Definitive Tests on the Heating and Cooling of Hub Motors

[John in CR]

Let's create what would be as long and steep of a hill as any one might realistically encounter on a bike ride, with a 1000 meter elevation gain over a 6km span, with a peak grade in the middle of 20%. That's pretty much like going straight up an appreciable mountain pass. With our prototype Grin hub motor, which has low thermal mass, and no extra cooling effects, if we do the climb at a steady 30 kph then the predicted motor temperature reaches over 330 oC and we've clearly cooked the hub

Almost a 17% average grade is awfully steep. The Pike's Peak road isn't nearly that steep.

If it's the same Grin hubbie in the simulator, then that little motor just needs a small enough wheel to make that climb. Wheel size make (ie gearing) makes all the difference when it comes to climbing.

 

[Offroader]

I thought water cooling a hub motor was one of the best methods to use?

Why isn't this looked at more here? It really isn't that hard to put some water blocks into the motor and make the side cover take a larger bearing for the water hoses.

Running a fuse hoses to some radiators on the bike isn't all that hard also.

So far I think Linukas has the most experience with water cooling as he has made and sold water cooling kits for various motors, including QS and mxus I think.
But I have never seen any data published regarding heat shedding capability on his systems. Nor do I remember seeing posts made of owners of water cooled hubs comparing the heat removal before and after water cooling.

Would be sweet to see some real world data on the matter of LC hubs.

Water cooling is probably the way to go because it isn't that hard to run a small water hoses to the front of the bike. You can easily add cooling blocks to the inside of the stator. As tests have shown the stator sucks in most of the heat really quickly and the water blocks will pull it out. You can add more thermal mass with the water that is distributed more evenly outside the bike. You can cool the motor while not moving.

Justin should really be doing some of his research into seeing just how effective water cooling is as I think it is the best solution and isn't all that hard to do. All it takes is having a custom side cover made and making a water block that can get maximum heat into the water.

 

[macribs]

Testing water cooling as well would suit the thread well, as water might work well also for power hungry freaks. Heck who knows FF might even add benefits to water cooling as well

Come to think of it, doctorbass also has a water cooled motor, a CL 5403?
Don't know if doc has got any test equipment but if he does maybe he can give it a go?

Or maybe Linukas has some data he can share, if by any chance he follows this thread?

 

[justin_le]

Justin should really be doing some of his research into seeing just how effective water cooling is as I think it is the best solution and isn't all that hard to do. All it takes is having a custom side cover made and making a water block that can get maximum heat into the water.

I can tell you with definitive certainty that I won't be investing weeks of my life testing out and experimenting with entire water cooling systems for hub motors, so don't hold your breath for that! Anyone who wants to is welcome to do so though, and I've posted some pretty clear methodology on how you can do all of these experiments with very little in the way of fancy lab test equipment. Just a field oriented controller, a CA, some thermistors, a motor stand, and a computer for logging. But what I will ask is that anyone who is inclined to post on that topic please don't just send glory pictures of your mods and anecdotal comments on how it performs. You must include before and after numeric test data, with actual values for W/Degree thermal conductivity terms. Otherwise it's not part of "definitive testing", and doesn't really belong in this thread.

Come to think of it, doctorbass also has a water cooled motor, a CL 5403?
Don't know if doc has got any test equipment but if he does maybe he can give it a go?

AFAIK he never ran the plumbing so it just acted as a glorified heatsink, which as I just posted can have remarkably good effects on reducing the heat rise in the motor in these high power runs. The post is here but image links seem broken:
https://endless-sphere.com/forums/viewtopic.php?p=1070059#p1070059

Of course piped water cooling will work very well, and even without a radiator just the latent heat capacity of 1-2 liters that you recirculate would be able to suck up a huge amount of heat. But is it necessary to run pipes and pumps etc. given just how effective the combined effects of passive air vents and/or ferrofluids can be? Other than in the most extreme usage cases I'd say probably not.

 

[justin_le]

Almost a 17% average grade is awfully steep. The Pike's Peak road isn't nearly that steep.
If it's the same Grin hubbie in the simulator, then that little motor just needs a small enough wheel to make that climb. Wheel size make (ie gearing) makes all the difference when it comes to climbing.

Yes, this is of course absolutely true and also one of the useful parameters to play around with on the trip simulation. For the same parameters, if I laced the Grin Hub into an 18" wheel instead of a 26" wheel, then the motor core temp only reaches 135 celcius, versus >330 celcius that it reached in the 26" rim.


