Why Do Lithium Packs Fail?

George S.

Well-Known Member
Why Do Packs Fail?

I bought 4 9ah LiFePo4 packs, rated at 12 volts. These packs were old and used. When I tested the 4 packs, they ranged from 6.4 amp hours to 7.7 amp hours. With different capacities, and 4 in series to make a 48 volt pack, you have to charge and discharge for the weak module. If you try to run the 6.4 amp hour module to 7 amps, the voltage will explode over the max, because it is fully charge. In the discharge phase, pushing the low capacity cell will drive the voltage down and ultimately kill the module. I tried to manage the charging but it was impossible. The weak module seemed to be really messed up. So I took it apart.

The theory I had was that the cells in these 4 modules were all ‘good’ but tired, so instead of 9 ah they had 6 or 7 amp hours. This is totally normal. When I took the modules apart, I found something completely different. There were 24 cells, and about 6 were essentially dead, very low voltage and no charge capability. And the other 18 cells weren’t really weak. They had capacities close to the nominal 1500 rating, maybe 1400 mah. I was surprised the module even worked as an ebike pack. It was impossible to keep it from overcharging and overdischarging, with 3 other modules in series. Plus this pack was not holding voltage well, at all, under load.

In this case the cause of the pack failure was revealed by tearing the pack apart and running capacity tests, charge and discharge, on 24 cells. That is a bit of work. In the end, what can be done? Getting 24 cells out of the welded configuration means tearing the nickel off, cell by cells, top and bottom. There’s no way to check the viability of a cell in the welded configuration. You have 6 cells in parallel, so that is the voltage you will get, that’s the smallest unit. Even if you could find a bad cell, or set of bad cells, then what? You can’t remove a cell from the matrix of nickel strips. Many people have yearned after a system where the cells could be removed, but it’s not really a tech that is out there.

If a pack stops working, do you toss it, recycle it? Well, it may be easy to remove all the (welded) cells from a plastic case, and see if there is any obvious problem. There are usually balance wires which isolate the parallel sets, and these wires go to the BMS. The BMS may have failed. You can trace power to the BMS, from the positive and negative sides, measure the voltage, see if the circuit will supply amps. You can check the balance, on each balance lead. You can check balance leads with the battery under load, as you reach discharge levels.

Batteries fail. Circuits fail, like a break in the nickel strips or internal wiring. BMS units fail or become erratic. Fuses blow. Cells short out or explode, but it’s not that common. I don’t know why a cell stops working and I’m not sure what the hazard is. These are relatively non-volatile LFP cells.

The final forensic procedure is the least attractive. Basically you take the pack apart to get at the individual cells, and test each cell looking for a pattern of dead cells or very low capacity cells. I’m not making an exhaustive list. I’m giving you a DIY’er perspective on the real world, but nothing more than a narrow slice. I have the time to just goof around with this stuff, though I tend to work outside, wear gloves, all that stuff.

There are ways to deal with this problem, that packs aren’t easy to analyze, or autopsy. There’s no real fix for many problems. Maybe there are two extreme ways to deal with the problem, with other strategies in between the extremes. Strategy #1) make batteries that are good enough, but super cheap. Or, (2): Go super tough, like bulletproof, but at a gold plated price. I can point to both, with examples, but there are so many other issues. If you buy an ebike with a battery, you don’t actually know which approach your seller uses. Strategy #1 means: If the cheap pack fails, it’s easier to just replace it. The bulletproof battery isn’t supposed to fail, so the seller should offer a multi year warranty. If the $100 Liito Kala pack is here to stay, totally automated mass production, it’s pretty attractive. I’m still using their hundred dollar special, four months, many cycles. If the pack failed in the next year, I would still take it apart, separate out all the cells, just to see if they were usable, or to see what might have gone wrong.

I have a Luna pack from their early years. I took it out of the case last week and threw out the BMS, which kept shutting down. Looking at the raw welding I’m not sure I should be totally pleased. See first picture. Second pic is the LFP pack partly disassembled.
 

Attachments

  • 20210831_135546.jpg
    20210831_135546.jpg
    646.6 KB · Views: 339
  • 20210824_184702.jpg
    20210824_184702.jpg
    358.4 KB · Views: 331
Did the LiFePo4 packs you bought each have a BMS?
Also if each of the 4 packs had 6 cells in series, the original nominal voltage for the pack would be more than 12V, so something there does not seem right. I think you ended up buying junk. Buying used cells is a risky business.
 
