My ES torque contest DIY motor build thread.

[John in CR]

LFP,

Regarding the winding difficulty, why not wind each coil as 2 pieces and join them in the center afterward to end up with both ends of your coil at the perimeter?

 

[Arlo1]

I will be just under 10khz coil freq at 100% duty cycle at 28,000rpm. But, just for kicks, lets see if we can get a ball-park guess on if the rotor will explode before 28,000rpm.

Diameter of magnet ring mass center ~0.148m
Velocity at mass center of magnet assembly at 28,000rpm: 216m/s
Total magnet mass = 0.54Kg.

This would put 176,600N of internal stress on my 6mm thick carbon fiber rotors. That's 18,020Kgf between the pair, or 9,010Kgf on a single rotor, and 429Kgf per 6.35mm width magnet array. If my weave of carbon fiber/epoxy has a rough tensile strength of 1034Mpa,this would mean at 6mmx6.35mm magnet supporting strip would handle a pretty shocking 4,019Kgf in straight tensile loading. But this wouldn't be straight tensile loading... that would assume the an infinite strength and modulus of the magnet mounting epoxy that attaches the magnets into the slots of the rotors, and it doesn't compensate for each loaded section being modeled as a wedge, and it doesn't compensate for the very real balance related vibrations and dark forces at work on anything trying to spin that fast. I will probably to to keep under 15-20,000rpm in it's installed application, which due to the x^2 relationship between RPM and destructive forces, should keep things at 1/4-1/2 of the stress it would feel in the above calculation. Also because the gearing becomes a PITA for very high RPM stuff on a bicycle.

If you have a machinest handbook there is a termaly velocity calculation for things like this. I used it to calculate a safe rpm for my 2 inch thick 18 inch dia steal drum for my motorcycle dyno. This project excites me alot if I had you money luke I would win for sure lol but because I dont I will watch closely from the sidelines!

 

[John in CR]

Arlo,

If you would "win for sure" then how about a few of us pool resources and build it?

 

[liveforphysics]

The more competition the merrier Arlo.

 

[Arlo1]

Arlo,

If you would "win for sure" then how about a few of us pool resources and build it?

Son of a bitch lol i never thought I would get that type of a response! Hmmmm well a side from spreading my self to thin I love this type of a compatition! I have dreamed about getting a bunch of my motorcycle building budies togetther for a contest like this (was thinking about a 100 cc limit and certain fuel for a go kart but other wise balls out engine building) But I might have just found whats going to keep me up at night for a long time! Ok so were do I find all the info? And I will read it and decide how I will proceed.
My only hold up is lack of free time but I am working on getting the boss to cut me back to 4 days weeks and I am almost caught up for my buisness! The only hold up is I am starting to date a prety cool hotty who, you will like this luke, she doesnt belive in having a tv in the front room and has no conection to the out side world on her tv just dvds!

 

[Miles]

Ok so were do I find all the info? And I will read it and decide how I will proceed.

Here's the Challenge thread: http://endless-sphere.com/forums/viewtopic.php?f=28&t=14484

 

[RoughRider]

BUT...my question is: why there are NO motors with halbach principle on the market?

IS the halbach-principle patented? So, its not allowed to sell motors with this magnet-system?

 

[Miles]

BUT...my question is: why there are NO motors with halbach principle on the market?

http://www.launchpnt.com/capabilities/electric-motor-design.html Something like this may be available next year.

 

[John in CR]

BUT...my question is: why there are NO motors with halbach principle on the market?

IS the halbach-principle patented? So, its not allowed to sell motors with this magnet-system?

LFP has no easy task to build 40 something (I still don't think that 21 per rotor is correct) halbach arrays with such powerful magnets. Figuring out an efficient way to do it may be a patentable process.

 

[liveforphysics]

Yep. Its been a considerable PITA to just play with the halbach configuration.

I had to buy a magnetic field viewer sheet, and an electronic polarity tester (I think its just a hall on a stick with a couple LEDs, but none the less very handy). I gotta run through the magnets indivdually and make the arangements, then check them on the mag field viewer, and there are always 2 that often need to be rotated to get it right, and the whole time you're doing this, the magnets want to jump apart with a lot of force and stick together in any configuration except the one you want. Lol

Its not very easy to setup.

John, this motor should be able to run submerged in the ocean. No iron in here, carbon shafts over carbon shafts, with ceramic bearings.

 

[liveforphysics]

Teaser:

 

[John in CR]

Eeek, square magnet's! I think I've got a fairly simple way to make the Halbach's using square in 2 dimension rectangles, but not for squares. I'm seriously considering converting one of my hub motors to halbach arrays for the magnets to shed nearly 5lbs just in the thick steel retaining ring. I found the right mag size to do it for about $100, but stronger mags should net me a more powerful, and lighter motor.

