Upgrading Magnets in Hub Motors

Sunder

10 MW
Joined
Sep 6, 2011
Messages
3,054
Location
Sydney, Australia
I came across this article here http://avdweb.nl/solar-bike/hub-mot...hub-motor-tuning.html#h0-1-6-stronger-magnets which talks about upgrading hub motors by using stronger magnets.

They suggest the cheaper hubs don't use the best magnets for cost considerations, but would be cheap to do the upgrade yourself.

Has anyone done anything like this, or know how hard it would be? There's some magnet manufacturers in Australia, that do custom N52 grade magnets in almost any size you want, so sourcing parts wouldn't be the issue. I'm looking at doing it on the Cute Q100.

Thanks.
 
For what purpose? Efficiency? Power?
 
itchynackers said:
For what purpose? Efficiency? Power?

Power density. I want the most power in the smallest motor possible.

Other things I will be looking at later, is how to cool the motor so I can put more power in for short hills.
 
change magnets, you change the behavior of the motor. I don't know if it would be more efficient or not, probably depends on how you use it, how it's geared, wheel size, etc.

AFAIK you would be changing both voltage and torque constants. May have effects on current needed to do the same work (don't know if it would increase or decrease).
 
amberwolf said:
change magnets, you change the behavior of the motor. I don't know if it would be more efficient or not, probably depends on how you use it, how it's geared, wheel size, etc.

AFAIK you would be changing both voltage and torque constants. May have effects on current needed to do the same work (don't know if it would increase or decrease).

Maybe the article is over simplifying it, but it does seem that just replacing them with bigger magnets gives it more power. Whether it consumes that power through more current or voltage, or reclaims some of the inefficiencies, it isn't all that clear.

I'm guessing this then, isn't a well researched and popular mod. I'm not sure if that makes me want to do it more, and report the findings here, or avoid it, as a bit of a time and money waster.
 
In that link, the guy's project is to make the most efficient bike, so he explains a lot of theory and the reasons for his choices and modifications. I don't think that he said that he'd change his magnets. It's possible that you already have high grade neodymium magnets in your motor because it's certainly very powerful for its size compared with others.

If you want the best from this motor, the sweet spot is 17 to 18 amps with 12s (44v nominal, 48v actual).

Or, if you've got the the time and inclination, try the N52 magnets and let us know the results, but be careful: You have to think about handling them because they're very strong and they jump into each other and break given half a chance.
 
d8veh said:
In that link, the guy's project is to make the most efficient bike, so he explains a lot of theory and the reasons for his choices and modifications. I don't think that he said that he'd change his magnets. It's possible that you already have high grade neodymium magnets in your motor because it's certainly very powerful for its size compared with others.

If you want the best from this motor, the sweet spot is 17 to 18 amps with 12s (44v nominal, 48v actual).

Or, if you've got the the time and inclination, try the N52 magnets and let us know the results, but be careful: You have to think about handling them because they're very strong and they jump into each other and break given half a chance.

I think I might at least take the cover off and see how hard it is to remove the magnets. If it looks easy, I might give it a go. Will be a while before I have time to do this though.

If it does look like they're easy to remove, I'm going to remove one and see how much weight they hold - the site I'm intending to get the magnets from gives approximate holding weight. I'll see if I can get them to stick to an overhead beam and put 100g at a time on it, and see if it even comes close to the claimed holding power of a N52 replacement. If it does, then I'll give up on the idea, if it doesn't I might give it a go, though I have no idea how I will measure or prove any improvements.

One reason I believe that stronger magnets might make a difference, is that he points out that RC motors are much smaller and lighter, but can generate more continuous power. Even Cute's specification sheet only claims 80% efficiency, when some RC motors are over 90%
 
Enough heat should soften the epoxy so you can pop out the magnets. Waste of time and money imo, since increasing the power of small motors will just melt em in less miles.
 
