Open Source DIY non-hub motors

spinningmagnets

Moderator
Staff member
Joined
Dec 21, 2007
Messages
12,952
Location
Ft Riley, NE Kansas
[This thread split off from the TPP thread, over concern about being able to get motors for ebikes if there is a trade war with China]

When it comes to hand labor, the part we'd have a hard time in the USA getting quality workers to produce at a low wage is the hand-winding of the coils on the stator. (just my opinion, I could be wrong). Using powdered-metal processes with minimal machining (with the machining being done by CNC, same cut over and over) Some of the complex parts can actually be easy to make.

Build motors in the U.S.? With labor who is willing to gain the RIGHT education to run the equipment?

These days, it can be on-the-job training (OTJ). The product design can be tweaked to emphasize more of the machine-run processes, and to simplify the remaining hand-work.

The Lebowski motor is coreless, so the coils can be easily machine-wound. A hairpin configuration dramatically simplifies motor construction. Segmented cores still use laminated metal sheets stacked to make the core, but if you segment them, they can be machine-wound, and then assemble the wound segments into a stator, with the laminations laser-cut locally.

I don't know which method would be "best", or at least the "most doable". One design might be cheaper per output watt, another might be more efficient or more powerful per output watt. A third might be "average" in every respect, except that it has the least amount of hand-labor involved, which...if USA-made, could be a deciding factor in order to make the motor affordable enough that customers would actually buy it.

Regardless of the method, I can order the parts, and then assemble plus test ten of them on a Saturday. However, I don't want to start down that road unless the design is expandable in a way that they can be affordably made on a larger scale by employees.
 
I have also been going down this road. There is a design option to waterjet the plates, and not wind the coils at all.
I will drop you a message about it as soon as I can Ron.

Sam
 
This is interesting.
Transmagnetics (I think) had already developped motors with individual interlocking poles. Hand-winding those would be quite easy.
Many companies in the USA specialize in cutting laminations, so that is not a problem.
I think (for a mid-drive unit) it would be relatively easy to design a series of motors that could be hand-assembled by the customer himself.
A kind of Kit for a certain diameter of motor, that you can build to the length you want, therefore giving you a choice of power level.
Desired kv would be attained by using appropriate # of turns & wire size.
Magnets on the rotor could be standard NdFeB that you epoxy end-to-end to fill the length of your rotor.
Then wrap the outside of the rotor with a couple layers of fiberglas or kevlar to retain the magnets.
Two machined end caps with bearings, and a cylindrical body cut to the length corresponding to the length of rotor you want to use.
That could be easily put in production and sold at reasonable price.
Have you seen the price of a (small & simple) Astro 3210 motor? Outrageous!
Here is the Transmagnetics:
 

Attachments

  • 7.png
    7.png
    129 KB · Views: 4,348
I'd love to be able to say that I thought up the idea of stator segments, but I also read about the Transmagnetics a long time ago. It could be poorly done, and also well done. Tiny motors have some benefit in their torque-to volume ratio by being an outrunner, but in the sizes I want...I believe the cooling options for an inrunner are well worth the effort to use that configuration. (GNG, LightningRods, etc)

Also, the Lebowski design loses one benefit by having no cores, but it is easier to manufacture, plus no cogging (axial, dual stator, triple rotor). Even when made as large as would fit, it would not be as power-dense as the steel-cored stators we are used to, but...maybe powerful enough? The axial motenergy and Agni motors are pretty powerful, but also large and steel-cored.
 
As high power is relatively easier to obtain by the use of a high-RPM motor, coupled to a high reduction ratio gearbox, the addition of a simple fan on the shaft might be quite efficient as a way to cool the motor. As I recall, a propeller's thrust is proportional to the square of the RPM, so even with a small O.D. motor, it would be possible to push a LOT of air in there, if the RPM is high.
I am presently building a bike with the BHT motor, which is similar to the LR Big Block. It is a nice motor, but still heavy.
An Astro with a gearbox would be lighter and more powerful.
But an Astro clone might be impossible to build with interlocking poles for simple winding, but something of around 4" in diameter, between the Astro and the LR Small Block might be attainable...

