered version has a higher winding factor. Does it produce better numbers?
(I have a feeling we have discussed this before.)