Sudden Unintended Acceleration, Chinese 3kw hub motor kit.

hmmmniek

100 µW
Joined
May 5, 2022
Messages
7
TLDR: Two times, my custom built bike suddenly unintentionally accelerated. Is this a common issue? What is the best way to move forward with this issue?

---

A while ago I bought this kit to build an ebike. I got everything up and running perfectly.
https://www.amazon.com/gp/product/B08R3TNCKT/ref=ppx_yo_dt_b_asin_title_o00_s00?ie=UTF8&th=1

The only complaint I had about this system, is that there is not much of a configurable "acceleration curve", if you give it throttle, it will just dump a crazy amount of power into the motor to get up to speed, so you have to be very careful.

Although this is definitely something you can get used to. But after about in total 80 or so km of riding on this thing, suddenly it would open throttle completely. Luckily I was able to keep control of the bike, and I also build in an easily reachable big battery cut-off switch which I could quickly use to save myself from... I don't know what would have happened if I didn't have that. :shock:

After some deep breaths, I turned everything back on again, and it all seemed to work fine again. I continued to use the bike for hundreds more kilometers, and even started doubting myself if this even really happened, maybe I just slipped my thumb on the throttle without realizing it? Maybe it was just my own fault?

Until yesterday, I was going about 20km/h, and mildly accelerating with the intent to reach 40km/h, when suddenly it went full open throttle again! Quickly grabbing the brakes while keeping the bike under control, and quickly after I managed to use the battery cut-off switch again. Luckily to me again, I had plenty of space to regain control.

It seems like someone else in the amazon reviews also had this problem:
https://www.amazon.com/gp/customer-reviews/R14553HIC6HWS8/ref=cm_cr_arp_d_rvw_ttl?ie=UTF8&ASIN=B08R3TNCKT

In that review, they say the problem seems to be the controller, which makes sense to me.
I tried contacting the seller, they say they have not encountered this problem before. I asked if there was some way to extract logging data, but have not received response back from them on that yet.

I rode over 300km on this bike so far, and the problem occurred twice. Which is a bit scary.

My question(s) which I would love your opinions on is:
- Is this a problem which happens more often with electronic vehicles?
- Is there any known common cause to problems like this?
- Any suggestion on how to move forward?
- Should I just bite the bullet and buy a new controller?
- If so, which one would you suggest for a 72v, 3000w peak system (I do have some dimensions constraints, the controller cant be wider than 110mm)
 
1.) No, this is not normal at all.
2.) Considering your breaks are not cutting off the power delivery even tho they should it's 100% the controller.
3.) Replace controller with a different one
4.) Absolutely, currently you are literally riding on something that can cost your life.
5.) Hall sensors seperate from phase wires is weird. I find it a very curious design choice. As for controller I'm not experienced enough in the high end segment to recommend anything, so I'll assume more experienced bikers will answer you on that one.
 
Danishblunt said:
2.) Considering your breaks are not cutting off the power delivery even tho they should it's 100% the controller.
Concerning that, I decided to not use their mechanical brake levers with throttle cut-off switch, since I want to keep the hydraulic brakes that were already on the bike.
 
hmmmniek said:
Danishblunt said:
2.) Considering your breaks are not cutting off the power delivery even tho they should it's 100% the controller.
Concerning that, I decided to not use their mechanical brake levers with throttle cut-off switch, since I want to keep the hydraulic brakes that were already on the bike.

that changes my deduction a little. Since it could be throttle that is misbehaving. I assumed you had the breaks installed that cut off the controller. You'll probably have to try removing the throttle and see if it still happens. While controller is most likely the culprit its best to isolate the other parts first.
 
Poor/intermitent continuity of the throttle ground wire will result in those symptoms. With no ground, the signal goes to 5V.

Check the continuity carefully, starting with the connectors. the problem could be there, or at the throttle, controller, or in the cable. For the throttle control, switching to a potentiometer based throttle, like a Domino, may provide more control.
 
