Stronger Motor Magnets

hias9

1 kW
Joined
Jul 11, 2018
Messages
422
On the usual larger Chinese hub motors, mostly magnet grades N35(S)H or N38(S)H are used.
As far as I know stronger magnets increase efficiency, but Eddy losses would also increase.

Would stronger magnets only be more efficient on lower rpms or also on higher rpms?
Is there a technical reason why they don’t use N45SH for example or are these simply too expensive?
 
For some designs, stronger magnets can also cause stronger cogging. Everything is a trade-off, you get more of one thing, if you sacrifice some of the other thing.
 
n45 have a temperature limit

scroll down on grin technologies all axle hub motor page
https://ebikes.ca/product-info/grin-products/all-axle-hub-motor.html

and youll read this

In addition to the mechanical changes we also updated from N35 to N40 grade magnets. Stronger magnets allow the motor to produce more torque for hill climbing with less heat loss, but they also increase the drag of the motor when riding without power. In our original design we erred on having lower rolling drag because at the time we felt this was more important than efficiency on hill climbs, but the Virtual Electronic Freewheeling feature in the Phaserunner/Baserunner controllers has shifted the balance and we switched to a higher strength magnet.
 
Yeah i would prefer the more wimpy magnet due to eddy currents getting worse and worse as the RPM climbs.

Low iron losses aka eddy currents are everything when you are running 20" wheels.. because the RPM is so much higher.

The all axle is a 27 or 28mm wide motor i believe? If so, given that a likely use is a 700c/29er wheel, stronger magnets does actually make sense so that you can produce higher torque at the combination of low RPM and high load..

A better motor for a tall wheel like that is the RH212 or a magic pie.. you can also make this higher low end torque with more poles.

...although a higher pole count will bite you in the ass with higher iron losses in a smaller wheel, just as stronger magnets will.
 
N45SH magnets would have a temperature limit of 150 degrees celcius (SH=super hot), N45UH would have a limit of 180 degrees Celsius (UH=ultra hot), so that would not be the problem.
Obviously Kv would be lower and drag would increase, but that's okay. I am trying to increase efficiency while the motor is powered.
However eddy losses would be a problem for me because max rpm while riding are about 1200rpm.
 
iron losses happen during power and not.. so the efficiency would decrease constantly until you are scaling a hill or accelerating from a stop.

You can get the most efficiency out of a right-sized motor and the ebikes.ca motor simulator will help you find that sweet spot.
 
I tried on the motor simulator a few months ago and according to the motor simulator, stronger magnets would result in higher efficiency also at higher rpms.
If the magnets are 10% stronger, I think Kv would be 10% lower and eddy losses 21% (1.1^2) higher, but still this results in higher efficiency.
 
How does one simulate stronger magnets on the ebikes.ca simulator?
I just looked it over and cannot find any options.
 
Custom Motor. I think 10% stronger magnets (10% more flux) would lower Kv by 10%, but increase Eddy losses by (10%^2 = 21%).
I am not sure by how much it would increase hysteresis losses.
 
Lowering kv will increase efficiency in some scenarios on it's own. You should isolate for that by not lowering the kV ( or otherwise monkeying with it until it matches the exact same RPM. )

Hysteresis losses are going to increase due to the stronger magnets. That's the force that the magnets create. You'll need to find out what that value is to calculate your penalty, lol. If you don't include this, you're not getting the full picture of all the variables.

Here are spec sheets for those magnet classes, if you know your magnet terms well enough.. ( i don't :lol: )

https://magnetsim.com/grades/N45SH
https://magnetsim.com/grades/N35SH
 
I found on the internet a common value to calculate hystheresis losses is that they are proportional to B ^(1.6).
Changing from N38H magnets to N48SH magnets would increase magnetic flux by about 10%.
For only simulating stronger magnets (Kv / 10%, Eddy losses * 21%, Hystheresis losses * 16.5%) efficiency and peak motor power per Volt (if nothing else is limiting) increases only slightly.
When (theoretically) rewinding the motor 10% faster, so the Kv will be back to the initial value, this would lower resistance and inductivity which will allow more peak motor power per Volt and also peak efficiency is now increased by a few percent.
 
