"Open source" motor?

[Fumesucker]

I was very interested in this post by axbor22 about the axial flux friction drive motor he built:

http://www.endless-sphere.com/forums/viewtopic.php?f=2&t=6762

A low rpm/high torque, high efficiency motor capable of driving the chainring with only a single gear, belt or chain reduction seems like a great idea to me.

Axbor22 says he used only a drill press (pillar drill) and a saber saw to make the motor, there are a lot of us here with that level of equipment.

I'd like to start a conversation about optimizing axbor22's motor design for highest efficiency, lowest rpm and highest torque.

 

[Fumesucker]

OK, I'll start off with a couple of ideas I have.

Rectangular magnets rather than round ones would be preferable I think, it would be possible to get a more even magnetic field that way with magnets just touching each other around the periphery of the armature. Trapezoidal shape magnets would be the best but they would have to be shaped from rectangular ones, this could be done fairly easily with something like a belt sander but keeping or getting the dust particles off the magnets after shaping would be a bit tricky. I'm guessing that it would be possible to pull off the dust particles with an even stronger magnet, does anyone know whether that would work?

Better copper "fill" in the coils, the coils should be nearly solid copper with the smallest possible hole in the middle of the windings.

Here is a quick and dirty drawing of what I have in mind. The square cornered trapezoids are the magnets with N and S pole alternating colors, the rounded corner trapezoids are the windings with the three phases different colors.

 

[PaulM]

It is possible to buy wedge shaped magnets that will form a perfect ring, but the rest of the motor has to be designed around this ring size.

http://www.forcefieldmagnets.com/catalog/product_info.php?cPath=23_37&products_id=92

They also sell parts for homemade wind turbines, which is basically identical to this type of motor.

Take the safety warnings seriously though, if they are allowed to snap together, count on at least one breaking!! Don't let them pinch fingers, and don't stick them to a metal surface unless you are able to slide them off an edge. I have a small 1/4x1/2x1 neo magnet on the fridge door that has a raised edge all around. If I stick it to the door using the 1/2x1 surface, it's all I can do to get it off!!

With regards to your other goals:

Lowest rpm: The only way is with more poles
Highest torque: Strong magnetic field -> big magnets, lots of turns on the coils with lots of current through them

Both of these above comments increase the size and weight of the motor, a big disadvantage for ebike applications.

Minimizing the air gap will help with the efficiency. I don't know about layering the coils, ie having three sets, one for each phase,each overlapped by a third of a coil. I think this would increase efficiency (or maybe just power density) but hopefully someone else can chime in on this.

 

[Fumesucker]

I used to be quite good with motors back in the days when electric RC planes were really just getting started, built a several two stage gearboxes, one with two motors, in order to get the highest power to weight ratio. Did quite a bit of rewinding and other motor mods for increased power and efficiency. Had to get out of RC because my vision just isn't good enough to see a plane well enough to control it properly at at any real distance. After rekitting a couple of fairly expensive planes thanks to poor vision I decided to turn my energies elsewhere.

Brushless is new to me though, I'm wondering if a motor such as I outlined would be compatible with commonly available brushless controllers, either RC or otherwise.

I have a fairly decent collection of neodymium magnets, a lot of them pulled from old hard drives, the problem is that they are mostly irregular shapes. I really had planned to buy magnets on ebay since I do a lot of business there and know that territory well and how to get a good deal.

Thanks for the link to the trapezoidal magnets, I knew I'd seen them before but couldn't find them when I went looking lately.

 

[Fumesucker]

One interesting point regarding axial flux motors is that you can vary the torque characteristics of the motor fairly easily by manipulating the air gap, narrow gap gives high torque, low rpm while a wider gap gives less torque but higher rpm.

A mechanism to allow the air gap to be changed while the motor is running would give you the equivalent of an infinitely variable transmission without having to have an actual transmission.

http://www.washingtonpost.com/wp-dyn/articles/A50140-2004Jun17_3.html

The axial flux motor is more like a pair of pancakes, with the rotor and stator shaped like disks and set next to one another. The gap between the two disks determines how powerfully the rotor turns, a force called torque. On a vehicle, you need a lot of torque to start rolling or to get up a hill, but you don't need much to coast along at a constant speed. That's why ordinary cars start in high-torque first gear and cruise in fifth.

