halbach arrays (was peak torque)

Sleep well...

Good morning! That's really useful, thanks! Looks doable, and very short endwindings.

I haven't found any thread with madacts motor project here, but there was one over at motoredbikes.

ES really good reference thread

Check out that page there, it has a pic showing rhomboid winding made from sheet, laid down on the pancake (perpendicular to axle).. Weird, because, if you take a disc of alu and spin it next to magnets, it will heat... not so with copper stator? It appears to be the same thing...

There is consistently more flux in my FEMM models for the 60/30 and 0/45/90 configurations (after I put enough poles in), but it isn't very big difference, it seems to me. For the e>0.98 people it must be huge though.
 
Here is a formula from a CSIRO paper I found. What is that last term there? Anyone got the Carter GW, 1967 “The electromagnetic field in its
engineering aspects” Longmans, 2nd ed., 254 and can check? Is it normally used for some sort of angular measurement?
 

Attachments

  • eddy.jpg
    eddy.jpg
    25.7 KB · Views: 1,579
<Yawn, slurp coffee, rub eyes>

Allow me to answer the one part my single active brain cell knows:

  • λ (Lambda) = coefficient of leakage permeance (or specific leakage permeance) as per the AF reference I previously have given.
  • Λ is the upper-case form of Lambda. I found a couple of references in regards to the coefficient of permeance that offer a couple of ways to view result. Many sources use the upper-case Lambda, while the first link uses Pc. Alas – with all things in scientific study involving multitudes of cooks the nomenclature often wanders from author to author.
    Regardless, I am glad you are spearheading this leakage investigation as it is overlooked in my process.
    Calculating Permeance Coefficient
    AF PM Wind Generator with A Soft Magnetic Composite Core
As to the shape of the winding, if you are overlapping then the shape is quite important to review and appreciate. On the flip-side, with no overlap – the winding should maximize the fill both in cross-section, and in the path shape… more or less the shape of the magnet perimeter with rounded corners where the up-down portion has the greatest tangential effect.

It might do for you to model both the middle and limits of the winding to better comprehend the interaction between the fields: Middle is implied; however it could be that the edges at the extreme of the winding produce less drag with particular shapes, which is naturally beneficial. Someone has a paper on that…

<more coffee, slurp…> KF
 
Did you find out who is going to make your magnets, or what grade (or who is cheapest)? If some more people were interested, maybe they would get cheaper. Not that anyone would agree on what size or shape, but still. Nblihe has something called M50, good for 100C.
 
Manufacturer: Biff recommended these guys http://www.nblihe.com/ which I think that is what you suggested :)

I haven’t dealt with them yet because I haven’t settled on the design. My plan is to use something close to 45SH; at 150°C that’s as low as I dare go temperature-wise. Wire will be rated at 200°C. This is after all AF and heat will be a problem, unless we’re running underwater :)

Production run: Interesting. What is the goal of your motor; is it for bicycle or motorcycle? Something else?

Did my previous post answer your question about the last term of the formula?

The sandman cometh soon, KF
 
Have to read and think about those whirls more, don't really understand them, but as far as the question goes I think you answered it.
Didn't think of a production run, just that maybe if 2 or three of us ordered magnets at the same time maybe they'd sell them cheaper.
Haven't finished designing either. It's for a terrain bike, I don't have rear shock yet so it's mostly slow technical stuff that is most fun to do with it. I do plan to run some water in it, a small amount in the bottom, which will distribute the heat to the walls, and boil off if it gets over 100c. Can't use air anyway, would be grit eating it up in not time.
 
Grit: Completely understood; once upon a time I considered venting that would open up as the heat index rose, sort of like how temperature relief values work; solid-state. However the fear though is what to do about low speed cooling, and for that we need to turn to a medium that is high-Q tolerant and has a high boiling point such as oil.

Canola has one of the highest-smoke points (240°C) of commonly available “natural” oil blends and it is also very light in grade which is what we want for wicking and small diameter through-holes. I don’t know the effect of canola upon PCBs should there be a leak.

Water on the other hand would boil at 100°C at STP – therefore it would require either a salt, glycol, and/or pressure to raise the boundary condition.

I envision that the cooling system would use a micro-pump activated by heat sensor autonomously, keeping the fluid moving even when the wheel is at rest, and directing the excess to external cooling fins made of a hyper-radiant clad-plate that is commonly available for heat exchangers. The fins on the front wheel would be on the opposite side of the disc brake (visually balanced) and approximately the same size. The rear fins would require a bit more creativity.

Alternatives to cooling hub motors, such as solid-state devices employed on CPUs and GPUs, are both expensive and consumptive; not the best use of technology.

