best order for cells with different voltage/ internal resistance

[Sparfuchs]

Hallo EV friends,

I'm about to build a 14s28p battery pack with "reclaimed" (sellers definition of reclaimed below) 21700 samsung 50e cells.

my main question is if there is a "best" cell combination order to consider by knowing the internal resistance and the voltage of each cell if they aren't all the same ?

Should I take the time to charge all the cells with a bigger difference than 0,1v to the voltage level of the others or is there a easier way like just combining the similar cells in the same parallel group ? If so, is there a order to start with ? Like first a group starting from negative bus with 28 parallel cells with 3,44v then 28 with 3,45v and so on or maybe exactly the opposite order ?

And do I have to consider the internal resistance somehow ?


I bought 400 of this cells but had the last 8 months no time to do build the pack so it they might self discharge a bit.

Now I checked the cells with my multimeter and most have about 3,57v to 3,59v but I also have some with 3,44v.
I also checked 7 cells with my miboxxer c8 smart charger, wrote down the values and checked the same cells again in different slots.
Unfortunately the values were quite different. values are first check/second check and (Value v) is the result from my multimeter.

C1 15/17% 3,57/3,57/(3,55v) 26/22mO

C2 16/16% 3,57/3,55/(3,55v) 35/31mO

C3 15/16% 3,57/3,59/(3,54v) 38/48mO

C4 07/07% 3,47/3,47/(3,44v) 39/30mO

C5 15/15% 3,55/3,55/(3,54v) 21/25mO

C6 14/15% 3,57/3,55/(3,53v) 52/31mO

C7 06/06% 3,43/3,45/(3,44v) 32/28mO

According to Samsung Datasheet the "Initial internal impedance" should be ≤ 35mΩ.
So do I measure wrong, is my charger unprecise or are the cells bad quality ?


Thanks a lot for your help.
Best regards Sparfuchs


Sellers definition of "reclaimed" cells:

These batteries come from battery packs in which welding errors have been found. The batteries have never been charged or discharged.

The cells we offer have:
- Minimal damage to cell
- Never charged during production (from the box) (each cell is tested)
- Resistance is according to datasheet, never worse (every cell is tested)

 

[docw009]

For the time/labor/cost needed for this pack, I think it's worth buying more test gear. I;d suggest the YR1035+ AC IR meter, about $40-45USD on aliexpress. The Samsung datasheet is quoting AC impedamce , which is what the YR3035+ measures. I have one, and it's accurate and fast.

AC impedance is probably more like 12-14 milliohms on a new 50E. That's what I was seeing on some slightly used 50E's and on new LG 21700'a.

 

[From-A-To-B]

Agree with Doc, above. With 400 cells to use, it’s worth getting the proper tools. Those little testers aren’t worth a damn for measuring IR.

Then build a battery so you use cells with as similar an IR as possible, setting aside the high AND low outliers.

IMO, when you build, cells shouldn’t be paralleled if they have more than 0.1V difference. So balance, then build.

 

[Hummina Shadeeba]

Leaving the cells charged for a period and then seeing the drop in voltage seems a good enough indication of the state of health of the cells, no?

Best to disperse the strong and weak cells equally within the paralleled groups.

Variation in voltage drop after 8 months seems a lot. Maybe they were stored at different temps

 

[jonyjoe303]

The IR on your cells are all over the place. If they are from the same Manufacturer they will be very similar in milliohms. I also recommend one of the yr1030's they will give you more accurate "repeatable" results. All it takes is one bad cell (high IR) to give you balancing issues down the line.

 

[harrisonpatm]

Leaving the cells charged for a period and then seeing the drop in voltage seems a good enough indication of the state of health of the cells, no?

It's part of it, yes, but not the whole picture. For a true measure of state of health of individual cells, capacity tests need to be performed, as well as IR tests. IR is good, but you can have good/equal IR on two cells that still have two different capacities

 

[Sparfuchs]

Thanks a lot for all your helpful replies !!!

I just ordered the YR1035+ with a slot for cells to make it easier to measure (picture below).

So I'll check all the 400 cells but how do I use the results ?