On the bottom left of the page there is button you can click to run a "simulation set" where a single variable is changed between a span of values, so we can set this to run a whole bunch of simulations and the cursor and section results are then output in a tab separated text file that you can copy and paste into a spreadsheet.


Doing that for the above scenario, varying the wheel size from 12" to 29", you can see how drastic the effect can be. In a 12" wheel the motor core only gets 75 degrees. In a 20" wheel it's 170 degrees, and by the time you reach a 22" wheel size the core is >200 celcius and cooked.

 

[Hyena]

That's very interesting about the heat capacity. A quick look at a table of heat capacities shows water would be better than aluminum for a given mass or volume in terms of heat capacity.

I had an idea a few years back to use a sealed water/oil bath along the lines of what Justin was getting at in the above experiement by adding more thermal mass.
Pics and details in this thread:
https://endless-sphere.com/forums/viewtopic.php?f=6&t=8482&p=744784&hilit=+hyena+air+oil+plate#p744784

But basically, sealing off the centre hollow part of the motor with a donut like plate and filling that with oil. It would allow the heat sinking equivalent of more metal/thermal mass but without the weight. No doubt running external hoses to a radiator would sink the most heat but this would be a neat and contained bolt on 'normal' motor. And being sealed you could run dual oil and air cooling for the best of both worlds (or now with ferrofluid as per Justin's experiements)

What do you think Justin ? It's sort of the opposite of what you were getting at with your potting experiements. Or, what about going the other way and potting right across from the windings to the axle but keeping the centre hollow and filling with oil ? You could use a balloon of sorts or easily dissolved styrofoam or similar inserts to maintain the cavity while pouring the epoxy. Better still, a thin soluble spacer above the windings so it left and air gap that would allow oil from the central bath to slosh directly over the windings.

 

[fechter]

Hyena's approach is interesting. It could work without some of the drawbacks of just filling the hub with oil. I think you'd want to keep all the wiring outside the oil filled part.

I agree that running plumbing through the axle to a radiator and needing a pump will be the best for heat removal but too much complexity. I wouldn't do it.

 

[justin_le]

What do you think Justin ?

I think I've shown in sufficient detail how you can quantify exactly the effect it would have. For instance, with a MXUS motor which looks similar to what you were dealing with, the core inner diameter is about 140mm, width is around 50mm, so your best case volume fill is ~750cc, but the axle and struts etc. occupy some of that so figure ~650cc of oil room. Typical oils are about 1.6 J/K/mL, so your fill in there would add just over 1000 J/K of core heat capacity. What effect will this have? Just choose a MXUS motor, then choose custom (all the parameters of the MXUS will stay populated), click "show advanced", and then change the core heat capacity from 2520 J/K to 3520 J/K, and see what effect that has on a given trip terrain, and you can decide if the benefits are worth the effort or not.

Similarly, if you want to see the effect of statorade PLUS the oil heat sink, then do the exact same thing but start with the MXUS 45XX + Statorade hub from the drop down. I think you'll find that putting 1000 J/K more sink in a motor that's already got 2500 J/K in the core isn't going to make much of an overwhelming difference, but if instead of oil you filled that chamber with water, then you're adding 2700 J/K to the core, ie more than doubling it. To first order appxomation in a high load scenario that would mean about twice the run time you can get from the motor before overheating, and as was already mentioned water has the extra plus of clamping the core to a firm 100oC max provided you give a little vent hole in the top for steam to escape.

It's sort of the opposite of what you were getting at with your potting experiements. Or, what about going the other way and potting right across from the windings to the axle but keeping the centre hollow and filling with oil ? You could use a balloon of sorts or easily dissolved styrofoam or similar inserts to maintain the cavity while pouring the epoxy. Better still, a thin soluble spacer above the windings so it left and air gap that would allow oil from the central bath to slosh directly over the windings.

In the end as I showed the copper->steel core conduction path is already pretty good, so touching the windings vs. touching the core doesn't make much difference in the final analysis.

 

[macribs]

Of course piped water cooling will work very well, and even without a radiator just the latent heat capacity of 1-2 liters that you recirculate would be able to suck up a huge amount of heat. But is it necessary to run pipes and pumps etc. given just how effective the combined effects of passive air vents and/or ferrofluids can be? Other than in the most extreme usage cases I'd say probably not.

Sure you are absolutely right. For most users water cooling will be overkill or too far out. I have high hopes for that radical vented motor with them large scoops, and if that motor will even keep the FF inside at full RPM some sort of ventilation might work out to keep debris from entering the motor while hot air escape and fresh cool air get "scooped" inside. Might still make for shorter FF lifespan but test data would be awesome.

