Building a motor

mxlemming

100 kW
Joined
Jul 17, 2020
Messages
1,116
So after completely failing to get Neumotors to sell me a motor (I really can't see what is so damned hard about 8038, 70kV, extended shaft version preferably with a keyway... any keyway will do...) and being 21 emails deep with Lisa Talia and still zero progress... I think I might try building a motor. I've kind of wanted to for a while.

Thing is, I don't really know much about this, so far I have done loads of controller related stuff, and have basically taken the little 8080 BLDCs and a 1500W hub as they come. This might be the motor learning thread for me.

I think the ideal motor for my ebike would probably be ~50kV, spins up to about 4krpm @20s battery (so with a ~7:1 reduction and 26" wheels I would get about 40mph) and weigh sub 2kg.

So thinking I would start with something like this (if they are really 2$ each I will find a hat to eat):
https://www.alibaba.com/product-detail/HIGH-Quality-Silicon-Steel-Brushless-Motor_1600116087634.html?spm=a2700.12243863.0.0.38403e5fEroHx8
Hopefully I can just get this completely off the shelf and stack them 2 high for 40mm of lamination
stator pic.jpgStator dimensions.jpg

And wind it like this:
LCM24,22.PNG
using something like 0.7mm wire 10 strands of, no idea how many turns... that's a calc for another day...
Seems like 22 and 26 pole motors use the same winding. Presumably 26 spins slower and has more torque.

So if I did this, with a 100mm air gap, 3mm or 5mm magnets 10mm wide * 40mm long (314mm circumference/26 = 12mm, 314/22 = 14.27mm which sounds about right for a motor magnet spacing), then in the 64mm hole in the middle shove a big old pair of 12mm bore bearing (6001 series 12*28*8?) and a lump of aluminium full of holes to force air through for cooling.

Total motor would be about 1.4kg plus the weight of the copper, which I guess would be significant, maybe 0.5kg?

Knocked this up in about half an hour in Solidworks. My day job is doing solidworks, so really doesn't take long.
Motor knockup SW.PNG
Adding a plausible amount of copper to it gives about 2.1 kg.
Motor knockup with copper SW.PNG

Worth trying for? Am I missing anything obvious?
 
X-sec showing the bearing layout etc.
Motor knockup X-sec.PNG
56.5mm depth, which would make it easier to put places.
 
Vbruun said:
Why do you want to go to 8038? :)

As opposed to 8057? Because the 8057 sticks out a lot. The motor at the moment is already dubious and drags on the floor when i crash and risks clipping my ankles.

As opposed to the current motor... I just want more power. The 80mm China special is pretty good, but however I gear it, it's never going to be a weapon. The videos and reports from madin88 and district9prawn look pretty promising.

But it's occurred to me that the bill for the 8038 will be about 600$ (8057 probably near 800$) once hall sensors and postage are considered, so if the stators don't break the bank there's a lot left from that for some CNC'd bits, magnets and wire... Plus the glory of building my own motor.
 
Am I missing something? I thought you already Ran an 8080 motor?

Apart from that I totally understand the want for more power. Even though I see that you are usually pretty good at getting things done I just want to add that it rarely gets cheaper to DIY with all things acounted for. However I really want to see a nice radial flux motor build here :)

What is the lamination thickness on that stator? - I couldn't figure it out from the ad.
 
Vbruun said:
Am I missing something? I thought you already Ran an 8080 motor?

Apart from that I totally understand the want for more power. Even though I see that you are usually pretty good at getting things done I just want to add that it rarely gets cheaper to DIY with all things acounted for. However I really want to see a nice radial flux motor build here :)

What is the lamination thickness on that stator? - I couldn't figure it out from the ad.

I see your misunderstanding. Neumotor count the size of the air gap. The Chinese ones and alien count the external dimensions. Current"8080" is actually more like 7040 in Neumotor speak. Which really doesn't seem that much bigger. Extra torque i guess roughly 80²/70² is 30% but then it also takes much more current and is meant to be generally much better quality so :?:

I think it will be cheaper to make one... Unless I count my time in which case it'll be like 10x as much. Rough estimate is 15$ copper, 15$ stators, 30gbp aluminum billet and I have a friendly machinist I reckon would cut it for me for<<200 for a few off. 30$ of magnets so total way less than the 600 the Neumotor would cost. If I can ever extort it out of them.

