Wheel resistance when controller plugged in

[oseloth]

Hi all,

So I have a problem and I believe it's controller fault. When I connect controller to battery my wheel show resistance in turning. Like one faze is active. I measured GND to fazes and + to fazes of the controller and I get around 10k ohm Ground to fazes. Display is not even turned on. just the controller plugged into battery.

What error could this be. If there is no short to ground. Hub motor is ok I did checked it. It spins free when not connected. No shorts to casing. Halls are ok all works fine. All cables are ok. Fazes of the hub are also ok, measuring 0.8ohm across fazes. Fazes connected in-between induce power and you can feel the similar resistance as when controller is plugged in.

Thank you all

 

[lnanek]

Maybe the controller supports regen so you are running a little generator? The Grin controllers have an option called virtual electronic freewheeling that can help by applying a little power in this case, I think.

 

[oseloth]

No regen option on this. Only breaks that disconnect their dedicated connectors, but for sure no regen. Regen is hardly efficient on a bike. Even in cars it up to 3% of total charge. So I didn't installed one with that option. Guess I'll just have to buy a new one and test it. It look to me like a software glitch stuck on one faze constantly. But tnx for the idea.

 

[E-HP]

What is the phase to phase resistance on the controller output conductors?

 

[oseloth]

What is the phase to phase resistance on the controller output conductors?

I'll go measure again but I think it was around 50k

 

[amberwolf]

So I have a problem and I believe it's controller fault. When I connect controller to battery my wheel show resistance in turning. Like one faze is active. I measured GND to fazes and + to fazes of the controller and I get around 10k ohm Ground to fazes. Display is not even turned on. just the controller plugged into battery.

Did the system ever work correctly? If so, then what specifically happened between the time the system worked and the time this failure appeared?

Are all the phase to phase resistance to B+ (battery positive) wire of controller equal and high? The FETs go to both, and if either side (high or low) of the bridge shorts then it will cause resistance to turning the wheel.

However, in both cases this will also happen with the controller *not* connected to battery, *and* connected to motor in the wheel.

If it requires that the controller be connected to the battery to cause the problem, it's not likely to be blown FETs, but it could be a gate driver problem in the controller that is causing it to turn on FETs and leave them on, which shorts the motor phase wires and causes wheel resistance--but this also may cause excessive heating in the controller and be detectable by hand, and/or it could then blow the FETs if the wheel is turned and the controller then tries to turn on other FETs while those are still on.

If it isn't that, then depending on system design, if the MCU gets powered even without the display on (not typical), it could be a function of the controller like antitheft measures or electric braking of some form, that has been turned on either by a setting from within a display or computer/phone based setup program, or connecting wire pairs installed in the controller for the purpose.

You'd need to check the controller manual for whether it has these functions, and how they are engaged, so that you can disengage them, if this is the case.

 

[oseloth]

Here are the measurements.

Black or red probe on - and between all fazes 9.8k
Between all fazes 18k

But:
Red probe on + and between all fazes OL
Black probe on + and all fazes 0 Ohm

Yes it worked properly before and than one of the Halls died. The middle one on a blue wire. I replaced that one and after that the wheel cogging started while controller powered on.

Anti-theft sistem in my case is just a key and I did check that. If key is not turned on the system is not turned on so that works also.
I disconnected breaks and throttle same effect. Just hall and fazes connected and that effect happens.

Also one new moment is that while that resistance is felt if I unplug hall cable the resistance is gone. I than plug it in and no resistance until disconnecting and reconnecting power again. But I tested all the halls, they all work good and in correct sequence and no shorts. with multimeter and with e-bike testing device.

 

[oseloth]

If it requires that the controller be connected to the battery to cause the problem, it's not likely to be blown FETs, but it could be a gate driver problem in the controller that is causing it to turn on FETs and leave them on, which shorts the motor phase wires and causes wheel resistance--but this also may cause excessive heating in the controller and be detectable by hand, and/or it could then blow the FETs if the wheel is turned and the controller then tries to turn on other FETs while those are still on.

