Destructive battery pack test suggestions

[mike_bike_kite]

I watched a short YT video on active balancers but each $20 unit seemed to work with 4 cells. I'm not sure how that would play out across a more complex 4p24s arrangement. Would it require a lot more active balancers which would add to the cost and a lot more complex wiring to install which would add to the failure rate. It also doesn't explain why other manufacturers wheels don't burst into flame or why the wheels burst into flame before even being used.

Maybe @Seba could add a battery monitoring function to his app that see's how hard a wheel is being used, whether it's run down to extremely low voltages (if that matters?) and how often it's fully charged and the batteries balanced (if that's possible to detect). I guess if the wheel is continually being charged to 98V but never 100V then that might be worth raising an alarm. 

[goatman]

the cheapy 1amp  balancers are more of a novelty gadget that i use for balancing the pack down around 2.7v. on my 17s4p packs  better off with the cheap 5amp balancers

cheap gadgets for hobbyists

unless your bms is able to be programmed to balance at a different voltage other than 4.2v i wont use them. an Ant bms is programmable and has bluetooth but $$$ and size

bluetooth programmable 20s+ active cell balancers  are big and pricey too like the Ant

not really made for euc

 

i dont know who seba is, these bms boards should be able to change the balance voltage to 4.1v or 4.15v and turn on or off the balancing through bluetooth for cell monitoring from the factory. so its simple and easily understood for people new to batteries

 

if you already have a weak p-group and then try balancing it with a bms at 4.2v itll just get worse. thats why i dont like a BMS, you force energy into a bad group trying to balance it while the other groups are burning off the amps, an active cell balancer moves energy from one group to another to level off the pack, if you do the active balancing at the top or bottom where there is less mah/0.1v its more accurate

[supercurio]

1 hour ago, mike_bike_kite said:

Maybe @Seba could add a battery monitoring function to his app that see's how hard a wheel is being used, whether it's run down to extremely low voltages (if that matters?) and how often it's fully charged and the batteries balanced (if that's possible to detect). I guess if the wheel is continually being charged to 98V but never 100V then that might be worth raising an alarm. 

Great idea, difficult to implement in practice due to volt-meter inaccuracies combined with the fact the user can stop charge before the end manually.
Might be worth investigating still.

Edited December 26, 2021 by supercurio

[RagingGrandpa]

Busy day... a board was sacrificed, as well as some "nickel" (plated) cell strips

But, I had trouble initiating thermal runaway.

My attempted method was severe overcharge, by applying a 12V 10A car battery charger to just one parallel group of cells in the 24s2p pack.
Starting cell temp was 50°F; starting voltage was 4.2V.
I only achieved about 180°F at the side of the cell can, and there was no immediate smoke or fire. The cell reached 12V, after which my power supply reduces current below 500mA, and temperature stopped increasing.
After removing the power supply, the cell fell quickly below 3.6V. (So where did the energy go??)
Then some time later (more than 2hr) during storage outdoors, the overcharged cells did pop their pressure relief caps and spit some hot gases and small flames (as evident from charring of the BMS PCB), but there was no "red-hot canister total meltdown" (because the cell shrinkwrap was not burned), and neighboring cells did not vent or burn. 

I really expected that 10A @ 5V would trigger runaway immediately. Perhaps if I started from a higher cell temperature, the 130°F temperature rise might get me there?

In earlier tests today, I sustained full-pack bolted short tests, involving discharge >150A for >1sec (yikes), so perhaps the cell CID's were already weakened before this overcharge?

Thoughts appreciated

Edited January 1, 2022 by RagingGrandpa
12V cells did eventually burn...

[supercurio]

Rather impressive performance from this cell group, able to resist this extraordinary stress!

So was that the over-charge protection as described in the spec doing its job, by dissipating all the excess energy in heat?
My main hypothesis for the fire has been over-charging of the healthy groups in a faulty pack by the other packs connected in parallel: pushing the voltage above 4.2V while the dead group in series self-discharge.

If this theory is valid, then maybe the fire happens after the cells spent weeks being pushed above 4.2V, which might defeat the over-charge protection instead of a lot of voltage for a short amount of time like you had in this first experiment.

