Definitive Tests on the Heating and Cooling of Hub Motors

jmz said:
The net effect is that I believe extremely good thermal bonding of the magnets and rotor backing to the shell will be required if field weakening is to be employed while increasing heat transfer through the rotor, hence potting of the rotor up to and perhaps over the magnet surfaces will be required when using FF.

"Required" is a really strong word here. In practice when you have FF in the gap area, then even under what I would call heavy loading situations the motor core is typically not going to get more than like 70-80 oC, the shell and magnets of course even cooler still, and that's way below the point where demagnetization even with field weakening currents would be of concern. If you're using the fact that there is FF in order to dump much more power into the motor core than you would otherwise do, then yes that would result in hotter magnets and you'd be prudent to have a temp sensor and do thermal rollback as always.

As has been said before, if you put the same input power in a motor with and without FF, the final steady state shell and magnet temperatures will be quite similar, even though the core in the non-FF case is much hotter. It's only during the transient stage as things are warming up that the motor with FF will show warmer magnets and casing.
 
flathill said:
Remember weaker magnets can make a motor more efficient (at certain loads/speeds) and run at higher speed but it will have less torque upon the second acceleration if the magnets are hot.

I think that is why the 2016 Chevy Volt has 2 electric motors. One motor has strong magnets and the other one has the cheaper weak ones.
 
I have had suppliers tell me the magnet was rated to 150c when it was really a 120c H grade. Just stick it in an oven for a quick test...but yes under operation the working point will move down. the oven test is just a simple test. you meed to do a load line analysis to know how much margin you have before hitting the knee on the BH curve. this is simple to do with FEMM

in any case with FF just keep the windings under 120c and you should be OK with any made in china hub motor and with good quality FF it will not turn to sludge

but yes if you are using controller field weakening you might have to be more careful about partially demagnetizing ur mags with FF in a high power vented motor
 
justin_le said:
If you're using the fact that there is FF in order to dump much more power into the motor core than you would otherwise do...

:wink:

I'm still pretty new to this, I thought I saw reports of stators quickly reaching 150+C in near-stall operating conditions.

In the PMSM world, strong field weakening is considered potentially hazardous even at conventional operating temps and well-defined magnetic parameters. Perhaps you can be reasonably sure that those stalled rotor power dumps are temporally separated from high speed, high-FW-ratio runs.

edit: To reiterate slightly, my concerns centre around the saliency of the motor in combination with the possibly variable/unknown/underspecified grade of magnet used, that might have only been designed and tested up to rated power in the Q-axis only.
 
yes there is a limit to how much field weakening you can use and it is totally dependent on magnet temperature and the working point

with FF the magnet surface temp may be different than thr shell temp. it is possible to partially demag magnets in situations like this which is why IPM motors are better the surface mount in some apps (see new Zero IPM motor)

with certain motors it it possible to demag at room temp with aggressive field weakening or a controller malfunction (fires at wrong time)
 
So I've now finished a round of testing on the Crysatlyte 'H' motor series just like was done with the MXUS earlier. With the three IR temperature sensors some decision needs to be made for what is the representative "shell temperature". For these results here I'm using the edge view IR temp sensor rather than the side plates for calculating the core->shell conductivity because this is more representative of where the heat is now flowing. The results aren't too different from the MXUS case up to 6mL. I was presuming it would similarly stay flat around here, but when I squirted in the rest of the syringe to give a full 10mL of till then there was another noticeable boots on the conductivity right to 8 watts/degree.

H3525 conductivity vs mL.jpg

Notice though that this is as measured to the hub shell and not the side plates. The side plates at this point were ~2oC cooler than the perimeter and if I assumed the conductivity to that temp as I did when running the MXUS experiments, then it would be more like 7 Watts/degree.

