DIY pack - dead cells or BMS issue?

[design_intent]

Could use some help to diagnose an issue with a custom 48V Li-ion (18650) pack - specs below.
I finished building it the other day & everything seemed fine - read 50.5V out of the BMS.
Wrapped it up and left it on the bench.
Went to test the motor and controller tonight and no voltage output from load wires!
Cut the heat shrink off to inspect. 46V total, 0V in 6th series bank (all others approx 3.8xV)
Bummer... Why?!
Any ideas what happened (these were all brand new cells and no external load has ever been applied to the pack)?
I'm now wondering do I replace the discharged cells and the BMS?...

- 13s3p (48V, 9Ah nominal)
- Brand new Sony VTC6 (3Ah) cells (were all reading 3.8xV when I built the pack)
- 45A BMS
- 8mm/0.2mm nickel (1 strip per parallel bank. 3 strips per series connection)

Thanks!

 

[amberwolf]

The most likely cause is the balancing shunt on teh 0v channel is stuck on, so it keeps continuously draining the cell group, and never stops. It's a somewhat common type of failure. Generally these only drain a few dozen mA at most, so let's call it 50mA, so they take a while to do this to a bank of cells. 9aH per bank, at about 3.8v, isn't quite full, let's call it 75% (I don't know wihtout an SoC curve, which I dind't look up for this cell model). 75% of 9Ah is about 6.75Ah. If it was being drained at 50mA, it would take over 100 hours to drain completely...so how long was it left sitting after finishing it, before resuming testing?

SInce I don't know the actual balancing current, and I don't know the actual state of charge of that group, that is a wild guesstimate, so it could take longer or much less time.


Anyway, you can test for this by replacing the cells in that group with cells that are below the balancing voltage level (usually 4.2v or so, so if they're 3.8v that's fine), and measuring the voltage across the balance resistor for that channel. If there is voltage across it, it is on and draining the cells. You can also check the other channels, and if any of them are draining they wlll also show voltage across the resistor. If none show a voltage there, then you have a different problem.


The second cause can be a bad cell that drains the cells connected in parallel with it, but this is not very likely with new cells. The rate of drain of this type of failure is very dependent on the specifics of the failure, and can happen very fast (making noticeable heat) or it can take days or weeks or more.

(Depends on their source whether they might actually not be new, and/or they might be rejects, even though new/unused. Hopefully not. )


FWIW, I would not recharge the cells that went to 0v; I'd consider them trash, and recycle them. Too much risk of fire if they're recharged and used normally after that deep a discharge. (sometimes they do work for a while, then something bad happens...sometimes they seem to work but have lots of voltage sag and get hot and don't have their original capacity. Sometimes they actually do work, with somewhat reduced capabilities and lifespan...but the risk of fire is always there.

 

[design_intent]

Thanks for the response amberwolf - much appreciated.

I finished wiring up the BMS last Saturday and then didn't come back to it until last night (Thursday) - right in your estimated time range.

I was worried this issue might occur with the other cell groups so I removed the BMS along with the discharged cells last night otherwise I would for sure take your suggestion and check the voltage across the balance resistors.

Your thoughts are worth a lot - I definitely will not recharge those three cells.

I'm also now concerned that in case of the second cause you mentioned being the root cause (or this occuring in the future), I'm considering cutting the nickel creating parallel connections and soldering fuse wire between the cells. What are your thoughts on such an approach and do you think 10A fuse wire would suffice?

My plans to fix are currently as follows:
- replace BMS with a 20A model (I only intend to draw around 20A anyway, maybe 25A max)
- replace the three discharged cells
- fuse wire between parallel connections on the positive ends?

Any thoughts are appreciated. Thanks!

 

[amberwolf]

I was worried this issue might occur with the other cell groups so I removed the BMS along with the discharged cells last night otherwise I would for sure take your suggestion and check the voltage across the balance resistors.

Really the measurment is just a confirmation of the problem, but if the time range is right, then that's a type of confirmation as well (though not a guarantee, it's the most likely).

Unfortunately it's difficult to repair the channels, unless you have a source for the SMT transistors and/or voltage detectors used, as those are the most likely things to fail and cause it to stick on, so replacing the whole BMS is easier (probably cheaper).

I'm also now concerned that in case of the second cause you mentioned being the root cause (or this occuring in the future), I'm considering cutting the nickel creating parallel connections and soldering fuse wire between the cells. What are your thoughts on such an approach and do you think 10A fuse wire would suffice?

Per-cell fuses would help in teh case of a complete internal cell short; it has to draw enough current thru the fuse wire from neighboring cells to cause the fuse to blow, or it doesn't do anything to protect anything. It's more useful in "massively parallel" packs; basically you'd have to see if there could ever be enough current in a 3P pack of your specific cells to blow the fuse; if not, it's not worth doing.

If you do want to do cell fusing, there are at least several threads just about that sort of thing, and multiple pack builds doing that--whether they've proven effective or not I don't know.

My plans to fix are currently as follows:
- replace BMS with a 20A model (I only intend to draw around 20A anyway, maybe 25A max)
- replace the three discharged cells
- fuse wire between parallel connections on the positive ends?

Basically the BMS current simply has to be lower than the current the cells x #P can handle, since it's job is to protect them. The lower you go with that current, the more protected the cells will be, but at the cost of system performance if your needs require higher currents.

The controller should be sized to provide the current the motor will need to do it's job...but the battery has to be sized to provide that. If the battery is too small / BMS is too low a current, it will shutdown whenever the controller pulls more current than the BMS is allowed to provide.