King Song S22 motor stator slippage issue: more severe than expected

[5Cauac]

Ok, I'm going to go against a few here but I believe they are the same "type" of clunk, all bad, with the 4th one slipping more than the others. 

Edited September 15, 2022 by 5Cauac

[oolong]

Personally I still think repeated testing can be a big problem.  If you keep running these test, your motor will eventually fail and possibly fry your motherboard in the process.  I hope it's not the case that people are running these tests regularly while still riding their wheel, because they would only be exacerbating their problems.   I think doing a test until you hear your first clunk is pretty much all you need in order to identify that your motor needs replacing.

It's like spraining your ankle and then dancing/jumping on it repeatedly to check on the injury.

[dycus]

On 9/14/2022 at 3:46 PM, supercurio said:

Do you know about a current probe which could be used to record phase current, up to 220A and possibly beyond?

[...]

Any suggestion of equipment and methodology welcome!

A clamp-on current probe could work and is easy to use. The biggest downside in this case is that for high-bandwidth ones (more than a few MHz), they tend to be very expensive. High hundreds to low thousands of dollars. The high current ones also tend to be lower bandwidth.

Though low bandwidth might be okay for this analysis. The control board uses high-frequency PWM (seems to be above hearing range, >20kHz) and the inductance of the motor to smooth the current out. So a much slower current clamp (several kHz) should be able to see the longer-term (hundreds of microseconds) current changes.

If it turns out higher bandwidth is needed, I'd suggest a cheap differential probe and probing across the phase current sensing resistor. Better yet, there's likely a current sense amplifier that's already amplifying the signal. If you're adept at tracing PCBs and reading datasheets, you could probably probe at the output of the amplifier and get much less noisy results.

Looking on Amazon, there are some cheap high-current probes that might have enough bandwidth.
Hantek CC-650: $100, 20kHz, 650A. I've used Hantek's CC-65 (same probe, just lower current). It worked well enough for the price, though the volts-per-amp marking on the meter were wrong.
OWON CP-05+: $225, 400kHz, 400A. Haven't heard of this company, just saw it on Amazon

Micsig makes cheap differential probes. DP10007 ($190, 700V, 100MHz) could work to probe the current sense resistors directly. I have a Micsig current probe (CP2100A) and it's great, very nice quality. I'm sure their diff probes are pretty good as well.

 

I don't know if the clunk would be from high-frequency (>1MHz) current spikes or if it's a lower-frequency effect that takes place during several motor commutation cycles (hundreds of Hz). I would suspect it's on the low-frequency end; very fast current spikes, even happening every PWM switching cycle, should be pretty well smoothed by the large flywheel mass of the rim/tire and the large motor inductance, and I wouldn't expect they would make such a pronounced, "long" sound like we hear with the clunks. (Long being hundreds of milliseconds versus the dozens of microseconds between PWM cycles)

[Elliott Reitz]

On 9/14/2022 at 7:52 PM, supercurio said:

Some observations on that:

...

@NErider the disagreement on what the "clunk" sound is means that the new "fixed" motor, still producing could still have defects, hence the high interest in investigating this fully before hundreds of motors get replaced.

If it's the controller: The firmware must be improved, but there's no hardware flaw.

If it's indeed the sound of the stator being loose, then there's a problem.

My 2021 S20 wheel working with new batteries and controller yet makes the clunking sound.  I shared a vid a few pages back.   It passed a 1x stress test yet sometimes makes the clunk (it comes and goes).

The clunk:

it comes and goes 

seems to be between the bearings and cover plates

It has "spring load" then a pop that's the clunk sound.  Like have to rotate the wheel to the spot but then no matter how slow crossing the spot the pop is the same.

Side-cover hammering can make the clunk appear (hit on the left side) or disappear (hit on right side last).  This was repeatable till I hit it really hard on the right side (just now).  Still I won't be surprised if it returns.  

I'm still riding it daily now.  400 miles since August 1st.  That's 600 miles on this motor.   