About three months ago I went through the same study using FEMM. The 9C motor roughly pencils out as having the magnet material of NdFeB 32 MGOe that best matched my torque and power usage. It’s possible my model was off a bit – however it feel pretty confident it is in the ballpark to begin the comparison.

During the study I presumed the magnets were about 27-28mm long x 2.2mm high and having a width of ½ inch. Through deterministic measurement, I concluded the Back Iron ring is fabricated from 8-inch Schedule 80 Steel pipe, likely 1018 alloy.

Steel Pipes Dimensions - ANSI Schedule 80

During the simulations where I wanted to keep the same stator – but with upgraded commercial magnets, I used the more commonly available size 1-inch (25.4mm) x ½ inch (12.7mm) x 1/8 (3.175mm) or ¼ inch (6.35mm). Trying to find sizes of thickness in-between was much more difficult.

The two problems that emerged through modeling were: Manufacturing & Output.

When you make the magnets thicker, that dimension eats into your back iron. Stronger magnets will require better magnetic support, so I went down that path – theoretically at first and then with practicality, and the results were less than whelming. This is when I figured out that the back iron was created by lightly machining the inside of 8-inch Schedule 80 Steel pipe. I modeled different alloys of back iron – presuming I could get my hands on 1010 or 1008 Steel, but FEMM is kinda fickle here and I could hardly make a dent in the results. But now we’re really far out on a limb talking about replacing the back iron of the motor to accommodate thicker and/or stronger magnets.

Next – I tried upgrading the magnets all the way to using NdFeB 52 MGOe – which is expensive and temperature sensitive. The safer number is between 45 and 48 in H or SH which can safely between 248°F (120°C) to 302°F (150°C) – which is the upper limit for many motor components. Plus – trying to find a manufacturer that fabricates a stronger magnet in a suitable size… well I only found one that was a close match – Magnets4Less.Com.

Neodymium Magnets N52 Rare Earth 1 inch x 1/2 inch x 1/8 inch Block - BrMax: 14800 gauss
Neodymium Magnet Physical Properties

Then came the modeling with all the presumptions – meaning I could afford it. I must have spent days on this - a week at least, and went through 1/8 inch thick all the way to 3/8 inch thick with big fat back iron in 1008 using N52: The best improvement I could achieve was 24% more output per Watt. Why would this be? :wink:

On the catbird seat, KF
 
Here’s another way to look at the problem: :)
Let’s say I’m not going to muck with back iron dammit; I just want stronger magnets! I’m going to buy off-the-shelf and DIY. Below is an illustration of the 9C motor physicality. Dimensions are in MM.

9C-2806-Hub-0.125MagnetsDetail.png

Back Iron OD = Light Gray
8-inch Schedule 80 Pipe OD & ID = Dark Gray
Original Magnets = Blue
New Magnets 1/8 inch (3.175mm) thick x ½ inch (12.7mm) wide = Red
Stator = Green
Dimensions/CenterLine = Yellow


The original magnets are about 2.2mm thick x 13.9mm wide x ~27 or 28 mm long. Compared to replacement magnets which are 1 inch (25.4mm) long we have the following discrepancies:
  • Replacements magnets have 70% mass relative to original units, so we’re already starting in a hole. :(
  • Torque Ripple is introduced because the replacement magnets will have a 3/4 mm gap in-between as opposed to nicely butted alignment. The ripple reduces the effectiveness of the flux density and essentially creates drag.
  • Because we elected not to machine the Back Iron, we will be forced to mill off 1 mm of the Stator radius so the replacement magnets can fit. This is not a bad way to go because we want more back iron and it forces the replacement magnets together. Conversely, if we machined the back iron to fit the magnets and preserve the stator – that pesky air gap becomes worse. Then again – machining away the stator will weaken the induced electric field and change the shape of the focus, so that’s a problem too. Also, we may need to bury the Hall Effect sensors in a bit deeper. The one real bugaboo I dislike about going this route is trying to keep the magnets equidistant during curing; however a removable precision spacer could work.
Note that it is entirely possible to craft replacement back iron – making it thicker so as to allow the replacement magnets to converge and reduce (although not quite eliminate) the pesky airgap by 0.25mm. But – at that point we’d have to make a new stator. :?