Counter to cooling is an efficient motor design that can operate with low-power, possibly in tandem with a 2nd wheel or more to reduce the thermal load and allow for ambient radiance in most conditions. That is actually my first priority of the design; split the demand between two wheels. Reminds me of a famous quote…

“I could help a bit, I could carry it, share the load!”
Samwise Gamgee to Frodo Baggins, LotR – RotK

Nothing like a good quote to start the day! <slurp coffee> :D
Sharing, KF
 
(Slurps fresh ground beans coffee from italy)

Counter to cooling is an efficient motor design..(snip)
Click to expand...
But, moving up a 15 deg hill at 5kmh, with lots of rocks, the hubmotor would either have to have lots of active mass, or an enormous diameter, or it could be a lighter, very well cooled motor that ran at low efficiency.
A rim motor would probably suffer from more rotor-stator clearance, unless the stator had some sort of (semi)floating mount and there was a film of oil between it and the rotor (or there could be rollers to guide the rotor or stator).

I wouldn't really want a higher boiling point - with water, 100c would be the hard stop, if the delta T was not enough then, it would just boil off and go somewhere to cool down (external reservoir/bike frame/espresso machine, or atmosphere).

Pumps - I'm thinking that the simplest options are
1: eccentric driven piston pump right on the shaft
2: a peristaltic pump right on the shaft
3: Acoustic pump (heat or solenoid driven)
4: Solenoid driven piston pump (ferromagnetic piston, coil around the cylinder, two check valves)
(5: No extra pump, use centrifugal force of the coolant, rotated by the rotor, to push it around in the cooling cirquit)
(6: No pump, slush the coolant around the housing by means of the rotor movement)
(7: No pump, use heat pipes)
(8: No pump, use more copper and magnets)

It looks to me that, without a fan, and a good deltaT, the convection heat transfer to ambient is going to be much less than radiation, which makes it easier to make a rough calculation about it.

Using this formula for convective heat transfer:
q = k A dT
where
q = heat transferred per unit time (W)
A = heat transfer area of the surface (mo)
k = convective heat transfer coefficient of the process (W/m2K or W/m2oC)
dT = temperature difference between the surface and the bulk fluid (K or oC)
Free Convection - Air : 5 - 25 (W/m2K)

And Stefan Bolzmann's formula for radiation heat transfer:
q = ε σ (Th^4 - Tc^4) Ac
where
ε = emissivity of the object (one for a black body)
σ = 5.6703 10-8 (W/m2K4) - The Stefan-Boltzmann Constant
Th = hot body absolute temperature (K)
Tc = cold surroundings absolute temperature (K)
Ac = area of the object (m2)
 
Kingfish said:
The rear fins would require a bit more creativity.

Alternatives to cooling hub motors, such as solid-state devices employed on CPUs and GPUs, are both expensive and consumptive; not the best use of technology.
Click to expand...

Rear fins: So you are going to use a disc brake rear then? I'm thinking the rear brake can be replaced by the motor, maybe.

Cooling: Were you referring to subcooling?

Snipped the copper resistance/temp chart from this paper.

Ways of subcooling:
1: Peltier stack
2: compressor (common refrigerator principle)
3: Thermoacoustic
 

Attachments

  • coppertemp.jpg
    coppertemp.jpg
    66.8 KB · Views: 843
Kingfish said:
Manufacturer: Biff recommended these guys http://www.nblihe.com/ which I think that is what you suggested :)

I haven’t dealt with them yet because I haven’t settled on the design. My plan is to use something close to 45SH; at 150°C that’s as low as I dare go temperature-wise. Wire will be rated at 200°C. This is after all AF and heat will be a problem, unless we’re running underwater :)
]
Click to expand...

In November Nblihe couldn't supply neo45SH, the best SH they could do is 42MGOe, that may have changed, but I doubt it. But that isn't necessarly a bad thing, most of the time you can just increase the thickness of your magnet marginally with the lower MGOe and you still get the same airgap flux dencity, for a minimal increase in mass. People get hung up on trying to get the higest MGOe number, but that really isn't necessarly the best thing for th job. There used to be a good paper on the subject on dexter magnetics page, but I can't find it anymore, I guess they removed it.

-ryan
 
Phyllis:
Without seeing your plans, I would suggest a high-gearing ratio to provide lots of torque for less power so the system (single or 2WD) is less prone to overheating. You’ve already identified the key to the solution which is make the motor larger in diameter; I bumped my design up a notch just for that reason. Every rotor/stator pair will add about 80-90% more torque and hugely increase the efficiency.

If pumps were employed, well – you’ve named a couple of my favorites:
  • #1: First time I saw this type was in lawn mower motors when I was a wee pup. That type is more useful than the splasher method at pushing oil around when the motor is turning.
  • #3 is my special favorite; compact, lightweight, few moving parts. What if it were a variable displacement piezo-electric micropump?!? Wouldn’t that be cool! 8)
  • The no-pump solution should not be discounted either. I found this little inspiration for veining the circumference.

