Successful axle change - KS14C

[andress]

So we all know KS14C has a bit of a trouble with the axle cracking. The first time it happened to me I went trough half a year long warranty process and apart from the time it cost me 290 USD as apart from the motor itself I needed new board (old one is not compatible with the new motors) and I also wanted new shell and I had to pay the shipping and taxes. 

I thought that with new type of motor they would have solved the issue but no... after 1400 km according to the app my axle broke again! 

This is how it looks from outside when your axle breaks:

detail after removing the pedals:

 This time I was so careful! I didn't jump the curbs I had barely any falls... but maybe it was caused by not using the spacers that come between the pedal holders and motor. I didn't use them because the ones from the old motor were not compatible with the new one and I didn't receive new one with the new motor. Tina said to proceed without them but I think that was the reason why the axle broke so soon because maybe I put them a bit further apart or something like that.

Anyway I didn't want to pay another hefty sum for a new motor to have it broken in few months again and I actually had a new 500W KS motor on hand since they sent it to me by mistake at first instead of the 800W one. First I tried whether it will work with my board meant for 800W motor. No surprise, it doesn't . But the axle is the same so I decided to gut the never used motor to get mine going again.

So I took both motors apart. And this is how i looks:

The way the wire are connected to the windings

Detail of the axle:

Detail of the hall effect sensor:

To remove the axle you need access to a hydraulic press. I tried to hammer the axle out but had no success, I only damaged the axle even more. I study a technical university so I just went to the building of mechanical engineering and after few minutes I had time setup when I could come to use the press. In the end the people there just did it for me because of safety concerns.

The press ( very old made in now nonexistent country of Czechoslovakia :D)

Video of the actual removal:

https://goo.gl/photos/nJ5HVcn3ogqr8ECB6

Detached axle:

To remove the axle I had to remove the cabling first and of course the cabling got damaged while being pulled trough the axle. The original phase wires were 1,22mm2 with plastic insulation the hall sensor wire were 0,22mm2 also with plastic insulation. The replacement cables I used were 1,5mm2 and 0,25mm2 with silicon insulation. So potentially higher quality but they were also thicker so now sheath to put them all together could fit in the axle so i kept them separate. 

Also as you can see they didn't have the right colors which caused me some trouble later on when connecting to the control board. I had them labeled but the tags fell while I was pulling the cables through the motor cover  So when finishing up there was a lot of experimenting with the right order. I knew  which wires were for hall sensors and which for phases from thickness and I knew which wires were the power for the hall sensors because these were the only ones with the correct color. But I still had two groups of 3 wires each. That gives me 36 combinations. Anyone who ever connected the wires wrong knows how crazy the unicycle acts if you do that :/ Luckily it only took me 8 tries! Those were some horrible couple minutes!

To pull the cables through the axle I used the same method as @Rehab1 - solder them all together and use single wire to pull the trough. Thanks for inspiration!

Unfortunately some of the hall sensors got damaged during  the axle removal so I had to change them too. The original sensors were MT4451-A (TO-92 package) as @esaj helped me discover from the markings. I could not get these in Czechia so I used TLE4935L instead as I have already used them once on different unicycle and they worked.

Damaged hall effect sensor:

And here is finally the finished motor with new wiring and hall sensors:

After putting everything together the unicycle works!

But there are some problems. It does wobble. When I ride it swings back and forth a little. It it pronounced way more while riding forwards then by riding the unicycle the wrong way with the brake light forward. Anyone has any ideas what could cause this? Here is a video of this https://goo.gl/photos/WfKaDb1iGzgadY1p6

Also the free spin maximum speed is different in each direction it is 40,6 km/h in forward (more wobbly direction) and 38,5 km/h in backwards (more stable direction). So IDK is this some mechanical problem ? the bearing perhaps? I used the motor covers from the 500W as the bearings spin much better. Could this be the cause? They seemed the same otherwise and they fit with no problem.

Problem with the board maybe? During the experimenting with the right wire order combinations some nasty thing were most likely happening to the board..:/

I will see how it progresses but please if someone knows what it could be let me know I would hate going trough all this trouble to have the unicycle break down again soon after.

So in total this cost me about 20 USD. 10 USD for the beers I bought the people who helped me with the press and 10 USD for the wires and hall sensors. It would cost more if I had to have the axle machined.

PS. There will be part two with the older motor from the first time my axle broke and new machined axle for that  

[US69]

For me the only reason can be the different hall sensors......

[SereneCry]

I think it's the metal plates that go in the pedalmounts on each side of the straight part of the axle. These get worn and thus do not clamp down tight enough on the axle. You can try using shims to fill the gaps, and tighten with a torquewrench (with red loctite on cleaned threads). Solder is a bad idea since it can come off while riding and short out your electronics.

