48V Li-Ion shuts itself down at ~45V

Joined
Jun 18, 2018
Messages
17
Location
alberta
My battery is a generic 48V 10.5Ah on an ebike with 5800 km on it.
Recently the BMS has started shutting the motor off whenever V drops to about 45 V. I can't tell exactly because my wattmeter resets itself when the supply drops to zero.
I can replicate its behaviour using a resistive load indoors. The following graph shows what happens.
discharge.PNG.png
I'm not sure why the voltage should be leveling off while discharging. Is that normal? Shouldn't the voltage be dropping off much faster?
What's wrong? BMS?

Angus

ps the preview shows a duplicate graph below that I can't delete. Sorry for the clutter
 
That's not too far off from the specification. Most likely one cell group is lower than the others and triggering LVC prematurely.
Try to measure the individual cell voltages.
 
? What battery ? more info please. What motor controller combo. Important. You can bring up a low cell. But first write drown all parallel cell voltage. like
1. 4.02
2. 4.11v
3. 3.40v


13. xxx volt
You my just have an out of balance cell or over working your pack or ?
How old ?
Try an answer all questions
 
angusinalberta said:
I can't tell exactly because my wattmeter resets itself when the supply drops to zero.
Most wattmeters I've seen have an Aux input for a battery to sustain them when mainpower is cut. (both teh WattsUps and Turnigy Watt Meters I have do..the Cycle Analysts don't need it cuz they save their data at shutdown once voltage drops far enough).


I'm not sure why the voltage should be leveling off while discharging. Is that normal?
If you're using a simple resistive load, then as the voltage drops, so does the current being drawn, so the load on the battery grows less and less, making the voltage level off.


The actual problem with the pack is most likely a cell group(s) that is not as capable as the others, and either drops farther than the others under load due to internal resistance problems, or due to lower capacity than the others (or both).

Leave it on the charger for a few days or longer, and the BMS may be able to rebalance the pack so the low group(s) are closer to the rest. That would fix it for a while, but you'd probably have to keep doing this to keep it balanced.

Eventually the group will get worse, and the pack's capacity will be limited by the weak group(s).
 
amberwolf replies:
Most wattmeters I've seen have an Aux input for a battery to sustain them when mainpower is cut.
It doesn't help in this case. Vmin is zero, not the value an instant earlier.
If you're using a simple resistive load, then as the voltage drops, so does the current being drawn, so the load on the battery grows less and less, making the voltage level off.
True of course. Embarrassingly so.
Leave it on the charger for a few days or longer, and the BMS may be able to rebalance the pack so the low group(s) are closer to the rest.
I've tried that a couple of times with the wattmeter connected. I expected that the battery voltage should sag a volt or two as the BMS discharged the healthy groups, followed by a charge cycle at 54.6V. But that never happened even after a couple of hours. The voltage stayed stubbornly at 54.65 V. At that point I disconnected the charger for fear of overcharging.
 
Then to find out what's actually happening in the pack, you'd probably have to open it up and check the actual cell voltages at the cell groups' balance wires.

I recommend checking both at the cell end and the BMS end, so that if there is a difference you'll know there is something wrong with the balance wire connection itself.

It is also possible that the BMS itself has a balancer resistor "stuck on", on the low-voltage group (if there is one), continuously draining it.

You should still see the normal cycling during charging, though, even though it wouldn't rebalance anything (because the low group is being drained at a faster rate than the other groups, since they're only drained when overfull, and it's drained all the time, even during charging).

Since you're not seeing any cycling, there might be something else wrong with the pack/BMS, though not sure what it would be.

Or nothing is wrong and the charger is now just keeping voltage at the same level all the time, just putting out a few mA (or few dozen), just enough to almost keep up with the shunt resistor drainage, so the low group keeps getting a bit more charge, and the already-full groups keep draining just a little bit, or at least don't get any fuller. (you'd have to measure current to verify that).

Or the BMS could be of the charge-shuffling type, that slowly moves charge from one group to another, instead of wasting it as heat thru resistors like the more common type.
 
Thank you amberwolf for citing all the possibilities. I hadn't even considered the charger.
Open-it-up day will have to wait for sometime after the 25th when the house gets quiet again. Unnecessary distractions would not be helpful.
 
The problem was a dead cell.

The procedure I followed to diagnose the problem:

1. Checked group voltages via the bms leads
Result: group 5 was lowView attachment 2

2. Tested group 5 with liitokala lii-500 using a dummy cell and the bms leads.
Result: 7.20 Ah This indicates that only 3 of the 4 cells were working
Capacity with four healthy 2600 mAh cells should have been more like 10.4 Ah

3. Group 5 was near the middle of the battery pack. So I had to totally disassemble it to get at the dead cell.
I used dental floss as a string saw to cut the cells apart. It seemed as if the cell sleeves were welded, not glued to one other.
Also I had the opportunity to test the individual capacities of the 51 remaining cells with the lii-500.
View attachment 1 It's interesting that it optimistically reports over 2700 mAh for year-old Samsung 26H cells that are rated as only 2600 mAh.

I'm currently rebuilding the battery using a vruzend 1.5 kit with a bluetooth bms and a Tesla replacement cell.

batt.sm.jpg
A big thank-you to all who responded and gave me such good directions.
I don't expect to be able to do any road testing before the snow disappears, mid April at the earliest. Cheers
 
Back
Top