Rear Fins: Hadn’t got there yet. The front was sort of a Buck Rogers winged retro concept. Loosely speaking – since both sides of the rear are occupied with brakes or gears, I kinda imaged dual wings or veined-plates mounted a little higher on a rear bike rack. Aesthetically these wings could be made of metal similar to the Shirron Plate Heat Exchanger Chiller for Home Brewing. That is some nifty cladding going on there with “316 stainless steel that have been brazed together with pure copper in an oxygen-free furnace”. If done correctly the fins could be protected by an outer shell of stainless. Just a thought...

Last one: Flash Evaporation. Might need to use distilled water or you’d have mineralization, or possible galvanic reaction without a sacrificial anode. Not my favorite for this application.

How about we design the best motor first and see how the heat develops! :)
Happy Friday, KF
 
Biff said:
<snip>
In November Nblihe couldn't supply neo45SH, the best SH they could do is 42MGOe, that may have changed, but I doubt it. But that isn't necessarly a bad thing, most of the time you can just increase the thickness of your magnet marginally with the lower MGOe and you still get the same airgap flux dencity, for a minimal increase in mass. People get hung up on trying to get the higest MGOe number, but that really isn't necessarly the best thing for th job. There used to be a good paper on the subject on dexter magnetics page, but I can't find it anymore, I guess they removed it.

-ryan
Click to expand...

Gotcha; good points. I use both the 48 and the 45 material when modeling; best- and worst-case.

Some of the Litz wire I've been looking at is rated between 120 and 150°C. Makes one think that if the wire can't take the heat why bother rating the magnet higher because it will be the windings that will become hotter :roll:

Best, KF
 
I want two pedaling ratios anyway, I like 1.2 and 1.8 (0.83 and 0.56 overdrive). So I'd like to use the same geartrain for the motor just with extra reduction. My CAD package (ms paint) is not as accurate as I'd like but here is the general idea.
View attachment biketranny.png

The big red pulley is on a freewheel mounted on the cranks.
The second freewheel hangs on the big red pulley, and holds the chainwheel.
If the rearwheel sprocket is 24 tooth, the chainwheel is 28 tooth, and this is the 1st pedaling gear ratio (dogs disengaged).
If I use 8mm pitch belt on the red pulleys, and 18 teeth is the minimum I can use, and 44 on the big red pulley, to match the ground clearance of the chainwheel, then the freerunning sprocket is 18 teeth.
The pedal gears are now 0.86 and 0.55 overdrive, and the motor gears are 1.33 and 2.1 reduction.
The motor diameter is about 250mm.

Preliminarily, I've chosen chain on the final drive because it might be less prone to skipping, so maybe I won't need an idler on the tensioned side (upside), and because of the limited width that is available on my current bike. The motor axle is going above and abit in front of the bottom bracket, to give better chain wrap on the freerunning sprocket and also because if I build a frame, the suspension pivot is probably going to be concentric to it.
I've chosen belt on the other drive because it is more silent and hopefully less maintenance.
Particularly that part I'm not sure about - could I use a smaller than 18t pulley? Then the motor gears could be a little lower.
What is the best belt? Gates polychain is advertised to be as strong as chain, width for width. It would be nice if I could use belts on both drives.

The dog shifting fork might be actuated by a magura rim brake piston, I think it might be more durable than cable.
The cranks would be shimano saint, moddified for the 1st freewheel mount, somehow. They are stiff and you can get them for 83mm BB to get more space, I have 73mm BB.

What I don't like about it, is the second freewheel in the path of the motor torque. Those things wears and fails even under the feet of strong humans, so with the motor I don't know if it'll last. I'd much prefer if I could find a simple way to implement a double set of dog teeth on the same hub instead, like in some of Thud's transmissions. Also, motor braking only in 2nd gear is ok but not super, I wonder if I could ditch the disc brake if I had motor braking in 1st as well [EDIT: not a good idea to ditch the disc brake, would introduce backlash in the brake system, no good for trials]. Probably lots of other problems I didn't think about. So any opinion welcome.
 
phyllis said:
Particularly that part I'm not sure about - could I use a smaller than 18t pulley? Then the motor gears could be a little lower.
What is the best belt? Gates polychain is advertised to be as strong as chain, width for width. It would be nice if I could use belts on both drives.
Click to expand...
I've used PowerGrip 8M GT3 with a 16t pulley - I doubt if PolyChain is so forgiving. For either, the recommended minimum is 22t.

This is for trials use?

I can't see how you can make enough torque for this set-up, with a coreless motor..?
 
If I'm interpreting your freewheel set-up correctly, you don't have the ability to overrun the motor drive when pedalling.....
Click to expand...
Hi Miles, quite right, if you mean I will have to drag around the motor when I am out of battery. Will be a big disadvantage especially for trials stuff, with the flywheel weight.