As it will be a 14s28p pack I'll have 14 different "voltage categories" so I can combine cells in parallel without any difference (like one parallel group of 28 cells with 3,44v, then a 28 cell group with 3,45v and so on)

• But is there a order for the serial connections ? (Like first group with the lowest voltage, and then going up until the last group with the highest voltage ?
• And what do I have to consider if I know the internal resistance of all cells for the right order ? 14s28p means I'll have to use 392 of my 400 cells and could not use 8 of them if they aren't as good as the others. But if there are more "bad" cells I'll buy a few more.

 

[harrisonpatm]

As it will be a 14s28p pack I'll have 14 different "voltage categories" so I can combine cells in parallel without any difference (like one parallel group of 28 cells with 3,44v, then a 28 cell group with 3,45v and so on)

I believe that this is where you need to take a step back and research a bit more to understand your project. The voltage of the batteries upon receipt has nothing to do with their state of health, their age, their capacity, ect. All it means is that they were charged or discharged to a certain voltage before they were shipped. When you build your pack, you can and should be rebalancing them anyway; ie, charging/discharging all cells to exactly the same voltage. Ignore cell voltage for now, it's not important.

If you are concerned about building a pack so that all the parallel groups are matched evenly, then you're talking about capacity, or the amount of energy, in mAh, that each cell is capable of storing. IR too, yes, I'm not disregarding that part, and it's good that you ordered a proper IR tester (your totally random and varied IR readings from your first post are likely just due to using a poor quality tester) . But capacity is more important.

These batteries come from battery packs in which welding errors have been found. The batteries have never been charged or discharged.

The cells we offer have:
- Minimal damage to cell
- Never charged during production (from the box) (each cell is tested)
- Resistance is according to datasheet, never worse (every cell is tested)

Does the seller state that they tested the capacity of every cell? If so, what is the capacity? And even if they did a capacity test on each cell, do you trust them?

Check out repackr, an online tool for assembling well-balanced batteries. To use it, you need to know all the capacities of your cells, enter all of them and the number of groups in series, and repackr will tell you what cells to put in what groups to make sure that each parallel group is balanced. You don't need to use repackr, I'm just using it as an example of how capacity measurement is required to assemble balanced parallel groups.

I apologize if this is repeating information that you already know. And if the seller can confirm that all cells were individually capacity tested, and you trust the seller that they capacity tested, then you can ignore it and build packs based on IR alone.

 

[harrisonpatm]

Just checked, and it seems like 21700 samsung 50e cells are rated for 5000mAh each. If all your cells are perfectly new and healthy, and you can confirm it, then you don't need to capacity test. The issue is that you know you're buying "reclaimed" cells, whatever that means, and you only have the seller's word that they're all close to new. If you have too many cells in your order that are 4500mah, 4250mah, 4000mah, ect, and you don't know which is which, then your battery will only have the capacity of it's weakest parallel group.

 

[Sparfuchs]

I believe that this is where you need to take a step back and research a bit more to understand your project. The voltage of the batteries upon receipt has nothing to do with their state of health, their age, their capacity, ect. All it means is that they were charged or discharged to a certain voltage before they were shipped. When you build your pack, you can and should be rebalancing them anyway; ie, charging/discharging all cells to exactly the same voltage. Ignore cell voltage for now, it's not important.

If you are concerned about building a pack so that all the parallel groups are matched evenly, then you're talking about capacity, or the amount of energy, in mAh, that each cell is capable of storing. IR too, yes, I'm not disregarding that part, and it's good that you ordered a proper IR tester (your totally random and varied IR readings from your first post are likely just due to using a poor quality tester) . But capacity is more important.


Does the seller state that they tested the capacity of every cell? If so, what is the capacity? And even if they did a capacity test on each cell, do you trust them?

Check out repackr, an online tool for assembling well-balanced batteries. To use it, you need to know all the capacities of your cells, enter all of them and the number of groups in series, and repackr will tell you what cells to put in what groups to make sure that each parallel group is balanced. You don't need to use repackr, I'm just using it as an example of how capacity measurement is required to assemble balanced parallel groups.

I apologize if this is repeating information that you already know. And if the seller can confirm that all cells were individually capacity tested, and you trust the seller that they capacity tested, then you can ignore it and build packs based on IR alone.

Thank you so much for your detailed reply Harrisonpattm,

I've already built many battery packs with reclaimed cells and the work great for now.
But as the cells were all the same voltage (+- 0,02) so I didn't consider (and also didn't have equipment) for further tests back then.