I haven't fully flushed out the stator->shell conductivity vs. RPM in this case yet, but based on some assumptions on that front I've put up the Crysatlyte H motors series with and without the Statorade fill on the trip simulator, so you can see in any kind of simulated load scenario how the core temperatures, wh/km, and all that compares with and without the ferrofluid fill:
H Motors on Trip Sim.jpg
http://www.ebikes.ca/tools/trip-simulator.html
 
flathill said:
When I make recommendations about leakage/wicking and lifetime, that is more directed to Justin who has to put his money and his customer satisfaction on the line, when and if he decides to offer an off the shelf motor with FF.

Thanks Flathill, these notes of warning stemming from all your experience of course have all been duly noted and are much appreciated. The material certainly seems to have phenomenal wicking powers to slither over surfaces, and I'll make sure we note that side plate sealing is essential. So far the stock factory seal on the Crystalyte motor I have under test hasn't shown any signs of fluid escape, but what goes on with daily use in the field could be another matter.

On a related front I did want to start some elevated temperature testing of these fluids in our lab oven. Can you comment if the gelling/gumming effects are in any ways affected by agitation and strong magnetic fields, or is it all the same just to have a sample sitting still in a container?
 
how many magnets does each motor have. more magnets means more "wipers" for a fixed airgap and diameter

my ultra motor has way less magnets which is proabably why I only had to use so little fluid

but more magnets means more wipers which means less thermal resistance
 
flathill said:
with FF the magnet surface temp may be different than thr shell temp. it is possible to partially demag magnets in situations like this which is why IPM motors are better the surface mount in some apps (see new Zero IPM motor)

thank you for bringing this up. Thats a very good point we should think about.
yes, the magnets probably may have higher temp as the back iron / spoke flange. Many motors, and also MXUS have rectangular magnets (no curved shape) what means they do not have much contact to the iron so not very good thermal flow.

Two days ago i had realized that I HAVE DAMAGED the magents of my mxus when i put the rotor into the oven for drying after painting it black.
The oven was set to 75-80°C and it was about 2h in there.
When i installed that motor to my bike i realized its performance was extremely poor. at low torque / power there was not much difference noticeable, but when i was trying to race up that hill i always do when testing a new setup, i could not climb it.
The winding temp jumped from 50°C to 120° in only 10-20s. First i thought there is a problem with controller settings, but i was not able to program it more efficient and than i thought about the oven and the magnets.. :(

THAT was a really stupid idea, but i had in mind that N42 Magnets are stable up to 120°C as manytimes mentioned here in board, but after doing some research i found out the max operation temperatur point of these magnets depends on the thickness, and thin types as used for hub motors have poor heat stability.
As my unintentionally "test" shows, they will get irreversibly damaged at only 75°-80°C.
N42SH types with this dimesions would have a higher operation point of 120°C or more, but does any of the popular motors have such?

here i have done my magent researching:

https://www.kjmagnetics.com/blog.asp?p=temperature-and-neodymium-magnets
https://www.kjmagnetics.com/blog.asp?p=magnet-grade

now i need to disassemble that motor and swap the rotor. Since i have seen it was manufactrued VERY POOR with the entire outer part of the iron ring beeing out of center (in relation to the inside or notch of sidecovers), it at least will not be wastful at all.
 
75-80C should not have caused them to demag...but maybe having only the rotor in the oven, not the complete motor could have caused the failure as the working point shifts to the left on the BH curve with a larger airgap. in your case there was the maximum airgap possible as there was no stator to return the flux

Note in some very specialized applications (halbach arrays where normal glue contaminates the apparatus from off gassing) the magnets are not glued, they are soldered in place to aluminum holders with the mags already charged. you may ask how that is possible if the solder melt temp is rated higher than the rated magnet temp. the trick is the provide no air gap at all (steel return) so the working point is as far right as possible when you put it in the oven to melt the solder. then the holders with the mags attached are assembled into the array by sliding them in machined slots. the the holders are fastened in place with set screws

also be aware that with cheap controllers that do not control phase current, it is often the weakening of the magnets that causes the controller to fry. You take it out on a run and everything is good. No you let the motor rest and the magnets heat. Now on the second standing start your controller pops (less back emf). most people never make this connection. This is why it is not a good idea to use RC controllers with ebikes. Matt gets away with it because he has SmCo magnets in his Astro motors but he likely doesn't realize that is the reason so he recommends RC controllers to everyone
 
flathill said:
75-80C should not have caused them to demag...but maybe having only the rotor in the oven, not the complete motor could have caused the failure as the working point shifts to the left on the BH curve with a larger airgap.

