Sporadically ignoring all user input and going to 100% power instantly

[riding_on]

I've got a 3000 watt ebike kit (it's pretty wide & powerful, so I think somebody just put a moped motor on a bicycle wheel) with a really dangerous problem: on extremely rare occasions, it stops responding to the throttle and immediately goes to maximum power. Since this thing literally does 60 mph at full throttle, it's basically a death trap. I'm interested in just how a product could end up with a problem like this though.

The setup: 3000 watt rear hub motor with throttle connected and working properly the vast majority of the time, pedal assist not connected at all, and rated for a maximum of 80 amps continuous draw from the battery, so I'm assuming it spikes to around 5000 or 6000 watts on demand. Sometimes, very rarely, maybe once every week or two with daily use, it -- with no warning whatsoever -- just takes off like a bullet. Completely ignores the throttle and power assist level and instantly maxes out.

The first time it happened I didn't have the brake cutoff connected yet, and I was barely able to slam both front & rear brakes down enough to bring it to a stop. Even motionless it wouldn't disengage no matter what I did. It only powered down once I smelled something that I can only assume was the motor overheating and probably triggering an internal cutoff. There's no indication of anything odd on the display when this happens, either (no error messages).

Now that I've got the brake cutoff connected, that signal does cut the power to the motor instantly no matter what. Putting aside the fact that I have a death wish for using this thing, I'm very interested in just how this behavior could possibly be a simple bug. Can anybody familiar with the internals of a three phase brushless electrical motor weigh in on just how something so dangerous could get past any company without being noticed? I'm assuming it's the controller and totally unrelated to the display, because when it happens, it definitely feels like it's accelerating slightly faster than even full throttle does.

 

[E-HP]

Carefully check the continuity of the black ground conductor on the throttle cable and connectors. An intermittent open circuit on the throttle ground wire sends the signal wire to 5V/full throttle. Some throttles use JST connectors, and the pins sometimes back out of the housing and don't make a proper connection, but may look fine without closer inspection.

 

[slaphappygamer]

Carefully check the continuity of the black ground conductor on the throttle cable and connectors. An intermittent open circuit on the throttle ground wire sends the signal wire to 5V/full throttle. Some throttles use JST connectors, and the pins sometimes back out of the housing and don't make a proper connection, but may look fine without closer inspection.

Good call!!

I had a similar issue with the throttle being unresponsive. First, it was intermittent, then it really became an issue when I was trying to keep up with traffic, going uphill, and the throttle gave up. PAS was still responding, luckily. The issue was at the JST housing. I didn’t really find it until I gently pulled on each conductor on each side of the JST plugs. The signal wire, the male pin wasn’t sitting right and on the occasional bump, throttle would cut out. I remade the connection, with a JST crimping kit I found on Amazon. Never had a problem since. Loose connections seem to be the culprit of intermittent issues. Ive been shooting trouble on copper wire for 25 years. I find this to be 80% true.

 

[Chalo]

Also note that a loose magnet inside a Hall throttle can make the thing fail full on, off, or in between.

 

[Hummina Shadeeba]

This throttle claims redundant power supply. Throttle controls | Domino OEM

I’m freaked by talk of a throttle breaking but don’t want to spend 500$. What safety features are available

I just use a three wire throttle

 

[harrisonpatm]

This style, while not necessarily better or worse in terms of quality or function, can at least be used with any sort of generic cable throttle, and tucked away somewhere else on your bike, somewhere perhaps less prone to water accumulation or physical trauma.

 

[riding_on]

Also note that a loose magnet inside a Hall throttle can make the thing fail full on, off, or in between.

Correct me if I'm wrong, but that problem wouldn't be corrected as soon as I cut the power with the brake signal, would it? The behavior I see is that once I squeeze the brake and release it, everything works correctly again.

Carefully check the continuity of the black ground conductor on the throttle cable and connectors. An intermittent open circuit on the throttle ground wire sends the signal wire to 5V/full throttle. Some throttles use JST connectors, and the pins sometimes back out of the housing and don't make a proper connection, but may look fine without closer inspection.