Yes this was my thought at first, the gate problem. But I feel no heat anywhere.

 

[amberwolf]

The unplugging hall connector removing the problem means it is the controller trying to do something incorrect based on the hall signal (or less likely the 5v supplying the halls is dragged down by some problem in the hall(s) which may drag down the MCU's 5v and cause incorrect behavior, etc.)

When you test the halls, are they connected to the controller with the controller powered on? What are the actual voltages you see at the 5v supply to the halls, and what do you see on each hall signal for a high, and for a low?



Just for curiosity, is the middle one physically flipped on your motor, or do they all sit the same direction (number side out toward the magnets)? Some motors flip the middle one to get 120 degree hall timing, some are spaced out to do this. If a motor is meant to flip it but it isn't, you get 60 degree timing which gives a different hall pattern (they still switch in sequence but the pattern is different) and the controller may respond badly to this if it isn't designed for it.

 

[oseloth]

Ok I will test the voltages. I haven't tested them connected to controller just via ebike test box and multimeter. But I was in diode mode just checking for beeps. I will do the voltages after work.
They all head a pre-carved slots and they were all facing written side out towards magnets so I couldn't switch sides. They are spaced out i would say 2-3 magnets apart. just from memory.

Are the resistance I measured ok? From the post above?

Thank you for your help!

 

[amberwolf]

Ok I will test the voltages. I haven't tested them connected to controller just via ebike test box and multimeter.

It's just important to do this test because the controller is what is exhibiting the problem when halls are connected; why it happens we don't know yet, but this might get closer to findng out.

They all head a pre-carved slots and they were all facing written side out towards magnets so I couldn't switch sides.

If they're in shaped slots that can't go backwards then nothing to worry about on that.

Are the resistance I measured ok? From the post above?

If the resistances are all the same on all the FETs then theyr'e all ok. It's when they are significantly different that it is time to worry.

But I doubt you have any such problem given the hall thing.

 

[oseloth]

It's just important to do this test because the controller is what is exhibiting the problem when halls are connected; why it happens we don't know yet, but this might get closer to findng out.

If they're in shaped slots that can't go backwards then nothing to worry about on that.




If the resistances are all the same on all the FETs then theyr'e all ok. It's when they are significantly different that it is time to worry.

But I doubt you have any such problem given the hall thing.

Ok. And that 0 Ohm resistance between plus and all fazes? They are the same but 0 when black probe is on the + and red on the fazes. But as you said the are all the same.

I'll write you later afternoon. it's 8.30 AM here

Thanks again

 

[oseloth]

I was just reading this article Testing Hall sensors

And I will also test this. Who knows if the polarization of the newly installed hall is correct at all.

 

[amberwolf]

Ok. And that 0 Ohm resistance between plus and all fazes? They are the same but 0 when black probe is on the + and red on the fazes. But as you said the are all the same.

FETs have a built-in parasitic diode that when reverse connected will conduct, so a meter that can supply enough voltage across it will turn it on and read a low or zero resistance. Diode test will read zero or infinite for off diodes, and a number representing the voltage drop for an on diode.

Note that if you do any resistance tests with the controller connected to power, the results don't tell you the same thing--they tell you which ones are being turned on by the controller (or if the voltage across the FETs is higher than the meter can handle in the mode it's set to, the meter will fail).

I'll write you later afternoon. it's 8.30 AM here

1AM here, but I sleep terribly and not much at a time; after particularly bad nightmares I can't go back to sleep for a while so I either work on my music (Bandcamp Amberwolf) or I come here and help where I can, until my mind changes gears and I doze off again, repeating this cycle until I have to get up for work (then I come home and repeat the cycle until time to go back to work).

 

[oseloth]

I always test it disconnected not to do any live shorts while probing.
I am more and more suspicious about that new installed hall. I'll go first check it's polarization activation. It might be it' on in reverse.

Damn that's bad to hear. You are in USA, some herbal remedies are legal there now Some good indica helps with sleep a lot

Are any of your work on SoundCloud or somewhere? I could listen to something new while at work.

 

[amberwolf]

I am more and more suspicious about that new installed hall. I'll go first check it's polarization activation. It might be it' on in reverse.

It could also be that whatever caused the hall to be a problem in the first place is still present, and causing a different version of the problem...

Are any of your work on SoundCloud or somewhere? I could listen to something new while at work.

Yes, thanks for any listening; it's in the Bandcamp Amberwolf link also posted in my mention of it above (and in my signature / profile). That site has stuff that is either "finished" or good enough to listen to, mostly in latest-stuff first order (the album Lies, Truth, and Assorted Inconveniences is composed of some newer and some older stuff).

There's also Soundclick Amberwolf which has experimental and unedited and older versions of the stuff that's on Bandcamp.

I do have a Soundcloud

https://soundcloud.com/user-792388526

, but iI recommend Bandcamp first, then Soundclick, before bothering with Soundcloud, since t has nothing new and nothing that is there is a recent version of itself, since they changed to paid-only for a lot of things, I can't update a song on there, can only delete it and then completely remake a page for it, losing any comments, likes, listens, etc., and any links anywhere to it being broken, I don't often bother putting up stuff there. (They also limited the space so I have to delete stuff to add new stuff).

(It isn't worth paying for any of their "features" since very few people ever listen to any of the stuff on any of the sites anyway, so I use them more to archive my stuff "off site" than anything else, in case of some tragic event that might destroy whatever I have locally.)

 

[oseloth]

I listened to several tracks. Quite good. I like that style of music. I'll continue tomorrow.

Update on the wheel. I doubled checked all the halls. They work properly. I get 4.8V on output and 0 when magnet cross. The testing device is also showing good on off sequence.

And again still the same problem. But this time i started disconnecting fazes whiles turned on and when disconnecting green and blue i could spin the wheel without resistance. Conecting any of the two made it resist again. No. Difference in strength if green or blue or both are connected. That must be controller fault.

I do have oscilloscope and I am tempted to see what is on the gate of all the FETs

 

[amberwolf]

With those results, I would guess that something in the controller is reading the halls in the wrong order (or the wiring to them is different order now than it originally was), and triggering FETs incorrectly based on that. (or the phase wires are in the wrong order, less likely)

When the halls are disconnected, it doesn't see any signals so it falls back to sensorless mode (assuming it will operate the motor without the halls connected), or simply stops trying to operate the motor at all (assuming it is sensored-only).

If it's the latter, then it could mean a fault in the gate drive or MCU so that when enabled or in sensored mode, it turns on one or both of the blue/green FET bridges when it isn't supposed to. It's obviously not just stuck on, or it couldn't go away when you disconnect the halls, so it is a "complex failure", but could still be a very simple cause that's easy to fix.

 

[oseloth]

For the hall side I am now 100% sure. I just changed them on the board so cabling is the same. orientation is the same. The all look in same direction towards magnets. They have slots cuted so you just can't turn them 180.

I double checked any shorts to casing. None also.


Bumer. I'll just order another controller and cross my fingers
I did had one older with less power and I did try to connect that one but my display isn't turning on.

What is strange though on the "broken" one is why is he working if the display is not tuned on. I mean do they work instantly when connected to battery. That shouldn't be the case. I mean, it draws some power and it could run the battery down.
I'll test the current consuption also with it plugged in, and with and without hall line connected.

 

[amberwolf]

I did had one older with less power and I did try to connect that one but my display isn't turning on.

Displays and controllers are not generally intercompatible, so unless you know for sure the display works with that controller, it's likely that it doesn't, and may even have different wiring that can lead to damage of controller, display, or both (because there are usually at least two wires with battery voltage, and two 5v-level data lines, and a ground, and sometimes other things--if any of the battery wires line up with 5v wires, stuff often goes POOF. ).

What is strange though on the "broken" one is why is he working if the display is not tuned on. I mean do they work instantly when connected to battery. That shouldn't be the case. I mean, it draws some power and it could run the battery down.

