Thanks John. That's useful info. It's certainly good to have plenty of headroom in the heat resistance of the magnets. We don't want to design for normal running at high temperatures, though. It's less efficient and high risk with such a small thermal mass....John in CR said:
Ultimate lightweight wheel-motor concept
- Thread starter speedmd
- Start date
I realised this problem just after posting this idea.
it would unscrew on regen. Even a non freewheel cassette would suffer this problem
locktite is surprisingly good but this idea need refining.
forget that. People will have problems if we go this route.
it will require careful installation and usage.
I started drawing the motor on the brake side.
to clear my chainstay the wheel will require a custom hub with heavy dishing this way.
I have a lathe and milling machine to make a custom center hub so I can make one
but its not practical for others that need a bolt on solution.
How much efficiency would I loose using cheaper magnet and steel ?
is there big savings to be made there ?
I would not mind loosing 2-3%, 90% is still efficient enough to me.
even 85% would be ok because heat managment will be much easier and range is way more dependent on throttle usage.
it s just that I am somewhat short on fundings for this project.
motor is bigger than it look. the high pole count and thin section mislead me.
it would unscrew on regen. Even a non freewheel cassette would suffer this problem
locktite is surprisingly good but this idea need refining.
forget that. People will have problems if we go this route.
it will require careful installation and usage.
I started drawing the motor on the brake side.
to clear my chainstay the wheel will require a custom hub with heavy dishing this way.
I have a lathe and milling machine to make a custom center hub so I can make one
but its not practical for others that need a bolt on solution.
How much efficiency would I loose using cheaper magnet and steel ?
is there big savings to be made there ?
I would not mind loosing 2-3%, 90% is still efficient enough to me.
even 85% would be ok because heat managment will be much easier and range is way more dependent on throttle usage.
it s just that I am somewhat short on fundings for this project.
motor is bigger than it look. the high pole count and thin section mislead me.
liveforphysics
100 TW
Going to 85% would be 3x the heat to reject than 95%.
This design works because it's efficient.
Fortunately, it wouldn't be nearly that bad of efficiency hit with lower cost materials. More of a weight impact and fairly small efficiency hit. Making the stator even twice the width won't buy you much for Pd though, maybe 10-15% to take a wild guess. You would gain ~20% heat from losing 1% efficiency on a 95% motor.
If you want to be economical and the attributes that make this motor special don't appeal to you, you could just buy a magic pie and save yourself tons of labor and hassles.
This design works because it's efficient.
Fortunately, it wouldn't be nearly that bad of efficiency hit with lower cost materials. More of a weight impact and fairly small efficiency hit. Making the stator even twice the width won't buy you much for Pd though, maybe 10-15% to take a wild guess. You would gain ~20% heat from losing 1% efficiency on a 95% motor.
If you want to be economical and the attributes that make this motor special don't appeal to you, you could just buy a magic pie and save yourself tons of labor and hassles.
speedmd
10 MW
Thanks miles for posting the three speed freewheel. Easy way to get free wheeling at the crank and with a freehub mounted motor we have a near zero drag motor when coasting the bike on slight down grades off throttle. With a standard freehub setup we will still drive the rotor when pedaling only while we are off throttle. This brings me to my next question I have been pondering.
Is there a way to electronically freewheel the motor by having it generate some of its own power when we are just pedaling the bike. Read a bit on reducing cogging, but more specific, by putting in say 50 watts of mechanical energy into it, can we generate and divert most of this (say 20- 40 watts) to spinning the motor freely though the controller when it is turned on. Would we be able to out smart a dumb motor to power itself once spinning to mostly eliminate its own drag. Worst case, we have the split free hub or second freewheel. Love the idea of some of the non cogging motor designs.
Is there a way to electronically freewheel the motor by having it generate some of its own power when we are just pedaling the bike. Read a bit on reducing cogging, but more specific, by putting in say 50 watts of mechanical energy into it, can we generate and divert most of this (say 20- 40 watts) to spinning the motor freely though the controller when it is turned on. Would we be able to out smart a dumb motor to power itself once spinning to mostly eliminate its own drag. Worst case, we have the split free hub or second freewheel. Love the idea of some of the non cogging motor designs.
I am not 100% sure I will make one. But if decide to make one, I dont mind using my lathe, milling, winding the stator and lacing a wheel.
It's more of a hobby project than a useful transportation device for now.
I already have a 2805 9C - 20'' wheel with 21S 2P lipo that is fun for me.
But this concept is so much better, I feel it's worth the effort.
regards
It's more of a hobby project than a useful transportation device for now.
I already have a 2805 9C - 20'' wheel with 21S 2P lipo that is fun for me.
