Issues with cheap controller and Panasonic hub motor

lukam

1 µW
Joined
Apr 30, 2020
Messages
4
Greetings!

A couple of months ago I've bought a used KTM e-bike, pedal assist only with all Panasonic electronics and hub motor. After some things happening I'd rather not post about I've come to the point of having to replace the controller on it hoping to keep the motor and the battery. The replacement controller is a generic Chinese 350W one and a twist throttle to go with it. After connecting everything the bike seems to function well on my stand, the wheel spins and responds to throttle as expected however after a test ride i discover that it has practically no power. Moving from a standstill with anything above the slightest twist of the throttle is impossible as the motor does a strange thudding skipping thing and refuses to move. It's also possible to lock up the wheel if it's stalled in an exact position making it stuck in place until the throttle is released. Some points and things I've tried:

  • I've tried many combinations of phase wire and hall sensor wiring and have concluded that they do not change the controller's behavior. After every combination change I'd run the "self learning" mode and would come to the same point as before - smooth rotation with no load but no actual power output (the self learning mode on Chinese controllers runs the motor in an attempt to detect motor phase and hall sensor layout, a calibration of sorts)
  • In the mentioned "self learning" mode the motor seems to run as it should (???): loading it by braking does not cause it to skip and thud, stalling it in certain positions does not cause it to lock up in place
  • I have not yet tested the battery current to see if that's what's limiting the controller's power output but I doubt that this is the issue as the bike had quite a bit of power with the original electronics (together with the previous point) and all the wires used on it are very thick, which makes me assume that the actual max current the battery is limited to is quite high

I'm hoping that someone has experienced and dealt with a problem like this as I'm all out of ideas at this point. Any help would be much appreciated!
 
Can you check the battery voltage at the controller? I think some old Panasonic used 24V. What is the voltage of your controller?
 
If the stuff you'd rather not post about has anything to do with damage to any part of the bike, in any way, then leaving that out may prevent us from being able to help you fix it.

The problem you have could be hall sensors in the motor, either damaged sensors themselves or wiring, or connectors. If the controller has a sensorless mode, you could try that, with the hall sensors disconnected.

Or you could have a defective controller, or one that is somehow incompatible with the motor itself (unlikley but not impossible).



The battery doesn't limit current. The controller does that. If the controller asks for more current than the battery can provide, first the voltage will sag, and if it is greater than the battery low voltage cutoff limit, the battery may shutdown it's output until the load goes away.

But it won't limit the current in a way that just lowers the power.
 
Well, 350 watts is a pretty no power controller. It may be working as it should. Was the old one stronger? Is the battery suffering?

Look hard at all your wiring, poor connections at plugs can limit the flow of power to one phase of the motor, resulting in it runs smooth, but can't push on that phase. Cuts your power by 1/3. Similar can happen if the wire is cut, and not many strands left on one of the big wires. Is repairing your wiring what you don't want to mention?

Runs smooth, sounds like your self learning worked.
 
ykuga said:
Can you check the battery voltage at the controller? I think some old Panasonic used 24V. What is the voltage of your controller?

Battery and controller are 48V. Voltage at controller is 52V with an almost full battery with no load, drops to 48V if I run the self learn mode and load it.

amberwolf said:
If the stuff you'd rather not post about has anything to do with damage to any part of the bike, in any way, then leaving that out may prevent us from being able to help you fix it.

The old controller was fried due to a short.

amberwolf said:
The problem you have could be hall sensors in the motor, either damaged sensors themselves or wiring, or connectors. If the controller has a sensorless mode, you could try that, with the hall sensors disconnected.

I checked the hall sensors, both with a multimeter and a scope. They all switch from low to high as the wheel turns and the waveform seems good. Runs the same with and without sensors, but not in the self learning mode. Without sensors it shows the symptoms in the self learning mode too. There's no way to manually switch modes using jumpers or anything, it's supposed to do it automatically.

amberwolf said:
The battery doesn't limit current. The controller does that. If the controller asks for more current than the battery can provide, first the voltage will sag, and if it is greater than the battery low voltage cutoff limit, the battery may shutdown it's output until the load goes away.
But it won't limit the current in a way that just lowers the power.

I was assuming the battery BMS might have some kind of cut off that would momentarily cut power if the current peaked too much. But again, the motor seems to function correctly in self learn mode and doesn't have trouble drawing more current then.

dogman dan said:
Well, 350 watts is a pretty no power controller. It may be working as it should. Was the old one stronger? Is the battery suffering?

Look hard at all your wiring, poor connections at plugs can limit the flow of power to one phase of the motor, resulting in it runs smooth, but can't push on that phase. Cuts your power by 1/3. Similar can happen if the wire is cut, and not many strands left on one of the big wires. Is repairing your wiring what you don't want to mention?