Edited September 16, 2022 by Elliott Reitz

[supercurio]

@Elliott Reitz that's the "knock" sound you are describing - the S22 is a wheel of many sounds.
They each have their own made up name, somewhat related to how.. they sound 😆

Clunk typically occur during a stress test, when the motor changes direction or sometimes when climbing stairs.

[Elliott Reitz]

21 minutes ago, supercurio said:

@Elliott Reitz that's the "knock" sound you are describing - the S22 is a wheel of many sounds.
They each have their own made up name, somewhat related to how.. they sound 😆

Clunk typically occur during a stress test, when the motor changes direction or sometimes when climbing stairs.

Well you're correct the sounds vary and lack name-definition.  The magnet-slip sound may be the slip sound.  The side cover knock or clunk is a little different.  I call it a clunk rather than a knock because it preloads via rotation then no matter how slow releases the pop of a clunk or knock.   Compared the slip sound the knock is very similar to hitting the side covers (as close as possible) to the hub with a hammer.  And hmmm doing so can make the knock worse or stop depending on which side gets hammered.  

PS:  I will be surprised if hammering the side cover can make it stop for good.  The way it comes and goes makes me think the side covers just aren't tight enough by design.  Here's an article I dug up about the shaft and bearing fit.  How to Determine Bearing Shaft and Housing Fit - Baart Group.  FTL:  "With a loose fit, bearings will creep – the unintended movement from or within their mounting place – or become damaged. A loose fit can also lead to a cracked raceway from lack of support to the inner or outer ring. When a fit is too tight, the bearing will experience a drastic loss in efficiency, operating temperatures will climb, and ambient noise increases."  --- looking at photos, you can see the bearing is part of the side-covers.  Its interference fit to the covers and thus clearance fit or transition fit to the hub.  So that means it's just a tolerable noise that's harmless but sometimes annoying.

Edited September 16, 2022 by Elliott Reitz

[onkeldanuel]

3 hours ago, Elliott Reitz said:

PS:  I will be surprised if hammering the side cover can make it stop for good.  The way it comes and goes makes me think the side covers just aren't tight enough by design

If so use Green Loctite and with that glue them Bearings in their seats, and you should be good

Edited September 16, 2022 by onkeldanuel

[supercurio]

 

King Song S22 motor failing during a well executed tough variant of the stress test.
Credits and big thanks to Andrei, a distributor from London: https://eucs.uk/

Motor production date: June 6 2022

As you can see it can take a long test and many rotation changes from maximum rotation speed in order to validate a motor.
This one didn't make it and it took 20s and 13 rotation changes to identify the failure.

[Elliott Reitz]

4 minutes ago, supercurio said:

 

King Song S22 motor failing during a well executed tough variant of the stress test.
Credits and big thanks to Andrei, a distributor from London: https://eucs.uk/

Motor production date: June 6 2022

As you can see it can take a long test and many rotation changes from maximum rotation speed in order to validate a motor.
This one didn't make it and it took 20s and 13 rotation changes to identify the failure.

As I can see... unless intentionally counter-rotated like that the stator would never get a force like that and clearly stands up to it at less than full speed.  Jumping a curb or rock or root wouldn't ever induce that much force on the stator.   I'd say that's like going fast in R reverse then shifting to D because someone said the clutch slipped in a new car model.  Such someone may have been a clutch-popper and a clutch popping test doesn't prove the car fails the use case beyond the firmware control-limits.

The stress test I did 1x on my wheel didn't reach full speed and I'm keeping it that way, and I'm continuing to ride it for commuting and trail fun!

[supercurio]

@Elliott Reitz note that motors failed in the real world months before a stress test was identified in order to replicate the issue synthetically, and two of the people injured we know of where beginners.

In this test, the forces on the motors are only the ones produced by the motor itself. When riding what's added in the mix is when the the motor acquires rotation speed and spins up in the air, until grip is regained. Then the forces can be significantly higher than what the motor itself can produce, dissipating the energy from the rotation inertia in an instant instead of one second.

It typically happens when jumping, but also more conventionally when accelerating and braking on bumpy surfaces. It is not an unreasonable test IMO. The landing test is orders of magnitude more demanding I think.