 

[MitchJi]

Hi,

I have two questions. It seems like a lot of the design of Luke's and Todd's motors is based upon the LaunchPoint motor:
http://www.launchpnt.com/capabilities/electric-motor-design.html

LaunchPoint has completed optimization and initial prototyping for a high efficiency, high power density, Halbach array electric motor. With a total weight of only 1.4 pounds, this motor produces 7 horsepower at 8400 rpm with 95 percent efficiency. At 5 horsepower per pound, this motor has a higher power density than any other motor on the market. The brushless, axial flux permanent magnet design is highly tolerant of temperature and centrifugal effects, and the ironless rotor and stator eliminate iron eddy current and hysteresis losses.

That's really great. 3.5 more power per pound than an Astro 3220 (4 pounds, 5.6 hp) and if both the Astro and LaunchPoint spec's are accurate its 2% more efficient. Its obvious that a 1.4 lb motor with 7hp at 8400 rpm is a lot better than a 4 lb motor with 5.6 hp at 7500 rpm. OTOH an Astro 3220 has more than sufficient power or torque per pound for most ebike users.

What excited me about Thuds and Luke's motors is that Todd mentioned a kv in the 20's and Luke stated (I assume the LaunchPoint motor also beats the Astro for torque as the minimum to qualify for this contest is almost 50% more than the 3220) :

This is why our contest is about specific torque. I'm shooting for my motor to have a KV in the 30's. It will be as near to silent as any air cooled motor can be, and it will work great for driving the wheel of a bicycle with a simple single stage reduction.

If we can get a motor with the same power and weight as an Astro 3220 with a kv in the 20's or 30's IMO that's the perfect Ebike motor. No complex reduction required and with a simple single stage chain reduction it will be as quiet or quieter than one of Matt's drives with a belt first stage.

The LaunchPoint power is specified at 8400 rpm and the Astro is at 7500 rpm. How do Todd and Luke plan to build motors based on that design, with comparable power without upping the voltage, with kv's in the 20's or 30's?

Does anyone have any idea how much difference the LaunchPoint use of custom pie shaped magnets makes as compared to Todd's round magnets?

 

[John in CR]

Mitch,

Hopefully those of us on the sidelines will be able to gain some insight into those answers. We can see that LFP's approach is to use optimum materials in order to lose weight, along with Halbach arrays to avoid even the weight of a steel backing ring for the mags. From what I've read building a motor that works is relatively simple, but a great one is in a whole different league. As far as determining the KV ahead of time of a coreless motor, my searches have come up blank.

I do think that the ideal bike motor will be a larger diameter than the RC stuff in order to give us the low end torque we need. The motors in the geared hub motors come pretty close, which are essentially short outrunners with planetary reductions to wheel speed in the 4-5:1 gearing range, so I have no doubt that high power, low Kv, below 10lb motors are possible. Because of their pancake shape, they can be mounted outside of the hub between the seat and chain stays. That leaves how to ventilate them without allowing dirt and debris into the motor, and a housing separate from the actual motor can easily accomplish that protection with very little weight penalty.

John

 

[MitchJi]

Hi John,

I need to clarify my question. Its not actually obtaining a low kv that's the problem. A 24 turn 3220 has a kv of 28. The problem is its rated power of 5.6hp is at 7,500 rpm which requires 288v! This doesn't buy us much if we want a low rpm (less reduction required motor) with good power.

So my question is if Todd and Luke create a motor with a kv in the 20's or 30's will it still have good power output (something anywhere close to the 5hp per pound figure)? Is this as simple as increasing the diameter of the motor? If so by how much?

…With a total weight of only 1.4 pounds, this motor produces 7 horsepower at 8400 rpm with 95 percent efficiency. At 5 horsepower per pound, this motor has a higher power density than any other motor on the market…

…the current design is only 6 inches in diameter and less than 1 inch thick, this high power density motor design can be scaled and adapted for larger applications…

If so am I correct in thinking the diameter of the motor would need to double to 12" to get the same power at half the rpm? The same (or approximately the same) power at about 1/4th the rpm would be sweet.

I think LFP's approach on materials is a feature allowed by the LaunchPoint design and the same approach used in their prototype (from their pdf):

Ironless rotor and stator
Avoids eddy current and hysteresis losses
Iron is heavy and not a very good structural material

Lightweight
Carbon fiber or aluminum can be used in place of iron for structure
The motor is magnet, copper, carbon fiber – each ideal for its purpose

If their claims are accurate (seems reasonable) ventilation shouldn't be a problem:

Large surface area for cooling
Power not limited by thermal issues, in most cases

I don't think a motor under 10 pounds is a problem. I'd like something under about 4.