Danishblunt said:
hmmmniek said:
Danishblunt said:
2.) Considering your breaks are not cutting off the power delivery even tho they should it's 100% the controller.
Concerning that, I decided to not use their mechanical brake levers with throttle cut-off switch, since I want to keep the hydraulic brakes that were already on the bike.
You'll probably have to try removing the throttle and see if it still happens. While controller is most likely the culprit its best to isolate the other parts first.

This is difficult though, so far I am unable to consistently reproduce the problem, since it only happened twice in over 300km of riding! I cant test ride it without a throttle :p

E-HP said:
Poor/intermitent continuity of the throttle ground wire will result in those symptoms. With no ground, the signal goes to 5V.
This sounds possible, but isn't that completely insane? I would guess it is a common problem that some wire would accidentally get disconnected. A single wire continuity problem, and the failure mode can be that the system goes full open throttle :shock:

In any case, I will definitely check this out before throwing more money at a new controller, thanks!
At the same time, I will probably add modification #4 from this post, to delay throttle voltage/smooth acceleration. At least in that case, if I delay voltage by some seconds, I would be able to remain control of the bike more easily when this problem occurs again (if the problem is indeed a throttle sensor problem, not a controller problem)
https://electricbike.com/forum/foru...r-throttle-operation-testing-and-modification
 
hmmmniek said:
E-HP said:
Poor/intermitent continuity of the throttle ground wire will result in those symptoms. With no ground, the signal goes to 5V.
This sounds possible, but isn't that completely insane? I would guess it is a common problem that some wire would accidentally get disconnected. A single wire continuity problem, and the failure mode can be that the system goes full open throttle :shock:

Having a fail point of just one bad wire does seem a bit much. Most better controllers will catch over voltage outputs on throttles on start up and going past maximum voltages. But when riding, or even stopped... It's what got me interested in Ebike throttles.


hmmmniek said:
Concerning that, I decided to not use their mechanical brake levers with throttle cut-off switch, since I want to keep the hydraulic brakes that were already on the bike.

This is where I would start! By all means get a motor cut-out brake or switch (Deadman or other wise) installed to allow safe and immediate shutdown if loss of motor output control is present. (thinking of throttle issue here...)
I appreciate the hydraulic brake problem. But there are products that have an external magnet and sensor that can be added to these...
https://electricbike.com/forum/foru...tall-your-bafang-hydraulic-disc-brake-sensors
Especially important if you're going to try some throttle modifications in the future. :wink:


hmmmniek said:
This is difficult though, so far I am unable to consistently reproduce the problem, since it only happened twice in over 300km of riding! I cant test ride it without a throttle :p

Double check all the throttles cables, connectors, and board connections. Pulled, smashed, damaged, cut cable(s) or connectors.
Next I would probably securly set the bike in a stationary position, with the wheel off the ground. And the throttle in a mid position.
Then gently tug around the cables and connectors looking for a jump in motor output. Perhaps cause some vibrations around the throttle and controller assemblies and connectors ETC.


hmmmniek said:
In any case, I will definitely check this out before throwing more money at a new controller, thanks!
At the same time, I will probably add modification #4 from this post, to delay throttle voltage/smooth acceleration. At least in that case, if I delay voltage by some seconds, I would be able to remain control of the bike more easily when this problem occurs again (if the problem is indeed a throttle sensor problem, not a controller problem)
https://electricbike.com/forum/foru...r-throttle-operation-testing-and-modification

Have a go at mapping your output voltages first. Always a good idea to see EXACTLY what your throttle is out-putting before putting a band aid on it. This may turn up a damage or defective throttle, or put the blame squarely on the controller. But if not, sometimes tweaking the starting and ending points are more beneficial.


u1s8Q5o.jpg




Best of luck.

Regards,
T.C.
 
E-HP said:
Poor/intermitent continuity of the throttle ground wire will result in those symptoms. With no ground, the signal goes to 5V.

I'm sure this is the reason for the OP's problem, but if I'm wrong let me know.

Every controller brand I own, five of them, will do this. I haven't tried my BBS02, but thats' a Higo plug and unlikely to fail.

It's one of the tricks I use to determine which pins in a controller are +5 and signal, if I can't otherwise tell, like when someone has grafted on a JST with different colors. I take a diode and test it between the three leads. When I get it on power/signal, with the diode pointing to signal. the motor comes on.