As far as pipes and radiators goes that will be for the most power hungry freaks, but it would still be amazing to see real world data on what kind of heat water would remove. Sure water cooling would make for added complexity, but as far as lifespan and durability goes I think pumps these days will not be the first thing to give. And at least water cooling loop is cleaner and less messy then oil cooling with ATF. Will it be hard to replicate your test rig or methodology for testing to have some kind of comparable data if one does testing of water cooled hub?

With those new Adapttos (mk II) pushing what 40% more peak power (14kw?) into the hub and also more current when use of OVS to run faster maybe the time is now for experimenting and testing with water cooling to avoid meltdown during hot summer days.

Anyway, huge fan of all your dedicated work for this thread and this forum. Finally we have real world test data to help us make the right decision for cooling.
Those smaller lighter motors will make for perfect commuters when FF is injected. This is by far my favorite ES thread. Mucho respecto!


Cool Hyena, now I got more reading up to do. At 3.30 in the morning.

Btw Doc's 5403 got the pics working, and new one added. Thx Doc!
https://endless-sphere.com/forums/viewtopic.php?p=1070059#p1070059

Man the mad skillz on the ES crowd.

 

[Punx0r]

Air-cooled sealed hub = winning
Air-cooled open hub = unreliable
Sealed hub with non-structural thermal mass = dead weight
Water-cooled with pipes, hoses, pump & external radiator = dead weight, expensive & unreliable

 

[madin88]

Air-cooled sealed hub = winning
Air-cooled open hub = unreliable
Sealed hub with non-structural thermal mass = dead weight
Water-cooled with pipes, hoses, pump & external radiator = dead weight, expensive & unreliable

+1

for larger hubmotor powered motorcylce (with two seat for instance) a watercooling system makes sense if not necessary even if the motor runs at 90% or above efficiency, but for ebike hubs it is OVERKILL
it is to complex and all the stuff requires lots of space. the radiator would need to be placed where now the controller often is mounted and where than install the controller? If one has a watercooling system he also wants to have a large battery so it would be saner to build a motorcylce straight away with motorcycle frame and parts instead of using ebike stuff.

 

[John in CR]

Air-cooled sealed hub = winning

Winning what? DD hubbies end up big and heavy, because the motor itself is way underutilized.

Air-cooled open hub = unreliable

Really? There have been almost no failures, though of course there are some conditions where a well vented hubbie would be a fail.

Sealed hub with non-structural thermal mass = dead weight

Agreed. Invest that weight in more motor for more benefit.

Water-cooled with pipes, hoses, pump & external radiator = dead weight, expensive & unreliable

I'd add that it won't work nearly as well as people think, because the coolant is too distant from the copper.

What truly "wins" is making less heat to begin with by:
-Using motors with proven high efficiency.
-Using smaller wheels for proper gearing to allow the system to run free of stress, which results in greater overall efficiency.
-Tuning the controller for the motor to make less heat without sacrificing performance.
For some reason the vast majority skip these entirely.

 

[macribs]

Water-cooled with pipes, hoses, pump & external radiator = dead weight, expensive & unreliable

Dead weight? Come on, the extra weight of a pump, radiator and rubber hoses. Negotiable at best, for many people it will not add much to the overall build. I mean if you got a 100 lbs curb weight already why obsess with a few lbs more? For those considering water cooling I am sure the extra "dead weight" will add to better cooling so all in all well worth it. Or at least I think it will be well worth it, as it stands we don't know yet as there are no tests done of water cooled hubs.

Expensive? A few feet of rubber hoses, an inexpensive pump, a radiator from a scrap yard and some aluminum piping to make a heat transfer coil inside hub.
If you consider water cooling in the first run, you can probably mustard the say the ≤ 50$ extra. On the other hand if you got money to burn you could go "all in" and and make a state of the art cooling system, with high powered silent fans, sensors that control fans, and a state of the art heat transfer block like Doc's. You can have sensors that spin the fans if needed to cool down the motor even if you sit dead stopped at a red light waiting after many hard stop and goes.

Unreliable? Really? Not so much. Pumps runs hundred and hundred of hours without failure. Your aluminum coil inside your hub (or state of the art CNC machined heat rtansfer block) will work to aid for better cooling even if you have a major crash and all cooling liquid exits the cooling loop. Just by the added "dead weight". So even in worst case scenario where you loose all your coolant, the water cooling heat transfer block will still work to cool down you motor, just by the added masses. How is that for reliable? On a side note those same arguments where used some 30-40 years ago - when dirt bikes and snowmobiles first started to become water cooled. Consensus was air cooled ruled because of the ease of use and "less chance of failure". Well, we know how that went.