I also cannot work out the lamination thickness. I've asked them. No idea if they'll even reply. Hope they do, this is dead in the water off i can't get stators, I'm not really up for making my own lamination stacks.
 
I see! Well that sounds like a good idea. I think that one would want to spin this type of motor really fast so thin laminations (0.2mm?) Seems like almost a must.

I agree with you that the BOM for one will be cheap, but try to keep track of the cost for all the iterations before the one you end Up using ;)
No matter what I think you should go for it!
 
I suspected something like this would eventually show up... Interesting project, though if I might be asking: why no inrunner ? Granted the extra rotor width of an out runner will give you some extratorque, but in runners are much easier to cool, and also simpler to integrate in a reduction setup. An inrunner design also allows for a larger axle and a wider bearing spreading, and better dust / dirt sealing which contributes to an overall much sturdier motor.
 
qwerkus said:
I suspected something like this would eventually show up... Interesting project, though if I might be asking: why no inrunner ? Granted the extra rotor width of an out runner will give you some extratorque, but in runners are much easier to cool, and also simpler to integrate in a reduction setup. An inrunner design also allows for a larger axle and a wider bearing spreading, and better dust / dirt sealing which contributes to an overall much sturdier motor.

Pretty much just about power and torque to weight ratio. As you observed, the air gap for a similar size inrunner is much smaller and they run much higher speed. So then you need gears.

The cooling is an interesting one. In the face of it, in runner looks better but you can always water cool the stator but you can't water cool the rotor which pushes back in favor of outrunner. For my current design, I'm trying to pump lots of air through an aluminum core which should get a decent amount of cooling.

I really don't think I could get a good enough power to weight ratio when including gearbox with an inrunner. The biggest problem for my ebike and probably madin88 and the others with belt drive and neu 80xx motors is that th neu really does seem to be the best motor for the task, but it's still far from optimal. Those of us who've ridden successfully with a design like this all much prefer it to hub motors.

TBC. One of the stator companies replied today but they want me to give them dxfs and then they'll laser out a small batch from 0.2mm steel which sounds expensive. Guess it's better make some dxfs and see how much.

Probably ought to fire up FEMM first and see how hard that is to learn.
 
Why not get a stator “off the shelf”? I have some sources if ur interested where I get them but can’t remember off hand or pronounce them.

I plan to use this:
https://www.antinnotech.com/collections/frontpage/products/mp-8074-sensored-version-brushless-motor-for-electric-skateboard-scooter-bike-go-kart-propulsion-system
And have a similar plan as u but I have two others I won’t use: Leopard 8072? And some bigger 80mm motor with a skirt bearing on back and cant remember it’s name or kv now or I’d wind the leopard to whatever u wanted for the usual retail cost.
 
mxlemming said:
qwerkus said:
I suspected something like this would eventually show up... Interesting project, though if I might be asking: why no inrunner ? Granted the extra rotor width of an out runner will give you some extratorque, but in runners are much easier to cool, and also simpler to integrate in a reduction setup. An inrunner design also allows for a larger axle and a wider bearing spreading, and better dust / dirt sealing which contributes to an overall much sturdier motor.

Pretty much just about power and torque to weight ratio. As you observed, the air gap for a similar size inrunner is much smaller and they run much higher speed. So then you need gears.

The cooling is an interesting one. In the face of it, in runner looks better but you can always water cool the stator but you can't water cool the rotor which pushes back in favor of outrunner. For my current design, I'm trying to pump lots of air through an aluminum core which should get a decent amount of cooling.

I really don't think I could get a good enough power to weight ratio when including gearbox with an inrunner. The biggest problem for my ebike and probably madin88 and the others with belt drive and neu 80xx motors is that th neu really does seem to be the best motor for the task, but it's still far from optimal. Those of us who've ridden successfully with a design like this all much prefer it to hub motors.

TBC. One of the stator companies replied today but they want me to give them dxfs and then they'll laser out a small batch from 0.2mm steel which sounds expensive. Guess it's better make some dxfs and see how much.

