40T Cells Voltage Drops While Not In Use

[rg12]

20S 7P pack made of brand new grade A Samsung 40T cells, now 6 month old, max current that was drawn from it was 140A peaks, got to 55C ANT BMS thermal cut off only once in it's life.
Then it sat on 3.9V per cell for a month with no use, suddenly cell row number 9 is at 3.0V while the rest are at about the same place they were when last used (around 3.9V) all perfectly balanced aside from row 9 which went down to 3.0V

So I charged row 9 to level with the rest and disconnected the bms to see if row 9 drops without the bms and it did.
It dropped 0.03V within 48 hours while the rest were perfectly balanced and dropped only 0.002V in those 48 hours.

Then I connected the pack to a load tested and pulled 80A from it for a few seconds and the max cell difference instantly went from 0.04V to 0.01V and stayed there after the load was removed.

What do you think can be the cause of that?

 

[Matador]

I think one of the seven cells in row 9 went bad and must have a higher DC internal resistance than all others. That would explain the voltage leak over 48h... But to be sure it would check voltages after a full week, as 0.03V could be almost normal drop ( from resting) if cells were just charged to 4.20V right before the 48h voltage drift test. But if you did not charge the cells before testing drift and they dropped 0.03V from storage voltage just by resting 48h... Not normal at all, the one faulty cell alone in that row of seven would drop 0.21V or so if it was singled out, not connect to anything.

I would dismantle row 9, test each cells for internal resistance using a SKYRC Imax B6AC v2.0 hobby charger ( or any other device that has an IR mesuring function ). You might find one of the seven cell might have an unusually high DCIR, and thats the bad cell. For example, 14, 12, 15, 12, 224, 13 and 14 milliohms. Then the cell with 224 millioms DCIR would be the bad one that's leaking voltage.

You already pretty much ruled out the BMS as a cause.

Matador

 

[eMark]

What about this ... good idea or not ?

Starting at one end of row 9 cut one bus bar at a time in the middle between the adjacent cell and check that cells voltage. Then do the same to each cell moving down the P-row. Then after removing and replacing the bad cell(s) you can use copper wire to resolder the bus bars except for the new replaced cell(s) in row 9. Would he be better to cut the bus bars on the negative or positive side or makes no difference?

Should he first unsolder all of the BMS jumper wires ?

 

[rg12]

I think one of the seven cells in row 9 went bad and must have a higher DC internal resistance than all others. That would explain the voltage leak over 48h... But to be sure it would check voltages after a full week, as 0.03V could be almost normal drop ( from resting) if cells were just charged to 4.20V right before the 48h voltage drift test. But if you did not charge the cells before testing drift and they dropped 0.03V from storage voltage just by resting 48h... Not normal at all, the one faulty cell alone in that row of seven would drop 0.21V or so if it was singled out, not connect to anything.

I would dismantle row 9, test each cells for internal resistance using a SKYRC Imax B6AC v2.0 hobby charger ( or any other device that has an IR mesuring function ). You might find one of the seven cell might have an unusually high DCIR, and thats the bad cell. For example, 14, 12, 15, 12, 224, 13 and 14 milliohms. Then the cell with 224 millioms DCIR would be the bad one that's leaking voltage.

You already pretty much ruled out the BMS as a cause.

Matador

The pack was sitting for a month on 3.9V per cell and row 9 dropped to 3.0V so I tested again from about 3.9V without the BMS and then the 0.03V drop happened so I'm pretty sure it's not a normal drift and 100% sure it's not the BMS.

I asked here to maybe avoid cutting out row 9 and testing each cell for IR but I guess that's what I have to do.
What worries me is why did it happen? 6 month old 40T pack that ran for 5C peaks and only once in it's life got to the ANT BMS thermal cut off of 60c.
This shouldn't happen ever imo.

What's weird is that the pack under 80A load is balanced and row 9 performs exactly the same as the rest, so maybe it isn't a high IR in one of the cells and just a semi internal short that depletes one cell (or more, who knows...).