Eventually, the engineers from GW figured out how to change the size of the gap while the motor was operating to respond to different demands. For high-torque acceleration, the coils are very close to the magnets on the rotor; for cruising speed, they back off. That helps the motor use energy efficiently, and allows a smaller motor to achieve a wider range of performance. Combined with a computerized controller, the motor has converted up to 95 percent of available energy into output, compared with about 85 percent efficiency for a typical electric motor and between 30 and 35 percent for a gasoline engine.

http://www.sae.org/technical/papers/2001-01-2493

http://www.sae.org/technical/papers/2002-01-1088

http://ieeexplore.ieee.org/Xplore/login.jsp?url=/iel5/25/28899/01300880.pdf?arnumber=1300880

Summary: High power density is the main advantage of the axial flux motors compared to the conventional radial flux motors. Therefore, they are suitable candidates for the power train of hybrid electric vehicles. In addition, operating speed range and efficiency of axial flux motors can be improved by changing the air gap of the machine.

 

[fechter]

Grinding or sanding Neodynimum magnets is very difficult. They are destroyed by heat also.
I cut a number of them using a dremel tool cutoff wheel and taking a loooong time. A wet diamond saw may be the best, but you need to lash the magnet to a substrate to keep the pieces from flying off. They'll stick like heck to a steel blade too.

Rectangular won't be bad at larger diameters.

Stronger magnets will give more volts/rpm
More turns will give more volts/rpm
Torque is proportional to amps x number of turns.
Flat spiral coils are good, but not necessary for good efficiency.
Multifilar windings reduce eddy current losses in the copper. Flattened wire does the same thing.

Making an efficient motor involves understanding where the losses are and trying to minimize them.

 

[Fumesucker]

Making an efficient motor involves understanding where the losses are and trying to minimize them.

Good point, that's why I'm asking for input, to identify where the losses are and decide how to minimize them.

After reading about variable torque in an axial flux motor via changing the air gap I think that it might be a better to forgo driving the pedals and drive the wheel directly through a simple chain or belt drive, changing the torque constant would give the same effect as driving through the gears and the overall complexity would probably be less (no crank freewheel needed for one thing). Changing the air gap could probably be done with something like the clutch actuation mechanism on a lot of motorcycles. Driving the wheel also allows for a lot faster turning motor than driving the crank, the wheel spins at somewhere around 500 rpm while the crank is more like 80.

At first I thought the magnetic attraction between the rotors would be overwhelmingly strong but after thinking about it for a while I realized that the rotors would be spaced far enough apart by the stator that the attraction might be able to be overcome with a relatively simple mechanism. I'm going to have to ponder this some more, Pinky.

I just tried grinding a neodym magnet on my belt sander, grind for five seconds and then dunk in water, it wasn't hard at all and I lost no magnet strength as far as I can tell. I don't have a grinding wheel on my bench grinder right now (I mostly use it for polishing brass, silver, aluminum and etc.) but I doubt it would be any more difficult on that. I collected a lot of the magnetic dust by putting a piece of steel in the path of the sparks given off and the dust stuck to it.

 

[Miles]

After reading about variable torque in an axial flux motor via changing the air gap I think that it might be a better to forgo driving the pedals and drive the wheel directly through a simple chain or belt drive, changing the torque constant would give the same effect as driving through the gears and the overall complexity would probably be less (no crank freewheel needed for one thing).

Varying the size of the gap wouldn't exactly be equivalent to using variable gears or a CVT...., certainly worth investigating, though.

 

[paultrafalgar]

At the risk of getting shouted at (I've said it before):
Why not stick the magnets to the rim of the wheel and "the other component of the motor" (electromagnets?) to a "brake caliper" (imagine it replacing the rubber of a caliper brake). You could then vary the "air gap" by operating the "brake lever"?
Please don't hit me!

 

[Mr. Mik]

At the risk of getting shouted at (I've said it before):
Why not stick the magnets to the rim of the wheel and "the other component of the motor" (electromagnets?) to a "brake caliper" (imagine it replacing the rubber of a caliper brake). You could then vary the "air gap" by operating the "brake lever"?
Please don't hit me!

I think you would multiple "brake calipers" around the whole rim, perfectly straight wheels, and a perfectly clean and dry riding environment.