I agree that the axial flux design is the simplest, but it suffers greatly in the power density department.

A guy from Transmagnetics has been active on ES in the past when he had a small batch of surplus motors to sell.
I wonder if he was a designer for their motors, or a sales person.
It might be interesting to hear the opinion of a good designer regarding the feasibility of manufacturing such a motor.

(Edit: Changed Small Block to Big Block)
 
The motor configuration that's "best" to manufacture locally may depend on what resources are available nearby, rather than settling on one design for all users in North America. Cost vs performance will always be an issue regardless of application and location of manufacture. In fact if some prototypes were made and tested, one configuration might be best for motors on the larger end of the scale, and another for the smaller motors.

However, I'd definitely want at least 1200W (if not more), to make it worth even trying. Clearly a non-hub. Anything below 2000W I'd want the motor to have the use of some gears (at least three, if not seven). Above 2600W, a 2-speed or a single run of chain to the rear wheel. Of course the broadest application would be the 750W-1500W range.
 
I completely agree with the minimum power of 1200-1500W. And also non-hub.
And indeed, a single gear ratio is sufficient if the motor is strong enough.

Now, would there be an interest for such a motor?
Nowadays, I think that people prefer turn-key, and not many of them will buy a motor and then fabricate the rest.
You would have to supply a complete package, that can be easily bolted to a bike.
 
If you are going to build a multi-part stator that bolts together, or is otherwise assembled in parts when installing. Than you should look into different stator designs. If you used curved magnets, and mushroom shaped stator poles, you would have more magnet surface area, and more magnetic force. Increased torque for simply curving the magnet. Not a huge amount mind you, but a 50mm magnet with a 90 degree arc is akin to using a 55mm magnet, 120 degree arc is equal to a 60mm magnet, 180 degree arc is equal to a 78mm magnet.

With a 40mm width, and a 180 degree arc, you would have a 68mm flat bar magnet equivalent.

Just something to think about.
 
I understand your point, atarijedi. Thanks.
I am no motor designer, I assure you, and I do not plan on becoming one, I simply do not have time at the moment. Many people on this forum already have advanced knowledge of motor desing, anyway.

One thing that is important regarding magnet shape, is that the waveshape of the back EMF should ideally be close to sinusoidal (as opposed to trapezoidal) for optimum performance with the recent FOC-capable controllers. I therefore assume that a constant gap between magnet & pole, across the whole width of the pole, might not be desirable. But that's just a guess. Someone who is proficient on the subject should chime in and correct me. Please.
 
BLDCs technically aren't sinusoidal designs, they are trapezoidal by their design. PMSMs (aka BLAC) are sinusoidal. Really it all comes down to whether or not you use overlapping stator coils or not. BLDCs do not use them, PMSMs do use them.

You can take any BLDC, hook it up to an oscilloscope, give the axle a spin, and you'll see a trapezoidal waveform, regardless of whether or not you then go on to feed it a trap or sine wave. Back EMF generated, with sinusoidal commutation, will still be a trapezoidal shaped signal. The only way to get a sinusoidal back emf is to have the coils arranged in a sinusoidal pattern.

For instance, when you look at the stator of a BLDC, the coils wrap around a single pole, but in a BLAC, like in an induction motor, the coils will wrap around the side of 1 pole and the other side of a pole 3 spots away. I don't know much about BLAC design though. I only ever barely touched on it in College.

I guess I should have also mentioned that this idea, of curved magnets, would be better for hub motors, but it will still work with non-hubs.
 
I've noticed overlapping coil-wraps in motors that were disassembled before. What are the benefits and drawbacks of that style? compared the common "GNG style".