Edited September 17, 2022 by supercurio

[Robse]

5 minutes ago, Elliott Reitz said:

As I can see... unless intentionally counter-rotated like that the stator would never get a force like that and clearly stands up to it at less than full speed.  Jumping a curb or rock or root wouldn't ever induce that much force on the stator.   I'd say that's like going fast in R reverse then shifting to D because someone said the clutch slipped in a new car model.  Such someone may have been a clutch-popper and a clutch popping test doesn't prove the car fails the use case beyond the firmware control-limits.

The stress test I did 1x on my wheel didn't reach full speed and I'm keeping it that way, and I'm continuing to ride it for commuting and trail fun!

I will grant you that such a situation as the test presents will not occur in reality, but on the other hand, I would expect the engine parts to be so properly screwed and or glued together that they can never fail like that regardless of what stress you offer.

[Forwardnbak]

This was designed/hyped to be off-road wheel and to be worked pretty extreme. 

I’d hope it would pass a pretty strong test to be confident in pushing it hard on jumps and trails.   

 

 

[Elliott Reitz]

21 minutes ago, Forwardnbak said:

This was designed/hyped to be off-road wheel and to be worked pretty extreme. 

I’d hope it would pass a pretty strong test to be confident in pushing it hard on jumps and trails.   

I agree with everyone it should be " so properly screwed and or glued together that they can never fail like that regardless of what stress you offer."

All I'm saying is I understand a companies' motive to use firmware to avoid such extreme cases while covering up the issue and dragging their feet fixing the units already in distribution while suffering repeated lockdowns.  Even if they have the ethics to address the situation they may not have the money to fix their mistake the right way.  So what should they and we-all do in that situation?  Is that this situation?  What about eWheels and other dealer's situation?  

I'm also choosing to continue to ride a prototype wheel with 600 miles on it without that particular issue (its still got the intermittent clunks).  

[Beachboy]

1 hour ago, supercurio said:

 

 

With such torture, I dont think any motor can survive it, thus it is a great way for the manufacturer to develop new enhancements for the motor in the future that can withstand such test.

anyways, I still have no time to break down my s22 motor, but I rode it many many times now and no such noise came up, and yesterday I rode it at 68-69kmh for several seconds (I did this 2-3x that day), wheel performs very well from 100% to 9% battery with no issues…

So…I think we might need to find better way in testing the wheel’s motor…btw, no more news from KS?

[supercurio]

7 minutes ago, Beachboy said:

With such torture, I dont think any motor can survive it, thus it is a great way for the manufacturer to develop new enhancements for the motor in the future that can withstand such test.

It took some time to find a stress tests capturing a brand new motor, out of the box failing, because a lot of other wheels pass.

Confirmed by KS: the test is expected to be harmless to a wheel free of mainboard and motor defect.

I have seen much longer tests of higher intensity passing without issue, including an internal one by REV Rides (30s straight and 20 rotation changes at max rpm)
Noting that although the motor and board passed, the motor was still producing clunk sounds. Then the motor was opened, showing no sign of slippage.

Therefore I think it's a pretty decent test if ran for long enough.
For extra confirmation, opening the motor afterwards make the result clear.

7 minutes ago, Beachboy said:

anyways, I still have no time to break down my s22 motor, but I rode it many many times now and no such noise came up, and yesterday I rode it at 68-69kmh for several seconds (I did this 2-3x that day), wheel performs very well from 100% to 9% battery with no issues…

So…I think we might need to find better way in testing the wheel’s motor…btw, no more news from KS?

There might be better ways, for sure. This one is very physically demanding which is problematic.
In terms of efficiency, seems okay.

[NSFW]

4 hours ago, supercurio said:

Confirmed by KS: the test is expected to be harmless to a wheel free of mainboard and motor defect.

This is worth repeating.

This test gives us the answer to a very simple question: is the motor strong enough to cause the stator to slip?

The answer to that question should always be no.

The interface between the hub and the stator must be strong enough to withstand the forces produced by the electromagnets. Because if it isn't - if the electromagnets are strong enough to break the stator free - then it's just a matter of time before the motor tears itself apart.

