Testing / diagnosing battery pack

[notvhere]

Hello, I've happily put on about 5000 miles over 5 years on a yosepower KT-based rear wheel system.
I've had one or two breakdowns in that time, but now I am having a time diagnosing/fixing this latest one. (Still very happy with the system)

It's a KT controller attached to a bottle-style 36V battery, with KT-LCD3 display.
On Friday, bike gave a judder and quit, electrics off. Holding the power button makes the display light up for almost a second, before it shuts off.

Thinking this was some type of short-circuit protection kicking in, I went into diagnosing mode, disconnecting, isolating, simplifying.

In all the tests in the next paragraph, everything else is disconnected, including motor, but also PAS, brakes, etc. I've tried with each of them connected, and the behaviour doesn't change, so assume they're all unconnected in the below.

I replaced the cable between the battery and controller. Same behaviour.
I replaced the KT-LCD3 with a KT-LCD4. Same behaviour.
I replaced the actual controller with a KT-compatible no-name one off amazon. Same behaviour!
All permutations between new and old parts have been tested as well. The LCD display can light up momentarily, almost every time.

The only thing left un-replaced is the actual battery pack. This battery pack fits is the bottle-style type, where the controller is in the mount. In the replaced-controller tests, I directly and correctly attach the new controller to the +/- of the battery pack.

I faked the battery by using the battery charger hooked up to the controller (42V/2A), with the controller and LCD display only. Obviously I can't/shouldn't hook up the motor as the current draw would be too much for the charger. This happily powers the LCD controller and I can press buttons on it, change settings, keep it alive for as long as I want.

The battery pack was partly depleted on Friday due to riding around. I of course charged it overnight. It charged up, the charger LED was green in the morning, from being red (charging) before.
The little 5-led display on the battery agrees it is fully charged.

A voltmeter on the +/- of the battery itself gives 42.1V, which should be perfectly fully for a 36V battery.

I was originally blaming the controller, since many posts on here talk of leakage current through the controller mosfets giving enough power to almost power the display, even though the battery itself can't give much actual power when turned on, but I replaced the whole controller with a different one.

How do I best test the battery? Have tools including a good fluke multimeter.
The blade fuse in the battery was good. I can test the cells inside for equal voltage. Can the BMS inside the battery be at fault here?
Can I just wire the battery pack to some incandescent bulbs in series (to drop the current to e.g. 5-10A) as a load test, or will that freak out the internal battery circuitry?

Thanks for reading this long post! I can add more details/tests if helpful!

 

[Nealh]

You wasted money on buying stuff you didn't need.
Any sign of a LCD not staying on and one should have investigated the power supply between the LCD and battery, inc the battery first.

Did you check any battery voltages at the discharge or try and check the ten cell group voltages ?

 

[notvhere]

The only part I didn't follow was the "voltages at the discharge". Do you mean at the battery output terminals (not the charging port)? If so, yes, exactly and only there.
Or did you mean voltages while trying to pull some power from the battery?

Testing each cell group is next up. I tested the battery first, got 42.1v at the output ports normally going straight to the controller, and then went down, "well if the battery outputs 42V, what is next down the line that could be bad?" route.

 

[saneagle]

There are capacitors in the controller that get charged by the battery. They have enough charge in them to power the system for about 1 second, which is what you're seeing. From that you can deduce that your battery is giving out some sort of power until you switch on, then it stops giving out power.

There are two possibilities: Either there's a short circuit causing the over-current to trip the BMS cut-off, or the battery is knackered and its voltage collapses when it tries to give current.

Short circuits can be anywhere but normally either the MOSFETs in the controller or the motor cable. The motor cable shorts only trip when you demand motor power, not when you switch on, and you've eliminated MOSFETs as a possibility by swapping the controller, so short circuit is low probability.

If the battery is knackered, the bike should run OK with the wheel in the air using the charge as the power supply. That's an easy test, so do that first.

 

[notvhere]

Great, thanks, will try that today and report back.

 

[notvhere]

Updates:

Tested each of the 10 cell packs inside the battery. 8 test at 4.17V, 2 test at 4.16V. The pack hasn't been charged since Friday night (it's Tuesday), so I don't think any pack is failing.

Running the e-bike using the charger instead of the battery pack (up on a bike stand with the wheel in the air so there's no real torque and therefore current draw needed) works just fine. Works exactly as expected.

I think this leads me to the little BMS board inside the battery pack. It has 4 N-mode mosfets on it, and I've located the data sheet for them ( CRSS052N08N ). Testing the source to drain in diode mode shows them all as shorts. Visually they seem fine. I think they test as shorts in circuit is because there's a resistor in circuit between them, so I'd have to take them out of circuit to test.

I can do that next, assuming people agree I'm chasing down the correct rabbit hole...

 

[notvhere]

New rabbit hole. Why don't I simply bypass the battery BMS? It's on the (-) line, (+) goes right out to the port. Jumpering from the (-) port to the (-) battery simply bypasses the BMS. I won't benefit from the balancing the BMS presumably does, but for a 10 second test...

 

[Nealh]

Balancing is only during the final 0.5v of the charge process.
For a test bypassing the BMS is fine for a short time but in use one needs the LVC/HVC and other back up services it provides.

 

[saneagle]

New rabbit hole. Why don't I simply bypass the battery BMS? It's on the (-) line, (+) goes right out to the port. Jumpering from the (-) port to the (-) battery simply bypasses the BMS. I won't benefit from the balancing the BMS presumably does, but for a 10 second test...

What you should do is run the bike on the stand with a voltmeter connected to one of the cell-groups or the whole pack. Turn the pedals or operate the throttle to make the motor turn. See what happens to the voltage, which should remain more or less the same with such a low load, then slow the motor down with the brake and see what happens to the voltage. That should show whether the cell voltages are collapsing or whether the BMS is switching off for some other reason.

 

[notvhere]

I know you've all been waiting with bated breath for my updates.

Found (and fixed) the issue. It might have been obvious to some, but it was a bit insidious. The positive wire inside the battery pack had a poor connection, that had started to blacken, to the output port. There was still connectivity, so there was still 42V at the output ports, but there was about 50 Ohms resistance just through that wire. I've snipped off the end, re-tinned it, dug through about 6 miles of glue to get to the pin, and re-soldered it and it has good continuity.
All fixed!

Thanks for taking the time to read and reply, and help walk me through.