Probably ought to fire up FEMM first and see how hard that is to learn.

Don't know how much you ride and in what kind og terrain, but I think that longievity might be an issue with an air-pumping outrunner sucking in the dirt.
But if you aren't riding all that much, I guess it will be fine :)
 
Hummina Shadeeba said:
Why not get a stator “off the shelf”? I have some sources if ur interested where I get them but can’t remember off hand or pronounce them.

I plan to use this:
https://www.antinnotech.com/collections/frontpage/products/mp-8074-sensored-version-brushless-motor-for-electric-skateboard-scooter-bike-go-kart-propulsion-system
And have a similar plan as u but I have two others I won’t use: Leopard 8072? And some bigger 80mm motor with a skirt bearing on back and cant remember it’s name or kv now or I’d wind the leopard to whatever u wanted for the usual retail cost.

I am planning on getting a stator off the shelf. Just waiting on them replying... What are your sources?

The motor you linked I already have but rewound for 50kV by alien power systems. It's not enough. Fun yes, but I can't get wheelie and speed with the same gearing and it overheats with now than about 60A phase in a few minutes and with 80A it'll overheat in less than 1 minute. I'm planning on building something that's just a different league to the ones you've listed.

Vbruun said:
Don't know how much you ride and in what kind og terrain, but I think that longievity might be an issue with an air-pumping outrunner sucking in the dirt.
But if you aren't riding all that much, I guess it will be fine :)

It's definitely an issue. The existing motors do not like the wet. One day I'll print a case to keep the rain off but then I'll have to limit the power to avoid overheating.
 
I Saw tom Stanton make a scoop and air filter setup for hos large outrunner. To me it seems like a great way to improve reliability whilst maintaining 90% power in an outrunner :)

I just read the thread about the grenaded aftermarket sur Ron outrunner that grenaded itself, which might make me unreasonably worried.
 
Vbruun said:
I Saw tom Stanton make a scoop and air filter setup for hos large outrunner. To me it seems like a great way to improve reliability whilst maintaining 90% power in an outrunner :)

I just read the thread about the grenaded aftermarket sur Ron outrunner that grenaded itself, which might make me unreasonably worried.

The grenaded surron aftermarket motor was just a completely retarded design. They'd cut big holes in the rotor shell for no apparent reason and the guy designing it had absolutely no clue how to calculate for centrifugal forces... He essentially didn't realize they exist, and went on to blame it on sensorless controller and...

Not to say it isn't concerning to me, but I do know how to calculate for centrifugal forces and would test it spinning up to twice my anticipated max speed (centrifugal forces proportional to square of rpm so this gives huge headroom). If I can get a controller to go that fast.

Air scoop a good idea, would work well with the design above I think. I've been thinking about this a lot but for my current motors, they pump essentially no air atall so it's futile.
 
I can get stators from some guy who calls himself tony. I can post his email here later if you want. He used to work for a place on alibaba but is solo now.

I got this bigger motor somewhere on alibaba as well. It’s the one with the skirt bearing. I forget what brand and details but maybe more the size u want.
 

Attachments

  • B410822D-1AF9-4EC3-A32F-9D00C6D255C7.jpeg
    B410822D-1AF9-4EC3-A32F-9D00C6D255C7.jpeg
    572 KB · Views: 1,864
Here are my motors.
IMG_20210817_170606125.jpg

They will both propel me to 30, maybe 35mph so plenty enough for legal ebike and general wizzing around, but there's nothing more to give, and I want the same kind of power as Galp's giant glory or the Russian illegalbike.

If you'd pm or post this Tony guys email address that would be useful. If he's ex alibaba supplier I presume he gets stuff for not so crazy prices?
 
rich_motor@163.com
He might sell u a couple for a good price and he had 80mm stators before and would stack to the height you want

Maybe a bigger gear reduction would allow you to use a smaller motor. Those “freewheeling cranks” seem great to me for that

But the bigger motor I have I forgot all its stats but it’s much bigger than the ant motor. if ur in USA I’d send it for my 20$ less than my cost (whatever it was) and I’d pay shipping if you’re interested. I won’t use it and plan to just use the ant motor

If ur getting a stator designed this seems a better design in the neu motor coming out. It’s claims of efficiency are great. I don’t know why but has different sized teeth and the LRK wind ends up better in that I can always get more copper on and easier and quicker to wind and also safer with the phases more isolated.


What’s the red motor in your pic?
 

Attachments

  • 6B1DF53D-6070-4B8A-A9AB-1E8FA2F82F0B.jpeg
    6B1DF53D-6070-4B8A-A9AB-1E8FA2F82F0B.jpeg
    20.7 KB · Views: 1,856
Red motor is a mcmaster 80m diameter.

Not really interested in your "bigger"motor. Same can size, 80mm probably 70mm stator... It's never going to have much more torque. Marginal increases at best.

Will give this rich guy an email. If I make a motor it's going to be 100mm stator, 24 slot 22 or 26 pole. looking to go more torque than the neu 80xx which madin88 found not quite enough for wheelies.

Not convinced by lrk instead of dlrk. Papers posted by madin and larsb show much higher harmonic content so will be unpleasant for FOC or other sinusoidal control.
 
i dont know what this bigger motor is or where i got it sorry its pretty vague and not trying your thread into a sales pitch but thought it might suit your wants.


mcmaster motor i dont see

reading this thread he does a thorough theoretical comparison of lrk vs dlrk but i think the biggest missing element is finding what will really fit. he did a simulation of the wires and stator but its hard to simulate and the difference between getting another coil on a tooth or not i think trumps the theoretical losses. the real test would be to wind it twice and just run it and see what kv, resistance, and no-load losses and tempted to do it to put the comparison to a real test for at least one motor.
https://endless-sphere.com/forums/viewtopic.php?f=30&t=89473&hilit=lrk&start=100

(are mutual and self inductances good or bad?)

it does show a slightly more square wave here https://endless-sphere.com/forums/download/file.php?id=225909
but i think it really depends on each motor and very subtle.
 
Looking for some help on magnetics...

A supplier offered to laser out 5x stacks of 0.2mm Kawasaki steel laminations for me, no price yet but crossing my fingers it's not stupid. I was going to buy the Neumotor at ~500$ for one off, so that sets the wet finger in the wind for my budget for this project.

Problem is, where I thought they would just sell me a stock stator, they want me to design it and send them DXFs. Fine,sure... but I have *absolutely no clue wtf I am doing* since it's like 12 years since I studied this at uni and frankly I did the bare minimum to pass that module.

My understanding is that I need to pretty much arrange the magnetic circuit:
magnet-> air gap -> iron spoke -> iron yoke -> iron spoke -> air gap -> magnet ->halback magnet

If I sum all the reluctance components given the width and length, equate that to the magnetic strength in amp turns equivalent, I can calculate the approximate field strength at any point in the circuit (not counting the corners...). With N42 magnets being like 1T ish, and iron saturating at 1.6T, and with a 1mm air gap and some pissing about in excel, I concluded that the spokes should be about half the magnet width. But that seems quite large - 5mm - which seems larger than most pictures of stators look.

1) I presume that if you have too thin iron, it saturates (bad) but how much leeway should I build in? The component from the coils current might need considering, but since its in quadrature to the magnetic field, I suspect actually it does not add to the total flux.
3mm spokes
Stator design 3mm .PNG
4mm spokes
Stator design.PNG
5mm spokes
Stator design 5mm .PNG
By comparison to other stators posted, about 3.5mm "looks" correct, but industrial and academic motors seem to use more iron.

2) I am struggling with how to estimate the resulting magnetic strength of the halbach and feed that into the equation.

3) All the stators I have seen have really sharp corners between the spokes and yoke. This seems intuitively (to a guy who spends a lot of time doing FEA on structural parts...) bad, since the magnetic flux is intuitively going to concentrate there and increase saturation in those regions, and therefore increase losses. I have therefore rounded them substantially - am I barking up the wrong tree? Obviously rounding them makes the winding harder.

