Soon the pack built will begin....YES

I am still testing and harvesting, and gaining some numbers.
I amtesting onIR and the SOH%

What is wise to start with, to sort them out to build the packs?
On IR, SOH, or mA
And the next sorting?

I think first on IR sorting then mA and then on soh%
My packs will be 100 cells each in a 7s.
(If i use some words wrong, feel free to rattle my chain.)

All pack's must be at the same amps? And the cells must be in one range? like 2400ma to 2600?
Per pack

You can not combine pack's that have different amps? Like one is 180 amp and the next in line is 200 amp(next 220) ect?

Thanks in advance

100kwh-hunter said:

I am still testing and harvesting, and gaining some numbers.
I amtesting onIR and the SOH%

Click to expand...

IR and SOH are kind of intertwined. If you have a reasonable IR you will have a cell that has a pretty good SOH (percent of capacity remaining)
Now IR is a measurement that is cell manufacturer and model number specific. If you are dealing with only 1 manufacturer and 1 model number then that is an easy way to separate cells right of the bat. If you are dealing with multiple manufacturers and model number the challenge becomes more difficult but not impossible. I am building a chart that shows my findings in what is a go and no go as far as IR is by manufacturer and part number. It can be found here in the IR by part number sheet. Look for the tab at the bottom of the page.It is by no means complete and will be massaged till I am satisfied with it but if you want you can have a preliminary look at it.
If the cell passes the IR test, I then test it for percent of capacity and that will give me the SOH, anything less than 80% goes to a special bin. anything above 80% gets set aside for recheck of V in ~30 to 60 days to see if there is a SD in the bunch. The ones below 4.11V get set aside for more testing. Most of those are usually already below 80% capacity from the original capacity check.


To build packs I would use the repackeras a guide and see what you come up with.
Preferably you want all your packs the same Ahs as the whole pack will only perform as good asyour lowest Ah pack.
So my sorting strategy would be mAh first( as long as you have set aside your lower than 80% capacity cells)as equally distributed across all the packs as possible and I am playing with the idea ofIR adding up also. Not sure about that yet but it is a work in progress. At present I am dismissing cells over 75m? no matter what the SOH is. Sony has some unusual GR G* series batteries that perform excellent but the IR is over 75m? with someof them running in the 80m? up to 95m? range with very good results.


Several members have the 100mAh difference bucket methode which seems to work for them to get reasonably balanced packs.

So that would be my strategy.

Wolf

I make is simple. If it takes to much time then its just stupid.

Sort all your cells in 100mAh buckets based on your max current. If you plan max on 1A then test at that! That will determine if the IR is good enough and if they will work.

Then you arrange your packs easily by just picking your biggest container and sort them random into each string. Ie if you got 3s system you take 1 cell into each of those and you continue untill its filled or you have to go to a lower capacity bucket.

So is this enough? YES its more than enough for a powerwall where you never will push a cell to its extents. How can it be that? Because of statistics and some decent work. As long as you have sorted them with a baseline system with same deviation for instance same tester ONLY you are good to go.

Do I have proof? Is 29 000 cells enough? Is 3 years of running enough and still packs within balance?
Will this work every time? It will work as long as your packs are big enough and you do a good work randomizing. I would say atleast 80 cells per pack for this to work decently. Smaller packs are more likely to have deviations.

Note that balancing with a bms or balancing is not to counter for uneven packs. You balance because the packs are drifting from each other. That can generally happen due to self discharging cells, cells with weird IR (This is removed by proper testing with higher current like 1A) or you are pushing the packs harder than you should

I would say that most common is self discharging or people using the packs on a way that isnt the best...

Enough with the ramble. There are plenty ways of doing packs. Before you even start to build you need to understand why you do it in certain way.

daromer said:

I make is simple. If it takes to much time then its just stupid.

Click to expand...

I agree totally if it takes too much time its a waste of time.
But sometimes it takes time to find the right answers. After more than 4000 cells recorded and tested,by no means 29,000 cells as daromer has, I at least have found some answers to make the build faster.
My cellsuccess rate is>90% since I adopted my IR test before charging strategy.
Its nice to know I'm not testing poor performance cells right of the bat. No wasted time and no wasted energy charging sub performing cells.
Its a great time saver.
Kinda neat when you have results consistently over 80% of capacity.





Not like when I started never knowing what I was going to get. Now that was a waste of time.




Wolf

The is as precheck is for sure doable. For me it takes more time than running the capacity check due to how I have set it up. The "wasted" energy in my case is not noticeable since i produce 100kWh per day and I sell more than 60kWh per day If I just save 5 seconds per cell thats more worth wasting that energy compare to what I can earn on my day to day job For instance I only write up numbers on the cell as "21" or "19" To save time haha....

Note that I dont consider time sitting in the charger as wasted time in my case since the only time i look at is the time "I" work on it. The other time im either sleaping or at work making money

For those looking at the Ir as pre make sure you have a proper 4-wire device. They cost around 50USD and up so its not much money.

Thanks for the this very good info.
My ir tester is still on its way, i think they still have to build it or something......

Apparently the way i am testing is good, for now its only soh%, later on i will also test the capacity between 3.3v and 4 volt when the packs are done.
I am also planning to make a second testing board to test each cell on the capacity between 3.3v and 4v.
To sort them out (on 100mah)i am planning to use the capacitythat i gathered with the soh
Every cell below 2000 and/or below 80% is for later use or different project.
Borders on ir,i thought between 35 and 75 ohms?

To use repacker, its going to be to much time consuming, typing in a couple of thousands of cells.
The way Daromer told, confirmed my thoughts, electricity will find the way of the lowest resistance.
The cells in front of the pack must work a little harder then the cells in the back of the pack.
My bus bars will be pointing in the same direction.
For now i am creating 3x 7s100p, when i have 3x more it will become 1x 7s900.
After that, per 700 good cells will be added up.
But still 100 cells per pack, to take each pack out of the system every year for checking up.

Gentleman, thank you very much for your answers, its really appreciated!

@ wolf, thanks for your xls sheets, downloaded some.
Would you appreciate that i pm/email you, my data when i start to collect data? for your charts to complete?

@ daromer, you wrote:Note that balancing with a bms or balancing is not to counter for uneven packs. You balance because the packs are drifting from each other. That can generally happen due to self discharging cells, cells with weird IR (This is removed by proper testing with higher current like 1A) or you are pushing the packs harder than you should.
Ofcourse i remove sd heaters and everything with a weird IR, and i am not pushing my packs harder then i should, did to much trouble for them.

But what do you mean by:Note that balancing with a bms or balancing is not to counter for uneven packs.
Sorry, don't mean to insult you, but i dont understand this one