To be exact, the throttle won't come on if the signal is at 5 volts. It has to be about .7V lower for max, which is your typical diode drop.
 
hmmmniek said:
E-HP said:
Poor/intermitent continuity of the throttle ground wire will result in those symptoms. With no ground, the signal goes to 5V.
This sounds possible, but isn't that completely insane? I would guess it is a common problem that some wire would accidentally get disconnected. A single wire continuity problem, and the failure mode can be that the system goes full open throttle :shock:
I don't know about insane, but this is a known failure mode of the commodity sourced ebike/scooter commonly available/used throttle devices combined with commodity commonly available/used motor controllers.

The system goes full throttle if the throttle ground connection fails-- not desirable in any imagined situation. Hence, the need for an easy kill switch. Brake lever switches suffice for most.

We like the relatively low prices of all this commodity equipment to power our bikes but an unintended consequence is that some poor engineering and design gets baked into the system and is difficult to initiate changes for improvement.

Another poor design baked into the system is (most) hubmotors reliance on insufficient axle flats set into the dropouts to secure the axle from rotating.
 
TommyCat said:
https://electricbike.com/forum/foru...tall-your-bafang-hydraulic-disc-brake-sensors
Especially important if you're going to try some throttle modifications in the future. :wink:
Looks really interesting, I will look into those add-ons for sure!

TommyCat said:
Double check all the throttles cables, connectors, and board connections. Pulled, smashed, damaged, cut cable(s) or connectors.
Next I would probably securly set the bike in a stationary position, with the wheel off the ground. And the throttle in a mid position.
Then gently tug around the cables and connectors looking for a jump in motor output. Perhaps cause some vibrations around the throttle and controller assemblies and connectors ETC.
Good idea, I'm going to try to do that sometime next week when I have time. Will report back if I find something interesting!

TommyCat said:
Have a go at mapping your output voltages first.
Also a great idea, good to know where we're at before making changes :p

docw009 said:
I'm sure this is the reason for the OP's problem, but if I'm wrong let me know.
I like the confidence, I hope this is the issue. Fixing/replacing a throttle is going to be much cheaper and easier than replacing the controller! I'll update this thread once I extracted more information from the system.
 
Almost certainly there is a poor connection in the throttle ground.

This is usually at the connector from the throttle to the controller, especially with open-back connnectors that can have poorly-crimped pins, or poorly-fitted into the shell so that the contact partly backs out of the shell as the two connectors are mated.

But it can be literally anwhere from the solder pad inside the controller all the way up to the hall sensor inside the throttle, and can be a bad solder joint, bad crimp, dirty contact, broken conductor inside the insulation (which may have no external sign of the problem).

So replacing the throttle may not fix the problem (it has at best a 50% chance of doing so).
 
For the responsiveness issue, that can be either throttle or controller.

As noted above, mapping throttle output voltages will help--if you get most of the voltage from around 1v to around 4v within a few degrees of throttle rotation, it's going to be difficult to adjust.

There are a lot of cheap throttles out there with various amounts of operational rotational range; unfortunately they dont' have any specifications for this so most don't tell you anything useful in this regard until you try them.

However, there are other options if you have the space for them. There are external throttle units that are pulled by a cable, and quite a few different ways of pulling that cable (including using a brake lever!). I use these on my SB Cruiser trike, one pulled by a brake lever and one by an ATV thumb throttle lever.



If the voltage change is over a wide rotational range, then the throttle is not the issue; it's the controller. That could be hardware, just the way it's built, or it could be a setting in it you can change within the display menus. But verify the throttle voltage vs rotation range first.
 
Connect a resistor (say 10K) between signal & ground, nearer to the controller (& before throttle connector) the better, safest is to solder on controller PCB.
 
Alright, so I had some time to test the theory that the throttle ground might have a poor connection. And that the system would go open throttle when the ground connection is lost.

First of all, I hooked up the throttle to a voltage meter. I graphed out the results which I got;
9FultMm.png

This all seems normal to me.
I added the (?) after "motor full speed", because I am not sure if this measurement makes any sense at all, it really depends on the selected "assist level", with a lower assist level, the motor reaches top speed (without resistance) at a lower throttle percentage.

I then taped the throttle at a fixed speed, and tried inducing some vibrations etc. No strange results here. Everything kept working as usual. Also tried tugging on the wires at the connectors of the throttle. No weird behavior.
To be sure, I wanted to replace the connector anyway, so I decided to do a bit of destructive testing before replacing the connector;
While the motor was running at ~30 percent throttle, I cut the ground wire of the throttle with some scissors. Based on the posts in this thread, I expected the system to go open throttle. But in reality, power was cut off and I just got error code "20H - Throttle abnormal".
[youtube]QNvubQ8n1es[/youtube]


To me, this feels like it rules out the possibility of it being a poor throttle ground connection.
At this point, safest bet is to get a new controller? It sucks for my wallet, but I also don't feel very comfortable taking this thing on the road in this state.

I had been looking at Kelly controllers, I saw they have bad rep in this forum for some reason. But they seem to be small enough to fit in the space I have available. None of the sabvoton controllers have a width smaller than 110mm.

KLS7218S - 10 seconds: 200A. continuous: 80A. Knowing they probably advertise phase-amps, I'm sure this would be able to handle 2000W continuous fine.. If I would even be able to get my hands on one, seems like there are no local web-shops selling these things..

I guess once I get a new controller, getting a new display is also required? Is there any way to check if my current display is compatible with a different controller? I seem to have this display: https://www.google.com/search?q=ukc1&tbm=isch
 
afzal said:
Connect a resistor (say 10K) between signal & ground, nearer to the controller (& before throttle connector) the better, safest is to solder on controller PCB.

Above works if issue is due to floating signal (signal line is cut) , don't know whether it will fix open ground problem, anyway it seems issue is not due to open ground
 
Hey hummmnick, great work on the data input! And that cutting the ground wire is inspired. :idea:

Just some random thoughts...

Although I'm leaning toward your assessment that the controller is at fault, I'd probably try a few more things before calling it quits.

Thinking out loud...
To me it would seem unlikely to lose the ground connection COMPLETELY, only to have it reset COMPLETELY after a reboot.
Then you have, just inducing the right amount of resistance in the ground wiring to cause a voltage spike. But not a total failure to keep it under the controller's throttle protection settings. Again resetting.
Or lastly, a possibility of the 5 vdc throttle supply line shorting over to the signal wire. Again resetting.

All seem a bit of a stretch, but things to look out for when trying the next steps if you so desire.

Your well done throttle voltage chart definitely shows some serious dead bands at the beginning and end of the rotation.
Just for fun I would open up the controller and have a look at the soldered connections between the throttle wire inputs and the circuit board. First to look for any cold or cracked solder pads, or frayed/loose wire strands. Paying very close attention to the ground circuit. And to see if there is a 4.3vdc pad that is available to use, instead of the 5vdc pad that the throttle input power seems to be coming from now. This would lower your upper voltage signal output down closer to the expected motor full speed voltage, flattening out the power curve a bit.
Then give that a try.

If the gremlin hasn't showed up yet, I'd open up the throttle assembly and take a good look at the hall sensors wiring connections. Also for poor soldering, or damaged wiring. At this time I'd probably install the #1 mod to eliminate to starting voltage dead band, shooting for ~1.3vdc with a closed throttle. Or perhaps the "double dipper mod" if the voltage input change mentioned above was not available, or is still too high.
This will lessen the excessive slope even more, and perhaps make it more usable and comfortable.

Of course, all this with a motor brake cut out wired in and ready to go in case any thing shows up.
But hopefully, I've found that by getting my hands on things, checking connections, and moving the wiring around. With a bit of luck, the true issue will show itself when testing the bike with it stationary.
Give you something to do whilst researching for a replacement controller. Stay safe!

But I totally agree that an unsafe system is not worth a continuous risk. I like the programmability of a Kelly controller and perhaps using a variable regen feature. But no hands on experience with them... yet. :wink:
Fany @ Kelly would be my go to source to confirm if your display would work or not.
 