But this concept is so much better, I feel it's worth the effort.
regards
let us make drawings for the parts with all dimensions, bearing fits etc. lets figure out approximately costs of every part fabrication and raw materials.
keep manufacturing as simple as it can be.
i have a friend who works in a company with mill / turn machines and he could fabricate some of the parts.
if we make a group buy for at least 10 sets of motor, it will hold down costs.
what do the others say?
keep manufacturing as simple as it can be.
i have a friend who works in a company with mill / turn machines and he could fabricate some of the parts.
if we make a group buy for at least 10 sets of motor, it will hold down costs.
what do the others say?
speedmd
10 MW
Hi madin88
Great you have a shop that can do small quantities. We are close to getting this going. Certain we can get ten to commit. For good fits I think we have to work out a few more details. Magnets, near exact stator dimensions for good case fit, torque arm and wire connections, etc. Miles, I am certain will want to at least look at the cogging and ripple a bit more. Controller choices / design. We also have not looked at any weight / power / heat trade offs in any depth. Great progress so far and I too am very excited to get something to play with, but still much to be decided to making a lasting design. I still am undecided on the rotor /stator bearing mount. Thinking now a good quality sprag bearing may be best for universal use so you are not stuck pedaling a direct drive system when the battery runs dry. Good solution on Regen is still to be found /decided also.
Great you have a shop that can do small quantities. We are close to getting this going. Certain we can get ten to commit. For good fits I think we have to work out a few more details. Magnets, near exact stator dimensions for good case fit, torque arm and wire connections, etc. Miles, I am certain will want to at least look at the cogging and ripple a bit more. Controller choices / design. We also have not looked at any weight / power / heat trade offs in any depth. Great progress so far and I too am very excited to get something to play with, but still much to be decided to making a lasting design. I still am undecided on the rotor /stator bearing mount. Thinking now a good quality sprag bearing may be best for universal use so you are not stuck pedaling a direct drive system when the battery runs dry. Good solution on Regen is still to be found /decided also.
Hi everybody,
Yesterday I played with Emetor and I was able to have 94% at 400 rpm which is close to 40km/h on a 20'' wheel. and use around 600W
The materials are more common and the active weight is 2100g
The only drawback is torque ripple but it is at high frequency so it maybe not a problem. Just a tiny bit of stator skewing should solve that.
For regen Justin did a good writing check this post : http://endless-sphere.com/forums/viewtopic.php?f=2&t=7891&hilit=regen+analysis
I attached the results. Before I get comments on BEMF please know that the better it matches the controller trapezoidal/square voltage the better.
So that is why the tops are flat. Also I kept the flux intentionally low to be able to use more regular materials.
I will continue mechanical integration drawings this weekend.
Yesterday I played with Emetor and I was able to have 94% at 400 rpm which is close to 40km/h on a 20'' wheel. and use around 600W
The materials are more common and the active weight is 2100g
The only drawback is torque ripple but it is at high frequency so it maybe not a problem. Just a tiny bit of stator skewing should solve that.
For regen Justin did a good writing check this post : http://endless-sphere.com/forums/viewtopic.php?f=2&t=7891&hilit=regen+analysis
I attached the results. Before I get comments on BEMF please know that the better it matches the controller trapezoidal/square voltage the better.
So that is why the tops are flat. Also I kept the flux intentionally low to be able to use more regular materials.
I will continue mechanical integration drawings this weekend.
liveforphysics
100 TW
Skewing the stator = Uber-fail.
It's easy to pop big efficiency numbers on a motor drive simulation. It doesn't count your harmonics against you yet, because they aren't against you until you go to power it with a real world controller.
The difficulty in making a useful and in-practice high efficiency and controllable motor is about designing out all the harmonic content and nailing perfectly sinus BEMF.
You can see Miles design did that well. If you skew a stator, you reduce torque ripple at the cost of ever being efficient or low harmonic. Think of harmonics as a paracitic that you don't see until you're actually powering the motor with a non - ideal - world controller like the simulator assumes.
It's easy to pop big efficiency numbers on a motor drive simulation. It doesn't count your harmonics against you yet, because they aren't against you until you go to power it with a real world controller.
The difficulty in making a useful and in-practice high efficiency and controllable motor is about designing out all the harmonic content and nailing perfectly sinus BEMF.
You can see Miles design did that well. If you skew a stator, you reduce torque ripple at the cost of ever being efficient or low harmonic. Think of harmonics as a paracitic that you don't see until you're actually powering the motor with a non - ideal - world controller like the simulator assumes.
First, I did not wanted to upset any one, I just find this thread interesting and wanted to discuss.
It's not too hard to change the design to have a more sinusoidal BEMF
I am trying to match BEMF and drive characteristic to take advantages of those harmonics.
Our bike drives are cheap on/off = almost square wave.