EDIT: Had a slight wobble on mine. Took it apart just now to do this shimtrick. Found the clampingplates have become askew (see picture attached) aswell as worn where they sit against the axle. This will ofcourse make the problem worse and is something to look out for..

PS: made a handy clamp to keep the wheel from tipping while working on it (see attached picture)

 

[andress]

I'm not sure this is what you mean, but KS uses different way to mount the pedals than most manufacturers. Have a look here. You just clamp the axle and tighten it using the two screws on top.

[andress]

1 hour ago, KingSong69 said:

For me the only reason can be the different hall sensors......

I hope not. On the other unicycle I used these as well because I didn't know what types were used there and it works well without any negative effects.

[US69]

Just now, andress said:

I hope not. On the other unicycle I used these as well because I didn't know what types were used there and it works well without any negative effects.

Yeah...but what was this "other unicycle"? Has it the same board as the KS14, Needs it the same sensors?

You have done a fantastic Job on the axle replacement and all...but i guess just using different hall sensors was a bit inconsequent!

[SereneCry]

41 minutes ago, andress said:

I'm not sure this is what you mean, but KS uses different way to mount the pedals than most manufacturers. Have a look here. You just clamp the axle and tighten it using the two screws on top.

Inside these two parts that clamp together you might find that the groove that fits over the axle is bigger than it should be because it has been worn. That would cause these parts to be loose on the axle even if you clamp tight

[andress]

7 hours ago, KingSong69 said:

Yeah...but what was this "other unicycle"? Has it the same board as the KS14, Needs it the same sensors?

You have done a fantastic Job on the axle replacement and all...but i guess just using different hall sensors was a bit inconsequent!

It was Freeman A4H and the specs of the sensors are so similar that I doubt that that is the problem. but I will try to find and buy the same type if no one has better suggestion. It will probably take a lot of time as they will have to be purchased probably from china. 

7 hours ago, Cryptonitor said:

Inside these two parts that clamp together you might find that the groove that fits over the axle is bigger than it should be because it has been worn. That would cause these parts to be loose on the axle even if you clamp tight

I will check even though the connection felt solid. I should have time to that over the weekend.

[Carlos E Rodriguez]

@andress did you measure the runout after installing the new axle?  

What i mean is suspend the stator with axle on two forks with the bearing also installed. 

Spin the stator and measure how much the stator wobbles/oscillated out of center.  If it wobbles then the axles is bent or the wheel casting was warped during the axles removal or axle installation. 

Next is checking the bearings. It is possible that a damaged bearing cage is causing one ball bearing to bind a little in one direction and a lot on the other direction. 

If you have a bad hall the motor would run very rough. The video looks like the wobble is once per revolution indicating bearing or axle or wheel casting. 

You mentioned that you tried to hammmer the old axle out. I think you might have bent you wheel casting causing the imbalance. 

[andress]

5 hours ago, Carlos E Rodriguez said:

@andress did you measure the runout after installing the new axle?  

What i mean is suspend the stator with axle on two forks with the bearing also installed. 

Spin the stator and measure how much the stator wobbles/oscillated out of center.  If it wobbles then the axles is bent or the wheel casting was warped during the axles removal or axle installation. 

Next is checking the bearings. It is possible that a damaged bearing cage is causing one ball bearing to bind a little in one direction and a lot on the other direction. 

If you have a bad hall the motor would run very rough. The video looks like the wobble is once per revolution indicating bearing or axle or wheel casting. 

You mentioned that you tried to hammmer the old axle out. I think you might have bent you wheel casting causing the imbalance. 

It is once per revolution I just verified before seeing your message  I did not try that, but I will. It seems like the most likely cause so far.

Also I just remembered didn't @electric_vehicle_lover say that hall effect sensors are only used in slow speeds? Because this happens at every speed...

[Hunka Hunka Burning Love]

Just curious - did you change out all three hall sensors or just the one broken one?    I don't know if maybe mixing and matching different hall sensors might be an issue?  Maybe they vary a little in sensitivity or trigger timing so that could be throwing the synchronization off?

[andress]

Just now, Hunka Hunka Burning Love said:

Just curious - did you change out all three hall sensors or just the one broken one?    I don't know if maybe mixing and matching different hall sensors might be an issue?  Maybe they vary a little in sensitivity or trigger timing so that could be throwing the synchronization off?

That's what I thought might happen so I did change all of them.