However in my theoretical study I ignored the airgap and had the magnets butted up snug. Why does the model fail to gain significantly in performance when I increase the flux density with stronger magnets?

Come on, step up to the plate and swing! :)
Without fear, KF
 
Kingfish said:
[*]Torque Ripple is introduced because the replacement magnets will have a 3/4 mm gap in-between as opposed to nicely butted alignment. The ripple reduces the effectiveness of the flux density and essentially creates drag.
:) There is an optimum range of spacing. Too much will increase torque ripple, as you say - I'm not sure how this will "create drag"? Too little means the flux at the edges short circuits, so butted is not what you want, either.
 
Kingfish said:
Why does the model fail to gain significantly in performance when I increase the flux density with stronger magnets?

Come on, step up to the plate and swing! :)
The stator core is close to saturation with the original magnets?
 
Hello Miles :)

Butted up: I believe the optimum would be – “not touching”, but as we see with 9C hubs – they do in fact touch. Not critical, but a noticeable gap decreases the flux field proportionally, and when you have back iron behind it – the shape of flux is reduced rather diabolically: Not what we want.

Stator Saturation: That’s part of it – absolutely. Good point! :idea:

The two key factors I was looking for are: Iron Core losses and Magnetic Strength which are directly related to performance. However – Stator saturation plays a large role!

The shape physical of the magnet determines the field shape. So back iron and Stator are designed with magnetics together as a system; they are optimized as one. I actually think that the weaker magnets work well with the wheel because the iron losses are reduced – so when we start adding more pull-strength, we’re creating a problem both for the stock stator and the back iron because it will want to resist rotation, like a stepper motor. The FEMM model shows much higher density at the choke-points when the fields pinch, specifically at stator corners and at magnet joins. This reduces the effectiveness of the stator to push against the magnets. I’ve tried increasing the back iron to ridiculous thicknesses, but the Stator appears to gate the gain.

Relative to ironless AF studies, even those with back iron rings, increasing magnetic field strength directly increases power and efficiency. I have to conclude, like you – that it’s partly Stator Saturation and Eddies. We can keep making magnets larger and more powerful, but we won’t ever see the full benefit.

The second part is the relationship between the physicality of the individual magnet and the magnetic strength. All things being the same – two magnets of the same material and grade, make one twice the volume (say by thickness) and you have essentially twice the lifting/holding power. This is a good thing; we want that! The other way to affect output is to change the material out for something stronger; N35 vs. N52. Let’s play with these factors for a moment…

I suggest that the 9C motors are using N35; a wild guess – I could be wrong, but allow me to make a point: It’s the weakest commercially available RE magnet we can lay our hands on and the least expensive. From the Neodymium Magnet Physical Properties previously linked, N35 is between 11.7-12.1 KGs; we’ll pick 11.9 KGs as the average. Now we want to replace it with the best that money can buy cos I won the Washington State Lotto Uber Mega Millions so we’re gonna order up N52 at 14.5-14.8 KGs, or 14.65 KGs as an average. The cost between these two increases exponentially – unlike the field strength; you don’t need me to tell you it’s not worth double the price. The fact is N52 is only going to improve the field strength about 23% over N35. N’uff said. :wink:

So now we’re back to changing the physicality of the magnet to cheaply increase the power, the torque/Watt ratio. I have modeled over 200 Ironless AF configurations and when I double the magnets’ thickness – you can just about bank that the output is going to double; it’s almost linear (there are associated losses which grow, even in the ideal world). The trade-off is we now have double the magnet mass, and a larger current draw to overcome momentum. :shock: I think this is why hub motor designers want to keep the magnets thin; find the right balance. With that in mind – if we could afford to create the custom sized magnet as directly replace it with stronger magnets, the problem with Stator Saturation would rear it’s pesky lil’ head. I can’t see spending money making a custom-sized magnet for a 23% gain, even if I could spend your monies. :)