Neu Motors and Xera Motors both have high efficiencies with a trapezoidal BEMF which more or less matches the hobby drives. (24 slots 8 poles)
They get 94% eff theorical but real world is more around 92%
I took a couple of months trying to understand : Performance Permanent Magnet Motor Design. ISBN : 1-881855-15-5
Mr Hanselmann explain that we can take advantages of the way drives are built.
The more closely you can match both drive and motor the better torque and the less ripple you get. because the difference between BEMF and drive voltage is smaller
The simulation can use a square wave drive or sinusoidal. I choose the square.
There is plenty of motors that use skewed rotor or stator. Mostly to reduce cogging and ripple at the expense of the rest.
Yes this design, if it see light, wont be silent.
That said, I don't have experience designing motors I just have just selected and used many for my job.
Just keep me posted when ready to built then.
It's not too hard to change the design to have a more sinusoidal BEMF
I am trying to match BEMF and drive characteristic to take advantages of those harmonics.
Our bike drives are cheap on/off = almost square wave.
Neu Motors and Xera Motors both have high efficiencies with a trapezoidal BEMF which more or less matches the hobby drives. (24 slots 8 poles)
They get 94% eff theorical but real world is more around 92%
I took a couple of months trying to understand : Performance Permanent Magnet Motor Design. ISBN : 1-881855-15-5
Mr Hanselmann explain that we can take advantages of the way drives are built.
The more closely you can match both drive and motor the better torque and the less ripple you get. because the difference between BEMF and drive voltage is smaller
The simulation can use a square wave drive or sinusoidal. I choose the square.
There is plenty of motors that use skewed rotor or stator. Mostly to reduce cogging and ripple at the expense of the rest.
Yes this design, if it see light, wont be silent.
That said, I don't have experience designing motors I just have just selected and used many for my job.
Just keep me posted when ready to built then.
speedmd
10 MW
Hi Joe90
Glad to see your posts. Much still to work out and any help is welcome trying to air out the concept before we start cutting metal. Much to learn here on my part also, and look forward to where this leads. If we can get a better than 90% efficiency in a lightweight slip on design, I see this a potential mass use product. Combine this with a sensorless Lebowski type sine wave controller with regen and all basses are covered.
Glad to see your posts. Much still to work out and any help is welcome trying to air out the concept before we start cutting metal. Much to learn here on my part also, and look forward to where this leads. If we can get a better than 90% efficiency in a lightweight slip on design, I see this a potential mass use product. Combine this with a sensorless Lebowski type sine wave controller with regen and all basses are covered.
This is a good conversation to have, trap vs sinus
Controller switching losses will be lower on a trap controller, but I think motor losses should be lower if the motor is designed to be use with a sinus controller
Given the many other benefit of FOC control, I think this motor should definitely have sinus BEMF
It is only a matter of time before we move away from cheap trap controllers
soon there will be wide variety of Lewbowski and InstaspinFOC controllers available to us here on the sphere
once you try a buttery smooth sensorless FOC controller in torque throttle mode, you will never want to go back to speed throttle trap controller with hall sensing
the sinus controllers can also clip to simulate trap controllers
Controller switching losses will be lower on a trap controller, but I think motor losses should be lower if the motor is designed to be use with a sinus controller
Given the many other benefit of FOC control, I think this motor should definitely have sinus BEMF
It is only a matter of time before we move away from cheap trap controllers
soon there will be wide variety of Lewbowski and InstaspinFOC controllers available to us here on the sphere
once you try a buttery smooth sensorless FOC controller in torque throttle mode, you will never want to go back to speed throttle trap controller with hall sensing
the sinus controllers can also clip to simulate trap controllers
The lebowski brain with sensorless start is awesome. this would be the perfect controller, but as far as i know the current brain board only is very large and it will be hard to mount it stealth like on your bike. It more looks to me it is designed for motorcycle use or very big and powerful e-bikes.
It would be nice if some here on ES would make a all in one unit like the Adaptto controllers
By now, the Mini-E is the controller to choose for this motor. Its about 350$ but its worth every cent!
It would be nice if some here on ES would make a all in one unit like the Adaptto controllers
By now, the Mini-E is the controller to choose for this motor. Its about 350$ but its worth every cent!
Hi Joe,Joe90 said:
Not a bad idea to develop a version for the cheap ebike drives.
The stator material that I used for my simulation was Sura 27 :wink:
Yes, you're never going to get torque ripple below about 15% with 'DC'.
If you want to save money on materials, you need to reduce your magnet arc from 180 deg. Also, you can simulate stock rectangular magnets in Emetor by using the breadloaf template and setting the magnet radius to a high value.
96t 112p, single layer, is not ideal for rotor losses.
parabellum
1 MW
Vice verseMiles said:
