18650 new cells to 0v!!

[jvens]

Hi all,

I have a problem that is repeating itself frequently.
I am working a lot with 18650 cells, and I assemble different packs.

Sometimes, after a while (within several weeks), one of the cells seems to be broken, to a voltage of 0V. So that brings trouble to the balance of the total pack.
It is not the CID (current interrupt device), because overall voltage of the pack is still intact. Meaning the other cells are getting to high in voltage to compensate the broken one.
I do use decent BMS modules on the packs, I do not believe this has anything to do with it. The cells that I use are NCR18650B (Panasonic), VTC5 (Sony), INR18650-30A (Samsung, a while ago). All these types had this issue sometimes.

My question is, can I discover that broken cell before I put the pack together?

It is very frustrating, it occurs very often recently, and cost a lot of money..

Thanks for your help!

Best regards,

John

[fourtytwo42]

I do use decent BMS modules on the packs, I do not believe this has anything to do with it.

Surely if the BMS was working properly its job is to prevent this happening!
What BMS is it and what are the charge/discharge regimes ?
Supplementary, are you sure the BMS contains charge balancing and are the cells in any way matched before assembly ?

[BravoV]

Adding above post, assuming the BMS is working, have you profiled say like do a full charge & discharge every cells, to have an idea they have similar characteristics ?  Like putting every one of them in a LiIon smart charger that can do charge recycling, laborious task though.

[capt bullshot]

As stated, cells to be used in a pack must be selected for tight tolerances in capacity and charge / discharge characteristics and must be of the same type and manufacturer. A typical BMS cannot compensate for gross differences, but should prevent discharging single cells below their minimum voltage or charging above their maximum by cutting off the whole pack. If that happens, the pack shouldn't be used anymore because that leads to the results you describe.

[Nusa]

What's the pack configuration?
How far are the cells allowed to discharge?
Roughly how many charge cycles do these packs experience before failing?

[jvens]

Thanks for thinking with me here

I think it has nothing to do with the BMS. Because I have checked and used the old BMSes again on new cells, and everything was fine.
It occurs mostly in the first 10 cycles. Or it was there before but the BMS cannot compensate enough and it is getting out of hand.

The packs that I build are various, from 13S8P, 6S4P, 7S4P, you name it..
And yes, a good BMS would prevent this from happening, BUT if a cell decides to give up, the BMS is not strong enough to correct something that cannot be compensated.

If the cells are matched? I don't know.. I allways use cells from the same batch. And I check the voltage of each cell. Image that I would match the cells to build a 14S10P pack for example.. What is the correct way to do that?

[sleemanj]

You said that other cell voltages are rising to maintain pack voltage, cells should not be allowed to rise above 4.2 unless you are building an incendiery device.  If your bms/charge circuitry is allowing this, I would not be surprised if this is the root of your problems.

18650 cells just do not fail with the frequency you are experiencing in normal usage.

Also, I hope those cells in parallel are fused or otherwise over-current protected, if one of those 8p goes short circuit, that would be pretty exciting for the other 7.

[hayatepilot]

Cells just dying after a couple of cycles should not happen with genuine quality cells. Those are individually tested at the factory finding a faulty one is probably less than one in a million.
Where do you get the cells from?
How much current does one cell need to provide? How hot do the packs get?
Do the cells fail short or open?
This sounds to me as a error in assembling the packs rather a fault in the cells.
Do you solder or spot weld the packs?

Greetings

[jvens]

@ sleemanj
Yes I have seen voltages rising above that. BMS manufacturer says the difference is that big, when a cell dies, so the BMS cannot compansate anymore.. Don't know if this is true though..
And no I have fuses in parallel.. Do you have an example of how it looks like? I use nickel/copper strips just over the complete pack.

@ hayatepilot
I get the cells from Nkon. The current is limited, I do think about the parameters, I read the datasheets for that. The packs are not getting hotter that allowed, until one cell is out of balance.
The cells fail short, it is not the CID.. I allways spot weld the cells, never solder on them..

Btw, I heard a cell can stop working if you weld on the exact middle of the (-) side.. is that a thing?

Here is an example of a BMS i could use https://www.alibaba.com/product-detail/13S-15A-PCM-BMS-for-48V_60684570837.html

[hayatepilot]

Btw, I heard a cell can stop working if you weld on the exact middle of the (-) side.. is that a thing?

Here is an example of a BMS i could use https://www.alibaba.com/product-detail/13S-15A-PCM-BMS-for-48V_60684570837.html

Can't imagine that the cells stop working if welded exactly in the middle. Some cells have a overpressure vent on the bottom which could be rendered useless depending on the weld.

That BMS may be functional but it shure doesn't look high quality. I would still suspect faulty BMSs to short the cells in your packs.
Could you upload some pictures of the packs?

Greetings

[jvens]

Here an example of a pack, 13S4P. One block of 4 cells was dead within 5 cycles.
What went wrong here, is the balance capacity of the BMS, I guess. The currents are to low for 4P to balance.

But in the beginning, all the cells were even. So how did the difference occur is what I wonder..

Another example, which I don't have photo's of, are pack for electric bikes. I have replaced the cells, also all even, and the same thing happened after a while.. so that made me think that the cells are just failing sometimes..