The only downside I see that is valid and worth considering is that water cooling will bring extra complexity to a build and builders will spend extra hours making/buying heat transfer blocks/pumps/radiators/hoses, mounting all the parts, removing trapped air from the cooling loop, and possible change coolant and pump every two years or so just to be on the safe side. Even a broken pump or leakage will not destroy your motor, the added mass of cooling loop will actually still cool the motor better then stock motor even with a broken pump and or leakage.

 

[Punx0r]

John, absolutely agree: Don't make the heat in the first place and that if material isn't contributing to torque generation, it should be minimised.

The standard I was setting for reliability is for a commercial product: ~10 years in all climates, including salted roads. Vehicle OEM. That's why Justin's tests on ferro-fluid in a sealed hub is so interesting to me.

Macribs, yes, it's considerable weight. If you did a really good job you might get total weight down to 1kg. In reality, it'd probably be more. Adding 1kg of extra copper and stator iron to the motor would be much better. Also consider there are motors out there that make 10kW/kg. That's potentially a lot of extra power capability to sacrifice to add a series of potential leaks.

Yes, expensive. Mass-produced products are usually costed to a fraction of a pence/cent. They also don't use parts from scrap yards. $50 is probably the production cost of a total 9C clone.

Unreliable, yes. water-cooling systems for ICE engines have received a century of development and are still a constant source of problems. All external parts of the system would be prone to corrosion and impact damage. Pumps die, coolant must be maintained (replaced at intervals) and if something can leak, chances are it will.

Again, that's why you should be excited about Ferrofluid, it's cheap, effective and simple.

 

[justin_le]

[EDIT, typed this before Punx0r's recent clarification. In the standards of a commercial product, I 100% agree with the tone of the summary -JLE]

Hey guys, this is not a place for arguments, and Punx0r you're being a tad provocative in phrasing this summary because that's not what the empirical and theoretical results show.

Air cooled open hubs have not been evidenced anywhere to be unreliable. As a potential motor manufacturer I'm nervous as heck to release a motor that is deliberately open to the external environment, but my own ebike has a vented hub motor with close to 10,000 km on it, which ride year round and only ever leave parked outside, on a marina dock exposed to all the salt water spray etc. And of all the drilled out side cover hubs we've sold, there hasn't been a single RMA associated with the drilling (much to my continual surprise).

And as for thermal mass being dead weight. Dead weight implies that it's not doing anything, but when you look at how the numbers crunch out it can have quite a surprising effect in real world scenarios that would otherwise lead to a motor overheating. When we've dealt with RMA'd motors that have burnt up, it's usually the result of users taking it on one particularly long climb with an extra load that just put it over the edge. And in those cases an extra kg or two of aluminum would have meant the motor happily absorbed the heat. Of course it's silly to design a motor with just extra thermal mass, you might as well make that into active steel+copper mass to get better motor efficiency while you are at it. But if you already have a given hub and want it to not burn up in a given high load scenario (like a single long hill climb on a trip, or a high intensity drag race) then it would be a surprisingly simple and effective strategy, even if counter-intuitive.

A wax that undergoes solid->liquid phase change might also prove really useful here. Have a look at this:
https://en.wikipedia.org/wiki/Phase-change_material
Most paraffins are on the order of 250 J/g, which means they'll absorb more heat in the solid->liquid state change than a similar amount of oil does going from 20oC to 150oC.

For an as-manufactured motor though, this sealed hub approach with FF to thermally link the stator to the shell for easier external air cooling is hands down looking like the best all-around solution. But for modders trying to really push power boundaries, the other techniques would perform better (thermal mass for short term power limits, externally circulated liquid cooling for long term power limits).

Air-cooled sealed hub = winning
Air-cooled open hub = unreliable
Sealed hub with non-structural thermal mass = dead weight
Water-cooled with pipes, hoses, pump & external radiator = dead weight, expensive & unreliable

 

[macribs]

@Punx0r sorry I did not read your post as in a context of a commercial available product. I was thinking for the average ES modder.
When I see your post in context to commercial products with warranties and a brand name to promote your claims makes sense.

For the hot rodders I still think any cooling mods that helps people achieve their desired level of performance and reliability is the right choice.
That could be open side covers, closed with FF, added masses or scoops, or even water cooling if that is what it takes to reach the desired performance goal.