That's the only way I can explain the fire that happened to a local rider in Stockholm who got his house burned, from the parameters he described.

Great job with the testing 👍

Edited December 29, 2021 by supercurio

[goatman]

energy went to heat into 

you can short cells without fire

https://www.youtube.com/watch?v=Ys5wwUJOz58

 

https://www.youtube.com/watch?v=oBlac5gWLfI

theres resistance in the wires

but dead short them

 

https://www.youtube.com/watch?v=nL2TT3darhY

if youve seen cell level fusing tests

https://youtu.be/BAPHF3Sq2t8?t=1116

 

is why i ask how this bms lead is attached to the plated nickel

 

https://i.imgur.com/1qS5DcQ.jpg

 

if they are using 18650 nickel on 21700 cells and that bms lead touches the outer neg rim. it could easily start a fire

 

 

 

Edited December 29, 2021 by goatman
photo

[supercurio]

Thanks for the great links @goatman, huge difference depending on the resistance in the short.
I'm blown away by how well this 30Q survived still.

How much variation in cell manufacturing? It looks like most cells are very robust to anything but a dead short.

Is it possible that the fires could be created by cells with manufacturing defects that still made it into the supply chain?
It's already a reasonable theory to think that the pack @RagingGrandpa inherited had some defective cells, which lead to a cell group at 0V instead.

[goatman]

anything is possible

when i was trying to find 30q and hg2 cells locally, they were all counterfeit and they all came from big chain vape stores

easiest way to find out is to do a capacity test and compare the discharge graph with the data sheet graphs

not sure if the lygte battery comparator has the m50 but it might be somewhere on the site

https://lygte-info.dk/review/batteries2012/Common18650comparator.php

i salvage alot of packs and typically its bad spotwelds that throw packs out of balance and 0 volt p-groups will show no sign of heat or burning, its like 1 cell will go bad from being overcharged or undercharged and start self discharging, taking the rest of the group cells to 0 volts

it just looks like a weakly designed pack that would flex and vibrate easily

if that little nickelstrip bms lead was spot welded on top instead of under the the nickelstrip it wouldnt be able to slip underneath the insulator ring of the cell wrapper but getting sandwiched between the cell holder and the battery, it wouldnt take long to cut through and dead short from vibrations or flexing. 

[RagingGrandpa]

3 hours ago, goatman said:

https://www.youtube.com/watch?v=oBlac5gWLfI

theres resistance in the wires

"
25C discharge on a 10s x 1P pack made of Samsung 25R cells by shorting the outputs through a long length of 12g wire. Enough current to cause the cells to vent and go into thermal runaway, but not so much current as to vaporize the tabs before the experiment could start. "

^ Smart. I vaporized mine in an instant.

3 hours ago, goatman said:

is why i ask how this bms lead is attached to the plated nickel

if they are using 18650 nickel on 21700 cells and that bms lead touches the outer neg rim. it could easily start a fire

I think it's prevented by the plastic frames, which cover the rim of the cell top at each positive end.
And the balancing lead is part of the plated strip- it was manufactured that way, so no need for an additional joint.

3 hours ago, supercurio said:

the pack @RagingGrandpa inherited had some defective cells

Hmm, well aside from the one dead group, the other groups were perfectly balanced, and were able to output absolutely massive currents (>80A/cell !!)

1 hour ago, goatman said:

not sure if the lygte battery comparator has the m50

Of course
https://lygte-info.dk/review/batteries2012/LG 21700 M50T 5000mAh (Gray) UK.html

1 hour ago, goatman said:

looks like a weakly designed pack that would flex and vibrate easily

Get your hands on one some time... they feel very rigid when all wrapped up, and get mounted with compression foam on 5 sides.
There's a blank fiberglass panel separating the two cell rows, with adhesive on it that makes the two rows resist flexure of the PCB... and the packs don't get loaded in bending with their mounting method in the EUC shell.