That's a good point about the working point being shifted without the stator present. In case others didn't see it that's a really great reference article about this here:
https://www.kjmagnetics.com/blog.asp?p=temperature-and-neodymium-magnets
Madin88, can you confirm for us that the no-load RPM of the motor running just a conventional (not FOC) motor controller has increased substantially since putting your rotor in the oven? If the no-load speed of the motor is still the same, then it's hard to be convinced that it's seen demagnetization rather than something else. If it's definitely spinning faster now then yes that's a sign that the MXUS motor magnets are not of a very high temp variety.

I've been using an oven at 100oC with one part thermally cured epoxy for bonding the magnets to the steel backing ring in my prototype motors, and hopefully haven't been loosing magnetization from them in the process! The magnets I'm using are N35SH grade which is supposed to be rated at 150oC.

On the thru-axle motor design I've got a 2mm gap between all the magnets, so in principle I'll be able to stick a thermistor in there between the pair of magnets without touching the steel ring and be thus be able to measure the magnet temp independently from the steel backing ring temp to see how different they are (at least when the motor isn't spinning).
 
If you were using a household oven, I've found they can float +-50degC of the temperature set on the dial when at low temps. The same oven works fine at higher temps.
 
pick urself up an old used F.W. Bell gaussmeter from ebay. Even if it is not dead on accurate it will still be useful for comparison purposes (such as measuring before and after oven cure). You may even be able to machine a slot in the back iron so you can measure the magnets without pulling the motor apart. This may be useful when you start experimenting with field weakening to see how hard we can push these motors (at x stator temp you can use y field weakening setting). that way you could tie your thermal rollback to ur field weakening setting. at least with FF you will know the ~surface temp of the magnet from the stator temp. real world motors are not going to have magnet temp monitoring so this is an advantage in a way
 
about magnet temp grades

a 120C N40 magnet is usually cheaper than 150C rated N35 magnet. The most expensive element in a neo magnet is dysprosium which give the magnet its temp resistance.

you really have to monitor your supplier to make sure they don't slip and ship you a cheaper temp grade magnet. this can happen to MXUS also. very common problem with certain neo suppliers
 
flathill said:
75-80C should not have caused them to demag...but maybe having only the rotor in the oven, not the complete motor could have caused the failure as the working point shifts to the left on the BH curve with a larger airgap. in your case there was the maximum airgap possible as there was no stator to return the flux

Got it in one.

Soooooo, anybody know some good compounds for rotor potting that will cure at room temperature? Otherwise a big steel core matched to the internal diameter of the rotor would be handy.
 
justin_le said:
At the moment I'm running a set of FF mL vs conductivity tests on a Crysatlyte H3525 hub motor, and in this one I just drilled a hole in the motor plate without removing the side cover and breaking the original seal. As a technique, assuming that you don't allow chips to fall into the side cover hole when you drill it (ie drill upside down so that gravity causes all the chips to fall out) then it means you don't even have to deal with sealing the side plates, you can leave the original seals in tact.