Runs smooth, sounds like your self learning worked.

Old bike electronics were rated at 250W (probably for legal reasons though, because of EU laws). I checked all the wiring, nothing is cut and no connections are heating up. I paid attention to the wiring as I was connecting everything so I wouldn't have issues with it now so I'd feel pretty dumb if the problem was just a wonky connection somewhere. I didn't want to go deeper into how the previous controller was fried.

Again, what's confusing me the most is that the controller is not displaying any of the problems (skipping when loaded, getting locked in place when stalled) in the self learning mode. Could it be that the controller is somehow stuck in sensorless mode while actually running even though the sensors are working?
 
lukam said:
Battery and controller are 48V. Voltage at controller is 52V with an almost full battery with no load, drops to 48V if I run the self learn mode and load it.
If this is with it off-ground, rather than riding it, then 4v of voltage drop means you have a problem with the battery (separate from whatever else is wrong), as it should not sag hardly at all while off-ground.

If this is while riding it, that's still a lot of sag, but probably about right for a cheaply-made battery pack of unknown cells.




The old controller was fried due to a short.
I didn't want to go deeper into how the previous controller was fried.
Details matter in troubleshooting. If you're embarassed about something you did to it, don't be. Many people here have done some pretty silly things. ;) But if you don't tell us what actually happened originally, it's possible that some detail in that will immediately point us, thru experience, to what is causing the problem. Might not tell us anything. Won't know till we know. :)

If this was a short in the axle wires, it may have damaged hall sensors. Since the test results below indicate they are not damaged, then probably not. Could've damaged phase wires; if it was a crash the motor may even have physical damage inside that's only apparent when under load riding it, not in offground learning modes.

However, if the short included the 5v line, it's possible for it to have damaged anything connected to it. (not likely, since the hall sensors still work, but depends on the voltage applied at the time--motor halls are often able to take up to 24-30v, other sensors like throttles may not be).

If this was a short in the battery side, the battery itself (it's BMS) could be damaged, but if you've replaced that whole thing, it doesn't matter, shouldn't be anything left on there to affect anything.



I checked the hall sensors, both with a multimeter and a scope. They all switch from low to high as the wheel turns and the waveform seems good. Runs the same with and without sensors, but not in the self learning mode. Without sensors it shows the symptoms in the self learning mode too. There's no way to manually switch modes using jumpers or anything, it's supposed to do it automatically.

OK. That probably eliminates one major potential problem.


It's interesting that the learning mode fails to work correctly in sensorless, but not in sensored, and that sensored mode doesn't work the same in learning as normal modes. But I don't see any reason it should do so, outside of the controller itself.

The only possibility (other than a controller design problem or other controller problem) that comes to mind is a problem with the throttle not actually providing it's full voltage range (or at least, not the full range the throttle needs).

What voltage range does the throttle read, from off to full rotation?


Another possibility is that if this controller was designed to work with a display, you may not be able to use anything other than basic low-level assist without the specific display/buttons it was made for, so you can turn up the level each time it is turned on. (I have some here like that by LiShui).




I was assuming the battery BMS might have some kind of cut off that would momentarily cut power if the current peaked too much. But again, the motor seems to function correctly in self learn mode and doesn't have trouble drawing more current then.

It might have a current sensor and do that, but the cheaper the BMS the less likely that is. More stuff, more cost. :/ If it did have a limit, it would just shut off until the load goes away, and you'd see any lights or display go off too. Voltage output would drop to zero with a load on it.

In self-learn mode, the wheel should only be being operated off-ground, not ridden, so if you're seeing high current draw then, too, there's something else wrong, either with the motor (or friction in whatever it is driving), or the controller.



Again, what's confusing me the most is that the controller is not displaying any of the problems (skipping when loaded, getting locked in place when stalled) in the self learning mode. Could it be that the controller is somehow stuck in sensorless mode while actually running even though the sensors are working?
Sure. Anything is possible, especially if something is wrong with motor or sensors or wiring that shows a problem only under load (when riding) and not when doing the offground learning stuff.
 
amberwolf said:
If this is with it off-ground, rather than riding it, then 4v of voltage drop means you have a problem with the battery (separate from whatever else is wrong), as it should not sag hardly at all while off-ground.

If this is while riding it, that's still a lot of sag, but probably about right for a cheaply-made battery pack of unknown cells.