All the connectors appear fine to me, and I tried replacing the throttle already -- it happens with both of them. I'll use a multimeter to double check later today though, just to be sure.

 

[amberwolf]

I've got a 3000 watt ebike kit (it's pretty wide & powerful, so I think somebody just put a moped motor on a bicycle wheel) with a really dangerous problem: on extremely rare occasions, it stops responding to the throttle and immediately goes to maximum power.

Normally that is a faulty ground between controller's MCU and the throttle hall, so that the throttle signal "floats" to some near-full voltage.

Since engaging and then disengaging the ebrake resets the behavior, it can't be that, unless there is some very weird wiring issue that allows the ebrake to complete the ground and "reset" the conditions (I can't imagine how it would happen without a very bad throttle ground connection and some stray capacitance that has to build up a certain trigger voltage, etc).

That leaves a firmware bug or an MCU hardware bug (command errata, etc). Perhaps a defective cruise-control feature.

There is also the tiny possibility that the PAS input (if there is one) doesn't have a pulldown or pullup resistor inside the controller, so the input is floating. Under certain conditions, it might get enough noise of sufficient amplitude change to be interpreted as a string of PAS pulses, and if the controller is setup for a mode that PAS, if engaged, would use full power (or full speed, etc), that's what you'd get. It *should* stop once the pulses stop, but if the noise is caused by interference from the phase wires, the higher current would actually make the problem worse and continue to make the system run.

Some systems are designed to override PAS with the throttle, but some obey whichever is higher input so if PAS is commanding faster speed but throttle is not, it'll obey PAS. Some are designed to override throttle with PAS. All of them would stop powering the motor when ebrake is engaged (actual results varying with design of ebrake input response).

If it is a simple PAS input issue, you can either ground the PAS input signal, or connect it to 5v, at the controller's PAS connector (or better yet at the PCB inside the controller). I recommend using a 1kohm to 10kohm resistor between the signal and whichever one you use. If the signal is tied to something, it can't float and so noise can't cause this problem.

I'm interested in just how a product could end up with a problem like this though.

Most software doesn't get tested thoroughly, and even things that are tested aren't necessarily fixed. I've beta tested various software (and software/hardware combos, including a motor controller) and seen what happens to bug reports--most are pushed off to future releases, and never get fixed, even if they are seriously affecting usability, etc.

There's also probably quite a lot of software that is unnecessarily complicated so that it is virtually impossible to test all permutations of input vs output, and even worse is built multigenerationally--the programmer (or team) for the software uses code they don't fully understand or know well to base their software on, and the people that wrote *that* software based it on or used code by other people, etc etc.

Testing of the end-product is likely to be limited to specific combinations of hardware, and specific usage scenarios, so there are probably quite a few untested usages / combinations that would trigger undesired behavior in the system--most of them probably won't be found by end users, but there will always be some.

Can anybody familiar with the internals of a three phase brushless electrical motor weigh in on just how something so dangerous could get past any company without being noticed?

FWIW, the motor itself doesn't have anything to do with the problem, in that it could be *any* kind of motor (assuming the correct kind of controller was used with it) and still result in the same thing, assuming the same problem with the system.

 

[riding_on]

Normally that is a faulty ground between controller's MCU and the throttle hall, so that the throttle signal "floats" to some near-full voltage.

Since engaging and then disengaging the ebrake resets the behavior, it can't be that, unless there is some very weird wiring issue that allows the ebrake to complete the ground and "reset" the conditions (I can't imagine how it would happen without a very bad throttle ground connection and some stray capacitance that has to build up a certain trigger voltage, etc).

That leaves a firmware bug or an MCU hardware bug (command errata, etc). Perhaps a defective cruise-control feature.

I'm thinking you're right: as I've continued using it (yes, I'm one of those special kinds of insane), I've noticed that it doesn't actually throttle itself to maximum. What it does is ignore the power assist level and immediately behave as though it's set to maximum, then ignore the throttle input and hold it steady at whatever position my thumb had it at when the "glitch" popped up. Squeezing the ebrake disengages it.