@Justin_le a while back there was a thread about adding an impeller to help cool the hub motor. Any thoughts on that? Think it would pair up with FF in a positive way? It seems that the thread just went to sleep without any conclusive data.
It seems like a low hanging fruit if it adds to cooling. Could be low weight and 3D printed locally.

 

[justin_le]

@Justin_le a while back there was a thread about adding an impeller to help cool the hub motor. Any thoughts on that?

Yes, I touched on it here:
https://endless-sphere.com/forums/viewtopic.php?p=1112026#p1112026
and here
https://endless-sphere.com/forums/viewtopic.php?p=1112341#p1112341

Short answer is that I think it will help compared to having just open holes, but not for the reasons that it is shaped like an 'impeller' moves air from the rotation. Rather it helps because it would form cavities that trap the much faster passing air as a result of forwards bike velocity, and help direct that into the opening hole. If I was to 3D print an under-disk part with the intention of moving air through the motor, I would shape it more like this:

And not like an impeller.

Think it would pair up with FF in a positive way? It seems that the thread just went to sleep without any conclusive data.

Stay tuned.

 

[John in CR]

Justin,

Does the Analogger record temp too if I'm using a CA3 and have the right temp sensor tied in? I ask because the blades on my motor bolt on quite simply, so I can do runs with and without them very easily and capture data to clearly demonstrate the dramatic difference they make in the amount of air flow through the motor. They definitely work as centrifugal fan blades, with no "scooping" action. Without them installed the vented motor cools only slightly better than sealed, but with the blades installed the ventilated cooling works so well that I ceased further pursuit of cooling mods, though FF and sealed interests me a lot for obvious reasons.

I'm confident that I'll be able to dispel your notions regarding blades, but I need input how to capture suitable data. My motors have a quite heavy stator that fills the rotor shell, so my thought is that I should first get the motor up to a healthy operating temp, the same temp to start each run. I have multiple long climbs near my house, so I'll be able to run high enough power to generate substantial heat without excessive speeds. I can also do some constant speed runs on the highway to see the impact of velocity.

I don't have any controllers with field weakening, so I won't be able to capture the same quality of data that you get from your test rig, but I think I can make the test conditions equal enough that the big temperature differences are obviously not testing errors. Temperatures are extremely stable here, and I'd plan to run tests early enough in the morning that the roads haven't heated up enough to distort the tests if one time is sunny and another cloudy.

Static tests using a friends dense smoke machine showed that the exhaust side blades start sucking air through the motor at only 100rpm. At higher rpm they create strong flow both through the motor and across the outer shell through the holes in the rim. Stator temp proves the throughput holds true while the bike is in motion, and the road grime pattern near the rim flange holes shows there's definitely flow across the outside of the shell too.

Something else I can test, is the effectiveness of my air scoops that direct more air toward the motor. I made them primarily for visual appeal, but at speed the 35cm2 cross section on each side of the swingarm definitely sends more air toward the motor. I can check with alternate sides blocked, which may make a difference since I intake air on one side and exhaust out the other side. Additional flow should increase velocity at the blade tips, so I believe the extra flow on sides is best, but never tested it. My motor runs so cool that I haven't had interest in optimization. eg I'm pretty sure an air dam on the intake side somewhat similar to your scoop would increase flow by slowing the flow that is now perpendicular to my multitude of small intake holes and slots, and increasing the air pressure immediately outside the air intake.

While I need to learn how to use my Analogger anyway, if I'm going to the trouble to demonstrate what I already know definitively works quite well on my bike, I want to generate solid data. My general idea is that if I start with the motor at the same temperature, and the same ambient conditions. Then I make the same long climbs using roughly the same energy and time, that the stator temperature (measured on the steel, not up in the copper where short temperature spikes create measurement noise) at the top or end of the runs should provide pretty definitive info, especially when the almost 10kg stator ends up with a 20°+ difference in temperature.

I'm all ears for any refinements of my plan.

John

 

[made_in_the_alps_legacy]

Glad to see that the idea about melting wax from april - p22 has made its way in your brain, justin...
i hope you might answer my previous question regarding axle temperature as an indicator how much heat keeps conducting along the stator, i kind of miss the IR shots and their rainbow colors... ahahah :lol:

 

[John in CR]

If I was to 3D print an under-disk part with the intention of moving air through the motor, I would shape it more like this:

I've always been afraid of arrangements like that, because it would deflect send grit and sand directly into the motor. With the vent holes commonly used what goes in stays in. I try to make intake as free flowing as possible, but to make sure no debris large enough to do damage gets in I use tiny drill bits for intake holes and thin dremel discs for intake slots. Size is made up for by quantity, with the limit being as many as possible without harming the side cover's structural integrity.

