pls HELP - only 2 cell groups charging, BMS showing 0 wire resistance

[brownj24]

Hello,

I built at 21s34p battery for my Bullitt cargo bike front bay area. Yeah, it's big!

Sadly, after only 24 cycles, my chargers were shut off by the BMS with the following error (see pic). 2 of the 21 strings show a little over 4.2v full charge, but all the other cells stayed around 3.7. The BMS also shows 0 wire (nickel strip) resistance on those two cell groups that charged up to 4.2v.

Prior to this, I did notice a hot spot on the battery pack beginning to melt through the blue shrink wrap (even while temperature probes showed overall battery temperature within spec for the EVE 21700 lithium cells used).

I'm looking for your advice/help if you can offer any suggestions as to likely cause of failure and what you would check first.

My plan: Check BMS connections and BMS wire connections, Check spot welds on the hot spot area on those two cell groups that are charged to 4.2v right now. Maybe try to bleed them down closer to the other cell groups if indeed the nickel strip welds have failed somewhere.

I'd appreciate any suggestions. Thanks!

Jeremiah

PS. Here's a video of me building the battery in case you're curious:

 

[amberwolf]

I haven't dealt with a BMS that attempts to measure resistance via the balance leads, so I don't know how yours is actually doing the job, so don't know what failure modes it's software or hardware might have that are different from typical.


But, typically, when you see cells that are that much different in voltage than others, those cells are very different in internal resistance and capacity as well, so will badly affect the performance of the entire pack (if not now, over time as they get worse).

A good BMS (there aren't many) will prevent operation (charge or discharge) of a pack with more than around a tenth of a volt difference between cells, because there is something so badly wrong with them that they could potentially have catastrophic failure modes upcoming, and it's not safe to keep using the pack until the problem is found and repaired.


Almost no cell vendor bothers to match their cells in properties when selling them, and if you don't use matched cells, then your pack will not perform as well (or as long) as it would if you did. Matched means identical (not similar, but identical) internal resistance, capacity, etc. You can test and sort cells that aren't matched to make groups that are at least similar to each other...but the results you are getting show that either this hasn't been done, or you have cells or interconnects within parallel groups that have failed.


If the BMS doesn't have a balance function, you can manually drain the high cells...but they filled up first when all the others are not even maybe half full because they have that much less capacity than the others. So this problem will not go away, it will keep happening, and the pack can only deliver the capacity of the smallest cell (group), so you will only get at best half the capacity the pack could otherwise deliver, based solely on that single picture of voltages.


BTW, anytime you have heat inside the battery, you have a big problem. If it's enough to melt things, you have a potential fire waiting to happen.

A properly spec'd and designed pack, properly built, will not heat up in use or during charge. If it does, it's either defective / failing cells that are heating up inside, has poorly built or spec'd interconnects, or is being used too hard for the cells and/or interconnects capabilities. (yes, lots of packs heat up in use...and they're either defective, broken, built wrong, or underspec'd for the usage they're in, even if they're made by big companies that "know what they're doing").

 

[brownj24]

Thanks so much for the reply.

I'll have to remove the insulation and wrap to have a look and I'll report back.

I'm thinking there is a localized issue with some of the inter-connections between the cells because I spec'd the build based on my max continuous current draw being well within the cell manufacturer listed specifications. On paper, there should be no heat like you say.

The cells were brand new from EVE.

Disappointing, but hopefully salvageable....

 

[docw009]

I'd say you lost some connections on those two P groups, a significant fraction, so they recharged very quickly. How that could happen, watching your video, is hard to fathom. Seems your connectivity is very diligent, Still, it's a gigantic pack. I didn't see how you how held the cells together. Some movement could have snapped off a few dozen welds? Yikes.

As for the zero resistance, the BMS must check the voltage drops on each group and divide by a known current. DeltaV/Current is resistance. With the cells at full voltage, deltaV is minimal and you get zero resistance,

 

[brownj24]

I'd say you lost some connections on those two P groups, a significant fraction, so they recharged very quickly. How that could happen, watching your video, is hard to fathom. Seems your connectivity is very diligent, Still, it's a gigantic pack. I didn't see how you how held the cells together. Some movement could have snapped off a few dozen welds? Yikes.

As for the zero resistance, the BMS must check the voltage drops on each group and divide by a known current. DeltaV/Current is resistance. With the cells at full voltage, deltaV is minimal and you get zero resistance,

Hey thanks for responding. That makes sense.

Funny, after another 24 hours the cells are now perfectly balanced again all around 3.76v. So the BMS must have rebalanced on its own. I also pulled off the shrink wrap and insulation where there were signs of heat penetrating through the insulation and blue wrap, but the kapton tape is in place and there are no signs of broken spot welds are burnt cells. Yet, the BMS still shows 0 resistance on cell groups 14 and 15. Otherwise, everything looks normal.