TommyCat said:
Just for fun I would open up the controller and have a look at the soldered connections between the throttle wire inputs and the circuit board. First to look for any cold or cracked solder pads, or frayed/loose wire strands. Paying very close attention to the ground circuit. And to see if there is a 4.3vdc pad that is available to use, instead of the 5vdc pad that the throttle input power seems to be coming from now. This would lower your upper voltage signal output down closer to the expected motor full speed voltage, flattening out the power curve a bit.
Then give that a try.

Actually I never measured voltage when using the controller as power supply. I just ASSUMED it was 5v, so I hooked up a different 5v supply to the throttle.. but it didn't cross my mind to first check the voltage of the controller to confirm it is actually giving 5v, that's a mistake from my side :p
I will check this next week, and see if the chart looks different if I use the controller as the supply when measuring voltage.

TommyCat said:
But hopefully, I've found that by getting my hands on things, checking connections, and moving the wiring around. With a bit of luck, the true issue will show itself when testing the bike with it stationary.
I really hope so! I'm sure there is a logical explanation. Its just hard to figure out how to reproduce it! :?

TommyCat said:
Fany @ Kelly
Is that an engineer/sales rep? Anywhere where I can find contact details?
 
hmmmniek said:
I will check this next week, and see if the chart looks different if I use the controller as the supply when measuring voltage.

Nice to have a "base" graph to compare to...
Curious though, did you cut the ground wire when using the controller's power supply?


hmmmniek said:
I really hope so! I'm sure there is a logical explanation. Its just hard to figure out how to reproduce it! :?

Intermittent that can't be reproduced reliably is always the most troublesome!

hmmmniek said:
Is that an engineer/sales rep? Anywhere where I can find contact details?


A helpful Kelly troubleshooter, hopefully he's still around.

"For custom software and configuration inquires, please contact sales@kellycontroller.com"

From their website...
https://kellycontroller.com/controllers/
 
Alright, last update from me in this thread for a while probably.

TommyCat said:
Curious though, did you cut the ground wire when using the controller's power supply?
I cut the ground wire when using the controller's power supply indeed.

TommyCat said:
"For custom software and configuration inquires, please contact sales@kellycontroller.com"
Fany is still around! I made sure to let him know the Endless Sphere community is thankful for his support:)


So anyway.. I took apart the mechanical brake lever + switch combo which was included in the kit which I had bought months ago.
I took the electrical switch (NC) and managed to attach it fairly securely (but definitely non-professionally :lol: ) to the existing hydraulic brake with the help of some hose clamps. Seems to work as expected. Hopefully this is going to be enough to make the bike a bit more safe to ride. I will be riding this for a while, hopefully the switch will actually still work when the system freaks out and goes open-throttle. :p
5I3Q6FF.jpg
 
hmmmniek said:
...hopefully the switch will actually still work when the system freaks out and goes open-throttle. :p
If applying the brake (squeezing the brake lever) cuts power normally now, then we can presume it would cut power during a (throttle lifted ground) runaway situation.

Replacing the big hoseclamp with some good quality black tywraps would clean up the look considerably, and reduce the obtrusive sharp edges.
 
99t4 said:
If applying the brake (squeezing the brake lever) cuts power normally now, then we can presume it would cut power during a (throttle lifted ground) runaway situation.

+1 Would be nice to separate a throttle from controller issue...



hmmmniek said:
Fany is still around! I made sure to let him know the Endless Sphere community is thankful for his support:)

:bigthumb:


hmmmniek said:
So anyway.. I took apart the mechanical brake lever + switch combo which was included in the kit which I had bought months ago.
I took the electrical switch (NC) and managed to attach it fairly securely (but definitely non-professionally :lol: ) to the existing hydraulic brake with the help of some hose clamps. Seems to work as expected. Hopefully this is going to be enough to make the bike a bit more safe to ride. I will be riding this for a while, hopefully the switch will actually still work when the system freaks out and goes open-throttle. :p

MacGyver will be asking you for tips! 8) May be enough to scare this issue away!



hmmmniek said:
Alright, last update from me in this thread for a while probably.

Will be hoping for all the best.
If you ever have the desire to map out the output with the actual controllers power, ground, and output connections... feel free.


Best regards,
T.C.
 
Back
Top