18650 parallel cell matching

CUDAcores89

CUDAcores89

Member
Joined
Dec 30, 2016
Messages
83
This isn't as much of a safety tip as it is a way to keep our cells more in balance. This is the method I used for my second ebike battery.​

The vast majority of powerwall builders are trying to get capacity as close as possible in each parallel group with the theory being that the parallel packs will tend to stay more in balance.​

We'll I decided to forget all of this and do my own thing.​

I decided to do something entirely different. Instead of placing cells of similar capacity in each parallel group, I matched the cell brands in each parallel group. For example, all the Sanyo UR 18650FM cells were in one parallel group, all the Sony US18650GR cells were in one parallel group, and all the Samsung ICR 18650 26F cells were all in one parallel group. all cells were above 2000mah, although that was all I had guaranteed. Some cells were 2200mah, 2300mah, a few 2500mah. My theory goes that each parallel pack will only charge to what it will take in, and then they will all bottom balance to the same end voltage. For example, if I take a 3S pack made of a parallel groups of samsung, Sony and Sanyo cells and I started charging them all from 3v open-circuit start voltage per cell, let's just assume they all have a fully charged voltage of 4.1v, 4.15v, and 4.13v. Then when I discharge them they will all draw out the same amp-hours. Then they will be charged up again to the same mismatched top-balance. But that's okay since they will stay that way and so on.​

Anyway, I went ahead and built two ebike batteries. One was with matching capacity (how we normally do it) and the other one was matching cell brands (my new method). What I found was while my new pack (parallel cell brands) was slightly more out of balance after riding, it stayed that way.This is actually what Jehugarcia did to some extent. He put all his sanyo cells in one parallel group, all the panasonic cells in one parallel group, and all the samsung cells in another parallel group and he says it has been working great for him. If this has been working for him, why don't we do it then?​

About my first pack where I matched capacity, this one did something we have all been used to. It started off perfectly balanced, but then the cells driftedfurther and further over time. Until it came to the point where I did actually hadto balance the cells.​

What do you guys think of this method? This is actually what flyabiru11 appears to be doing with his packs (judging by his last video). He hasn't deployed them yet so we don't know how it will do. But at least from my very limited testing it has revealed that matching identical cells together in parallel while maintaining a minimum capacity has kept my cells in balance far better than capacity matching.​

Think any of you guys can do your own testing on some small packs? I can say at least for me from now on I will be using this new method, but I want to see what the community thinks about it.​
 
The closer to a matched capacity you have between them, the more capacity you can extract without going outside the safe SOC,

Your method is a tad strange, as realistically you would want a mix, so that one pack doesnt end up with only high esr cells and run hotter than the rest,

The drift you saw in capacity comes from differences in battery ESR, the higher the ESR, the lower the efficiency of charge, say you charge up at 0.5A, and measure a capacity of 2440, depending on the esr this will read lower when discharged, as some energy is being lost as heat. if its a nice new cell it may discharge 2420mah, while if its a much poorer cell, it could be quite low, say 2200mah,

It probably would not be too hard to cook up an excel calculator to re-rate cells based on there ESR for matching, but for now the true capacity of these cells can be found by charging the cell with a mah counter, discharge it then take the average of the 2 counters.
 
Makes perfect sense if you look at the charge/discharge curves in the PDF datasheet for each cell brand/model.
I've been considering going one further and making one entire battery for each cell brand, eg 3p7s of Panasonic, 4p7s of Sanyo cells, etc.
Give each battery it's own battery management system, and isolation relay.
Not enough cells to do that sort of thing at the moment.
 
Just a thought on why this method might actually work.

Different brands have different manufacturing processes and or ingredient sources for the very chemicals used to make the batteries.

Perhaps what he is seeing where there is no significant change in the unbalanced state is because with the matched chemical composition of the individual cells they just keep plugging along in unison?
 
I'm making my packs 96P from 2.1-2.5 Ah (I set the higher and lowerones aside) and keep them all the same brand. I haven't really done any extensive testing yet but that seems like an acceptable range to me. The majority of goodcellsI've been recovering has been in the 2.3-2.4 Ah range.

My strategy is I will eventually be able to match 7S LG with LG and Sanyo to Sanyo ect... when I have enough 96P packs made. Or at least match the ones closest to each other in the spec sheets.