That's cool about the 16t. where can you get those pulleys? How wide was it? So power grip has better grip than polychain?

Yea, trials and such. I don't know ofcourse if it will have enough torque, I tried to design it so it could break a normal bike chain set up on 18t chainwheel. That is normal (oldschool) trials chainwheel, and they do break of course, so people change them often and use heavier chains. Normal chain is 1200kg rated so that's 350 NM on a 15t sprocket (1200*9.8*0.0254*15 / (4*pi)).

For the motor I used a formula I found in Shane Colton's paper, torque NM = 2NIBLR, where N is number of active turns (turns per tooth * number of teeth * 2/3), I is amp, B is peak tesla in the airgap, L is the length of the magnets in meter, and R is mean radius.
His simplified formula for calculating B is Br*th/(th+g), where Br is remanent tesla, th is magnet thickness, g is airgap, in this case half the distance from magnet to magnet. I had to multiply the formula by 0.8, and then it matches FEMM pretty decently.
So I made a spreadsheet where I can specify the motor geometry and input current and voltage, and it gives me number of turns, wire thickness, phase resistance, torque, max rpm etc.

I've tried to optimize the transient torque versus the copper temperature, not taking any heat transfer into account. I've allowed 100K temp rise and a duration of one second should be more than enough. So that's the reason for the water slushing around inside, it needs to cool down quickly to 100c for the next kick.
To take a 1 stator, 250mm diameter motor as example, here are some parameters:
magnet length: 35mm
magnet thickness: 3mm
stator thickness: 3mm
field: 0.62T
wire diameter: 1mm
length/phase: 14.5m
phase R: 0.32 Ohm
copper active mass: 267g
total copper mass: 300g
total active mass: 1360
torque for one second: 230NM @ 132A, 84V (acceleration force with 2.05 red. 142kgF)
Max Rpm: 530 (max speed with 1.33 red. 50kmh)
Or with 1.2mm wire, 330NM for 0.5 seconds, 275 A, max rpm 760.
In my model, a double stator motor with the same dia and active mass would be a fair bit better, but it would depend much on endwinding length for which I don't have any good idea about. The CSIRO looks super efficient in that regard but I think I would be better off with flat or square wire for this rpm range and load. For those numbers above the fill was 0.5 through the inner diameter so a bit less throughout the gap.

If you think I'm way off please let me know.
 
phyllis said:
If I'm interpreting your freewheel set-up correctly, you don't have the ability to overrun the motor drive when pedalling.....
Click to expand...
Hi Miles, quite right, if you mean I will have to drag around the motor when I am out of battery. Will be a big disadvantage especially for trials stuff, with the flywheel weight.
Click to expand...
So, why not connect the cranks to the output pulley/freewheel and hang the motor pulley/freewheel from that (supported by a bearing)? Too much crank offset?
 
phyllis said:
That's cool about the 16t. where can you get those pulleys? How wide was it? So power grip has better grip than polychain?
Click to expand...
M8 PowerGrip GT belts can run on HTD pulleys. You can get 16t HTD pulleys from stock but mine was custom.

PolyChain can handle greater torque, I was doubtful about using it around such a small radius. I've no experience of using it, though.... PolyChain uses a different tooth profile to PowerGrip.
 
Miles said:
phyllis said:
If I'm interpreting your freewheel set-up correctly, you don't have the ability to overrun the motor drive when pedalling.....
Click to expand...
Hi Miles, quite right, if you mean I will have to drag around the motor when I am out of battery. Will be a big disadvantage especially for trials stuff, with the flywheel weight.
Click to expand...
So, why not connect the cranks to the output pulley/freewheel and hang the motor pulley/freewheel from that (supported by a bearing)? Too much crank offset?
Click to expand...

I don't understand... is motor concentric to crank spindle? Don't quite see it.
 
phyllis said:
Miles said:
and hang the motor pulley/freewheel from that (supported by a bearing)? Too much crank offset?
Click to expand...
I don't understand... is motor concentric to crank spindle? Don't quite see it.
Click to expand...
I meant the motor drive pulley (large one) & freewheel .

The inner part of the motor drive freewheel is connected to the outer part of the crank/output freewheel.

The inner part of the crank/output freewheel is mounted on the bottom bracket spindle.

See: http://www.endless-sphere.com/forums/viewtopic.php?p=356222#p356222

More complicated to do than your present plan but it would give you dual isolation.
 
You must log in or register to reply here.

Similar threads

dogman dan
E-BikeKit rewiew, 36 volt front hub
2 3
Replies
64
Views
18,291
dogman dan
dogman dan
Reid Welch
"Extreme MY1018" archival, closed out
Replies
11
Views
17,762
GIGATT
G
Back
Top