Inrunner vs outrunner power losses

[gabry]

Hi guys, i've done multiple builds using outrunner motors (hub) and i never had overheating issues when testing those motors at no-load condition.

Now i got a batch of inrunners (IPM) both hairpin and normal, geared and direct and all of them when tested at higher speed and no-load they draw relatively high amount of power (5/10% of nominal power) and heat up a lot.

One motor in particular, the biggest (10kw) does reach >80°C after 5 minutes running at 18kerpm (the maximum it reaches at 60V) which is the equivalent of 65 km/h with a total gearing ratio of 7.5 (given the 2.9 of the motor).
Also the motor doesn't heat upin a homogeneous way like the other ones, but on one side reaches 80°C and the other one is still at 60°C.

The power losses seem to be from eddie currents, and eddy current loss is proportional to the square of the frequency so should be correct that the faster it turn the more heat is generated, but it's normal this much and so irregular?

I'm not familiar with this kind of motor, but wasting 6/800W just to spin the motor and reaching so high temperatures wasn't expected, my last build had a qs273, 110kg build, and while cruising up-hill at 50km/h @ 4kW i was at 50/60°C motor temperature (oil cooled).

If it does reach 80°C without load and with a fan blowing air on it, wil be capable or carrying a 100kg motorcycle + 70kg rider + gear without overheatting?

By doing a bench test at high currents (100A) it doesn't heat at all (3mohm phase resistance), but running at 20A phase and high speed it becomes a toaster.

Photos of the motor temperatures after running 10 minutes at 11kerpm.

 

[amberwolf]

Uneven heating could be the way the stator is mounted to the casing; better thermal connection where the heat is higher.

Could be the covers aren't machined to center the rotor in the stator, so the airgap is different on one side vs the other, and this could affect the motor to make it operate less efficiently, and probably affect the eddy current heating.

Could be the stator is wound incorrectly, too few or too many turns on parts of it, or parts of some windings are put on the wrong teeth, etc.

Can't think of anything else ATM.

 

[gabry]

Uneven heating could be the way the stator is mounted to the casing; better thermal connection where the heat is higher.

Could be the covers aren't machined to center the rotor in the stator, so the airgap is different on one side vs the other, and this could affect the motor to make it operate less efficiently, and probably affect the eddy current heating.

I just asked the factory and they showed me the stator on this particular motor is made of "modular" stacks of plates, I assume it's the way this e stacks interconnect that changes the Eddy currents maybe

Could be the stator is wound incorrectly, too few or too many turns on parts of it, or parts of some windings are put on the wrong teeth, etc.

I checked inductance and resistance of each phase and it's the same so windings should be correct.

They also stated that under load it shouldn't heat like without a load, but I don't understand how it can be.
I bench tested in open loop (no load) at up to 100A and the heat was the same as 20A.
I'll build a bench test with two motors to test it under load and free spinning.

 

[amberwolf]

Either the wrong way, or more likely they didn't use anything between the windings and the stator laminations, so that one or more turns' insulation coating was damaged during winding, and is shorting to the laminations to some degree. It's most likely that it is multiple points, on each of the darkened-teeth-windings sections, and that it is more than one phase invovled, so that current from the phase itself is passing thru the laminations and not just hte windings, creating some amount of short circuit between phase FET bridges to the controller.

That would create excessive phase currents on those shorted phases, but might not be a direct enough short to cause a shutdown or controller damage, depending on how it's designed and any limits it has / monitors.

I would expect an FOC controller that monitors all three phases for current to throw an error in this kind of situation, because of unbalanced currents, but if it only monitors two and those are both involved in the short, it might know there's a problem. Or a non-FOC controller that doesn't monitor anything but battery current wouldn't know there was a problem either, until one of the phase bridge FETs blows up. :/

 

[amberwolf]

BTW, some phase shorts to stators (or other phases) only show up while at a higher (operating) voltage and won't show up wiht just an ohmmeter. In these cases you need a "megger" or high-voltage tester. I had an MXUS 3k (450x) with this kind of issue, and used a meter I don't recall the name of now but is in the SFOC5 testing thread and the SB Cruiser thread from 8 or 9 years back, I think, along with the fixes I did to correct it. (sometimes you have to rewind the motor to fix it, sometimes you can use things like CoronaDope to do so).