If i take your calculation ( eddy current up 16.5% and magnetic force up 22% ), we get1% efficiency loss at a reasonable cruising load.

https://ebikes.ca/tools/simulator.h...&bopen=true&wheel_b=26i&wheel=26i&hp_b=0&hp=0

Generally it seems like the lower the load, the higher the efficiency loss is.

https://ebikes.ca/tools/simulator.h...wheel=26i&hp_b=100&hp=100&grade=12&grade_b=12

At very high loads, the efficiency loss disappears, but we don't see a power advantage. Maybe the motor simulator can't model that. Maybe the motor simulator just isn't good for modeling any of this, ha.
 
It's the other way round. Eddy losses increase proportional with B^2 and hysteresis losses with B^(1.6).
Also you did not lower Kv. Without lowering Kv the results make no sense. If you leave the Kv (so you assume rewinding the motor also) you need to adjust resistance and inductivity.
If you just increase the losses and leave all the other parameters unchanged obviously efficiency will get worse.
Regarding max power / Volt you don't see a difference because it is limited by the controller in your example.
 
Okay then the efficiency loss would be worse if i flipped the numbers.

I did not lower kv because we're trying to make an equal efficiency comparison. Changing the kV changes the rpm and thus the load point and will lead to unpredictable results where the efficiency will either go up or down based on the motor's characteristics.

So basically you need to properly calculate 5 different variables to properly simulate this, but we can't even be sure the motor simulator is taking everything into account.

I can add the magnet losses but i don't get a performance gain in any situation with the simulator.

You can adjust my examples in the links i provided to rule out the controller amp limit issue. I toyed with it a lot. It doesn't really make a difference.
 
If you increase just the magnetic losses, but don't lower Kv this makes no sense at all. What exactly are you trying to simulate that way?

10% stronger magnets without rewinding could be simulated by: Decrease Kv by 10%, Increase Eddy Losses by 21%, increase hysteresis losses by 16.5%. Only increasing Losses, but leaving Kv unchanged makes no sense at all.

If you want the same Kv with 10% stronger magnets, you would have to rewind it 10% faster which would decrease resistance and inductivity by 10-20%.
 
It makes a lot of sense to keep the RPM constant when you are comparing two motors because eddy currents scale with RPM. And efficiency is effected by RPM and load.

Two things happen when you lower your modified motor's kv:
There are less magnetic forces per mph that obscure the efficiency loss.
The load point of the motor changes.

This can make the difference in both power and drag disappear for the motor that is not adjusted to the same kV, so without taking this factor into account, you'll end up with a biased conclusion. You can even end up with lower efficiency than you expected. It completely clouds the 1-2% difference you are looking for.

You can run the other faster spinning motor at a lower speed using the throttle, however this will begin a much faster rate of PWM which will decrease the faster motor's efficiency and bias the comparison. This may not happen on the simulator but zombieSS has proven that this happens in real practice with a common trap controller.
 
If I compare the stock motor with the motor with stronger magnets (without rewinding, so different Kv) at the same speed (using throttle on higher Kv motor), the motor with stronger magnets has higher efficiency.
The PWM losses of the controller are probably not considered in the simulator, but they would be higher on the stock magnets (=higher Kv) motor.
It would not bias the comparison, but increase the difference a little further.

I really don‘t see a disadvantage of the stronger magnets except when rolling (higher drag) and maybe in field weakening area (more fw current required because of lower Kv).
 
Let's see your simulator link ( aka now it is your time to show your work ). I want to know how you managed that. It doesn't make sense. Even Justin says the efficiency goes down due to higher magnetic drag. ( however, with that specific motor, it increases power in high load so he considers it 'worth it' ).

PWM losses are lowest at full throttle and so tiny that it's very hard to measure them, we don't need to split hairs that fine :)

Field weakening has a very sizeable negative effect on efficiency so you'll want to pick your motor based on the kv being 10% lower, not make it up with controller settings.
 