I'd like a topography that is slightly scalable to being a little bigger and a little smaller for models V2 and V3, but... for me? I have narrowed the first version down to:

Non-hub

in-runner

Small enough for efficiently providing 750W at 36V, and big enough for 1200W at 52V (20A to 25A).

I would compare new designs against a MAC with the gears removed, and used as a non-hub (the stock MAC is an outrunner). If I had to copy only an existing design, the Transmagnetics holds my current top choice.
 
A true sinusoidal motor driven with a true sinusoidal controller will have constant torque. We can get close with BLDCs and a sinusoidal controller like Adaptto/Sabvoton/Kelly, but there is still torque ripple and inefficiency because the motor is designed in a trapezoidal fashion, regardless of any smoothing done by the inductance in the stator coils.

I've never designed an actual motor before, I've only done small exercises at different times throughout a semester, involving discrete parts within a motor. Maybe I'll pitch my curved magnet/stator idea to Transmagnetics and see what they say.
 
Atarijedi, I'm very pleased to learn that you ARE knowledgeable about those motors.
Big thanks for the heads up on the trapezoidal/ sinusoidal bemf waveforms.
I'll be very interested in any eventual exchange between you and Transmagnetics.
This thread is getting very interesting. Hoping others will join in.
 
For my first e-bike, last year I modified a MAC motor to use it in a mid-drive configuration.
I ran it with the Adaptto all of last winter. However, the particular application of a fatbike running in snow, on hilly terrain, made the little MAC quite thermally overwhelmed, all too often. Not to mention its lack of power, although I had an additional reduction ratio of 3:1 going to the rear wheel. Also, when the MAC is used as a mid-drive, its internal gear ratio of 5:1 becomes an effective 6:1.
That's why I bought a BHT motor to replace it for this coming season. But the Adaptto can't drive it. (I would really like to understand why the Adaptto can drive only hub motors. Is it because the coil inductance of mid-drive motors is too different?)
Anyway, a bit off-topic for this thread. Here lies my quest for a good motor.
 

Attachments

  • P1060454.JPG
    P1060454.JPG
    211.6 KB · Views: 4,257
I am most interested in a motor design that is "produce-able", rather than cutting edge. That being said, if a small design tweak will result in a dramatic improvement, I am eager to understand the benefits/drawbacks...

Also, my interests aside, this thread is free to discuss two or three designs, and at any point...if a certain design shows great promise, it can be split off to allow a more focused examination. I believe this discussion will benefit North America, but...it will also benefit any country that desires ebike motors, and is NOT China (EU, S. America, Africa, etc...).
 
Altair said:
For my first e-bike, last year I modified a MAC motor to use it in a mid-drive configuration.
I ran it with the Adaptto all of last winter. However, the particular application of a fatbike running in snow, on hilly terrain, made the little MAC quite thermally overwhelmed, all too often. Not to mention its lack of power, although I had an additional reduction ratio of 3:1 going to the rear wheel. Also, when the MAC is used as a mid-drive, its internal gear ratio of 5:1 becomes an effective 6:1.
That's why I bought a BHT motor to replace it for this coming season. But the Adaptto can't drive it. (I would really like to understand why the Adaptto can drive only hub motors. Is it because the coil inductance of mid-drive motors is too different?)
Anyway, a bit off-topic for this thread. Here lies my quest for a good motor.

Inductance shouldn't be an issue with the Adaptto, since it is a sine wave controller. Inductance only becomes an issue when you are using PWM control as the stator coils in the motor act like a low pass filter and smooth out the signal you are feeding the motor. It's called integration.

Here is an example of ideal PWM integration, in won't actually be anywhere near as nice as this sine wave, in fact it might look closer to a triangle or shark fin, but it gets the idea across

1107InterfaceConnectionFig4.jpg


So the lower the inductance, the less smoothing that happens, and you get current ripple, current spikes, etc.. It becomes hard on the controller and the motor as heating issues will develop.

