Definitive Tests on the Heating and Cooling of Hub Motors

Excellent work :)

The idea seems almost ready to roll out. The only remaining question I have is whether the ferrfluid is stable and resistant to separation over long-term use?

For potting and halls, I assume you could just mask or shutter off the small area of the stator containing them.
 
Justin,

Awsome work as usuall. You the real king of ebike innovation !!!

One of the discusstion caught my attention regarding the poting idea. I think that in fact it might change a bit the inductance of the motor and all related other parameters. I remember i saw a video a few years ago showing test of stator fatigue in the industrial AC motor, I think it was made by ABB or Siemens. It is about end winding vibrations.
This video was clarely showing that on high phase current the stator winding tend to move Inside the stator iron core when strands are not properly tightened together and also that are loose in the hole of the core!!

My guess is that if the winding move it is in opposite direction of force applied as well and make it torsion and on a long period of time might result to varnish damahe due to rubbin against the core winding slot. and otehr phase. it move at the AC frequency the current that is applied to it as well.

So in that case if the windig move in the stator core and there is less work that move the magnet and a bit to move the windig in opposite direction , it might reduce the inductance noh?

By poting the stator this will stop any possible movement and probably also change the inductanc and efficiency of the motor to!

Here is a document pretty interesting about end winding vibration:
http://lib.tkk.fi/Diss/2010/isbn9789526032863/article5.pdf


here is one video similar to the one i watched this one is less spectacular than the other video i saw but it is still impressive even with the stator strongly attached together :

[youtube]lPNU2vniUIc[/youtube]

Doc
 
BSATOM said:
I am still curious about FF "vs" ATF, so I am watching for that post.

Here be that post. Since the motor side plates were already nicely sealed I decided not to bother opening it up and flushing out the small amount of ferrofluid in the motor but just added the ATF in addition, 140mL, making the total fluid fill about 155mL. I wanted enough to clearly wet and submerge the windings and this looked to be enough.

ATF Injectoin.jpg
ATF Fill Level.jpg

I ran the test both at 200rpm and 400rpm, in case there was any substantial difference in the splashing/cooling behavior in the motor between these two speeds, and here are the raw results
ATF Raw Results.jpg

In the 200rpm comparison, the 140mL ATF fill had just over 8 Watts/degree core to shell conductivity, versus 7 Watts/degree for the ~5mL ferrofluid. So it's it's a bit better for sure, but really not by nearly as much as I was expecting. It means that with say 200 watts of continuous heat generation in the motor core, the ferrofluid would have a 28 degree core temperature rise, while the ATF fill have a 25 degree rise, a whooping 3 degrees cooler.

As for the extra drag on the motor from the ATF fill, here's what that looked like:
ATF No Load Drag.jpg

It could be that the presence of the residual FF in the core made that a bit worse than a test with pure ATF, but but mixed liquid had the magnetic properties so diluted that I doubt it's very far off from pure ATF.

As for how this compares, the drag with 140mL of ATF is lower than with 16mL of ferrofluid, but quite a bit higher than the minimal FF required for thermal conduction.
Drags Compared.jpg

And of course no experiment with oil filled motors is complete without a tale of oil leakage. I did seal the small injection hole, but over the course of the testing quite an oil puddle formed all around the connector assembly, with the oil having mostly worked its way out from inside the cable sheathing. There wasn't any leakage from the side plate seals.
ATF Leakage.jpg
 
Doctorbass said:
By poting the stator this will stop any possible movement and probably also change the inductanc and efficiency of the motor to!

It's amazing the windings don't fail sooner with that much movement. I don't think potting them would change the inductance significantly and the efficiency should improve slightly if energy is not wasted in moving the windings around.

On the ferrofluid, having ferrofluid in the gap should increase the magnetic field strength slightly. This may offset some of the drag losses. Since the ferrofluid is attracted to the magnets, there should be less tendency for it to run out.

Was there any leakage with the FF?
 
fechter said:
Doctorbass said:
By poting the stator this will stop any possible movement and probably also change the inductanc and efficiency of the motor to!

It's amazing the windings don't fail sooner with that much movement. I don't think potting them would change the inductance significantly and the efficiency should improve slightly if energy is not wasted in moving the windings around.

Yeah, that's some crazy video footage Doctorbass. I'll have more details on the potted motor core soon, I'd considered that there would be overall ruggedness improvements but thought that would mostly be help with road vibrations and not magnetic forces moving the windings like this.