Just set Kv 10% lower and increase hysteresis losses by 16.5% and Eddy losses by 21%.
Here is an example comparing at same speed:
https://ebikes.ca/tools/simulator.html?motor=cust_10_0.12_0.2_23_1_0.02_0&batt=cust_85_0.03_16&cont=cust_459_459_0.02_A&frame=mtn_tuck&hp=0&mass=120&cont_b=cust_459_459_0.02_A&motor_b=cust_9_0.12_0.2_23_1.165_0.0242_0&batt_b=cust_85_0.03_16&frame_b=mtn_tuck&mass_b=120&hp_b=0&bopen=true&throt=9.43&wheel=26i&wheel_b=26i

On steep hills it would be more significant:
https://ebikes.ca/tools/simulator.html?motor=cust_10_0.12_0.2_23_1_0.02_0&batt=cust_85_0.03_16&cont=cust_459_459_0.02_A&frame=mtn_tuck&hp=0&mass=120&cont_b=cust_459_459_0.02_A&motor_b=cust_9_0.12_0.2_23_1.165_0.0242_0&batt_b=cust_85_0.03_16&frame_b=mtn_tuck&mass_b=120&hp_b=0&bopen=true&throt=13&wheel=26i&wheel_b=26i&throt_b=12.4&grade=15&grade_b=15

Can you show me an example where the weaker magnet motor (with Kv 10% higher, but less magnetic losses) is more efficient?
 
Ok.. in your scenario, motor A has less efficiency.
If i take the throttle difference out of the equation and adjust the winding equal, motor A is more efficient.

https://endless-sphere.com/forums/posting.php?mode=reply&f=2&t=109658

If i take motor B and add more volts, motor B is even more efficient.

The change in nm-A is what seems to make the biggest difference in your scenario though.

The higher efficiency while climbing makes sense but the better efficiency during cruise just doesn't make sense.

I'll concede and let someone with more knowledge on this topic than me speak :p
 
Can you please show me the link where the weaker magnet motor is more efficient? If you assume the motors have the same Kv (stronger magnet motor rewinded) did you lower resistance and inductivity by 10-20% for the faster winding?
Let‘s please compare at the same voltage. The idea is only upgrading to stronger magnets, no change to the voltage.
Why should better cruise efficiency not make sense?
 
Better cruise efficiency does not make sense because the same forces that cause cogging during freewheeling are acting on the motor when under power too. And extra power needs to be used to overcome that.. more is needed the faster the motor spins.

GrinMotor_Drag.png


0.1nm of difference is tiny, but it requires a few extra watts to overcome.. so this should be reflected as lower efficiency until you start climbing, like grin says.

I don't think we can simulate this properly without some of the fancier tools Justin has.
 
But you need 10% less motor current for the same torque (before considering losses) so less copper losses and this probably outperforms the increased magnetic losses.
This diagram is just about cogging losses which will definitely increase.
 
In some windings that can happen and efficiency will be 1-2% higher for windings but this depends more on total copper fill. Some winding combinations dictate the use of a bit more copper as the wire is thicker to achieve the desired amount of turns. At higher amperages there can also be a loss in the power bus which is often un-accounted for in people's simulations.

A good example of A LOOOOOTTT of copper fill is the big crystalytes of yore and QS hub motors.. this can indeed make more torque per amp.. but go too far with copper and you end up with excessive end turn losses and you lose efficiency that way.. and too little copper for a motor will have the eddy currents becoming a dominant source of loss.

But in this scenario there is no difference in the windings.. so okay, i am rambling. :lol:


We can compare the grin v1 and v2 motors here:

https://ebikes.ca/tools/simulator.h...20_GA&bopen=true&throt_b=97&grade_b=5&grade=5

The simulator says the v2 motor is 1-2% more efficient in pretty much every scenario you can roll it in.
It disagrees with Justin's statement on the webpage.

If you are correct then this is an interesting avenue for increasing efficiency on any motor.

I guess i would like to be more sure because replacing magnets is a pain in the ass job.. i also use 20" wheels on all my bikes so my rides are a lot more sensitive to losing power from eddy currents than say, a 26" wheel.

I feel like from middle to high loads, this makes sense.. but at low loads, you likely see the efficiency loss.. what's ideal for you depends on your vehicle design i suppose.
 
The copper losses in this case did not decrease because there was a change to the windings. They decreased because with the stronger magnets you need less amps for the same torque, and copper losses are proportional to I^2. So it’s probably not just 1-2%.
I am just wondering why they only use N35(S)H or N38(S)H magnets on the Chinese hub motors? A little increase in drag would not really matter I think. Or is there maybe a significant price difference?
I use a 24inch wheel, but up to 1200rpm, so Eddy losses also play a role for me and this is deep inside field weakening area where the advantage (without rewinding the motor) would probably be less.
 
Back
Top