 

[Fumesucker]

At the risk of getting shouted at (I've said it before):
Why not stick the magnets to the rim of the wheel and "the other component of the motor" (electromagnets?) to a "brake caliper" (imagine it replacing the rubber of a caliper brake). You could then vary the "air gap" by operating the "brake lever"?
Please don't hit me!

It's really quite a logical idea, I can see two problems though.

The first is that magnets on the rim are going to pick up ferrous debris from the road. The second problem is that the coils really need to be around the entire circumference of the rim to fully utilize the magnetic field of the magnets.

Another issue is unsprung weight, such an arrangement would probably end up weighing as much or more than a big hub motor, if you are building a bike with no rear suspension then it wouldn't be an issue but it will really louse up the handling on a suspended machine.

 

[ekline309]

A low rpm/high torque, high efficiency motor capable of driving the chainring with only a single gear, belt or chain reduction seems like a great idea to me.

Me too!!!

To me this seems like the only way build a bike because I want the option to remove the motor and batteries and have my normal bike at its core.

I envision a model where you clamp your low RPM brushless motor right above your freewheeling sprocket, pop on a longer chain, clamp on your bats, controller, and throttle, and you're off to the races!!!

Full Disclosure: I have never ridden a hub motor.

 

[Fumesucker]

I'm about to order some magnets for a proof of concept motor, I'm going to get 50 1/2 dia x 1/16 round magnets, I already have all the materials I need for the rest of the motor save for the wire.. I'm going to trim the sides of the round magnets a bit so as to make the magnetic field more dense and shrink the size of the rotor some.

I can have up to 24 magnet pairs on my motor, what I need to know is the best choice for number of magnet pairs versus number of coils..

Would 24 magnets and 21 coils be good?

I'm also wondering how an axial flux motor such as I'm about to build will work with an RC brushless controller, ISTR that RC controllers are looking for a more or less trapezoidal back emf and I doubt that the pancake motor will have that waveform..

 

[fechter]

Don't worry about the BEMF waveform. Just about anything will work with the controller. Axial or radial does not really change the BEMF waveform. With your setup, the waveform should be nearly trapezoidal.

There are a wide variety of valid magnet/coil combinations.
My BMC uses 16 magnets/18 coils. The Puma uses twice this (32/36).
24/21 might be OK, but I'm not sure.

Here's a publication that might be helpful:
View attachment brushless_motor_phase_pole_slot configurations.pdf

 

[lawsonuw]

Don't worry about the BEMF waveform. Just about anything will work with the controller. Axial or radial does not really change the BEMF waveform. With your setup, the waveform should be nearly trapezoidal.

+1 on this. I've taken a small 6-wire stepper and wired it so it looked mostly like a BLDC motor. Hooked it up to my chepo controller and it started and ran just fine.

Marty

 

[Fumesucker]

There are a wide variety of valid magnet/coil combinations.
My BMC uses 16 magnets/18 coils. The Puma uses twice this (32/36).
24/21 might be OK, but I'm not sure.

Here's a publication that might be helpful:

Thanks for that.. It looks like 24 magnets and 18 windings.

I found wedge shaped magnets for an 8" diameter disk on ebay but I don't want to spend that kind of money yet..

http://cgi.ebay.com/Wind-Genertor-Neodymium-Wdege-Magnets-22-5DEGx1-4-TH_W0QQitemZ170271873660QQihZ007QQcategoryZ121837QQtcZphotoQQcmdZViewItemQQ_trksidZp1742.m153.l1262

 

[Fumesucker]

After drawing it out I now understand why 24 magnets and 18 coils works.

Take a look at the pic below, the circles are the 24 magnets with N and S poles alternating while the wedge shapes represent the 18 coils with the three phases three different colors. Note how each color wedge has the same magnet orientation in it all the way around the motor, in this manner all the torques from six identical phases (18/3) add linearly.

I think what I'm going to do for experimentation purposes is just wind three coils, one for each phase and place them in three adjacent positions. Any measurements of torque or current I then take can be simply divided or multiplied by six to get a ballpark figure for what the entire winding filled motor will do.

I now have a couple more questions.

First, can someone point me to a resource that explains what the theoretical differences between a delta and a wye configuration are as to how the motor will operate.

Second, I'm thinking that a demagnetized ferrite magnet from an old speaker might make a decent core material for the windings, does anyone have any thoughts on this?