I'm investigating the JK BMS manual and info available online.

 

[brownj24]

As for the zero resistance, the BMS must check the voltage drops on each group and divide by a known current. DeltaV/Current is resistance. With the cells at full voltage, deltaV is minimal and you get zero resistance.

So far, this thread shows others having the same issue without resolution so far: Wire resistance 0 JK BMS

I'll contact JK and post updates for others who may run into the same issue if I find out what's causing it.

 

[amberwolf]

Funny, after another 24 hours the cells are now perfectly balanced again all around 3.76v. So the BMS must have rebalanced on its own.

How much balancing current does the BMS do per group? Usually it's on the order of 50mA, some big BMS may go up to 100mA.

Let's say it's 100mA. 24 hours of 100mA is 2400mAh, or only 2.4Ah. If those groups went from over-full down to half full with only 2.4Ah (or less if it actualy drained quicker than 24h and you just didn't notice it yet), it means the entire groups have at most twice that capacity, or less than 5Ah.

Not knowing what capacity your cells are, we'll assume they're 5Ah each. That means either 33 of 34 parallel cells in those groups are disconnected, or all those groups' cells are pretty well trashed if they're all still connected, or some combination of the two, or there is a lot of missing information, still unknown / untested, about what is actually happening.

The cells were brand new from EVE.

Unfortunately, in and of itself, that doesn't mean anything about their quality or matching of properties.

Sellers and manufacturers have been reported to sell multiple "grades" of cell, and not necessarily tell a buyer which grade they get (or to lie about it), or simply to avoid ever specifiying the details of what a particular grade is. They also have been reported to send cells that vary widely in properties, even within a single box of cells (which would presumably be from a single batch, but that alone doesn't guarantee matching properties).

Hopefully none of that is relevant to your cells, and you just have an interconnect issue...but i thought you should be aware of the problems of getting good and matched cells for projects.


The "easiest" way of getting matched cells is to only use large-format cells (not little cylindricals) salvaged from large-EV battery packs, since the builders of those packs are going to do what they can to perfectly match each cell to the others, so the pack performs it's best for the longest possible time, so the manufacturer doesn't have to deal with expensive warranty issues.

(small-EV makers don't really care because they either don't warranty the battery or limit that warranty severely enough that they can get out of most of the repairs...harder for large-EV companies to do).

 

[amberwolf]

So far, this thread shows others having the same issue without resolution so far: Wire resistance 0 JK BMS

I'll contact JK and post updates for others who may run into the same issue if I find out what's causing it.

My best guess is that the hardware in the BMS doing the measuring (or allowing it to do it) is failing at the channel level, if it always results in a zero reading. If it resulted in just a changed reading, I'd suspect a connection (perhaps internal to the BMS or cabling).

 

[brownj24]

My best guess is that the hardware in the BMS doing the measuring (or allowing it to do it) is failing at the channel level, if it always results in a zero reading. If it resulted in just a changed reading, I'd suspect a connection (perhaps internal to the BMS or cabling).

Thanks so much for taking the time to think about this and respond - much appreciated!

By channel level, do you mean within the BMS hardware? I'm not sure what "channel level" means in this context.

 

[amberwolf]

By channel level, do you mean within the BMS hardware? I'm not sure what "channel level" means in this context.

I mean the individual sets of hardware in the BMS that are used to measure individual seriesed cells (or groups). If it's a 14s BMS, there are at least 14 such sets of hardware.

It's likely that the actual measurement is done at the main MCU, and the measuring device / pin there is simply switched between each individual cell/group in the series of cells by a separate piece of hardware (or set of such), and that the zero readings indicate channels of that hardware that have failed.

Without a reasonably complete schematic of the BMS (never available) I couldn't point to a specific circuit, but if you find analog multiplexers (either chips or discrete components) on it these are probably involved.

If it's a dedicated battery-monitoring chip rather than a generic MCU, then it probably has individual inputs on the chip itself for these, and that chip itself would most likely be the failure point.

 

[brownj24]

Ok, thank you. I hope that's the case, honestly, so that I can swap out the BMS and still make use of all those cells!

Thanks! I'll provide an update when I have one, in case it's helpful to others.

 

[amberwolf]

Remember you have (probably) two separate problems: Cell groups that have either interconnect or cell problems (or both!) causing lower capacity, and a BMS issue of some sort causing whatever incorrect operation is happening within it.

 

[brownj24]

Remember you have (probably) two separate problems: Cell groups that have either interconnect or cell problems (or both!) causing lower capacity, and a BMS issue of some sort causing whatever incorrect operation is happening within it.

Ok, thanks for clarifying.