 

[docware]

So I charged row 9 to level with the rest and disconnected the bms to see if row 9 drops without the bms and it did.
It dropped 0.03V within 48 hours while the rest were perfectly balanced and dropped only 0.002V in those 48 hours.

It´s common that cells have little bit higher voltage drop after charging, depending on charging parameters.
Lets sit the pack few days without BMS and measure voltage of all groups. BMS is still the first suspicious.

However, your effort to kill the cells by high load inevitably will bring undesirable consequences.
You still didn´t comprehend that you need different cells, automotive grade, high load cells.

 

[rg12]

So I charged row 9 to level with the rest and disconnected the bms to see if row 9 drops without the bms and it did.
It dropped 0.03V within 48 hours while the rest were perfectly balanced and dropped only 0.002V in those 48 hours.

It´s common that cells have little bit higher voltage drop after charging, depending on charging parameters.
Lets sit the pack few days without BMS and measure voltage of all groups. BMS is still the first suspicious.

However, your effort to kill the cells by high load inevitably will bring undesirable consequences.
You still didn´t comprehend that you need different cells, automotive grade, high load cells.

I'm guessing you didn't read the thread before replying...

and 5C for short bursts out of 8C high quality cells is not a common cell killer.
Most will pull the rated C rate and that's a different story.

 

[Matador]

Hard to say exactly why one cell would go bad (whether from high DCIR or partial internal short (which would be off the chart high DCIR).

Sometimes it's possible you get one cell with minor defect ina batch. And if you pull high loads on occasion, that defect may become more manifest.

I happened to me before, albeight with lower quality chinese generic 18650 cell (which I admit might br of questionnable quality), but all other cell except the failed one were all at same DCIR +/-10%. But on cell was way off the chart at 1500 mOhms (shorted)... Although that pack was a 7S1P made of 3C (6A) 2000 mAh from a cheap eskateboard.

So I would not necessarly expect that to happen on quality 40T cells, but who knows. I'm thinking high loads (although burst) did not help, but that cell probably was bon with a defect. Other possibiliies ? One bad spotweld ???? Would not explain voltage downward drift.

We sure can emit hypothesis, but the best way to solve this IMHO is to take the scientific approach and test each cells in that row. If you dont have a device to measure DCIR, you can measure voltage drop and load amps (simultaneously with two seperate multimeters). Calculate DCIR with ohms law with for two different loads (eg 5A and 20A) and eatimate DCIR. DCIR for each cell will be a rough estimate but the bad cell will still stand way out of line in terms of DCIR compared to other 6 cells

Matador

 

[rg12]

Hard to say exactly why one cell would go bad (whether from high DCIR or partial internal short (which would be off the chart high DCIR).

Sometimes it's possible you get one cell with minor defect ina batch. And if you pull high loads on occasion, that defect may become more manifest.

I happened to me before, albeight with lower quality chinese generic 18650 cell (which I admit might br of questionnable quality), but all other cell except the failed one were all at same DCIR +/-10%. But on cell was way off the chart at 1500 mOhms (shorted)... Although that pack was a 7S1P made of 3C (6A) 2000 mAh from a cheap eskateboard.

So I would not necessarly expect that to happen on quality 40T cells, but who knows. I'm thinking high loads (although burst) did not help, but that cell probably was bon with a defect. Other possibiliies ? One bad spotweld ???? Would not explain voltage downward drift.

We sure can emit hypothesis, but the best way to solve this IMHO is to take the scientific approach and test each cells in that row. If you dont have a device to measure DCIR, you can measure voltage drop and load amps (simultaneously with two seperate multimeters). Calculate DCIR with ohms law with for two different loads (eg 5A and 20A) and eatimate DCIR. DCIR for each cell will be a rough estimate but the bad cell will still stand way out of line in terms of DCIR compared to other 6 cells

Matador

I guess I will take that row apart and do some testing.
I have all the equipment for that so will check for IR and if nothing is obvious after that will go and charge them each and let them sit and see which one drops.