Did the LiFePo4 packs you bought each have a BMS?
Also if each of the 4 packs had 6 cells in series, the original nominal voltage for the pack would be more than 12V, so something there does not seem right. I think you ended up buying junk. Buying used cells is a risky business.
The 6 were welded across, top and bottom, so 6 in parallel, then four in series. So 12 volts. Nope, no BMS. Not something I wanted, but it might have made a difference. They might have been junk. The cells are DLG energy, which is a real Chinese company. I probably will switch to new cells for my ebike packs. The cost difference is not that great. The Chinese are running the recyclers to ground. The two 'salvage' companies I follow are selling a fair amount of new cells. So I guess you must be right about the risks. You have to remember that 5 years ago a lot the DIY scene was harvesting very sketchy laptop packs and soldering them together. The gap between used and new pricing was huge.
 
I appreciate that you and another member posted unfavorable "reviews" for purchases from another member that prominently posts as a self proclaimed battery expert giving entirely unvetted and dogmatic opinions. I'd take both of those reviews as 1 star Amazon equivalents and proof that spending more doesn't guarantee higher quality or better performance. Common experience is that higher price sellers don't always provide better after sale customer service or warranty either. I'm not ever going to try to build my own battery so my solution has been to buy low cost batteries with good reviews from Amazon. If there is an early failure I would take the issue to the seller first (1 year claimed warranty) and if no satisfaction then to Amazon. In my limited experience Amazon has stepped up and did the right thing when a seller wouldn't (on one or two occasions, not ebike battery purchases). So far my low cost batteries have performed very well, on par with a Yamaha battery that would cost $1000 to replace, $1250 if a charger where included. One of my batteries is going on 3 years old without any noticeable degradation in performance. I could buy 6 batteries for the cost of one of the yamaha batteries. Unfortunately I can't get an inexpensive Yamaha battery due to data port requirements.
 
I appreciate that you and another member posted unfavorable "reviews" for purchases from another member that prominently posts as a self proclaimed battery expert giving entirely unvetted and dogmatic opinions. I'd take both of those reviews as 1 star Amazon equivalents and proof that spending more doesn't guarantee higher quality or better performance. Common experience is that higher price sellers don't always provide better after sale customer service or warranty either. I'm not ever going to try to build my own battery so my solution has been to buy low cost batteries with good reviews from Amazon. If there is an early failure I would take the issue to the seller first (1 year claimed warranty) and if no satisfaction then to Amazon. In my limited experience Amazon has stepped up and did the right thing when a seller wouldn't (on one or two occasions, not ebike battery purchases). So far my low cost batteries have performed very well, on par with a Yamaha battery that would cost $1000 to replace, $1250 if a charger where included. One of my batteries is going on 3 years old without any noticeable degradation in performance. I could buy 6 batteries for the cost of one of the yamaha batteries. Unfortunately I can't get an inexpensive Yamaha battery due to data port requirements.
I also highly appreciate your honest although unfavorable reviews, its hard to hide from the truth.

I have a different philosophy than EMGX and try to buy what I consider the best available batteries for ebikes which for the DIY market are EM3EV and Luna Wolf packs both of which have individual fused cells. Im probably paying more than I should for this privilege but thats my choice.

We are each allowed to pick the solution thats works best for us.
 
Last edited:
The 6 were welded across, top and bottom, so 6 in parallel, then four in series. So 12 volts. Nope, no BMS. Not something I wanted, but it might have made a difference. They might have been junk. The cells are DLG energy, which is a real Chinese company. I probably will switch to new cells for my ebike packs. The cost difference is not that great. The Chinese are running the recyclers to ground. The two 'salvage' companies I follow are selling a fair amount of new cells. So I guess you must be right about the risks. You have to remember that 5 years ago a lot the DIY scene was harvesting very sketchy laptop packs and soldering them together. The gap between used and new pricing was huge.
OK that make sense, 6P4S for each 12V pack. They may have been used in an application without a BMS and run very hard so that would explain why some got out of balance want were killed, but I wonder if there were connections for the BMS to be connected externally, as it's strange that some cells pithing the same 12V pack would survive, something externally must have been shutting them down when the voltage dropped too low. SO many horror stories of people buying used battery packs hoping that the remaining capacity is still good enough. Generally that is not a good idea for bikes since the weight penalty is huge, it's usually better to use a chemistry with a higher energy capacity. Leave the LiFePo4 for stationary storage.
 
I have seen three of my batteries fail. When I looked inside, all were caused by a bank of cells failing to hold charge, with the remaining banks checking out OK. The first time, two rows had drained to zero, and I suspect it was the BMS "upgrade" I had installed two years earlier. Usually, a BMS only draws power from either one bank, or equally from all of them. Seeing two non-sequential banks fail suggests a circuit issue.