That said, I've never heard of the BHT before, don't know any of the specs, and I don't know the internals of the Adaptto. You'd think it should be able to control it, does the BHT have hall sensors? I don't think the Adaptto can do sensorless.
 
The BHT is sinilar to LightningRod's Big Block. It does have Hall sensors.
https://endless-sphere.com/forums/viewtopic.php?f=28&t=57483
In fact, the Adaptto does drive it, like other mid-drives, but roughly. It impossible to get a good control at low throttle, and also at high speed.
The Adaptto also has a sensorless mode.

That size of motor would be a good candidate for what we are aiming at, in this thread.
 
I got nothing to contribute with in this thread when it comes to motor design or technical aspects, but I will subscribe none the less so that I might learn something about motor design. Seems like a great idea to open source a motor design. And it would be good to see alternatives to the far east motors we usually enjoy.

The one thing I don't grasp fully here is what is the overall design goal? If certain choices are to be made, will quality trump cost? I mean I would be paying extra (within reason of course) for added quality. Is there are a way to go the extra mile on the design? Use better quality parts? Higher quality copper, better magnets, thinner laminations, better and more robust bearings, etc or is the idea to make motor a cheap as possible? Who are the targeted buyer? Any bloke on the internet reading about e-bikes or dedicated current heads that spends too much time on the ES? Who is the targeted buyer will likely impact the willingness to shell out the $$. Heck I would even pay extra just to use an open source motor designed from the ground up within the community. But would my brother in law that knows e-bike only via me?

I think these are questions that needs to be answered to set a goal, and a baseline for the project. Most likely we can't beat the prices from China. But could we have a better motor design? Use better parts or do anything that will set this design apart from other design?

In a DD hub motor as I am most familiar with them, there are a few points that I can think of that could have been done better. Curved magnets already been mentioned. But there is room for lighter design in regards to back iron between flanges, lighter and stronger axles by using lighter material (ie Ti) How does that translate into a mid drive motor? I remember the latest motors for Zero motorcycles used a clever design for magnets, like a pocket for magnets rather then relay on adhesive. What benefits does such an approach hold over traditional designed motors?

As there are so many different manufacturers of motors I think most of the low hanging fruits are already taken, to improve on performance, efficiency, or cooling for a new motor design I think we got to be Ok with the price increase over china motors.
 
As long as China continues to crank out inexpensive motors, there will be no affordable domestic production of motors (of any type). However, if there is an interruption of the motor supply due to a trade war (or any other reason), then what are the options for people who want to jump on the ebike bandwagon?

I just don't see domestically-made hubmotors being a quick fix. Now...for enough money, anything can be made, but...a non hub is much more viable (IMHO). The question is...what is the most "doable" non hub motor? A "best case scenario" design can be scaled up for more power, or scaled down for the efficiency crowd. But, the initial design must have reasonable performance, coupled with reasonable efficiency, with the biggest influence being that...it is something that can be actually produced.

48V X 25A = 1200W seems to be a popular middle ground. A low kV can have a single chain to the rear wheel, and a higher kV can add a jackshaft to make the system more power-dense. Cooling mods could boost the amps to 35A (or more), which I feel is not unreasonable, before going to a V2 larger motor.

"Let not the perfect, be the enemy of the good". There will always be upgrades available for those who wish to have some part be better in some way. A non-hub has no issue trying to squeeze thicker phase wires through a hollow axle, so...having a much thicker phase wire cable is easy and cheap.

Thinner laminations? The cost is maybe $10 per motor in order to have lams so thin that users can upgrade to a higher voltage with no eddy current heating issues. Thin 0.35mm lams are a requirement from the beginning.

Bearings. Larger are more expensive, smaller has less inherent friction. Non-hub motor shafts (axles?) don't take the brunt of jumps (like a hubmotor), so axles don't need to be super strong or super thick. 12mm, 1/2-inch (12.7mm), and 15mm are the common choices that would avail us of mass-produced pulleys and sprockets...along with a wide variety of bearing choices.