 

[gabry]

Either the wrong way, or more likely they didn't use anything between the windings and the stator laminations, so that one or more turns' insulation coating was damaged during winding, and is shorting to the laminations to some degree. It's most likely that it is multiple points, on each of the darkened-teeth-windings sections, and that it is more than one phase invovled, so that current from the phase itself is passing thru the laminations and not just hte windings, creating some amount of short circuit between phase FET bridges to the controller.

That would create excessive phase currents on those shorted phases, but might not be a direct enough short to cause a shutdown or controller damage, depending on how it's designed and any limits it has / monitors.

I would expect an FOC controller that monitors all three phases for current to throw an error in this kind of situation, because of unbalanced currents, but if it only monitors two and those are both involved in the short, it might know there's a problem. Or a non-FOC controller that doesn't monitor anything but battery current wouldn't know there was a problem either, until one of the phase bridge FETs blows up. :/

Yes there was a current imbalance of about 10% I could monitor and it did happen just at stator voltages great her than 30V (so when spinning fast enough).

 

[amberwolf]

Yes there was a current imbalance of about 10% I could monitor and it did happen just at stator voltages great her than 30V (so when spinning fast enough).

Then my bet would be a winding short to the stator teeth from insulation damage--probably on both of the discolored teeth's windings.

If you want to try to save the motor (assuming no warranty to void), get some Coronadope and "drizzle" it down into the whole winding sets that are discolored. It will take a while to cure, and will be very smelly, so I'd do this outside--but if the weather is cold it may not cure properly.

There are several versions of it by different companies; I have used both of the types in the linked page for motor fixes, but the clear one worked better for drizzling down into the windings.

https://www.amazon.com/MG-Chemicals-Insulating-Varnish-Liquid/dp/B0BMNR4L3F?th=1

I don't remember which place I actually bought mine from but there may be a link in my SB Cruiser thread or the SFOC5 thread in the part that talks about fixing that MXUS hubmotor.



If possible, keep moving the stator around at all sorts of angles in all axes to "force" gravity to help wick the CD into all of the windings and between them and all the laminations as it's being applied, and as it cures. Otherwise it might miss some of the spots that could be causing the problem, and / or just run off the "bottom" of however you have it oriented during application and curing.

If you needed to do this for a lot of different applications (like mold-making, etc) you could make a motorized multi-axis rotation device to hold the curing part and slowly rotate it in all three axes continously until you turn it off, but it's too much work for just doing this one thing. (I once ran across some 3d-printable versions of this kind of thing but I don't have the links anymore, and apparently don't have the files on this computer).

 

[gabry]

Then my bet would be a winding short to the stator teeth from insulation damage--probably on both of the discolored teeth's windings.

If you want to try to save the motor (assuming no warranty to void), get some Coronadope and "drizzle" it down into the whole winding sets that are discolored. It will take a while to cure, and will be very smelly, so I'd do this outside--but if the weather is cold it may not cure properly.

There are several versions of it by different companies; I have used both of the types in the linked page for motor fixes, but the clear one worked better for drizzling down into the windings.

https://www.amazon.com/MG-Chemicals-Insulating-Varnish-Liquid/dp/B0BMNR4L3F?th=1

I don't remember which place I actually bought mine from but there may be a link in my SB Cruiser thread or the SFOC5 thread in the part that talks about fixing that MXUS hubmotor.
View attachment 381057 View attachment 381058


If possible, keep moving the stator around at all sorts of angles in all axes to "force" gravity to help wick the CD into all of the windings and between them and all the laminations as it's being applied, and as it cures. Otherwise it might miss some of the spots that could be causing the problem, and / or just run off the "bottom" of however you have it oriented during application and curing.

If you needed to do this for a lot of different applications (like mold-making, etc) you could make a motorized multi-axis rotation device to hold the curing part and slowly rotate it in all three axes continously until you turn it off, but it's too much work for just doing this one thing. (I once ran across some 3d-printable versions of this kind of thing but I don't have the links anymore, and apparently don't have the files on this computer).

The windings are already glued together with some sort of resin, which is the one that got the darker color.
I'll try to check continuity between the stator and the windings, but it's weird as there is a plastic insulator I could see between the windings and the metal core.

 

[amberwolf]

It might be a winding-winding short instead, but those are rarer except when windings are oveheated enough to darken the whole motor first.

