Help me understand my lithium(LFP, LiFE) HOT BBQ FAILURE

[hour]

As title states, LIFEPO4 and thankfully things just got really REALLY hot, and one pack is fried - no damage to my pop up camper which I was using as a workbench for this process.

I have (had) two identical 40-cell 26650 lifepo4 battery packs, 10 cells paralleled, 4 series, lovely 12v portable cases. No issues for several months. I returned from camping two weeks ago with some pretty low packs, 12.5v each, and let them sit in my garage. There is SOME constant drain on the packs as they're each outfitted with tiny LED voltmeters and BMS units. This might tie in to the story.

Anyway, went to charge them both up several hours ago for a trip tomorrow and noticed one of the units voltmeters was off, the other reading 12.2. "Guess the BMS has shut one pack off" I told myself...

Charged up the one that was resting at 12.2v without issue, then moved on to charging up the one that appeared completely dead. The victron charge controller (with 24v power supply feeding it as 'solar') took off doing its thing, but my blunder was not throwing an RC cell monitor on it first.

Ran some errands and when I returned, noticed this pack's voltmeter had turned back on but was reading 10.9v then dropping to 9, cutting off, then rebounding to 10.5, 9, over and over. BMS clearly kicking off. Figured at that point I had a bad cell group. Smelled like hot science and kapton tape, so I disconnected the charge source and went on to investigating.

Threw the RC cell monitor on it at that point and it only recognized it as a 3S battery: 0.4v, 3.45v, 3.45v, [no fourth cell group]. So, I tore it apart and got my individual 10P modules separated, then hit with a proper multimeter.

One group is reading in the millivolts. I broke that apart and 9 of 10 cells are also reading in mv on the multimeter, but ONE single battery is 3.33v. How the hell?

I figured that was the group that had completely dropped off causing the RC meter to only detect a 3S battery... thus I expected to find a 0.4v group, and two 3.45v groups. Lo and behold, now that they're not connected to each other anymore in series, I ended up with THREE 3.33v 10P groups.. and of course the one completely bufu'd group of ten.

Question 1: how the hell is a single battery out of a group of 10P still reading acceptable voltage, and the other 9 fatal at under a volt?

Question 2: how the hell do I have three 10P blocks that are now reading acceptable in the 3v range when I expected to have that 0.4v one?

Nothing visibly shorted, no wires melted, and the BMS survived and probably averted major disaster. Long jumbled post, thanks for reading and any ideas. I think I'll go ahead and rebuild the pack with all new cells, but that leaves me 31 individual batteries that seem good. I might just call that 30 because I'd rather not use the ONE battery out of a 10P block that didn't go fatal, just because. But the entire assembly was extremely hot, meaning I may not wish to repurpose those 30 survivors either.

 

[john61ct]

Rebuild with new for sure.

Proper CC load discharge tests to see if any old ones worth repurposing, 75-80% SoH being go/no-go.

And only for non-critical, portable powerpak for screen gadgets maybe.

Or just toss the lot, if you don't see this as a fun hobby.

Lessons learned?

Ideally battery bare cells get fully 100% isolated from BMS, all gadgets with any drain at all.

Storage charge sufficient to guarantee never get close to 12V. Usually 30-60% is fine, but

Check on stored packs every couple/few weeks, top up as needed - but not above selected storage voltage - more frequently early on to **test** the previous rule, less frequently only once certain.

Store cool as possible.

 

[luthj]

Did the pack get pulled near the minimum voltage? If so the weakest cell group got pulled so low, that it reversed polarity. Upon recharging you were in essence reverse charging that cell group. This can cause electrolyte breakdown, heat, etc. The single cell that survived had a high resistance connection to its peers. This is why it survived the excessive discharge. However this may be the reason that that cell group itself was weak and got pulled low before the others.

 

[luthj]

With one cell group down 10%, the other cells would be at ~3.2-3.5v when the low group hit zero soc. The pack could have shown ~11v. It's also possible the low cell group got overcharged/over bolted, and lost electrolyte, this failing similarly.

 

[jonyjoe101]

I build my own lifepo4 also, as soon as I notice one battery voltage is too low I always check everything. I also monitor balancing when charging, I never trust the bms to stop the charging.

When I had something similar happen to me on a 3s50p pack li-ion pack, when stored two cell drained below 2 volts. I had 2x 3in1 lcd meter connected the whole time, I suspect that was what cause my problem. those 2 meters never worked again thats why I think they drain the battery. Now I disconnect everything including bms when storing.