 

[Arlo1]

Teaser:

Im a little lost? Where is the magnet wire?

 

[amberwolf]

See those copper sheets? ?)

 

[Arlo1]

See those copper sheets? ?)

I do but... realy? I got to see what he is thinking.

 

[amberwolf]

Well, I forgot what he said (if he did) but if those are copper PCB, it's probably going to be etched coils. If it's just copper sheet, it's probably going to be cut into spirals, slightly expanded to make it a "wire coil" and layered together in sandwiches with the CF.

Exactly what it is we probably won't find out till he's done. He likes to tease way too much.

 

[liveforphysics]

Well, I forgot what he said (if he did) but if those are copper PCB, it's probably going to be etched coils. If it's just copper sheet, it's probably going to be cut into spirals, slightly expanded to make it a "wire coil" and layered together in sandwiches with the CF.

Exactly what it is we probably won't find out till he's done. He likes to tease way too much.

You are definately on the right track there AW. At least that WAS part of my plans... Now BigMoose has got my convinced that at my frequency range, the reduction of eddy current losses will far out weight the advantages of more complete copper fill that could be gained by using the flat/solid wire methods that I was planning... So now I'm browseing litz wire. lol

 

[Miles]

................So now I'm browseing litz wire. lol

:wink:

 

[liveforphysics]

If I want to have a PWM frequency in the controller that offers just 5 blocks of commutation resolution at max RPM and not take a signifigant skin or eddy loss, I shouldn't have any conductor strand diameter over 30-33awg, and all the strands making up the wire need to be electricly isolated from each other.

That is bad news bears for copper fill %, and SEEMS like bad news for coil resistance, but according to the stuff I'm reading it will be LESS resistance at the PWM frequency the motor will be running with.

For copper at 10khz, you've only got the outside 0.66mm layer that is helping you to flow any power. The materials inside that 0.66mm layer is just causing eddy currents and coil heating and inefficiency.

A high RPM motor with this many poles will need to have a commutation freq around 25-50khz at the least to perform with some timing resolution in relation to rotor position (unless you're going to rely on it staying in sync always, then you would only need to run the commutation freq, but this seems unrealistic not to have at least some resolution).

This means you're closer to the 0.4-0.5mm range of useable copper thickness before going thicker just means more heat and higher resistance at these freqs.

This totally destroys my plans of using thick solid coil wires. Part of me is sad for wasting so much time experimenting with finding thick coil winding solutions, and part of me is just happy to have found out now rather than after potting thick material coils into a custom CF stator.

For the purpose of this competition, it would be optimal to simply do big thick solid coils and run at a low RPM with a low PWM freq, and Rac will be almost equal to Rdc at the low frequency, and you will have an advantage in lower winding resistance. But, this would exclude the motor from being able to have a high power density, which is very much important to me and it's ultimate application.

So many tough choices!

Thank you BigMoose for your help and guidence as always, and Happy Birthday my most brilliant friend!

-Luke

 

[liveforphysics]

Hmm... other wire impedence calculators are saying 22awg is the max I can go for 50khz... That would enable quite a bit better copper fill rather than paying the insulation thickness cost on litz made of 30-33awg wires...

This would mean I could go with something easily available like this:

http://cgi.ebay.com/LITZ-WIRE-TYPE-2-7X22X36AWG_W0QQitemZ220369306877QQcmdZViewItemQQptZLH_DefaultDomain_0?hash=item334f08c4fd


But it would put me at worse copper fill than Litz made from thicker stranded stuff. I found some type 7 square Litz and rectangle litz wire as well, but it had thick outside insulation on it, which seems to kind defeat the advantages of being square to begin with...

Anyone have 50ft or so of 7-12awg litz made from 22-30awg strands with ulta-thin high temp insulation laying around they want to sell?

 

[Miles]

If you use strip/ribbon at the optimal thickness, can you get a better fill factor? The thinnest Polyimide insulation available seems to be 64 microns.

 

[liveforphysics]

If you use strip/ribbon at the optimal thickness, can you get a better fill factor? The thinnest Polyimide insulation available seems to be 64 microns.

I will re-work the fill % numbers while now having the knowledge of the maximum conductor thickness and see what I come up with. It may very well work out better.

You can feel free to work them yourself and post them for me if you're feeling mathy. I've gotta get some stuff finished at work so I can try to get out of here on time at the moment. lol