 

[Sparfuchs]

So now I've got the 400 cells I bought and WILL use 392 of them however their quality is.

I do also have the mentioned Miboxxer C8 smart charger, a multimeter and soon the YR1035+ resistance tester.

I just want to get the best result with my given options.

So could anyone give me a "step by step guide" of what to do before spotwelding ?

(If step 1. is buying an other tester/charger to get accurate mAh values to check capacity I will !) any product recommendations ?


Edit:

and just to complete:
the miboxxer c8 charger shows a "percentage" of the capacity of each cell but this values seem way to low and I've no idea how it comes to the value. And as the charger can't discharge I also doubt that values are helpful for my project ?

and btw...I'll use a Smart Ant bluetooth BMS for the pack.

 

[harrisonpatm]

(If step 1. is buying an other tester/charger to get accurate mAh values to check capacity I will !) any product recommendations ?

You'll need a discharge tester for measuring capacity. The opus or liitokala are great for 18650's and I think they're supposed to fit 21700's, but I haven't done it myself so someone else will have to confirm.

So could anyone give me a "step by step guide" of what to do before spotwelding ?

I was responding to @Eastwood a few weeks ago with similar advice:

Eastwood’s first battery build

I pulled this photo off a facebook forum, It shows why p-groups cannot be glued together. They need a separating insulator/ This pack is still insulated, but now that the PVC on the right cell is ripped, a tear on the left side will short circuit the right hand cell, The top of the left hand...

endless-sphere.com

Check out secondlifestorage.com for lots of other good advice and/or procedures to follow.

 

[harrisonpatm]

the miboxxer c8 charger shows a "percentage" of the capacity of each cell but this values seem way to low and I've no idea how it comes to the value

It's a guess based on voltage, but it is just a guess, and the only way to be sure is a proper capacity test.

 

[CONSIDERABLE SHOUTING]

I had a Miiboxxer until I realized it's meant as a hobbyist recharger and not a battery tool.
My Opus can fit 2170s but only 2 at a time.

 

[Sparfuchs]

I just read a bit and realized that even a good charger/discharger/testers have to fully charge them and then discharge them to know the mAh of each cell, am I right ?

I've 400 cells with a average voltage of 3,5v -3,6v at the moment and I absolutely haven't got the time to charge all 400 cells.

So I need a solution to get the best possible cell order that can be figured out within a few hour !
so the options I've got for now are:

• to test the internal resistance and the voltage of all 400 cells.
• In case a few of them would have worse internal resistance then the others I could sort them out and buy a few new ones if needed.
• and I could make sure that every parallel group contains cells with same voltage.
• edit: or charge just the cells below 3,5 Volt (and stop charging at max 3,6v). but is it even better than just sort groups by voltage ?

So is there anything else I should do, know or buy to find my best solution within a day ?

Edit:
You said I should sort the cells by capacity and not by voltage. But it's also known that you shouldn't have more than 0,1v difference inn each parallel group.
But I still don't know if the voltage order (14s lowest v group to highest v group or opposite) or the voltage difference in series is important ?

 

[999zip999]

Wow that's going to be a lot of time and effort. Can you wait out the pros and cons. Maybe just spend your time in building this pack otherwise it's going to be 2 to 3 months to go through those cells put a data sheet on them compare them separate them and sort them.
You roll the dice on buying used cells 400 so just maybe build the pack. As that's going to end up taking a lot of time. How many amps is your controller. If using it 100 amps or under there wouldn't be much demand on 28p 21700 cells.
I'm just trying to balance the pros and the cons.
I mean 140 ah pack you can only charge to 80-90precent and discharge to 20 percent and not worry about it as much.
I believe this must be for a motorcycle.

 

[Sparfuchs]

I just did a try at repackr with a fictional 7s3p battery pack with values from 3500 mah to 3600 mAh (+5 mAh each) and it seems like it's not about heaving as similar mAh values of each cell in the parallel group, but about heaving as similar mah values of each parallel group in the pack right ? And the max. 0,1v difference per cell are just to avoid to much charge/discharge within the parallel group ?