[capt bullshot]

So the BMS is supposed to disconnect the pack from the load / charger when one cell has UV / OV. The BMS has to monitor each cell  (or multiple P cell pack) voltage. If it doesn't, there's most probably at least one root issue.
Can't tell from your picture if your wiring was correct, and can't tell if the BMS does so. So you'd have to check if your BMS does disconnect the load if one cell has too low a voltage, and disconnects the charger if one cell has too high voltage. You should also check the BMS itself for uneven leakage currents through the cell sense wires.
The load / charger should stay disconnected until the opposite action (charging if UV occured / discharging if OV occured) starts. And yes, these "passive" BMS cannot equalize larger differences in cell voltage / capacity, so your failure is as expected if there's a significant difference in the cells.
Did you check the maximum allowed cell discharge current capabilities matching you load demand? There're many cell types that cannot cope with a load a high as an electric bike, to that might be the root cause here. For a jellybean LiIon cell, one isn't supposed to discharge them at more than a 0.5 ... 1C rate (about 4A ... 8A for a 4P pack). For higher currents, you'd need cells that are specified for your load current, e.g. for power tool usage.

[jvens]

Good advice to check the BMS, unfortunately I have no broken one to test with atm.

Yes I allways check the specs of the cells. This one was allowed to discharge till 6,7A, and fused with 5A.

I think uneven cells are the cause indeed, but I cannot understand why something would happen (apart from a bad BMS).

What can I do to get the cells equal to each other, before connecting the BMS? Or shouln't that be necessary..?

[capt bullshot]

There are at least two things to do:
1. cells must have the same (real) capacity or pre-selected by the manufacturer
2. cells must have the same state of charge when assembling the pack - just fully charge them before assembly

There's another point:
Don't rely on the BMS for disconnecting the load at end of discharge. A typical BMS turns off the load as a last resort to prevent permanent damage. This voltage is too low for repeated usage. For normal operation, there should be another charge / discharge control circuitry that shuts down the load when the battery is discharged. This usually has a higher threshold (voltage per cell or "fuel gauge" based) than the BMS. And the same is true for charging, the BMS usually disconnects at a too high voltage for normal usage, the charge controller is responsible to end the charge.

[hayatepilot]

Another example, which I don't have photo's of, are pack for electric bikes. I have replaced the cells, also all even, and the same thing happened after a while.. so that made me think that the cells are just failing sometimes..

That doesn't make sense. You must be doing something wrong. New cells just don't drift or fail like that.
Even without any BMS good quality cells will stay within a couple of tens of milivolts after dozens of cycles. Either you are connecting them wrong or those are counterfeit cells.

[BravoV]

The devil is in the detail, OP has not exposed enough details involved here, imo its hard so know what was happened or happening as now. 

[capt bullshot]

Beeing able to log the voltage of each cell while charging / discharging in the real application would help in diagnosis.

[jvens]

@ BravoV, yes correct. I had some issues and got my conclusion out of that. But I seem to be wrong, and therefor I need to do new researsch, whenever the problem occurs again..

@ bullshot yes that would be great. Do you know the tooling for this, that is affordable..? I have profilab but it's not my favorite thing to work with..

[capt bullshot]

@ bullshot yes that would be great. Do you know the tooling for this, that is affordable..? I have profilab but it's not my favorite thing to work with..

On a budget: A multimeter, paper and pencil plus a trained monkey
Otherwise: Anything that does allow you to measure n voltages (e.g. a multiplexer and a DMM), is remote controllable and some logging SW on a PC, maybe a multi-channel recorder, ...
Advanced: build some uC based hardware that is incorporated into the battery pack sending data to e.g. a raspi recording the data
Anyway, I don't have any recommendation on something that one can buy because there are many options from some near scrap multi channel data recorder from the EvilBay up to a 40kEUR instrumentation ...

[Fraggie]

Do you test the cells , to see if they have a capacity close enough together?
because i think that's your problem..

https://youtu.be/seap6_v5Ck0

[james_s]

Are you soldering directly to the cells? That will often damage them. The proper way is to spot weld them with nickel strip, or buy tabbed cells.

[Kilrah]

I use this for individual cell checking:
https://www.amazon.fr/ARINO-Universel-7%EF%BC%8BBatteries-42%EF%BC%8BTailles-Application/dp/B01D1DG8WQ

For pack stats, a good RC charger that's appropriate for your cell count, with logging capabilities, mine is:
https://www.progressiverc.com/icharger-3010-b.html

But anyway, 3 possibilities in the order of likeliness IMO:
- Cells are trash in the first place, counterfeits, high reject rates, who knows (would kinda favor that, seems that site sells cells (hah) at about a half of the prices I'm used to seeing, there must be a reason for it...)
- You're damaging them by soldering to them, if not using spot welding you must be EXTREMELY careful with your technique. I use desoldering braid and VERY short heat times. If you're making that many get a spot welding machine, can get one for $150 or so.
- The BMS is crap, or badly connected

Sometimes, after a while (within several weeks), one of the cells seems to be broken, to a voltage of 0V. So that brings trouble to the balance of the total pack.
It is not the CID (current interrupt device), because overall voltage of the pack is still intact. Meaning the other cells are getting to high in voltage to compensate the broken one.
I do use decent BMS modules on the packs

This simply cannot be. The BMS's purpose is precisely to cut all current flow through the battery both if a cell is too low, AND if others are too high. So having both conditions and still having current flow is simply impossible if you have a "decent BMS".

Of course before assembly you must charge all cells to the same level (hard to do it anywhere else than full, with all the caution it then implies during assembly), and yes try to match the internal resistances.