My science may not be perfect – but it is how I understand it given I’m not a physicist or chemist or mathematician. I’ve goofed with FEMM long enough to trust some of what it reports, however it falls down at times with back iron material changes. You’d think there would be a noticeable difference between 1018 alloy and 1008 but the deltas were lost in the noise. I have to conclude that the 9C motor is optimized for one magnetic shape and charge, and the only reasonable change we can easily make is in the amounts of winds/turns. We can’t really affect these very much either cos the copper used is pretty good quality. If we look to change laminations – we’ll now were remaking the core, and if we go that far, I’d begin fresh and make it the way I’d want: Now we’re off topic. :roll:

However I do wish to say to the OP that I had exactly the same idea some months ago, and one day decided to go through the rigor and model it thoroughly and completely as a cheaper alternative to Ironless AF. On that note, when I win the lotto for real, I’ll be making AF motors... right next door to my brewery. :twisted:

Now if I could just find the winning ticket stub…
Full of flux, KF
 
Kingfish said:
I suggest that the 9C motors are using N35; a wild guess – I could be wrong, but allow me to make a point: It’s the weakest commercially available RE magnet we can lay our hands on and the least expensive.
When I did the analysis of the UltraMotor stator for Farfle, I came to the same conclusion.

http://endless-sphere.com/forums/viewtopic.php?p=609412#p609412

With N32 magnets:

file.php
 
Miles - I remember that thread and it was such a good read! Do you have the link to the original backstory? Where'd the magnets come from?

On the AF studies I was taking mine all the way to 2.3 or 2.4T, but then only had back iron to deal with on the outer rotors for return path and leakage control; no electric steel - could be 1018. So simple.

~KF :)
 
Kingfish said:
Miles - I remember that thread and it was such a good read! Do you have the link to the original backstory? Where'd the magnets come from?
Original backstory? I don't know of any info that's not in the thread: http://endless-sphere.com/forums/viewtopic.php?f=30&t=41576
 
Wow, I read through KF's post, and the first bit made sense, then I lapsed into "engineering talk I don't understand".

I want to understand it, and I am going to do more research when I have time (Currently at work), but thank you so far for your musings, models, and thoughts.

One thing that I would like to clear up though, and maybe KF or Miles can answer me straight: Dogman said something about more power = faster melting. I would have thought in terms of power consumption, pretty well the only variability would be in the thickness, type, and winding of the winding wire. So if power was gained, it would actually be from inefficiency - as we know almost all inefficiency eventually translates to heat - so wouldn't having more power through stronger magnets, actually cool the motor?
 
Yeah, I'll step up and take a swing at the short answer...You can't polish a turd. Sure Mythbusters made some shine, but they were still just turds, which is the point of the saying.
 
John in CR: Yeah, exactly! :lol: Except I like my 9C, warts and all. I mean – it’s about the best we can affordably have, although far from perfect.

Sunder, in theory – having more powerful magnets should yield better efficiency because the repelling force of the magnets should push harder against a smaller current. That’s the pipe-dream!

Presently, we push current through the motor and there is some resistance. Push a lot of current and it really heats up. The pipe-dream suggests stronger magnets gets us away from these heat issues. The problem is that it is not a panacea for every motor, and some will actually perform worse. I believe the 9C is a good example of a motor that could be modified with off-the-shelf magnets as I described, however there is a strong possibility that it will perform worse after the mod in whichever direction the magnets lay (back iron mod or stator mod). Looking at the 9C hub, they (those of them that made it) cut about every corner to make it affordable, including lack of rust-protection. :evil: The bearings & seals are the lowest quality, the back iron is likely the cheapest iron they could find… 8-inch pipe, and to prevent overheating they run skimpy wire out of the axle. When cornering hard, the hub covers flex, and my disc brakes sing loudly in protest. If I get on it from a start, the hub covers flex. And I don’t think my setup is a hot rod; I’m pulling 1.2-1.5 kW/wheel.