On the ferrofluid, having ferrofluid in the gap should increase the magnetic field strength slightly. This may offset some of the drag losses. Since the ferrofluid is attracted to the magnets, there should be less tendency for it to run out.
Was there any leakage with the FF?

Have a close read of this post again, reading every word right to the end!
https://endless-sphere.com/forums/viewtopic.php?p=1106508#p1106508
 
Punx0r said:
Excellent work :)

The idea seems almost ready to roll out. The only remaining question I have is whether the ferrfluid is stable and resistant to separation over long-term use?

For potting and halls, I assume you could just mask or shutter off the small area of the stator containing them.


For that I guess temp logging over time would help you notice any significant changes to internal motor temp. As I can picture any separation will not keep FF in place and that would lead to higher internal temps. I guess real world results will tell us over time.

Btw. What about FF viscose? Will thicker FF be a negative thing? And what kind of oil are they using with the FF? Will FF work even in sub zero temperatures?
Or will it be frozen like a brick a rip magnets apart when wheel start turning?
 
The data sheets don't typically state a lower temp limit but if you bug ferrotec enough you might get a viscosity vs temp chart. I'm sure it's linear so you should already be able to extrapolate from justin's drag vs temp data. Definitely won't rip off the magnets or anything like that. We are talking turning from motor oil consistency into honey consistency at very low temps. I can tell you real world this has not been an issue as things that use ferrofluid tend to warm up very quickly when used/moved
 
I think I will test this FF stuff out on my cromotor.

I know that ATF fluid worked really well in the hub-motors, with the only drawbacks being leaking and magnets becoming unglued.

Since I had to reglue my magnets in my cromotor I used Hysol EA9394 expoxy which is made to not be affected by stuff like oils.

I never wanted to try ATF fluid because of all the people who had leaks no matter how well they tried to seal it. But if the FF doesn't leak then that may be the way to go.
 
Their is lots of inrunner ebikes out their if you dont want to test for FF maybe at least just do the ATF then we can see how it effects the cooling and no load losses. Also how did you come up with the ATF minimum level ? U could try with less and test no load at the point where it starts going up as i bet it splashes at lower then winding level so u could get good cooling with less drag . Also u could try owt thinner oils at same level and see the difference.
 
Nathan said:
Their is lots of inrunner ebikes out their if you dont want to test for FF maybe at least just do the ATF then we can see how it effects the cooling and no load losses. Also how did you come up with the ATF minimum level ? U could try with less and test no load at the point where it starts going up as i bet it splashes at lower then winding level so u could get good cooling with less drag . Also u could try owt thinner oils at same level and see the difference.


It's not a lack of his interest to test, it's the mechanism of function.

It clings into the gaps between magnets. If you have an outrunner spinning, it can continue to dwell in this void. If you have an inrunner spinning, it is slung off the rotor once it spins up.
 
flathill said:
The data sheets don't typically state a lower temp limit but if you bug ferrotec enough you might get a viscosity vs temp chart.

Indeed, I did just that but the most technical response I got was this quote "We have a general rule of thumb that the viscosity of oil based ferrofluids will decrease by a factor of 2 for every 10oC the temperature rises, so what you are seeing is fairly normal"

If we go back to my original drag power vs. temperature plot, the baseline power for no fill at 430 rpm is 55 watts, so drawing a line on the graph here we can see how much additional drag is a result of the fluid viscosity


The best fit line I got by setting the viscosity to decrease by a factor of 2 with ever 22oC increase in temperature, rather than every 10oC. Although we should note that I was monitoring the copper core temperature in this plot here, rather than the fluid temperature which is likely somewhere between the core temp and the shell temp. So taking that into account, the actual max fluid temperature was likely a lot less than shown in this plot, perhaps 50oC or so, which would have things closer inline to their 10oC halving.

Kodin said:
Quick thought here, have you done any testing with thicker ferrofluid solutions to see if there's any difference? Honestly if we are simply filling until light contact is made, a thicker solution may not have any drawbacks since we're riding the threshold of thermal contact vs. drag... Any consideration toward testing different viscosity options?

Yes some other types are on order to get here around Oct 25th, although my interest in the more magnetically reactive and higher viscosity materials isn't so much for putting it in the air gap, but to see if we can make little magnetic holders that could sustain a bulge of FF to wipe around the copper end-turns on the sides of the stator, where there isn't already a magnetic circuit preset to hold the fluid there. It's a bit hair brained but seemed worth a try.
 