That said, I'll admit that I plan to ride my S22 for a while after the replacement motherboard gets here. But I'll be riding it gently. And when the rainy season starts I will conduct that test as vigorously as I can. That way I'll have a whole winter to deal with the motor swap. And next summer I want to be able to ride it without holding back.

[Elliott Reitz]

22 minutes ago, NSFW said:

This is worth repeating.

This test gives us the answer to a very simple question: is the motor strong enough to cause the stator to slip?

The answer to that question should always be no.

The interface between the hub and the stator must be strong enough to withstand the forces produced by the electromagnets. Because if it isn't - if the electromagnets are strong enough to break the stator free - then it's just a matter of time before the motor tears itself apart.

That said, I'll admit that I plan to ride my S22 for a while after the replacement motherboard gets here. But I'll be riding it gently. And when the rainy season starts I will conduct that test as vigorously as I can. That way I'll have a whole winter to deal with the motor swap. And next summer I want to be able to ride it without holding back.

Yea, I'm riding it fairly mildly till my 2nd S22 arrives from eWheels (after they test it).  Once that wheel arrives, I intend to disassemble this AR-S20 (its motor/hub are S20, not S22) wheel and extra-epoxy glue it myself.  I may figure out the clunk in the process.  And wow these sliders sucked today on super-size gravel (the stones averaged about 3" dia, some 6", mostly <2").  The fender-nails and wire-guide parts to the fender are surely part of that slider friction.

[BIGboiFASTboi]

I mentioned this earlier but nobody replied.... Anyone willing to try a pendulum test?  Rock back and forth while standing along a wall, and torque the wheel 200 times in quick succession to stress test the motor and control board.

[Elliott Reitz]

3 hours ago, BIGboiFASTboi said:

I mentioned this earlier but nobody replied.... Anyone willing to try a pendulum test?  Rock back and forth while standing along a wall, and torque the wheel 200 times in quick succession to stress test the motor and control board.

If you just mean normal pendulum I doubt that would over-stress anything. 

Meanwhile I can happily report that I did a max I could seat-grab torque traffic stop from about 15mph to 0 yesterday. 

I torqued the wheel harder than I've ever torqued any wheel and to my surprise it stopped perfectly in the space of less than a car length. It was truly confidence inspiring because all this time I don't like my power-pads for breaking.  I've also noticed I can knee-squeeze the back of the seat like I do on my MSX 100V.   Anyway, that seat-grab stop exceeds the max-torque I believe I will ever ask of this wheel in any circumstance.  Thus I'm not going to worry about it's stator slipping.  Being that it was one of the original USA-40 batch I'm guessing it may have been more carefully assembled.  Or maybe just lucky.   

So in total, a 1x spin test and now an ultra-torq seat-grab-stop convinces me that this particular wheel is fine for daily use.  

[Planemo]

1 hour ago, Elliott Reitz said:

If you just mean normal pendulum I doubt that would over-stress anything. 

To be fair I'm not convinced your seat-grab torque test will either.

It's sudden application of torque which will test the stator interface. For example if the wheel ever gets airborne and wheelspeed changes before hitting the ground again and trying to balance the rider. Braking from a high speed whilst skipping over a large bump would be another good example.

If you don't do jumps/kerb drops etc then I'm sure you will be fine with the test you have done, I just didn't want others to think the grab test will be conclusive.

 

[5Cauac]

5 hours ago, Elliott Reitz said:

So in total, a 1x spin test and now an ultra-torq seat-grab-stop convinces me that this particular wheel is fine for daily use.  

Perhaps but..........

I performed the stress test three times and pass (although not as aggressive as others) and a few drop tests by jumping, spinning up the wheel and landing on tabletop jumps at the bike park and passed. HOWEVER..........this past weekend I hit a root with my pedal while taking a berm, lost control and stepped off the wheel. The wheel came to an abrupt, immediate stop as I was tethered to the suspension arm, and burned the mother board.

I heard the "loud snap/squeak" of the stator slipping after reviewing the video and determined it was the cause the burned out MB. 