 

[Sparfuchs]

If my new "understanding" is right would that mean:
If I made sure the internal resistance of all my cells is fine and have to build the pack just by knowing the voltage, it would even be a really bad idea to put all 3,4v cells together in a parallel group because there is a good chance those cells have less mAh and so this group would have way less mAh then my highest group and so the whole pack is influenced by that.
So it would be a even better Idea to absolutely randomly (for a good average mAh) take 28 cells per group but only make sure there is not more than 0,1v difference from the highest to the lowest cell in the parallel group ?

Did I just have a breakthrough or messed up even more ?

 

[harrisonpatm]

I just read a bit and realized that even a good charger/discharger/testers have to fully charge them and then discharge them to know the mAh of each cell, am I right ?

Yes

I absolutely haven't got the time to charge all 400 cells.

I've gotta ask, why don't you have time? Is this your primary form of transportation? What's the hurry?

If you need to build a pack immediately, right now, then don't worry about IR, put them in whatever order you want. It'll work just fine. How long will it work that way? Probably not long.

If you want to do something with a concern for quality and longevity, then take the time and follow procedures that ensure you're putting together a pack with adequately balanced cells. Good things take time and patience.

and I could make sure that every parallel group contains cells with same voltage.

Again, you should be doing this anyway, with any battery, regardless of what cells you're using or whether you test them. Ignore voltage when sorting. Just balance the cells before you spotweld. You can charge/discharge cells to whatever voltage you want them to be, from 2.5 to 4.2v.

You said I should sort the cells by capacity and not by voltage. But it's also known that you shouldn't have more than 0,1v difference inn each parallel group.

This is still correct, but, this doesn't change that you need to check capacity if you're concerned with balanced P groups. They only need to be the same voltage right before you assemble your pack.

I just did a try at repackr with a fictional 7s3p battery pack with values from 3500 mah to 3600 mAh (+5 mAh each) and it seems like it's not about heaving as similar mAh values of each cell in the parallel group, but about heaving as similar mah values of each parallel group in the pack right ? And the max. 0,1v difference per cell are just to avoid to much charge/discharge within the parallel group ?

This is all correct. Unfortunately, your hypothetical test at repackr involved all cells within 100mah of each other. That's not a huge deviation. If you were actually building a 7s3p pack, and all your cells are 3500-3600mah, that's going to be a well-balanced pack already, regardless of how you sort them. My concern is that within the 400ea 5000mah cells you've purchased, used, you will probably have most of them, testing at 5000mah, but you may have some testing at 4000mah, maybe 3500mah. 4000/5000mah=80% capacity loss, and with used/aged cells, this is possible, and you won't know which ones are the lower capacity ones unless you test them first.

Wow that's going to be a lot of time and effort. Can you wait out the pros and cons

Sure. The pros of skipping capacity testing now and just building a pack is that you have a pack sooner. The cons of skipping capacity testing is that one P group out of 14 is composed of excessively aged, lower capacity cells, which limit the capacity of the entire battery, decreasing its range and lifespan.

It's up to you. Nobody is making you capacity test every one of the used cells you've purchased. I'm just recommending it.

 

[harrisonpatm]

If I made sure the internal resistance of all my cells is fine and have to build the pack just by knowing the voltage, it would even be a really bad idea to put all 3,4v cells together in a parallel group because there is a good chance those cells have less mAh and so this group would have way less mAh then my highest group and so the whole pack is influenced by that.

Yes. You have it now. A 14s battery will have the capacity of its weakest parallel group.

You're making 14s28p. Each P group of 28 cells, rated at 5000mah, will have a capacity of 140,000mah, or 140 amp hours. If they're actually all new and at their rated capacity. So you'd build the pack, expecting it to have a capacity of 140ah. But if you have terrible luck and you actually end up sorting all 80% aged cells into a single P group, then that P group will have a capacity of 112ah. Therefore, the entire battery will have a capacity of 112ah. It doesn't matter that the 13 other groups are great cells, they've been bootstrapped to their weakest link.

But, if you were to distribute all those 80% cells evenly into each P group, then each P group would have a capacity of, say 130-135ah. Much better.

So it would be a even better Idea to absolutely randomly (for a good average mAh) take 28 cells per group

Randomness and even distribution is exactly what repackr does when you give it the values of your cells.

And as I said in an earlier post, you don't have to use repackr.

This is one way to do it manually.

but only make sure there is not more than 0,1v difference from the highest to the lowest cell in the parallel group ?

Stop worrying about voltage. The current voltage of a cell has absolutely no relation to its capacity.