So I agree with John, with a caveat: It may be a turd, but it’s my favorite turd and I wouldn’t poop it out. Not until we have a better choice. :wink:

Out of TP, KF
 
Report this postReply with quoteUpgrading Magnets in Hub Motors
by Sunder » Sat Oct 27, 2012 7:29 pm

I came across this article here http://avdweb.nl/solar-bike/hub-motor/permanent-magnet-dc-hub-motor-tuning.html#h0-1-6-stronger-magnets which talks about upgrading hub motors by using stronger magnets.

They suggest the cheaper hubs don't use the best magnets for cost considerations, but would be cheap to do the upgrade yourself.

Has anyone done anything like this, or know how hard it would be? There's some magnet manufacturers in Australia, that do custom N52 grade magnets in almost any size you want, so sourcing parts wouldn't be the issue. I'm looking at doing it on the Cute Q100

Q100? Have you looked inside yet? If it's like the Q128 it's a inrunner. Magnets spinning inside the stator. Relativly small diamier. Maybe 2 inches.

How old is the Q100? If it's yaer or so old it' probably has better magnets than the newer ones out there now.
The first 3 128's had strong magnets the second batch I got early this year had crap for magnets. The new ones could be spun in either directions with little differance in drag. freewheeling or against the motor was about the same drag!
So if it's a new one it might be worth the gamble. It will also need a lot less magnets than a 9c does.

Just a thought.

Dan
 
DAND214 said:
Q100? Have you looked inside yet? If it's like the Q128 it's a inrunner. Magnets spinning inside the stator. Relativly small diamier. Maybe 2 inches.

How old is the Q100? If it's yaer or so old it' probably has better magnets than the newer ones out there now.
The first 3 128's had strong magnets the second batch I got early this year had crap for magnets. The new ones could be spun in either directions with little differance in drag. freewheeling or against the motor was about the same drag!
So if it's a new one it might be worth the gamble. It will also need a lot less magnets than a 9c does.

Just a thought.

Dan

Thanks Dan - It's so new, I haven't even installed it yet. May be worth a look at.
 
Don't the small diameter motors in geared hubbies use curved magnets? If so, matching them is going to be tough.

A rewind that achieves a better copper fill seems a more sure route to improve such a motor.
 
John in CR said:
Don't the small diameter motors in geared hubbies use curved magnets? If so, matching them is going to be tough.

A rewind that achieves a better copper fill seems a more sure route to improve such a motor.

I had thought about that, and if I can buy a junk motor to rewind and still have it successfully work afterwards, I would consider doing it.

One thing at a time I guess.
 
Its hard to understand a lot of the technical stuff here.

I have removed all my magnets from my cromotor and the thought came to me, why not replace them with better magnets? Since I have them out already and can measure them to order custom magnets, why not replace them with something better?

My question is, can I purchase better magnets than what was in my cromotor to get better performance out of my motor? Possibly more torque or something?

I also have a 2nd cromotor of the same exact type and can compare the differences.

Thanks
 
HTB1CmOHegZC2uNjSZFn761xZpXau.png_.webp


I bought this older 17" QS hubmotor that was advertised as 1500W continuous and 3000W peak with a kv of 9,3. Different casting but might be similarly wound and with crappy magnets as in the picture above. Is there potential to be gained by swapping magnets and upgrading the puny little phase wires? Or is this an actual lemon that needs to be rewound to gain anything? Pretty hilarious power to weight ratio at 3kW/20+kg.
 
Back
Top