[Hunka Hunka Burning Love]

Do you have some spares?  Maybe some are defective or vary in specification.  Esaj mentioned how he got a bunch of low priced components from AliExpress, and how some might be counterfeit, out of tolerance or factory rejects.  Maybe one or more are defective?  Also it looks like you skipped insulating the legs on the hall sensors whereas the OEM ones had yellow sleeves.  Maybe there is an intermittant short somewhere?

It's also interesting to note the KingSong motor has little green PCB boards that act as a junction to solder to rather than soldering directly to the little sensor legs.

[andress]

I tested all of them after installing using arduino. They all work. Shorts are close to impossible the sensors are glued in the socket and the legs are soldered to the mini pcb which is also glued to the motor. I bought them here in Czech republic in reputable electronics part shop. I really don't think they are the problem. There is tens of times that the hall sensors change their output per revolution but the wobble happens only once per revolution.

[Hunka Hunka Burning Love]

You mentioned the wheel performs differently riding it backwards than forwards.  Have you tried flipping the outer hub in the other direction?  Maybe  the way the stator teeth and outer rim magnets are aligned it works better when set up in a certain direction due to variance in the angle and airgap?  Just putting out random ideas here.  I believe that you had mentioned that you forgot the orientation at one point?  Maybe use a Sharpie marker to mark Face A to Face A on the rim.  Label wires with one line, two lines, three lines, etc for reference.

Did you label the motor covers and their screw holes?  I don't know how symmetrical they are, but did you use shims to even out the airgap before reinstalling the first motor cover?  Maybe try remounting the covers and ensure the airgaps are even.  Perhaps the outer hub is installed slightly skewed so the air gap on one side is larger than the other resulting in uneven magnetic forces.  How precision fit are the screw holes on the motor covers?

[electric_vehicle_lover]

7 hours ago, andress said:

It is once per revolution I just verified before seeing your message  I did not try that, but I will. It seems like the most likely cause so far.

Also I just remembered didn't @electric_vehicle_lover say that hall effect sensors are only used in slow speeds? Because this happens at every speed...

Changed my opinion about that, please read here:

 

[electric_vehicle_lover]

On 20/06/2017 at 9:22 AM, andress said:

But there are some problems. It does wobble. When I ride it swings back and forth a little. It it pronounced way more while riding forwards then by riding the unicycle the wrong way with the brake light forward. Anyone has any ideas what could cause this? Here is a video of this https://goo.gl/photos/WfKaDb1iGzgadY1p6

Also the free spin maximum speed is different in each direction it is 40,6 km/h in forward (more wobbly direction) and 38,5 km/h in backwards (more stable direction). So IDK is this some mechanical problem ? the bearing perhaps? I used the motor covers from the 500W as the bearings spin much better. Could this be the cause? They seemed the same otherwise and they fit with no problem.

Thanks for sharing all this great information!!!!

About the issue, I had the same when developing my OpenSource firmware, that the motor would run faster to one side and with a bit less noise... and that happen when I were using no FOC control, where the FOC angle wasn't adjust automatically and was hardcoded -- and depends only on the hall sensors signals timmings. And this makes sense in my experience, if that angle is not "like at the middle", the motor will run with different speeds on each side.
I would say the hall sensor you used for some reason outputs his signal with different timing. I would try to change all the 3 for the same hall sensor reference.

[Marty Backe]

Wow, great inspirational work and your pictures are very nice. 

[SereneCry]

On 20. juni 2017 at 9:09 PM, andress said:

 

I will check even though the connection felt solid. I should have time to that over the weekend.

Attached is a picture of how I solved the problem with the two clamps not tightening. Abit hard to see since the problem area is under the nut. I simply put them in place and put threadlock in the gaps (on the flat parts of the axle, and on the clamps where they meet the axle) clamped tight, let it cure for 24 hours, wiped off excess. PS: Trying blue (medium strength threadlock) and nylon fishingline wound around the threads this time. Might've been better with just red loctite but time will tell.

[edwin_rm]

I'm having a similar problem with the KS16 now....

edit: Actually, this video is actually the same as my problem with the KS16 (wondering if it is the same issue described above for the 14c):

 

 

[andress]

On 6/21/2017 at 10:38 AM, Hunka Hunka Burning Love said:

You mentioned the wheel performs differently riding it backwards than forwards.  Have you tried flipping the outer hub in the other direction?  Maybe  the way the stator teeth and outer rim magnets are aligned it works better when set up in a certain direction due to variance in the angle and airgap?  Just putting out random ideas here.  I believe that you had mentioned that you forgot the orientation at one point?  Maybe use a Sharpie marker to mark Face A to Face A on the rim.  Label wires with one line, two lines, three lines, etc for reference.