A bms I recommend for your lifepo4 is the chargery bms8t, I been using it on my 220ah 4s40p lifepo4. It has its own lcd screen, temp sensors,audio alarms, works with any charge method especially solar. It uses mechanical contactors (relays) to stop charging. Mechanical relays don't bleed voltage like mosfets do. I'm a big believer in audio alarms thats why I like this bms.

I would never use a regular bms while charging with solar (I use them before to charge with pwm/mppt), because when it triggers to stop charging, the solar controller will create voltage surges as it tries to forcibly charge your battery, what if the mosfets blow, will current flow again. I be hesitant to reuse your bms after this incident. There was an incident I read about, where someone accidently used a 70 volt charger to try and charge a 50 volt pack, a bms should protect against that but the bms let the amps into the pack and it caught on fire. That why I don't trust the bms to work everytime. BMS should be the last line of defense.

This is a picture of my 2 year old 110ah lifepo4, notice the LEDs I use to monitor cell voltage, now I just use the chargery bms to check cell balancing.

 

[john61ct]

Just put a cheap lead battery in parallel to your LFP, will buffer all spike/surge/load dumps at contactor disconnect from the charge source.

Or put your cutoff contactors upstream of the immediate charge source, in this case between the panels and their controller.

 

[hour]

A bms I recommend for your lifepo4 is the chargery bms8t, I been using it on my 220ah 4s40p lifepo4. It has its own lcd screen, temp sensors,audio alarms, works with any charge method especially solar. It uses mechanical contactors (relays) to stop charging. Mechanical relays don't bleed voltage like mosfets do. I'm a big believer in audio alarms thats why I like this bms.

Thanks, I'll look in to this unit. Before breathing a sigh of relief after building both packs, I charged and discharged them more times than I care to admit, to make sure that my solar settings on the Victron were perfect (staying in a very comfortable range). This was total user error, I should have connected the balance charger I own when I noticed the voltmeter off, or at least the RC cell monitor. The surviving pack is fine and performed admirably this weekend. It's comprised of new cells, where as the others were advertised as new (and same cell) but had wildly different characteristics and arrived as crap quality re-wraps. Made me wonder...

john, luthj, thank you so much for the input. I relied on both of you (and jonyjoe101) immensely when building these things, and the house battery in my truck. This was a great learning experience and I have 40 new (NEW... NEW) cells arriving tomorrow to rebuild this box. Takeaways: 1) use the balance charger at home, I have it - why waste it? 2) quick disconnect of everything that drains power when not in use - the voltmeter doesn't do any good when the packs are just loitering in my garage waiting for the next trip and disconnected at storage voltages is king, 3) when in doubt AT LEAST connect the RC cell monitor first to get an idea of what's going on.

While I've charged/discharged over 30kwh on my DIY lifepo4 house battery for my fridge in truck this season, since I built it late Spring (fine tuned - SUPER OCD - BMS-less) I may just revamp that at the end of the season after this experience. It's extremely well balanced with the same 6(10?)amp balance board that jonyjoe has used, but they're used cells, and my $50k truck doesn't need to go boom. Battleborne may be in the future there, just for the peace of mind.

 

[john61ct]

A decent ammeter and known-accurate voltmeter / DMM are the most basic and essential tools.

 

[luthj]

Does your BMS disconnect if one cell group goes wonky? Just curious.

The cells that were not zeroed out during the failure, are likely just fine. It sounds like you just had a BMS or other failure. Given that this is about an 8 out of 10 on the Lithium failure scale, the LFP cells demonstrated there tolerance for abuse. No fire, no melting, etc.

If you build another pack, I would suggest putting a good load on it. Then checking each cell connection (both pos/neg) for voltage drop. If any vary more than 10% or so, you know there is an issue that needs checked.

 

[deeznutz206]

Were these the K2's? I sort of recall your thread about them.

I built a 4S12P and 4S8P using the K2's. I am not using a BMS on either of the packs because I balance charge them every time.

 

[hour]

Does your BMS disconnect if one cell group goes wonky? Just curious.

The cells that were not zeroed out during the failure, are likely just fine. It sounds like you just had a BMS or other failure. Given that this is about an 8 out of 10 on the Lithium failure scale, the LFP cells demonstrated there tolerance for abuse. No fire, no melting, etc.

If you build another pack, I would suggest putting a good load on it. Then checking each cell connection (both pos/neg) for voltage drop. If any vary more than 10% or so, you know there is an issue that needs checked.

Just finished rebuilding the pack tonight using new Valence 2500mah 26650 cells. The BMS may have disconnected originally (hence built in voltmeter off) when one cell group plummeted below the minimum. Why that group never recovered, and ate itself, is beyond me. The voltmeter did kick back on when I began charging it so the BMS must have thought things were somewhat okay - but they were clearly not. I didn't check if the voltmeter immediately cut off when I removed charge, as I was plugging in the RC cell monitor at that point and then quickly dismantling the pack.