To test for winding-winding (phase-phase) shorts, you have to disconnect the central Y (wye) point where all three phase windings connect to each other.

(unless t's a delta winding, in which case you have to disconnect the three points where each pair of phases connect to each other, which is also where the phase wires to controller connect to them).

Otherwise a regular meter can't tell the difference between the low winding resistance and a short between them.

But either way, if the problem only happens at a higher voltage than say, a handful of volts, a regular multimeter is unlikely to read the short--it may only even *be* shorted (an arc) at the higher voltage, which usually requires the "megger" / hipot tester / etc I mentioned before to find.

 

[gabry]

It might be a winding-winding short instead, but those are rarer except when windings are oveheated enough to darken the whole motor first.

To test for winding-winding (phase-phase) shorts, you have to disconnect the central Y (wye) point where all three phase windings connect to each other.

(unless t's a delta winding, in which case you have to disconnect the three points where each pair of phases connect to each other, which is also where the phase wires to controller connect to them).

Otherwise a regular meter can't tell the difference between the low winding resistance and a short between them.

But either way, if the problem only happens at a higher voltage than say, a handful of volts, a regular multimeter is unlikely to read the short--it may only even *be* shorted (an arc) at the higher voltage, which usually requires the "megger" / hipot tester / etc I mentioned before to find.

Ok so I've opened the motor again and done a reverse engineering of the windings, everything seems good to me, looks like a dual Three phase winding with 12 slots, 6 each.
A and B phases are wired together, A and B star centers are not.
Resistance seems to be fine, tried with 60V and didn't notice any voltage leak.

I measured 2.6/2.7mohm from phase lead to star center, done for all 6 phase to star center.

 

[amberwolf]

If none of the ohter teeth are swapped around, then that is probably why it's not running right, since there's no inter-winding or winding-tooth shorts.

If the windings for each tooth are individually wound and have their own "taps" to connect ot the other teeth, you can simply swap those two electrical point sets.

If the are not individually wound and the windings continue from tooth to tooth, you would have to cut the *right* wires coming from those teeth and swap those endpoints.

Are both sets of phase winding swapped on those teeth?

 

[amberwolf]

Flipping the polarity wouldnt' usually fix this, just actually swapping the "order" of those two.

I'm unable to follow the diagram vs the numbering to figure out how you created the numbering scheme to determine two are swapped and still call the diagram correct...if one is wrong the other must be too, as far as my brain will allow me to see.

 

[amberwolf]

if it isn't the order that's wrong, and there's no wiring faults or winding shorts, then I am unable to think of anyhthing that would cause darkened windings on just specific teeth*** or cause the current draw you're seeing****

*** (other than factory using different or damaged wire on those, *or* the entire motor was overheated and damaged and then just the non dark teeth were rewound for some reason)

****if the problem occurs with sensorless as well as sensored, then it can't b ewrong positions of the halls or other encoder vs the teeth

 

[gabry]

Flipping the polarity wouldnt' usually fix this, just actually swapping the "order" of those two.

I'm unable to follow the diagram vs the numbering to figure out how you created the numbering scheme to determine two are swapped and still call the diagram correct...if one is wrong the other must be too, as far as my brain will allow me to see.

It's a motor with two three phase subset.
I called them A and B.
Each subset has three phases, 1, 2, 3
Each phase is composed of 2 windings, for a total of 2subset*3phase*2windings = 12 windings.

Each windings is wrapped around a "tooth", each tooth had a numeration from the factory, 1 to 12.

The circle in the schematic represents the stator.
The inductors in black are the windings.
1A, 1B, ecc, are the phase wires (that get merged 1A +1B = phase 1 )
The red bridges are the taps from winding to winding.

The schematic from a theoretical point of view looks correct, if I physically swap 1 and 12 I get a wrong schematic, that's why I'm confused. On the stator there is written 1 and 12 in the wrong order, but then the winding seems to be connected correctly.

At this point this motor is to be trashed, but would've been good to understand what's the cause

 

[amberwolf]

, each tooth had a numeration from the factory, 1 to 12.

ah, that's the key there--the markings are not your markings, they are factory ones. that wasn't clear in your previous posts. since these teeth are modular they probably screwed up and put the ones iwth the wrong marking in those places, and it probalby has nothing to dow ith the windings.

i sitll can't think of anytihng other than the previusly listed reasons for the issue.