I speak too soon! This was true for all the conductivity tests that I did at 200 rpm even with 9mL of fluid. But when I then ramped it up to do the 400 and 500 rpm tests, lo and behold there was a line of spray marks down the tunnel. So the stock seal that comes at least on the Crysatlyte H motors isn't guaranteed to be fluid tight.
H3525 Leakage, edge view.jpg

In this case, it was only the left side cover that had some liquid escape, and looking closely it seems that there are a few mars on the side cover face like you'd get if the cover was pried off with a screwdriver at some point. On the right side cover there was no signs of any leak in the gap.
H3525 Leakage, side view.jpg

So the lesson being if you do have an unopened motor you can't necessarily count on the factory seal to hold an overfill of FF in place when the motor is spun at fast RPM's. I do wish I had run this exact test first with just the 5mL of fill.
 
hillzofvalp said:
liveforphysics said:
If you were using a household oven, I've found they can float +-50degC of the temperature set on the dial when at low temps. The same oven works fine at higher temps.

I was going to say the same thing.. they don't have smart pid circuits usually.

Totally agree.. Conventional oven are not the best...

My best advice to all of you guys that want to bake epoxy or paint is to make a poor man oven using cardboard box that you can fit the part inside and use hair dryer output that you thru in a hole you make in the box!! this make an excellent convection low temp oven!!

Usually our wifes dont like alot seeing our painted motor rotor in the oven and have the entire kitchen smell like toxic !! :lol:

I have found that i can get very stable temp to about 70 degree C withc can cure 3M scotchweld epoxy in 2 hours and meet the shear spec after that ! ( see my DP420 shear test video about that ... you can see the diy oven i think! :wink:


Now about magnet that loose magnetisation... i did not had any loss in power after i used the oven at my work that is +/-0.5 degree stable at 80 degree C... Even... 4 years ago i had my Giant with 5303 that reached 260 degree C at the stator after i have bring it to someone that never watched the temp alarm... the motor was smoking like hell!... and surprizingly i only felt very little loss in performance... magnet was probably over 150 degree C on the Clyte! But magnets a are thicker on these heavy motor i think...

Doc
 
or you can just make ur motor an oven and run it with a dc power supply

adjust v until internal ambient stabilizes at the cure temp of the epoxy paint

pre drill the FF fill hole when spray painting and leave it open while the paint cures

coat the whole inside of the motor with flat black paint after cleaning with IPA

take time after painting to seal the side covers and also the screws by the heads (using Threadlocker but just a tiny amount, remember excess threadlocker never cures if standard loctite blue and can mess with the FF, or use CA type plastic safe threadlocker)

after curing paint and sealant and letting air out for a day add FF

Or dont use a heat cure epoxy paint!

for gluing magnets you may need an oven

Note many structural adhesive epoxy that is of the heat cure type will still cure at room temp it just takes much longer

The reason many are heat cure is to speed production

But some are really truely heat cure and never reach fully strength until heated (read the data sheet)

if you do it right ur motor will never leak and the FF will last over 15 yrs
 
flathill said:
or you can just make ur motor an oven and run it with a dc power supply

adjust v until internal ambient stabilizes at the cure temp of the epoxy paint

pre drill the FF fill hole when spray painting and leave it open while the paint cures

coat the whole inside of the motor with flat black paint after cleaning with IPA

take time after painting to seal the side covers and also the screws by the heads (using Threadlocker but just a tiny amount, remember excess threadlocker never cures if standard loctite blue and can mess with the FF, or use CA type plastic safe threadlocker)

after curing paint and sealant and letting air out for a day add FF

Or dont use a heat cure epoxy paint!

for gluing magnets you may need an oven

Note many structural adhesive epoxy that is of the heat cure type will still cure at room temp it just takes much longer

The reason many are heat cure is to speed production

But some are really truely heat cure and never reach fully strength until heated (read the data sheet)

if you do it right ur motor will never leak and the FF will last over 15 yrs

That's a good idea.. :wink: but problem is that when rreassembling the motor after painting, you dont want to scratch the fresh painted magnets and stator when puting everything together... :x so paint must be dry...