The measured 4v of voltage drop is under very high load (near stall) that I apply with the rear brake. It's how I simulate riding it while I'm stuck inside. I noticed the issues I'm having when I went outside for a test ride but when I'm working on the bike inside I have it on a stand and I simulate riding it by applying the rear brake. The battery pack is also Panasonic (both the cells and the BMS), it went through quite a few cycles but I know the capacity is still decent from riding the bike before.

amberwolf said:
Details matter in troubleshooting. If you're embarassed about something you did to it, don't be. Many people here have done some pretty silly things. ;) But if you don't tell us what actually happened originally, it's possible that some detail in that will immediately point us, thru experience, to what is causing the problem. Might not tell us anything. Won't know till we know. :)

If this was a short in the axle wires, it may have damaged hall sensors. Since the test results below indicate they are not damaged, then probably not. Could've damaged phase wires; if it was a crash the motor may even have physical damage inside that's only apparent when under load riding it, not in offground learning modes.

However, if the short included the 5v line, it's possible for it to have damaged anything connected to it. (not likely, since the hall sensors still work, but depends on the voltage applied at the time--motor halls are often able to take up to 24-30v, other sensors like throttles may not be).

If this was a short in the battery side, the battery itself (it's BMS) could be damaged, but if you've replaced that whole thing, it doesn't matter, shouldn't be anything left on there to affect anything.

The short was between the old display and the controller, not on the motor side. Either the display or the controller electronics were damaged beyond repair and there's no replacement parts. Not really relevant to this problem so I didn't mention it.

amberwolf said:
It's interesting that the learning mode fails to work correctly in sensorless, but not in sensored, and that sensored mode doesn't work the same in learning as normal modes. But I don't see any reason it should do so, outside of the controller itself.

The only possibility (other than a controller design problem or other controller problem) that comes to mind is a problem with the throttle not actually providing it's full voltage range (or at least, not the full range the throttle needs).

What voltage range does the throttle read, from off to full rotation?

I haven't measured it yet but i'm fairly certain that the throttle is working correctly as there's still a complete range of speed wiht it, from barely rotating the wheel to spinning really fast.

amberwolf said:
Another possibility is that if this controller was designed to work with a display, you may not be able to use anything other than basic low-level assist without the specific display/buttons it was made for, so you can turn up the level each time it is turned on. (I have some here like that by LiShui).

Probably doesn't mean much but the local seller I got it from told me that it should work as is, with just the throttle connected. I don't know the exact model since all the writing is in Chinese but it's the type you'd get as one of the first results when googling for an e bike controller.



amberwolf said:
In self-learn mode, the wheel should only be being operated off-ground, not ridden, so if you're seeing high current draw then, too, there's something else wrong, either with the motor (or friction in whatever it is driving), or the controller.

There's slight current draw when the motor starts and then it drops to a lower constant current, so normal behavior.


amberwolf said:
Sure. Anything is possible, especially if something is wrong with motor or sensors or wiring that shows a problem only under load (when riding) and not when doing the offground learning stuff.

Again, when working on the bike with it on the stand I simulate a load by squeezing the brake. There definitely is a difference in the way the controller behaves in the self learning mode and in regular mode, both under load and no load.
 
I'll bump the thread with more info since I did some more testing. I found a thread here about the type of controller I have. I opened it up and found where the debug led should be, soldered one in and found out that the controller is indeed stuck in sensorless mode. It's not reporting any other problems like damaged mosfets or anything like that. I also measured the current draw and the battery is definitely not the problem since I've measured a max input current of 18A (when the motor starts stuttering and acting up). I also rechecked the hall sensors and they do all indeed switch from low to high as the wheel turns.
 
If it's stuck in sensorless, it may have a hard time starting up from a stop. Sometimes it may even try to go backwards first.

Unfortunatley there's probably no way to make it work in sensored mode; it's probably programmed to be that way, or some other hardware thing (even though it has wires for halls, and a learn mode).

I've got one controller that isn't even supposed to be sensorless-capable, but it turns out that it is really *only* sensorless. It has hall wires on it, all green (so are the phase wires), and a self-learn function. But even if the halls are not hooked up at all, it does exactly the same thing as if they are--it does not respond to them at all, and always runs in sensorless. It doesn't have as much trouble starting up sensorless as most do on my SB Cruiser trike (which is very heavy), but it has to be helped forward in many situations to get going instead of just grinding.

Like others, it also, in offground testing, may spin backwards a small amount before it starts spinning forward.

There are a number of labelled but unknown function pads on it, and I've experimented with them all thru a resistor to ground or 5v, with only a few having any response (EABS, reverse, "three speed" switch, that isn't really three speeds, but three modes that affect speed/etc). There's a thread here about this controller if you're curious (though I don't know if any of the stuff "whatever" posted is applicable or not):
https://endless-sphere.com/forums/viewtopic.php?f=2&t=68501
It's running different motors nowadays, but it's still working.
 
Hello Lukam,
As I encounter the exact same issue as you do I'm just wondering if you found out what was going on in the end?
I must say that spending so much time to end up stuck so close to the end is pretty depressing :mrgreen:
 
Back
Top