 

[fechter]

I've gotten quite a few magnets from this place:
http://stores.ebay.com/Emovendo_Magnets_W0QQcolZ2QQdirZQ2d1QQfsubZ5QQftidZ2QQtZkm

They don't have any wedge shaped ones, but a pretty good selection of discs and blocks in higher grades.

The ferrite used in speaker magnets is designed to retain magnetism, which would be bad for a coil core (huge hysteresis losses). You need the kind of ferrite that does not retain any magnetism after being exposed to a high flux. The stuff in switching transformers is more like it. Finding a disc might be hard, but you can find all kinds of torroids. A torroid won't be ideal, but might be pretty good. Use it as the core for a flat coil. There just won't be any flux in the center. You can also find ferrite in rod form, which could be sliced into discs with a cutoff wheel or diamond saw.

On delta vs. wye, both configurations can work well. Theoretically, you can get circulating currents in a delta configuration that result in loss. In practice, these losses are very small.

One thing on your pole configuration; you want to avoid having too many poles align at the same time or you will get a lot of cogging. Not sure how much this affects efficiency, but too much cogging results in torque ripple losses and can be annoying for other reasons. I think your configuration will be OK, but if there was another one with fewer aligned poles it might be better.

 

[johnrobholmes]

The higher the magnet/slot count ratio gets to one (without hitting one), the lower the cogging torque aka pole attraction.


Acceptable wind patterns and ratios in blue, for series wound. http://www.powercroco.de/Kombinationstabelle.html

 

[disadvantage]

I searched for "ferrite" at mouser.com, and discovered a company called Fair-Rite that sells ferrite bobbins:

http://www.fair-rite.com/cgibin/catalog.pgm?THEAPPL=Inductive%20Components&THEWHERE=Open%20Magnetic%20Circuit&THEPART=Bobbins
Click the Power Applications tab near the top.

Would these work for winding coils onto?

 

[fechter]

I searched for "ferrite" at mouser.com, and discovered a company called Fair-Rite that sells ferrite bobbins:

http://www.fair-rite.com/cgibin/catalog.pgm?THEAPPL=Inductive%20Components&THEWHERE=Open%20Magnetic%20Circuit&THEPART=Bobbins
Click the Power Applications tab near the top.

Would these work for winding coils onto?

Yep, those should be pretty good. Having a spool shape might be better than just a cylinder.

 

[Jeremy Harris]

Following that link to the neo magnet store gave me an idea, but I'm not sureif it would work.

Inside an auto alternator the rotor is an axially wound core, with iron pole pieces that form a skewed set of multiple magnetic pole pieces. The coil is usually just a simply wound bobbin in the middle.

If one of the iron pole pieces was slid off the shaft (I think these are normally just a press fit onto a serrated section of the shaft) then the coil could be removed and replaced with a stack of ring shaped neo magnets, as on that magnet store site. The iron pole pieces could then be slid back giving a PM rotor instead of a separately excited one.

My guess is that a decent sized alternator modified like this would make a reasonable big in-runner three phase brushless motor, but I've no idea how efficient it would be.

Jeremy

 

[Fumesucker]

Jeremy,

Here is a thread about converting an alternator into a brushless motor.. I do believe someone posted a link to a controller specifically made to energize the rotor as well as commutating the stator windings although I don't have time to look through the thread again at the moment. I read so much it's hard to keep track of what I've seen where..

http://www.rcgroups.com/forums/showthread.php?t=905411

Someone also mentioned that there are some motorcycle alternators with PM rotors that can be very easily converted into brushless motors.

 

[Fumesucker]

I pulled apart an old PC power supply and found a nice toroid that would be just the right size to take slices off to make cores.. 1.3 inches OD x 0.625 ID x 0.450 thick.. I have several old supplies in my junk room that should yield toroids of the same or similar size. The first coils I make are going to be coreless, I'm not sure that a core is even needed for this type of motor.

I ordered a Castle Creations Thunderbird 54 controller to experiment with, should be here by the beginning of next week, along with the magnets.

If the motor works well with the 1/16 thick magnets, I'll order some considerably thicker ones after I run some tests and get a handle on what the Kv is likely to be.

 

[Jeremy Harris]

Many thanks for that link, it looks like there's some really interesting stuff buried in it. That RC Group seems to have some pretty inventive members.

Jeremy