How to build a pack with mixed cells?

Hi all,

this is my first forum post, probably not the last.
I have some 48V-plans for the future, but I will write something about that in a separate thread.

By a lucky coincident I got a bunch of cells today (really unplanned) with a total of 923.4Wh.
A large portion of that are 30x 21700er cells with around 18Wh each.
But unfortunately the 21700er are from 4 different suppliers,
mostly 3.6V 5Ah but also some with4Ah and 4.9Ah.

Can I put all the 21700er together into one pack?
Surely I will do some cell analysisto test the remaining capacity.
As I am aiming for a 48V configuration, a 14S or 13S configuration would be the way to go here.

Thanks for helping out a beginner.

Best,
nurF

nurF said:

Can I put all the 21700er together into one pack?
Surely I will do some cell analysisto test the remaining capacity.
As I am aiming for a 48V configuration, a 14S or 13S configuration would be the way to go here.

Click to expand...

I'm assuming these are Lithium-ion (3.0v to 4.2v) soI'd recommend 14s (14 in series) for a 48v configuration - its what most of us do as this operates nicely with most48v equipment - giving you the broadest set ofoptions for charge controllers,inverters, BMSs, and just 48v equipment in general.

Yes you can mix them. You want to test them before making a distribution. You might find that the4000mah only have 3600mah (e.g. 90% of original capacity) and the 3600mah have 3600ah (100% of original capacity). After test -I would distribute them evenly by capacity (in 100mah increments) and manufacturer as reasonable - its OK if things are not perfect. You want each of the 14s packs to have the same ah(s) as reasonably as you can - with the different brands distributed as is reasonable.

If you have 923wh of cells, then you should havearound 65 cells.5p cells per pack * 14s would be 70cells. If each pack averages 5cells in parallel * 3800mah each on average then you have 19,000mah (@3.7v) in each pack. If you could get the 14 packs so the largest delta is within 400mah or less (e.g. <= 2% of a pack's capacity) you should have a perfectly manageable 14s battery for a typical Solar charge/discharge profile. If the battery must deliver high current such as a motorized 'something' then it get's more critical to keep it tight.
My own packs are 130ah to 133ah - 2% difference - in a solar application and they stay balanced with no effort.

Of course you need to look into a BMS as well.

I wanna mix my cells too but not using the fixed number per pack


"Arrange cells in to packs so each have similar capacity but vary the number of cells in parallel and keep cells in a pack roughly the same capacity" don't work for me

for example, 2 series return 1 pack only.
all tests I tried goes wrong using this option

You should have same amount of cells in each pack. Make sure to sort Them evenly and by statistiska in terms of brands and AGE and that Will work out the best.

Having different amount of cells just makes it more complicated to get even current on each cell

Agreed with Daromer. Same number of cells per pack. This is for amp balancing. If 1 pack is 10 cells, and the other is 15 cells, then 10 cell pack with see a 50% increase in amps per cell.

20A load
20 / 10 = 2A/cell
15 / 10 = 1.5A/cell

This will not only stress the low cell count packs more, thereby reducing their overall life cycle, but it also makes it so that you need heavier gauge wire on that pack and you're risking popping fuses on that pack (if you use fuses, which is advisable)

LC-2nd said:

for example, 2 series return 1 pack only.
all tests I tried goes wrong using this option

Click to expand...

Please clarify this statement. This doesn't make any sense.

Korishan said:

Agreed with Daromer. Same number of cells per pack. This is for amp balancing. If 1 pack is 10 cells, and the other is 15 cells, then 10 cell pack with see a 50% increase in amps per cell.

20A load
20 / 10 = 2A/cell
15 / 10 = 1.5A/cell

This will not only stress the low cell count packs more, thereby reducing their overall life cycle, but it also makes it so that you need heavier gauge wire on that pack and you're risking popping fuses on that pack (if you use fuses, which is advisable)

LC-2nd said:

for example, 2 series return 1 pack only.
all tests I tried goes wrong using this option

Click to expand...

Please clarify this statement. This doesn't make any sense.

Click to expand...

I guess repacker have some bugs in the variable number cells option
I tryed many times and nothing goes right.
using cells of 20A for tests, more problems. repacker ignore em all (ok, no cell have 20A but I tried it also)
I wanna make small packs of used 18650. modular system. maybe each with 30A. if one goes bad it's easy to find the problem or replace it. used cells = many with different amperage and I wanna use em all. and I know the risk of some goes bad in the future.

please, just test repacker with variable number of cells. here in my PC the results are a nosense

Korishan said:

Agreed with Daromer. Same number of cells per pack. This is for amp balancing. If 1 pack is 10 cells, and the other is 15 cells, then 10 cell pack with see a 50% increase in amps per cell.

20A load
20 / 10 = 2A/cell
15 / 10 = 1.5A/cell

Click to expand...

Do you mean...
20A / 15 cells = 1.3rA /cell
20A / 10 cells = 2A / cell

?

If you use the 21700 in the smaller count packs, that would balance out somewhat.