The BMS was tested on initial build for over and undervoltage. I'm using it currently on the new build and tested that the overvoltage part still works, at least. And the active balancing. I'll do a controlled discharge test tomorrow and make sure it cuts off on the low end still, too. In reality I never use the packs to the point of low voltage cutoff, and I plan on just recharging when I get home and unplugging anything that will cause drain. If I don't intend on going back out for a bit I'll charge to storage voltage and then unplug everything.

Were these the K2's? I sort of recall your thread about them.

I built a 4S12P and 4S8P using the K2's. I am not using a BMS on either of the packs because I balance charge them every time.

They were advertised as K2s, and the same model I had purchased a set of 40 (new) for the other pack when I initially built both. I would have gotten 80 of the K2s new from the good source, but I guess I got the last batch. That lead me to ebay where I was sucked in to a listing for the same 10.5wh K2s advertised as new, but received crap quality rewraps that didn't behave like the new K2s at all. Ahh well, live and learn.

I just charged up the new pack (and living K2 pack) @ 185w, and they both did well using a Q6 Pro hobby charger with balance leads connected. I couldn't begin to tell you why I put a Victron MPPT in a mini pelican knockoff case for charging these portables. I didn't have any intention of solar charging them ?‍♂️.

Ordered some aviation plugs - 5 pin panel mounts with opposing ends so I can have a way to balance charge without popping the cases open and using the JST connectors. I'll throw the Q6 Pro in a toolbox and mount a 120v->24v power supply on the bottom. Run that setup off the generator if needed, or at home, and also be able to switch inputs to the Q6 to feed it from my house battery (9-32v input I believe). And always have the balance leads connected. Never again!

 

[john61ct]

So assuming you can reliably get K2s from a trustworthy supplier

authentic not rewrap
official front-door factory output, graded A quality
fresh new mfg date

are K2 considered good quality, for both high performance extended longevity?

comparable to A123?

how does Headway fit in?

other LFP makers in small (under 10Ah) cylindricals?

 

[deeznutz206]

Just finished rebuilding the pack tonight using new Valence 2500mah 26650 cells. The BMS may have disconnected originally (hence built in voltmeter off) when one cell group plummeted below the minimum. Why that group never recovered, and ate itself, is beyond me. The voltmeter did kick back on when I began charging it so the BMS must have thought things were somewhat okay - but they were clearly not. I didn't check if the voltmeter immediately cut off when I removed charge, as I was plugging in the RC cell monitor at that point and then quickly dismantling the pack.

The BMS was tested on initial build for over and undervoltage. I'm using it currently on the new build and tested that the overvoltage part still works, at least. And the active balancing. I'll do a controlled discharge test tomorrow and make sure it cuts off on the low end still, too. In reality I never use the packs to the point of low voltage cutoff, and I plan on just recharging when I get home and unplugging anything that will cause drain. If I don't intend on going back out for a bit I'll charge to storage voltage and then unplug everything.



They were advertised as K2s, and the same model I had purchased a set of 40 (new) for the other pack when I initially built both. I would have gotten 80 of the K2s new from the good source, but I guess I got the last batch. That lead me to ebay where I was sucked in to a listing for the same 10.5wh K2s advertised as new, but received crap quality rewraps that didn't behave like the new K2s at all. Ahh well, live and learn.

I just charged up the new pack (and living K2 pack) @ 185w, and they both did well using a Q6 Pro hobby charger with balance leads connected. I couldn't begin to tell you why I put a Victron MPPT in a mini pelican knockoff case for charging these portables. I didn't have any intention of solar charging them ?‍♂️.

Ordered some aviation plugs - 5 pin panel mounts with opposing ends so I can have a way to balance charge without popping the cases open and using the JST connectors. I'll throw the Q6 Pro in a toolbox and mount a 120v->24v power supply on the bottom. Run that setup off the generator if needed, or at home, and also be able to switch inputs to the Q6 to feed it from my house battery (9-32v input I believe). And always have the balance leads connected. Never again!

Ah you got the Valence's. I was thinking about getting the Headway 38120's when they had them in stock but I already have 3 packs and can't justify another one..........yet.

 

[john61ct]

Since Lithium Werks B.V. took A123 Systems over as well as Valence and Super B, I think it's safe to say the production quality of all those brands is good, right?

Problem is when looking for better value pricing, why I'm asking about K2.

And continuity reliability of supply over time is also critical, want to know the same cells will be available, ideally through the same supply chain as I need to build more packs, or replace bad cells at maintenance time down the road.