An interesting test would be to run a shaftless stator underneath a pan of ferrofluid to see whether or not you get odd oscillation patterns. Might help profile whether or not that's practical. Could be that with the 3-phase A/C switching going on the fluid might behave in an unexpected way. Plus you know, it looks cool... :p

I assume this is with the potted motor? I'd think an un-potted motor will likely induce far more drag.
Are you thinking similarly to an axial-flux motor design where you have magnets mounted on the side covers in-line with the windings to at least hold the fluid...?
 
try testing with the magnets on the side covers without fluid to check for eddy current heating of the windings.

with the FF the return path will be better defined so maybe it might not be an issue, but the no FF would have the most stray flux so that would be the worst case baseline.

there also may be an optimal orientation, not alternating like the rotor mags, but one south ring and one north ring that will form a ring of fluid. really interested to know if the fluid stays in place at high speed on the side walls or will migrate due the electromagnets back to the stator/rotor gap
 
Right when I begin thinking I couldn't be more impressed with you, Justin, you go and do something that continues to blow me away. Not only is this ferro-fluid heat-bridging a completely new thing that has unexpectedly come out of left field (at a time when most ebikers have not even heard of ventilating or oil-cooling yet), you are running pure research, and publishing the raw data.
 
justin_le said:
Have a close read of this post again, reading every word right to the end!

I managed to miss the entire page!

Great tests and test results. x2 heat transfer is a major improvement.

When the stator is installed, the pattern of FF will likely be different. The gaps between magnets should no longer be the most attractive spot. Most of the flux will be concentrated between the magnets and the stator iron. It would be interesting to look at the area around the stator teeth with FF in the gap.

The stuff they use for loudspeakers should be pretty low viscosity.

One other thing is some of the losses due to drag may be offset by increased flux in the gap.
 
Punx0r said:
Excellent work :)
For potting and halls, I assume you could just mask or shutter off the small area of the stator containing them.

Yes, we had some discussions here about how easy or hard it would be to replace the halls in the potted stator. Personally I'm quite excited to have the hall chips be fully potted, because the largest cause of hall failures/hall issues is water ingress corroding the leads, and the occasional case of vibration fatigue causing a leg to break off. Both of those are eliminated. But there is always the chance of electrical failure if people have an axle spinout/wires severed and short the hall signal lead to V+ via one of the phase wires. But in principle (and some day in practice I hope) the hall sensors would have a small protection circuitboard interface that was fully tolerant of any kind of wild voltage exposure and then there's no reason you'd ever need to "unpot". In any case, for this particular experiment the hall sensors wires were in an annoying location for the mold and so we just snipped them all off.

We really had no idea how easy or hard it would be to encapsulate the stator of an existing finished motor with resin, but as a first attempt we machined a mold cavity from a thick sheet of UHMW plastic, with an geometry that was a snug fit and on the insides went right up to the spokes of the aluminum stator support. We poured thermally conductive epoxy resin in the bottom of the mold...

Mxus Potting Mold.jpg

Then inserted the stator, with the phase wires on the upper side so that he bottom half of the stator going into the mold didn't have any wire channels to worry about...
Pressing in Stator to Potting Mold.jpg

It was a pretty snug fit, so we needed to use a proper press to get the stator all the way snug on the bottom of the mold..
Press Stator Fully.jpg

Then we used tape and plastecine to damn the inside of the top half of the core and seal where the phase wires come out, and continued pouring resin in from above. Even with the thick viscosity of this thermally filled resin, it seemed to have little problem navigating down between all the copper wires and stator teeth and really saturating/filling the whole thing by just sitting for an hour with no vacuum degassing or anything like that.
Fully Poured.jpg

After an overnight oven assisted cure, things looked pretty good for our moment of truth. The stator came out of the mold with just the slightest effort on the press and looked glorious..
Potted Core with Mold.jpg

Potted Core.jpg

We've got it now installed back in the motor shell and ready for testing to see what effect this has on thermal results, but are currently all out of the ferrofluid samples and are waiting on the big batch to arrive in order to continue the tests.
One thing I realized when I had the MXUS motor apart is that the thermistor we had installed was glued touching both the winding and the core, and so would be showing an average temperature of sorts rather than the pure copper temperature. I think to better discern the effects of potting vs. not potting we'll want to repeat some of those experiments with one probe embedded only in the copper, while the other temperature sensors is only on the stator laminations.

flathill said:
try testing with the magnets on the side covers without fluid to check for eddy current heating of the windings.