 

[Punx0r]

Apologies if my previous post was provocative - my assumption was that this was research was primarily biased towards feeding back into commercial production.

Justin, I'm pleasantly surprised how well your vented motors have apparently held up. Do they use a protective coating?

Adding thermal mass niggles me slightly because it feels like half a solution, in that it does help average temperature under varying load, but does nothing to increase continuous power rating.

 

[macribs]

Adding thermal mass niggles me slightly because it feels like half a solution, in that it does help average temperature under varying load, but does nothing to increase continuous power rating.

True, but as of today if you "need" more power then what can be had by mxus 3K, qs 205 or similar the next thing is almost double the weight and more then 20 + kilos.
There are a few other options, like hubmonsters, dual/triple/ astros etc but they work best in those funny looking small wheels or they really set you back financially. If the new revolt 160 pro turns out to be gentle for the controller that could be a high powered motor that does work well for mid drives. But I have not seen any builds completed yet using 160 pro yet.

So as it stands there are not many options to have more power without adding significant more weight to the hub motor. In light of that I feel cooling said hubs are the best way to go about it as of now. As proper cooling will allow to do more repeated runs, climb steeper hills and get more fun per buck spend. Each man must decide his own power goal and do the best cooling to suit their goal.

 

[madin88]

instead of adding weight to the stator (we should not forget that this is bad UNSPRUNG weight), i would carry a bottle of water in a back pack with me or in a cycling bottle and douse it over the hot motor case. should have the same effect and it REMOVES the heat instead of keeping it inside the motor.

what about winding the motor with larger rectangular shape strands for higher copper fill? any thoughts how many percent higher fill could be achieved?
or what about stronger magnets?

as i have damaged the magnets of that MXUS 45 in the oven i think of replacing them. i do not know if it would make sense to install stronger grade like N45SH or N48H in view of performance. in view of heat stability it certainly would make sense.
I got a contact address to a chinese magnet manufacturer from a friend (his company ordered a large amount of custom magnets from them) and i think of making a quote request there.
Dos someone have interest?

 

[macribs]

Madin, if you are going to change magnets why not go for the same samarium-cobalt magnets used on Astro motors? Those magnets can take 200 Celsius of heat without demagnetizing.

IIRC stronger magnets was topic a few pages back, and there was no conclusive consensus that stronger magnets would translate into stronger motor. It would affect Kv and the characteristic of a motor. I am sure others can fill you in on this, as my memory is fading.

 

[madin88]

during my little research about magnets i found out that cobalt samarium types have lower field strengh as types made of neodymium. additionally they are very expensive.
as i would be happy if the motor can withstand 120°C without damage, the great heat stability of SmCo simply would not be necessary. Also i think it would not be clever to swap the magnets for ones with less strength.

I am no expert (maybe someone with better knowledge can comment), but if we replace the magnets in a motor with stronger ones, shouldn't it than have higher kT (and lower kV)? Or with other words shouldn't it than produce more torque at given amps?
I know that magnet strength alone is not all and other things like saturation of the stator core and back iron return must be considered, but shouldn't it work like this in general?
At the moment i have magnets in my motor with lowered strength and it overheats extremely fast when i have demand for high torque.
Because of this i don't think that stronger magnets will make the motor poorer, but as already said i am no expert in this..

EDIT:

i should read more exact. the answers i can found here in the PDF:

https://endless-sphere.com/forums/viewtopic.php?f=2&t=48753&p=1114023&hilit=magnet#p1114023

and from luke

However much magnet field moves past the teeth is how much BEMF you get. Decrease the magnetic strength, decrease the magnitude of BEMF.

It works backwards as well of course, stronger magnet makes lower kV (higher BEMF) which makes more torque per amp, but at the cost of reaching max RPM for a given pack voltage sooner.

Stronger magnets mean more core losses, but potentially less copper loss for a given amount of torque produced, so it really comes down to trade-offs for the given applications needs.

thanks guys for the infos!

conclusion:
stronger magnets should make the MXUS better above a given required torque. with an increase of eddy losses i can live with because i anyway will not often move at the peak efficiency point (who really does?). and as the stator is made of 0,35mm lams i believe the increase will not be that much..
the 9C you tested Justin had 0,5mm, right?