I'm not quite sure what you mean.

If you mean the motor runs with the display off, that's normally impossible, because the display, when turned on, is usually what passes battery power to the controller LVPS to make 12v and 5v to run everything (MCU, gate drive, etc) except for the battery voltage that goes directly to the FETs for phase voltages that go to the motor.

If it operates differently with the display on vs off, but still operates, it means something is wrong in the wiring between display and controller or something broken in the display, such that the display can't turn off the power to the controller's LVPS.

And it means that the display is probably preventing the controller from operating, probably due to a limit exceeded as chosen within the settings in the display.

Whatever is going on, it just gets stranger with every symptom.... :/

 

[oseloth]

Displays and controllers are not generally intercompatible, so unless you know for sure the display works with that controller, it's likely that it doesn't, and may even have different wiring that can lead to damage of controller, display, or both (because there are usually at least two wires with battery voltage, and two 5v-level data lines, and a ground, and sometimes other things--if any of the battery wires line up with 5v wires, stuff often goes POOF. ).

I opened them both and connected all correct wires. red to red, gnd to gnd, tx and rx to tx rx, and the last cable is probably throttle. But it didn't turn on. That was just a test. When I pluged the old one it did turned on so nothing is fried in this attempt.

I'm not quite sure what you mean.

If you mean the motor runs with the display off, that's normally impossible, because the display, when turned on, is usually what passes battery power to the controller LVPS to make 12v and 5v to run everything (MCU, gate drive, etc) except for the battery voltage that goes directly to the FETs for phase voltages that go to the motor.

If it operates differently with the display on vs off, but still operates, it means something is wrong in the wiring between display and controller or something broken in the display, such that the display can't turn off the power to the controller's LVPS.

And it means that the display is probably preventing the controller from operating, probably due to a limit exceeded as chosen within the settings in the display.

Whatever is going on, it just gets stranger with every symptom.... :/

hehe. My life is full of strange problems. This is the way...

The stuck motor symptom is showing with hall and faze lines connected to controller immediately when I plug the battery to controller. I don't even have to plug in display. And If I plug it and not turn it on it still shows that two fazes working.

That is a mystery

I think I spent enough of your time on this. There must be some glitch in the Matrix there. I'll order another one and try it like that. If the symptoms continue than I will go deeper into the problem.

 

[amberwolf]

I opened them both and connected all correct wires. red to red, gnd to gnd, tx and rx to tx rx, and the last cable is probably throttle. But it didn't turn on. That was just a test. When I pluged the old one it did turned on so nothing is fried in this attempt.

Displays don't usually have the throttle connected to them, and don't normally have a wire in their cabling to the controller that includes throttle signal--it's normally just two battery voltage wires, a ground, and two data wires. One battery voltage wire is *from* the controller to power the display, and the other is *to* the controller, only powered on when the display is turned on, thus providing power to the controller's LVPS.

If that battery-level output wire from the display is connected to the controller's throttle input, it would prevent normal controller operation because the battery is higher than the throttle "pedal high" limit (a safety cutoff some controllers have to prevent runaway throttle operation) or the throttle "already on at power up" limit (a safety cutoff some controllers have to prevent unintended controller operation at startup). Depending on the input hardware of the controller, could even reach and damage the MCU.

A few systems (TSDZ2, some proprietary OEM-only like at least early Stromers, etc) do connect the throttle to the display, and then the display feeds the signal to the controller either directly or by converting it to data and sending it over the tx/rx lines.


I wish it was otherwise, but colors don't necessarily mean anything other than that they are that color. Matching colors may work, or may lead to damage or faults. If you know for sure from either tracing the wires to PCB pads inside the devices, or from manufacturer documentation stating specific colors are for specific functions, you can go with that...but otherwise, making sure *before* making the connections is a good idea. (even finding color info from other users of the device isn't a guaranteed thing, as wire colors are too-often chosen by whoever builds a particular item, based on what wire colors they have there at that moment, or on how they feel like wiring it that day, etc. )