For sure, this was one of my main concerns. I think it won't be too bad for this application of cooling the end turns in a DD motor since the steel core that the magnets are moving past is laminated to reduce eddies, but the other area of usage that intrigues me a lot is to use small pockets of contained FF to help shed the heat out of geared motors. In a geared hub, just putting FF in the air gap won't do wonders because the rotor is still has an additional air gap to get to the motor shell. But if we could find a way to have small magnets in the motor shell that support a FF thermal wiper to pull the heat off the rotor without inducing too many eddy currents in the metal, then that could prove fruitful.

there also may be an optimal orientation, not alternating like the rotor mags, but one south ring and one north ring that will form a ring of fluid. really interested to know if the fluid stays in place at high speed on the side walls or will migrate due the electromagnets back to the stator/rotor gap

Yeah, when I have the fluids I will run some experiments, but my hope is that you could design the internal geometry such that if the FF gets flung off the side plate magnetic support, that it would reconnect with it later and not wind up in the air gap space. Or if it does wind up in the air gap, that it gets pushed out (as we saw happened in my first test when the side plates weren't sealed) and then drawn up by the side cover mag. structure. This could all be a moot discussions because we might find from doing the potted core test that there are little extra gains to be had versus just a small amount of FF on the airgap and that's it, so we'll see.

spinningmagnets said:
ot only is this ferro-fluid heat-bridging a completely new thing that has unexpectedly come out of left field (at a time when most ebikers have not even heard of ventilating or oil-cooling yet), you are running pure research, and publishing the raw data.

Thanks. What's awesome is that this came pretty much out of left field for me too directly as a result of ES. I've been familiar with FF's for quite a while but it wasn't till Fetcher made mention in this thread here that it clicked how insanely ideal this could be as a thermal heat flow solution for direct drive motors. So yes here's to sharing ideas and results, and I'm glad to be able to contribute back with the tools and motivation to validate some of these concepts that others have experimented with.
What's really staggering (and kinda sad) is just how little R&D of any kind at all goes on at these giant factories and overseas companies that turn out ebike motors and other parts by the millions. It's all about moving the most amount of products with the least expense. There is collectively way more research and experimentation done by amateurs here on ES than all the ebike companies I've ever visited combined.
 
Major improvements most time come from many small improvements stacked together. With e-bikes, you have a ideal development platform. You have economy, weight, efficiency and cost considerations to name some of the constraints that need to be balanced for most. Pushing in any one direction is a balancing act of sorts. This FF improvement is one of those that seem to be a significant one with little down side. Look forward to seeing how the potting adds/ works to pull some additional heat. Great work.
 
looks sweet. another benefit of potting is it makes the motor slightly quieter. good job getting the potting down in between the teeth that should reduce windage losses with or without FF
 
Offroader said:
Can someone explain what exactly potting a hub motor does?


Offers the potential to improve cooling performance (and hence improve both burst power and continuous power handling), as well as weatherproofing reliability.

Potting also doesn't leak or weep or drip unlike many fluid based solutions (with FF being an exception due to clinging into the gaps between magnets.
 
Anyone know how abrasive ferrofluid may or may not be? I'm curious as to the long term impact on the stator/rotor gap.
 
Samd said:
Anyone know how abrasive ferrofluid may or may not be? I'm curious as to the long term impact on the stator/rotor gap.

I'm sure it's quite abrasive if between two rubbing parts. In the air gap, it should not be abrasive at all due to the lubricating effect of the oil.

Interesting points from a manufacturer of ferrofluid:

The thermal stability of a ferrofluid is related to particle density. The particles appear to behave like a catalyst and produce free radicals, which lead to cross linking of molecular chains and eventual congealing of the fluid. Catalytic activity is higher at elevated temperatures and, therefore, ferrofluids congeal more rapidly at these temperatures.

High magnetization ferrofluids are of interest as they produce volumetric efficiencies of magnetic circuit designs leading to lightweight and lower cost products. They can also be used to reduce reluctance of magnetic circuits and fringing field thus increasing useful flux density in the air gap. The domain magnetization of magnetite ultimately limits the maximum magnetization value that can be realized in a ferrofluid.

Thermal conductivity of a ferrofluid depends linearly on the solid loading. Fluorocarbon based ferrofluids have the lowest thermal conductivity of all commercial ferrofluids, therefore they are the least desirable materials for heat transfer applications.
 
What brand and type potting liquid did you use ? and what temperatures and how long did you bake it for ?
 
Agreed about no abrasive impact in the air gap. But evidence above suggests that it bridges the gap.
 
Back
Top