Video of the whole ordeal and the rescue from the forest with my S18 will be out soon. Currently working on it. 😎

[Tawpie]

11 minutes ago, 5Cauac said:

this past weekend I hit a root with my pedal while taking a berm, lost control and stepped off the wheel. The wheel came to an abrupt, immediate stop as I was tethered to the suspension arm, and burned the mother board.

This is consistent with early MB failures that happened before "we" knew about the stator slippage. I still think there's something not quite right in the firmware when it's under load and tips over.

What was your tilt cutoff angle set to? I'm thinking that if the wheel leaves the ground under load, it'll try to spin up… but at some point the tilt sensor will try to shut it down with a full stop command (my KS wheels don't seem to coast to a stop when they tip like they do when overspeed turns off the power). Large tilt angle before shutdown gives the wheel more time to gain momentum and that adds to stress on the stator attachment when the brakes are applied.

[5Cauac]

@Tawpie I was thinking along the same lines as you were regarding the Tilt Cutoff Angle, mine is set to 70° however I don't think to comes to a "Hard" stop when activated. It just cuts the power. 

[Rawnei]

Have to understand the nature of the motor problem, it's not just that the stator suddenly slips from any amount of stress applied, it's not completely loose in there from factory, it's more likely that the glue, friction or whatever forces are holding it will gradually loosen over time, that is what the repetitiveness of the stress test is supposed to expose.

Nobody wants their motor/wheel to break, but would you rather it happens with you on it at speed or in a safe test?

[Elliott Reitz]

1 hour ago, Rawnei said:

Have to understand the nature of the motor problem, it's not just that the stator suddenly slips from any amount of stress applied, it's not completely loose in there from factory, it's more likely that the glue, friction or whatever forces are holding it will gradually loosen over time, that is what the repetitiveness of the stress test is supposed to expose.

Nobody wants their motor/wheel to break, but would you rather it happens with you on it at speed or in a safe test?

Agreed. However, something with a friction-fit and glue may NEVER loosen too.  A 1x stress that doesn't break it doesn't prove that to be the case, but the way static vs sliding friction works that's likely to be the case for the press-fit-friction element.  As for glue it can "break" in a similar manor.  So passing a stress test of any form is a good indicator, but not a proof.

5 hours ago, Planemo said:

To be fair I'm not convinced your seat-grab torque test will either.

It's sudden application of torque which will test the stator interface. For example if the wheel ever gets airborne and wheelspeed changes before hitting the ground again and trying to balance the rider. Braking from a high speed whilst skipping over a large bump would be another good example.

If you don't do jumps/kerb drops etc then I'm sure you will be fine with the test you have done, I just didn't want others to think the grab test will be conclusive.

I do lots of jumps and curb drops, etc.  That's why I needed suspension to start with. 

The hard-breaking that I did is only another form of a 1x stress test.  IDK if its as stressful to the motor as jumps or free-spin-reversal.  But it is much more real-world to me where I don't normally put that much torque on the wheel.   That particular time I had an emergency stop situation with a new seat to grab.  It worked better than I had imagined it could have. 

Again not proof.  

1 hour ago, 5Cauac said:

Perhaps but..........

I performed the stress test three times and pass (although not as aggressive as others) and a few drop tests by jumping, spinning up the wheel and landing on tabletop jumps at the bike park and passed. HOWEVER..........this past weekend I hit a root with my pedal while taking a berm, lost control and stepped off the wheel. The wheel came to an abrupt, immediate stop as I was tethered to the suspension arm, and burned the mother board.

I heard the "loud snap/squeak" of the stator slipping after reviewing the video and determined it was the cause the burned out MB. 

Video of the whole ordeal and the rescue from the forest with my S18 will be out soon. Currently working on it. 😎

Agreed.  Stress test only indicates it hasn't failed yet while risking immediate failure.  Immediate failure may be better than unexpected failure, especially at speed.  However, stator slip is torque induced not speed induced.  Sure at highest speeds the motor is consuming the most power - but that doesn't translate directly into the stator interface force.  Thus I'm more concerned about slippage on steep climb or emergency breaking situations.  And hay note that the S20 fire of U-Stride/Saing was preceded by a breaking induced slippage sensation.