Doc
 
If u let it dry a couple days before assembly it will be hard just not full cure (depending on the exact type)

A good way to put a motor together without the stator being sucked to the rotor is to use a drill press

But anyway we should mot sidetrack the discussion with paint talk

---

About testing FF in a dish in the oven...this not a comprehensive lab tyle test but is useful for comparing fluids. After a week at 100c or whatever put the fluid back in the motor and compare to fresh fluid (if it has not gelled). Can also check for evaporation by weighing before/after. In use in the motor is much worst case as the fluid may mist/splatter from movement, accelerating evaporation; magnetic shearing is also at play in a running motor, but the oven test allows you to at least do some sort of bench test
 
justin_le said:
Madin88, can you confirm for us that the no-load RPM of the motor running just a conventional (not FOC) motor controller has increased substantially since putting your rotor in the oven? If the no-load speed of the motor is still the same, then it's hard to be convinced that it's seen demagnetization rather than something else. If it's definitely spinning faster now then yes that's a sign that the MXUS motor magnets are not of a very high temp variety.

i plan to do this but first i need to get hands on a trapez controller. with the adaptto and the influence of all the settings i may probabaly not get accurate data - even without field weakening. what i now can say is i have not noticed any strange behaviour in RPM. I got 90-95kmh with the wheel off the ground what should be normal for a sine controller, or maybe a bit higher as normal (80V batt, 22" wheel, 12kv motor)

@ flathill

in that very informative blog about magnetics there is written with BIG FAT TEXT that max temperature operation point depends on the shape (thats the performace coefficient). that is with the magnet in free air without back iron or other things, but i believe it will not make much differene. Do you know more about that?
One thing is clear: a N42 magnet with thin shape in free air can get partial damage at 80° or even lower :!: there is an example in that blog.

What i do not know is the influence of time the magent is exposed to that temperature. the question is does it make a difference if its only 1 min or 1h.

liveforphysics said:
If you were using a household oven, I've found they can float +-50degC of the temperature set on the dial when at low temps. The same oven works fine at higher temps.

yes the 75°C i have set the oven to may not be that accurate, but not that extreme. when i took it out of the oven it did not felt hotter as i expected and i was able to touch it shortly without burning my fingers.
to be sure i will bake a piece of metal at same temp setting and measure the temp.
 
1 m or 1 h does not make a difference. think of it like wind- up spring toy. if you wind it too much it doesn't matter how long you keep it in the overwound state before you let it go. the damage is done

another aspect of FF viscosity that may turn out to be important is that thicker fluids are less likely to splash (be ejected from the gap, micro droplets)

if you start beta testing your motors in the field it would be best to pick testers who will use the motors worst case (a form of accelerated life testing)

1) skinny tires pumped hard (road or commuter type bicycle)
2) no suspension
3) climbs up a long steep hill everyday (1+ kWh battery)
4) rides on a cobble stone road on the hill or after big hill relatively slowly (when the motor is hot and fluid viscosity is lowest, less centrifugal force at low speed)
5) optional, uses field weaking
6) repeats the same trip at least 4-5 days a week (commuter)

Lucky with a hub motor if if some fluid does find it's way out of the gap, it will most likely find its way back into the gap given the centrifugal force. the bigger issue with splashing is the fluid may age faster

a normal splash test is done by dropping the ferrofluid in a test jig from a few feet onto paper lined concrete, and then examining the droplet formation, but this is not as relevant to a hub motor
 
Suddenly there is another factor to read up on, the secrets of magnets. Even more technical english. :? :oops: :cry:

I saw one thread here where one did swap some magnets in a motor because magnets cracked somehow. Would it be possible to swap out all magnets with different higher quality magnets? What would be potential benefits swapping magnets? Other then higher temp handling.

What happens if one swap magnets with higher quality and stronger magnets? How will the characteristic of the motor change? Lets imagine we use the mxus 3K for this. Will swapping to stronger magnets produce more or less torque? Would KV change?

Is the choice of type of magnets to use in a motor just matter of price? Would hubmotors benefits from using rear earth magnets like Astro uses?
 
Back
Top