1 hour ago, goatman said:

typically its bad spotwelds that throw packs out of balance

An interesting theory... as I showed, the welds at the bad group were very solidly attached, and there was still excellent continuity, both in the high-current path and also all the way up to the BMS balance resistor pads.

Some corrosion was visible. Did the corrosion come before the cell failure; or after (e.g. venting)... hard to say.

 

Test results posted below:
https://forum.electricunicycle.org/topic/26199-worst-case-fault-current-gotway-battery-destructive-test/

Edited December 29, 2021 by RagingGrandpa

[goatman]

im surprised that bms lead isnt spot welded, be alot of wasted metal to stamp it all in one piece. 

the mystery continues

[mike_bike_kite]

What happens if you deliberately replace a good cell with a bad cell? Do the other cells try to rebalance it and put all their current through the bad cell?

PS I have no idea what I'm talking about

[RagingGrandpa]

25 minutes ago, mike_bike_kite said:

What happens if you replace a good cell with a bad cell?

It depends on the condition of the "bad" cell.

• Installing a high-resistance cell means the parallel good cell would be continually cycling, discharging and recharging with it's out-of-sync parallel, likely leading to accelerated aging and loss of capacity.
   
• Installing a low-capacity cell (e.g. aged cell, or a smaller cell) would have the same out-of-sync rapid aging problem as above.
   
• Installing a totally open / inert cell (or just removing 1 cell entirely) would send the full pack current through its parallel good cell, overdischarging that good cell very quickly.
   
• Installing a 0V cell (without internal shorting) likely results in the parallel cell raising it back to a normal voltage (but without any substantial energy storage), and so the remaining parallel would be overdischarged quickly during use, the same as above.
   
• Installing a 0V cell with internal shorting (like my test pack had) would immediately discharge the parallel cell, dissipating the energy as heat. If this heat doesn't cause runaway, continued use of the pack would result in further heating, both during charging and during discharge. Risky.

Why did it come to mind?

[mike_bike_kite]

23 minutes ago, RagingGrandpa said:

Why did it come to mind?

I thought that if a cell couldn't charge to it's full capacity (4.2V?) then the parallel cell would keep pumping electricity into it to try and charge it up. This might then cause it to overheat and set the neighbouring cells off. It obviously goes without saying that I honestly have no idea about any of this stuff but I think it was one of the early thoughts about possible causes.

[goatman]

some people call them "heaters", if you use an infra red camera on a battery pack you can find them because the bad cell is warmer than the rest of the cells

Edited December 30, 2021 by goatman
changed a word

[RagingGrandpa]

Ok, I learned a little more respect for "dead cells" today...

I had a wrong assumption previously, that low-voltage cells are 'inert' because of very little stored energy. 
But in fact, I discovered that cells with less than 1V present (and even 0V) can still burn suddenly and violently if punctured.
Just because a cell is under 1V and disconnected, does not mean it is fire-safe!
Perhaps the good news about low-voltage cells is: they are unlikely to start fires from shorting, because almost no current will flow.
But they're still a hazard if physically damaged. (And probably also a hazard if current is forced through them.)

Another piece of good news is: the two original dead-and-internally-shorted cells from the pack did not react to piercing nor exposure to a torch flame. Something must have happened to them chemically during the many months of storage with an internal short, that made their guts inert. Glad to see a stable outcome is possible... just wish I knew how to cause it.

Later, I took one of the undamaged cells at 4.1V, and immersed it in a 5%/wt solution of sodium carbonate saltwater to discharge it. After 8 days of immersion, bubble formation slowed to a trickle. After rinsing and drying, the cell was resting at 2.2V, and could only manage 6A of short-circuit current (falling rapidly). Afterwards, the cell did not respond to piercing or smashing (good news).

 

I also tried to replicate the single-cell shorting fire referenced by @goatman above, but was unsuccessful. Even with the cell wrap removed, getting a good connection between the button and the rim was difficult. Using an undamaged cell at 4.1V, I managed to get a piece of copper wire to pass substantial current by pressing it against the rim with a tool. The M50T emitted smoke, and eventually burst its pressure relief cap, but did not emit flames nor proceed to violent runaway. (Perhaps it would be different if the cell were hotter before the test...)