I have decided to go for Halbach array, not for performance reasons, but because I can then easily 3D print the rotor for prototypes, saving me a lot of hassle and expense with back iron of specific size and I can iterate easily, changing air gaps by just pressing print again. Who knows, maybe it will prove to be strong and light enough I never need machining. I suspect heat will actually be the nail in the coffin of the 3D print rotor.
Motor layout.PNG
Motor air gap.PNG

Using these:
https://e-shop.magsy.co.uk/neodymium-magnet-prism-20x4x2-n/ for the halbach elements and https://e-shop.magsy.co.uk/neodymium-magnet-prism-40x10x5-n/#/22-quantity-16_pcs for the main magnets. This intuitively seems like a hell of a lot of magnet, about 60$ worth.

Hummina Shadeeba said:
i dont know what this bigger motor is or where i got it sorry its pretty vague and not trying your thread into a sales pitch but thought it might suit your wants.
Respectfully, I am basically not interested in further 80x china motor optimisation, this thread is about building a substantially new motor optimised for mid and belt drive ebike conversions, for which there really is nothing good available. If you want to talk about them, you can PM me...

Many people have sent me many suggestions, the only one even approaching optimal is the Neumotor, which I suspect Madin88 wrung everything possible out of and then a bit, and getting one is like pulling teeth, they are crazy expensive (2x more than a QS205???) and the specs given are in fact bullshit. The various gear motors, enourmous 5kg inrunners etc... just no, they all fall well well short on various parameters; size, weight, torque. We need something better.

To make the kind of drive me, Madin88, district9prawn and others have built, I think we really need motors that enable ~6:1 reduction ratio (~16tooth to ~96 tooth), which in turn enables use of HTD8M belt single stage reduction and massive, near silent power transfer capability. On a 26" wheel, say 60mph, that's ~15 rev/second = 90rev/second on the motor shaft = 5400rpm. Targetting 13PP => 70kerpm, 1100eHz at flat out, which is about the limit for proper FOC.

Increasing from Neu 80mm stator air gap to 100mm air gap gives a torque increase of 56% (ratio of diameter squares), and increasing the pole pairs does also increase torque, less clear how much by in practice though.

The other functions that need substantial improvement over available RC motors are:
1) shaft bearings - they are not designed to be loaded with gears or belts, the bearings are inset quite far and small, so die
2) stack height - the available 120mm od china motors pack an enourmous pile of useless crap - prop mounts, skirt bearings, wasted pointless gaps etc
3) For proper FOC, you need much better position detection. Sensorless is flakey as hell and does not play well with field weakening, MTPA etc, and traditional hall sensors in the stator slots go massively out of position when you start pushing current. I found this when I started running a sensorless and hall algorithm together and monitoring the phase difference. The simple solution seems to be to just have another $ of magnets mounted in the big hole in the middle of the motor where not much else is happening and put a 2$ JLCPCB with SMT hall sensors under them.
Hall magnet location.PNG
 
I get using the magnets that are there already for triggering but if the halls go all wonky pushing current, it seems optical could be the way to go. Is it worth adding provisions for them?
If you're pressing in an aluminum core would the press fit warp the laminations on you? You could go slip fit, cut slots in the hub, add some keying / drive nubs in the laminations, make removable bearing end caps, and clamp the sandwich all together with a few bolts through the bearing cups. Kind of like a floating disc brake on a motorcycle. Adds some complexity but should make changing a bearing easier with less risk of damaging sensitive parts in a press. Or remove and drop the bearing cups in an oven and avoid a press completely. This could also setup a convenient way to hold it for a winding setup.
A few holes for some light spring pins or for slip fit alignment pins just for assembly should help you hold the stack aligned. If you find you don't need the spring pins or you just use for the initial alignment you could leave them empty.
Could also add some holes in the laminations to try to get airflow in there to suck the heat out. If you got fancy with staggered patterns you could kind of add a fan blade into the laminations to suck air through on its own.
 