It also, for whatever reason, powers on the display if I pedal fast enough without the battery connected at all. I've read up to 14 volts just using my hand with it inside held up by my arms. It's a very odd little motor, but I do so love imagining what look somebody will have behind their motorcycle helmet when I beat them in a race (because it's a bicycle, and I don't abuse ignoring traffic laws). In short, I love the thing. But it was a death trap before I learned to adjust my riding style to it.

As always, dear sir, you have my gratitude and warm sentiment for your expertise.

 

[amberwolf]

I'm thinking you're right: as I've continued using it (yes, I'm one of those special kinds of insane), I've noticed that it doesn't actually throttle itself to maximum. What it does is ignore the power assist level and immediately behave as though it's set to maximum, then ignore the throttle input and hold it steady at whatever position my thumb had it at when the "glitch" popped up. Squeezing the ebrake disengages it.

You can replace the controller with a different one to fix this problem. You may also get other features you may want by doing so, by choosing a controller that has them, if there are any you don't presently have.

For instance, the ability to use the pedals to actively control the system not just as on/off like most PAS do, so you don't have to use a throttle for typical riding--just set it up to make your legs "bionic". (at least one way to do this is with the Cycle Analyst in between the PAS sensor and the controller, but there are others, probably some I don't even know about by now).

It also, for whatever reason, powers on the display if I pedal fast enough without the battery connected at all. I've read up to 14 volts just using my hand with it inside held up by my arms.

Any motor is a generator as well, so if you spin up the motor it generates voltage (faster = higher voltage). The controller acts as a rectifier to convert that to DC to feed to the battery, which is also how regen works (that's just controlled, vs the uncontrolled way you see when the system is "off" or has no battery connected).



Note that without a battery (or if the battery BMS shuts off, disconnecting it from the controller), then if you spin the motor fast enough (equivelent to how fast or faster that it would spin at max normal output from the motor) it will generate a voltage higher than that of the battery's full charge, and if the controller/etc isn't designed to handle that voltage, it's possible for damage to occur.

Unless you're going downhill in a hurricane it's unlikely to happen by pedalling, though.

 

[riding_on]

Any motor is a generator as well, so if you spin up the motor it generates voltage (faster = higher voltage). The controller acts as a rectifier to convert that to DC to feed to the battery, which is also how regen works (that's just controlled, vs the uncontrolled way you see when the system is "off" or has no battery connected).



Note that without a battery (or if the battery BMS shuts off, disconnecting it from the controller), then if you spin the motor fast enough (equivelent to how fast or faster that it would spin at max normal output from the motor) it will generate a voltage higher than that of the battery's full charge, and if the controller/etc isn't designed to handle that voltage, it's possible for damage to occur.

Unless you're going downhill in a hurricane it's unlikely to happen by pedalling, though.

I feel a little embarrassed to admit that I've forgotten my display is damaged -- it automatically turns on when power is introduced to it, but once turned off, doesn't turn back on. When it was new, it stayed off until you powered it on manually. In the back of my head I forgot that a voltage wouldn't normally turn anything on at all. No wonder I never noticed the voltage produced when the motor turns on other ebikes

 

[neptronix]

Hey, i had this problem with a phaserunner v2 ( very old ) and it happened due to issues with the halls in one of my motor. It would randomly stop responding to the throttle, and a few times go WOT on me unexpectedly.

My CA direct plugin model would also freak out during these episodes & there was no response to the ebrake lever.

Ended up making something i could pull to do an emergency power cut to the controller in case it unexpectedly went WOT on me again.



Problem doesn't occur in sensorless mode or with a different motor that had working halls.

Interestingly, an old infineon controller from 2010 shows a similar symptom, but it has random blips of power instead of going WOT continuously.

I forget the result with my VESC w/halls connected but i remember it never going WOT during test rides with the halls.

Grin says future revisions of the Phaserunner have increased safety.

Your issue may be throttle related or hall related. Does your controller have sensorless mode? if so, that could help narrow down the source of the problem.