The other two batteries got unbalanced. I'll assume the reader knows what unbalance means. Neither of these batteries had a BMS with balance capability, or they might have kept going. I rebalanced one of them manually and put it on light duty on my wife;s bike, and it seems to be stable. I had originally run it on my BBS02 mid drive, even though it wasn't a big enough battery as far as current. so I think that stressed out one cell,

The second one also was rebalanced manually and useable, but I noted the unbalanced cell group had drained down to 2V while sitting around, making it potentially unsafe, as that's below the safe operation of an 18650 lithium cell. I'm not using that battery anymore. It's been replaced.
 
Last edited:
OK that make sense, 6P4S for each 12V pack. They may have been used in an application without a BMS and run very hard so that would explain why some got out of balance want were killed, but I wonder if there were connections for the BMS to be connected externally, as it's strange that some cells pithing the same 12V pack would survive, something externally must have been shutting them down when the voltage dropped too low. SO many horror stories of people buying used battery packs hoping that the remaining capacity is still good enough. Generally that is not a good idea for bikes since the weight penalty is huge, it's usually better to use a chemistry with a higher energy capacity. Leave the LiFePo4 for stationary storage.
Actually, yes, they had balance leads in a connector like a hobby LIPO pack, but it was a totally different connector. When I took the pack apart I did not like the construction. There were thin strips of nickel and very small wires or minimal nickel strips to connect the 6 cell parallel modules. The nickel is hard to remove because it shreds, so I am trying to leave the nickel in place and split the packs with a dremel. To some extent I classify this as 'fun' but there are better ways to go. There's a lot of nuanced design people might try with ebike packs. I think the only way to recycle cells would be to automate the whole thing, strip the cells, test the cells, make some kind of useable pack or module that would be really cheap. But there probably isn't enough cost advantage. Jehu (Jag35) sells Pinyou cells for $1.30. I think a thousand BAK (known brand) cells can run around a buck a cell from China. A lot of the 'good story' used stuff is around there, and who knows how well it will strip out of wherever it is installed. If you figure 52 cells in a basic 13s4p 48v pack, it's around $70 in parts.
 
Ideally, our batteries should just age gracefully, getting older and delivering less AH. I've got batteries from 2015 that have behaved like this, but they still have enough capacity for my daily usage, when I use them.

There's a really good discussion where someone has done the work, using a pretty rigorous test. This cell dropped from its original 2600 maH rating to 1800 mah after 200 heavy discharge cycles at 20A,

https://endless-sphere.com/forums/viewtopic.php?f=14&t=103092&start=525

wearout.jpg


I would say that on my unbalanced packs, running the first one at 150% of its rated current caused one or two cells to lose enough capacity to pull down its parallel neighbors. By reducing its load, as I did, it's able to keep up.

On the second one, who knows. Probably not the best cells, certainly not a Panasonic 25R, and with time one lower quality cell started pulling the others down.
 
Last edited:
Ideally, our batteries should just age gracefully, getting older and delivering less AH. I've got batteries from 2015 that have behaved like this, but they still have enough capacity for my daily usage, when I use them.

There's a really good discussion where someone has done the work, using a pretty rigorous test. This cell dropped from its original 2600 maH rating to 1800 mah after 200 heavy discharge cycles at 20A,

https://endless-sphere.com/forums/viewtopic.php?f=14&t=103092&start=525

View attachment 98879

I would say that on my unbalanced packs, running the first one at 150% of its rated current caused one or two cells to lose enough capacity to pull down its parallel neighbors. By reducing its load, as I did, it's able to keep up.

On the second one, who knows. Probably not the best cells, certainly not a Panasonic 25R, and with time one lower quality cell started pulling the others down.
There is a lot of hard truth to be distilled from what you say. The Sanyo GA chart shows a very significant drop in capacity even over 100 cycles. (Google Sanyo GA datasheet.) This is a little confusing because you pay the big bucks for the GA and it is considered a premium cell. But after 100 cycles you are in Chinese basic cell territory. You really don't want to push the pack even to the limits of the cell. Not when cells are cheap and you can easily add a row or two, go from 13s4p to 13s6p. That saves you. Even a 3c cell with 6 sets can give you 15 amps, vaguely the legal limit of a 48v ebike pack. I was bouncing stuff off a cell vendor, via email, about how a 14s4p pack with the Tesla cells, the 21700's, would be 40 amps, 900 wh. Jehu rates the Lishen 21700 he sells at 13.5 amps max. So 50 amps with 4 sets. With 56 cells that is $210 for the pack, DIY. My guess is that a 21700 pack with these cells could be charged to 90% and last a long time. And then you have the question whether one can buy an assembled pack from a US warehouse/Chinese vendor for a few dollars more. There is a need to look at the nuanced view of capacity, and how to get value from any pack.

sanyo ga.png
 
Back