Jrbe said:
I get using the magnets that are there already for triggering but if the halls go all wonky pushing current, it seems optical could be the way to go. Is it worth adding provisions for them?
If you're pressing in an aluminum core would the press fit warp the laminations on you? You could go slip fit, cut slots in the hub, add some keying / drive nubs in the laminations, make removable bearing end caps, and clamp the sandwich all together with a few bolts through the bearing cups. Kind of like a floating disc brake on a motorcycle. Adds some complexity but should make changing a bearing easier with less risk of damaging sensitive parts in a press. Or remove and drop the bearing cups in an oven and avoid a press completely. This could also setup a convenient way to hold it for a winding setup.
A few holes for some light spring pins or for slip fit alignment pins just for assembly should help you hold the stack aligned. If you find you don't need the spring pins or you just use for the initial alignment you could leave them empty.
Could also add some holes in the laminations to try to get airflow in there to suck the heat out. If you got fancy with staggered patterns you could kind of add a fan blade into the laminations to suck air through on its own.

All good points/ideas.

I started modeling in a renishaw encoder but after several died unexplained at work I've lost enthusiasm for this until they resolve/explain. Incremental ones... I downloaded a few steps and they were all quite huge and the discs seem hard to buy. If you have any model suggestions... Obviously the on axis ones are problematic for outrunners.

Doubt you'd get much airflow through the laminations by the time you consider the copper windings covering things but I'll definitely be adding location features before I order. I was planning on making the stator to core like 0.1-0.2mm clearance and then bonding it with loctite 620. Spring pins would be nice and secure though. I'll Look into that. Want good contact for good heat transfer to the aluminium core which will have lots of air pumping through it.

Re the bearings, I'm hoping to completely avoid ever changing bearings. In the usual 8080 motors they need changing because they're piss small no brand ones and simply cannot take the load required for any application. With some nice big actual skf bearings, possibly even doubled up or a needle bearing, I suspect this will outlive 99% of applications.
 
You could likely clearance the hub for the sensors to flush them and maybe even seal off the optical area with a 3D printed cap. Or a 1 or 2 piece cap that doubles as an impeller.
You could do the chopper disc inside out and have the sensors inside, not sure if there are any benefits but I think it's worth considering. You would probably have to remove the chopper disc before splitting the motor to not damage the sensors I think. Could do a 2 piece (mirrored / cut in half) chopper disc so you can remove one side, rotate it properly, and then pull it apart. This could leave everything else in place and not mess up your angle adjustment if that helps.

If the chopper is ferrous you could switch to halls if needed.
Only electrical reason I've seen opticals fail is too much current or unexpected transients in the power, signal, or 0v lines. Did the ones that failed on you have tvs diodes? I don't have any recommendations besides the obvious stick to name brands recommendation.
I've noticed some leds that have static protection built in. I've been using constant current drivers lately on optoisolators so they have a wide input voltage range. They look like diodes. Just have to make sure they don't pwm.
I'm not sure what the other side of your motor will look like or if thats all of it. Might be able to add this stuff easier there. Could make this pretty easy I think.. Is the shaft keyed, D shape or splined? Not sure how to align the chopper on the open side if it's an option.
Where does this bolt to a mounting bracket?
A bolt on timing gear bolt pattern and centering boss could be a nice to have in the 8 spoke / hub area. I'm not a fan of set screws.
I keep struggling some imagining this opposite my normal inrunner thinking, might be some bassackward derp in these ideas.
 
The horrible image I attached is what I meant for making airflow with the laminations. If the red oval slots are cleverly placed /aligned with the winding T shapes you could rotate each lamination 1 position to make a screw / fan shape. It's exaggerated a bunch but helps visually what I meant.

As for the chopper disc it can just be flat. Same guy cutting the laminations can cut it. I attached a simple chopper disc I did for a robot design.
 

Attachments

  • Stator design.PNG.png
    Stator design.PNG.png
    23.5 KB · Views: 1,469
  • 20150805_083505~2.jpg
    20150805_083505~2.jpg
    238.7 KB · Views: 1,467
Think it'll need a lot higher resolution than those discs. Normally i think they image and etch them.

How is this screw/fan shape causing air flow? The laminations dont spin. I'm generally fairly keen on aluminum in the middle... 1) lighter and 2) higher thermal conductivity.

I hadn't thought of different laminations which i guess is possible it they're laser cut... If it went to production (no plans really...) It would need to be stamped.

The encoders that blew were brand name. Chances are I did something terrible to them. Not sure what tvs etc they have, sealed unit. Put me off though. Absolute encoders.

I'll probably key the shaft. Not thought that far ahead.
 
Back
Top