After you sort cells by capacity, you can line them all up in parallel, connect + an - to a resistor, and they'll balance themselves to exactly the same voltage. Will take anywhere from a couple hours to a week, depending on capacity and resistor rating. And guess what, if you balanced all 400 cells to exaclty the same voltage, then check voltage again, and find one or two cells that are still lower voltage than the others? Congrats, you found a self-discharging cell, which would negatively impact whatever P group it ended up in. Now you have another testing method up your sleeve.

 

[999zip999]

The I.R. of each parallel group should be the same.
Tha capacity of each parallel group should the same..
No when I say that you have to be reasonable you can't put a dead cell in each parallel and hope it to be good.
1. What is this battery for. ?
2. How many amps are you demanding from the controller ?

 

[999zip999]

Plus buy a Bluetooth BMS and you can hook it up before you wrap up your battery. Then you could charge it up put a minor load on it then charge it up again and you'll see what's happening.
A big pack I would use cell level fuses. Something like this.

NICKEL FUSE 2P-6P WIDE CONTINUOUS ROLL BY THE FOOT! 18650 CELL LEVEL FUSING

Each foot equals 15 cell connections long by 2, 3, 4, 5, or 6 wide. For example 2p wide would be 30 cell connections per foot (2 wide x 15 long) and 6p would be 90 cell connections per foot (6 wide x 15 long) Introducing custom cell level fused nickel design that can be spot welded with ease...

batteryhookup.com

But it's hard to say because we don't know how many amps you can draw the pack or what's it for a golf cart.
Ok 28p x10 amp. 280 Watts the max. If these are 10 amp cells if not we'll have to lower it or raise it.
No matter what that's a lot.

 

[Sparfuchs]

I'm so thankful for all your great help !
Just one question left:

I checked all the 400 cells for internal resistance and voltage and had to charge a few of them a bit so they are all between 3,53V and 3,63V (I marked those cells to spread them in the paralell groups just to feel save). Now all cells are sorted by 0,01V. So all 3,53v cells together, all 3,54 cells together and so on up to 3,63v.
Now I'm about to place them in 14sx28p in the cell holders. As all the cells were awesome from internal resistance I assume the voltage might be (don't have to be) a indication of their capacity or even quality. Even if it's not I will try to spread the cells by voltage as even as possible in the 14 parallel groups. So for example, if 140 cells are at 3,56v, each of the14 groups will get 10 of them.

So my question is if it matters if I start to spread the highest or the lowest cells for parallel ? Or should I try to spread the voltage inside the groups ?

 

[harrisonpatm]

I assume the voltage might be (don't have to be) a indication of their capacity or even quality

It's not.

Do me a favor. Pick one of your cells. Doesn't matter which one. Record it's voltage. Charge it for an hour at a safe current, whatever method you feel comfortable using. Check its voltage again. Did it change?

Now I'm about to place them in 14sx28p in the cell holders

So you're going ahead building your pack without testing any of the cells' capacity? Just want to make sure you're okay with that.

 

[Sparfuchs]

It's not.

Do me a favor. Pick one of your cells. Doesn't matter which one. Record it's voltage. Charge it for an hour at a safe current, whatever method you feel comfortable using. Check its voltage again. Did it change?

So you're going ahead building your pack without testing any of the cells' capacity? Just want to make sure you're okay with that.

So IF! this are all brand new cells, that were all about the same Voltage when I bought them (didn't check back then..just assuming from all the other "reclaimed" cells I had and checked for now), how it comes that some of them do now have way lower voltage then the others ?! If the only thing I can/ or am willing to check is internal resistance and voltage, I think of the possibility that there might be slightly "quality" differences from the cells that have held there voltage and the cells that lost some of it. I can't tell if that is even a relevant difference, and absolutely normal also for brand new cells. But I'd also be ok if those cells aren't same quality than new ones.. they cost less. But according to the test of the internal resistance and the voltage (after 8 months) made me quite optimistic that I've made a good deal.
So anyway I will not buy a new charger and carge and discharge of the 400 cells and will try my best from the values I've got.

So whatever their quality or capacity might be. So it's not about making sure I combined them best possible way, it's about increase the chances that I did with the values I have. So should I spread all the cells I've sorted absolutely randomly all over the pack or is there something I could do to increase the chances of getting good balance ?