Edited January 9, 2022 by RagingGrandpa

[enaon]

Quote

I also tried to replicate the single-cell shorting fire referenced by @goatman above, but was unsuccessful.

nice ending, like a scary movie  

maybe the short was released soon, also it may need some obstacle on the sides to not let it release pressure, to emulate the cell being among others. Nice tests.

Edited January 1, 2022 by enaon

[alcatraz]

Hey RagingGrandpa. I'm curious about the dead cell group. Are they all positive poles facing downward (with the wheel upright) and are they up against the outer tape/heatshrink?

Signs of corrosion?

[RagingGrandpa]

6 hours ago, alcatraz said:

the dead cell group. Are they all positive poles facing downward (with the wheel upright)

Sorry, I can't be sure: the EUC (Nikola) wasn't mine, and it depends on if the pack was installed in the left-side shell, or the right-side shell (a different edge is 'up' on each side).

There was no evidence of water intrusion, and the owner did not report instances of immersion or other flooding.

7 hours ago, alcatraz said:

are they up against the outer tape/heatshrink?

No cell-ends directly touch the shrink, because the frames serve as a spacer and cap. But yes, the dead group was facing 'outward' as shown above.

7 hours ago, alcatraz said:

Signs of corrosion?

Yes. The dead pair appear to have leaked some amount of corrosive (vapor?), which corroded the weld strip near only those 2 cells, and also created a small amount of powdery residue evident all around the pack.

[RagingGrandpa]

Round 2!

I have a fully-functional LiTech 900wh pack pulled from an RS due to suspicion of water exposure... 

I'm thinking of testing:

• Will be BMS perform an overcurrent trip separate from the melting fuse? At what value?
• Is the overcurrent detection symmetric? (Charging vs discharging?)
• Does the output shutoff when the temperature threshold is exceeded? 
• Does the output shutoff due to cell undervoltage? 

Other requests welcome  

@Chriull@Tawpieetc

Edited May 1, 2023 by RagingGrandpa

[sbb]

Charge the pack. Alligator on to one cell, discharge it down to 85%, then see how the pack handles the imbalance when you toss it on the charger?

[Chriull]

On 4/29/2023 at 12:18 AM, RagingGrandpa said:

I'm thinking of testing:

•  
• Does the output shutoff when the temperature threshold is exceeded? 

"5x temperature probes with buzzers to alert at a 65°C level & shut-off when the temps exceed 75°C"

from https://forum.electricunicycle.org/topic/24490-new-gotway-extreme-bull-commander-leaked-datasheet/?do=findComment&comment=386114

On 4/29/2023 at 12:18 AM, RagingGrandpa said:

• Does the output shutoff due to cell undervoltage? 

Hopefully not! This was with the first wheels when the used the common ebike bms - many faceplants happened due to undervoltage shuttoff. There was a famous topic from @hobby16 on how to shunt the protection mosfets...

On 4/29/2023 at 12:18 AM, RagingGrandpa said:

Other requests welcome  

The 50e have a max charge current of 4.9A (not for cycle life) specified. So 9.8A for 2p. I was always wondering if / how the protection mosfets on the bms stand such a current temperature wise. In a closed plastic wrap. In a more or less closed conpartement. For some hours...

Same for ~30A (~max non continous current for 2p 50e) discharge current. Maybe with some ir pics during this ~18min discharge. If there are some "weak spots" warming up?

Great would be some automatic charge discharge cycling with such numbers (10A charge/30-40A discharge) with single cell group voltage logging How the cells degrade, how much the bms can balance, ...

[RagingGrandpa]

On 4/29/2023 at 1:05 AM, sbb said:

discharge [one cell] then see how the pack handles the imbalance when you toss it on the charger

Surely it will disable recharging... but yes, I could discover what the cell-overvoltage threshold is. 

[0000]

Any update?

[RagingGrandpa]

Busy season's here... it's suffering confinement in my outdoor BBQ for the time being.

edit: it's done: https://forum.electricunicycle.org/topic/34852-litech-pack-teardown/
 

Edited December 11, 2023 by RagingGrandpa