Anyone who reads this thread is free to order a batch of motors in any configuration they like (for whatever reason). However....I still believe the main choices are:

Inrunner, for inherent heat-shedding, and the ability to easily add cooling mods later.

Diameter? the smallest that will fit 1200W, to enhance torque-per-watt, and fit-ability. Outer shell can be an aluminum tube-section.

Side plates should be flat aluminum plate, easy to water-jet/laser-cut. The shape can be a larger diameter than the perimeter shell, so cooling mods can be easily enshrouded (air/liquid).

Due to the end-turns of copper wire on the stator, the rotor will be narrower than the widest part of the stator. The shaft bearings can be mounted externally for ease of construction, and easy maintenance. Diameter of shaft should avail builders of a common size of bearing. This way, builders have a choice of cheapest available, or upgraded ceramic/etc...shielded/sealed/etc.

How wide? Wider is better for torque, narrower is better for fit-ability once a pulley-belt is attached to the shaft (chain output is narrower, maybe provide a stator in two or more widths?). Shaft should extend out both sides. Of course motor is reversable, but non-used side is easily cut off (for turn-key motor). Maybe builder wants unusual configuration using both sides? Shaft should be order-able in a variety of lengths for non-assembled kit.

Maybe parts kit should have wound stator parts available in several widths? Customer orders shaft width according to their needs, and side-plates fit any width of motor?
 
Correct me if I am wrong but the Astro motors are made in the US? Sure they are not pennies on the dollar but then again they do deliver long lifespand high quality motors that really can take abuse even for e-bike usage. There are a few others as well that are not made in china, typical for all of these are price in not in any way their selling point. Nor should it be as it is impossible to compete with chinese manual labor costs.

So I see two options. Either design a motor that can be machined and milled or even robot assembled to avoid costly man hours doing grunt work. Maybe not viable for entry level motors, but could be worth considering for mid to high power high quality motors. Ie use copper bars rather then copper strands. Simplify where possible and avoid as much manual labor intensive work as possible. Initial costs will likely run high, setting up production lines, or even robotics. Probably not financially sound unless aimed to sell bulk. And is the market ripe yet?

They other is to in fact do the manual labor like most motor manufacturers. Here simplified design and assembly will go a long way. But as long as there is no Chinese hands doing the assembly price will not match the chinese motors. So here one would be targeting high quality parts, high quality assembly and a very strict quality control. How much would you pay for a better motor? Iirc CL had a motor priced at > 1.000 $ for a while. Then prices fell by 40% or so. So guessing demand for > 1.000 $ hub was not really there.

So neither one of these options pins out, one has to high initial costs and the other gets a too high cost end product. Is there a way to still make it work?
Well if we can't do anything about initial costs or end product costs we need to think volume. Large volume. More volume then e-bike alone. Like industrial volume. Multi usage of the final product. And SM you made agreat point earlier in the thread about modularity. Making a design that can scale.
Need more then 10kw? Sure no problem. Just make it longer, or larger diameter. Still not large enough? Use multiple motors.

But the headache does not stop there. Are the usable controllers that will work with the motor design? Controllers that are available and tunable?
Or should the motor be developed from the ground up with a special controller design or controller type in mind.

Very exciting thread and I would love to see this open source motor come to life.
 
Yeah, Astro's are definitely made in US. They are small enough that they require a significant reduction. I suppose if this motor can't be made cheaper than an Astro, then its not worth producing, and this is all just a technical exercise.

Maybe the remaining design choices should focus on low cost?
 