Did you label the motor covers and their screw holes?  I don't know how symmetrical they are, but did you use shims to even out the airgap before reinstalling the first motor cover?  Maybe try remounting the covers and ensure the airgaps are even.  Perhaps the outer hub is installed slightly skewed so the air gap on one side is larger than the other resulting in uneven magnetic forces.  How precision fit are the screw holes on the motor covers?

So I disassembled the motor again and there were some marks on the magnets as if from rubbing by the stator so I guess that's the cause for the once-per-revolution wobble. I tried to put on the original motor covers. Those I had marked the exact positions they were in before ever taking them off. It didn't help. It was actually slightly worse because of worse bearing that were on these covers. So I flipped the stator in the other direction against the outer rotor and that helped! Even with the new covers the once-per-revolution wobble is gone!

On 6/21/2017 at 5:23 PM, electric_vehicle_lover said:

Thanks for sharing all this great information!!!!

About the issue, I had the same when developing my OpenSource firmware, that the motor would run faster to one side and with a bit less noise... and that happen when I were using no FOC control, where the FOC angle wasn't adjust automatically and was hardcoded -- and depends only on the hall sensors signals timmings. And this makes sense in my experience, if that angle is not "like at the middle", the motor will run with different speeds on each side.
I would say the hall sensor you used for some reason outputs his signal with different timing. I would try to change all the 3 for the same hall sensor reference.

Even though the wobble mentioned above is gone you are probably right. It still rides different in each direction. In backwards direction it actually feels better than the original (as far as I remember, no other KS14C around to verify). In forward direction there are some oscillations when there is a bump. I will leave as it is for now. It works and I'm kinda tired off messing with the motor and not riding the unicycle. But if I manage to get the same sensors that were there before (unlikely cannot find them anywhere) I will change them  
 

BTW Does anyone know whats the normal maximum free spin speed for the newer board and motor version of KS14C with full battery?

On 6/23/2017 at 1:19 PM, Cryptonitor said:

Attached is a picture of how I solved the problem with the two clamps not tightening. Abit hard to see since the problem area is under the nut. I simply put them in place and put threadlock in the gaps (on the flat parts of the axle, and on the clamps where they meet the axle) clamped tight, let it cure for 24 hours, wiped off excess. PS: Trying blue (medium strength threadlock) and nylon fishingline wound around the threads this time. Might've been better with just red loctite but time will tell.

 

Good that is works for you but as I said I have entirely different way the pedals holder attaches to the axle... It is much simpler way and if it isn't the actual cause of the axle breaks it seems to be better way as I had never had a problem with it getting loose.

 

Thanks again for everyone's suggestions on the cause! As it is right now I call it great success and you can expect another project with the older motor during the summer.

[Hunka Hunka Burning Love]

Glad to hear that you got it working.  Just curious, do you use shims in between the metal facings and magnet when mounting the motor covers, or do you rely on the screws to move the stator housing into proper position?  I'm just wondering if the cause was misalignment during cover install or an actual magnet face to stator face lateral misalignment that necessitates a certain hub and stator arrangement.

If the bolt holes in the cover are larger than the bolts, some alignment play could occur during reassembly.  Maybe regular use of shims (eg. Playing cards) is a good idea considering how small the airgap is.

[andress]

I tried to use the shims but found them pretty useless. Even when using them the cover realigned the axle during tightening of the screws- I assume this by the sounds it made.The covers are not aligned by the screws but by the rim that goes around the circumference of the cover. It only goes in the right positions when pressed to the outer hub so that's why it seems that it is the screws that align it but they are there really just to keep it all pressed together. So the bigger screw holes allow only for a angle shift which should not affect the axle at all. I tried to take a little video of what I mean but I'm not sure that you can see it. Basically I try to demonstrate that the cover stays in one position even without any screws at all.

 

https://goo.gl/photos/RYUDr45ZcEiQ7RCK8

[Hunka Hunka Burning Love]

Thanks for the video.  I do see that inner fitment rim and see what you mean.  Is there any play between the rim and the outer hub?  If not then maybe the magnets/stator facings are laterally skewed slightly resulting in the friction contact previously.  Interesting.  Any ride history of harsh drops that may have warped the outer hub?

[Rehab1]

On 6/21/2017 at 11:23 AM, electric_vehicle_lover said:

I would say the hall sensor you used for some reason outputs his signal with different timing. I would try to change all the 3 for the same hall sensor reference.

Well stated! @Andreas just be careful when removing the hall sensors as they are epoxied in. Cover up the stator windings next to the sensors with multiple layers of tape to prevent any damage to the copper windings! Been there and it sucks!