No I don't think this is just a technical exercise. I think a high quality product can still be made in the western world if there is enough things that set end product apart for the run of the mill similar products from similar manufacturers. For an inrunner that will be anchored to the frame weight is not the end of the world, sure if motor turns up really light nice. But weight savings for the sake of chasing the last gram might not be worth it as motor will be sprung weight. So we can accept chubby, what can set it apart from competitors products? Better magnets, both shape and type of magnets. Some has higher "melt down" point the others. What about adhesive? Possible to use one that can withstand more heat? Sensors? Could optical sensors outperform hall sensors? Less prone to failure maybe? What about casing/motor shell. Is there other materials that are cheaper, easier to handle, require less machining?

Well there have been a few Astro builds here on ES despite that those motors cost a pretty penny.
I think the 3220 is around 700$ or so shipped from Astro, a little more from Matt as he does larger axle and a fan add on. As long as people feel they get their moneys worth I think they accept to pay a little extra for premium products. Cromotor was not cheap when first released, been a few of those around too.

Would it be possible to avoid using copper strands winding for an inrunner? What kind of smart design can help reducing the assembly time, and hands needed to complete a motor? The fewer parts the faster 8) To be serious one thing that could bring costs down is to contact a manufacturer that already got an assembly line running. Maybe they can do assembly in between their own runs. Don't even need to be a motor manufacturer as long as the hands on deck are willing to learn.

Same for parts, it could be savings in having various sub contractors for parts. Companies that already got equipment has already taken the initial costs and could likely do parts cheaper then starting from scratch.
 
If you do a teardown costing on a motor the difference between western and eastern/low cost production is the labour for the windings.
There is a way to remove the cost of winding the coils by hand and make it cheap to wind up in western countries.
The secret lies with fellow wind turbine designer Sam Chamas' very clever serpentine design for his VAWT wind turbines. You just switch the design to axial flux and wind the large loops by hand and kink it in.
No more paying peanut wages to radial wind up motors.

The way to do this is to make the design axial flux based Ron and make a simple serpentine coil jig on a table. And to waterjet the side plates, magnet holders and bearing seats. Or use cheap bearing blocks.
Watch the full movie: https://www.youtube.com/watch?v=RavLRtywIhI
Here's an indication of Sam Chamas's design. If anyone cross posts this please credit Sam with the idea.
Those dowels could be on screwthread runners like a drill chuck.
Capture.JPG

I would be interested in waterjetting up a few pieces of an axial design and distributing to a few builders around the globe during October. We could group buy some magnets and standard bearings. I also have my cheap 12fet 3077 mosfet based sine controllers that might suit - same manufacturer as the powervelocity one. I would donate a few for trialing. Might be a nice match. Maybe we even run it twin three phase like John in CR's much bigger motor.
What do you say hive minds?
 
Sam, as far as assembling a DIY motor, that axial-flux with the wave windings seems brilliant. I recall Lebowski used two stators and three rotors on his axial, so I imagine we could do that also for this reversed VAWT 3-phase generator design? Is there any way to guess at the power and kV of a motor of that style, if we choose a diameter, number of magnets, size and shape of magnets, diameter of wire and number of wraps, plus any other parameter needed?

Twisting it around into the wavy shape would be very easy if a jig was made where half the blocks are stationary, and the other half all move from an outer perimeter towards the center at the same time (via a linkage under to jig). Very nice that each of the three windings only have two wire ends each...hard to screw that up.
 
Glad you are aligned Ron. There is a small drop in performance from serpentine winding but the cost reduction is key.

Sam seems to stagger the windings in order to reduce turning resistance and cogging. Much in common with bikes.

There are a bunch of axial motors about we could use as a guide for power per unit of size/volume.

But I am thinking it might also be useful to start a target power level of say 2 to 4kw zone and consider a controller choice as well as a target cost. Maybe a standard one inch shaft to keep the architecture of the bearing etc simple between metric and imperial? The vawt guys use wheel bearings for simple mounts. I notice a lot of cheap plummer blocks online these days.

And maybe start hunting some decent neodymiums in a commodity size. Maybe 50mm again for unit simplicity.


Sent from my iPhone using Tapatalk
 
Back
Top