Tested Ah is testedAh though right? I understand a higher rated 3.0AhAh battery that tests @ 2.5Ah may not have as many cycles left as 2.5Ah ratedbattery that tests @2.5Ah and thats the reason I keep them all in 96P packs ofthe same type. I will eventually have to swap out the 3.0Ah rated pack when it reaches too many cycles.

I'm definitely no expert but I've done some research and this makes sense to me. If I followed the most prevalent 'professional' advice I've seen outside of the forums and never once made an effort to think for myself, I wouldn't be doing this at all. :)
 
this is a very interesting topic, I also thought about it before combining my cells. But I never tested your way, so I don't really know what's better...

I can imagine, that, even if you take cells with the same capacity from the same manufacturer, they might have a different age and you never know how many cycles they already went through. And the more capacity you want, the more cells you need - therefore it's getting harder to find enough cells to combine from the same manufacturer...

Parallel connected cells will balance themselves, so if you connect let's say a cell with 2500mAh and another one with 1500mAh, they will always adjust to the same SOC by transferring energy between each other. I noticed that when soldering parallel packs, some of the batteries got a little warm, some didn't. My powerwall now has a 200p7s configuration, the parallel blocks are split up into packs of 20 cells, all having 42Ah +/- max. 1Ah. I started with 5 packs in parallel and added more as soon as had enough cells tested and packs soldered (always 7 packs with 20 cells), and when adding a fully charged pack to the ones already in production (not fully charged), I meassured a balance current of sometimes more than 10A flowing out of the new pack, depending of the SOC difference, so I could see 'automatic balancing' of all parallel connected cells.

But this won't happen with cells connected in series, that's why we need balancers. A difference in capacity in series of cells also means different voltages, that won't adjust without balancers. So, from my point of view, it depends on the balancing system you use, if your cells stay in balance or not - or at least how long it takes to balance them... I use balancers with max. 10A balancing current and my 7 blocks of 200 cells (for two months in production now) never have a voltage difference higher than 35mV. Even when charging with higher currents (60-80A) or running power consuming devices like a washing machine or dish washer (which take 2000-2500W and pulling >100A out of the batteries), the voltage rise or drop is always even on all 7 serial blocks.

So from my experience, having series with same capacity works great. But anyway, would be very interesting to get some test result of you way.

Have sun!
Oliver
 
Regarding balancing you should not need to balance more than 1 time... In ideal world :)

For instance when you top charge it and it should stay that way. Yes it will go out of balance when you discharge due to all packs not equal ir or capacity... But that is easy fixed by moving some cells if needed.

Due to its 2nd hand cells and we only know the capacity/ir/manufacturer type and not cycles or any other i like the random fact and mix. because then it "should" be equal in the end... Though yet to be prooven...
 
Since this is a good topic and one I have been reading and wondering about for some time now. I have a question about mixing new and old cells in the same packs. I have older used cells ranging from 1800-2100, no idea how many cycles they have been through or have left obviously. Can I mix "new old stock" cells that are testing out to be 2300-2500 in the same pack or would that be hard on my new cells? I only have roughly 180 new old stock cells and the rest well say 500+ old cells.
 
Personally i woudlnt mix them as in mix them... but i would not have a problem running them in paralell as long as you monitor them. Ie create 1 string of new cells and 1 with old. Then paralell those.

Thats how i have my cells arranged. Randomg old in one string and then 2 strings with what i know is old cells but never used.
 
daromer said:
Personally i woudlnt mix them as in mix them... but i would not have a problem running them in paralell as long as you monitor them. Ie create 1 string of new cells and 1 with old. Then paralell those.

Thats how i have my cells arranged. Randomg old in one string and then 2 strings with what i know is old cells but never used.
Click to expand...

Thanks for the reply. There are so many ways and ideas of doing it wasn't sure what was best. I will do that then so hopefully I don't destroy a whole pack of new cells. so then with the old cells just take them all and try and match packs as close as possible?

I have noticed on some cells also that they will say full charge on my 4s but when I check them with the multimeter they will say something less like 4.1 or 4.09 or something like that. Would these cells be ok to use as long as they don't self discharge too low after sitting?
 
Leave them sitting for a week or 3. If they dont drop then they may be good but do a 2nd charge to see if you get them bumped up. They generally sit down to 4.12-4.14 rather fast.


Note that with 2nd hand cells there are no "right" ways. its merely better ways :)
 
well, if you have a bad cell, it will cause problems, no matter where it is... ;-) From my experience, it doesn't matter if you mix old cells with new cells in a parallel cell block. Each cell will take as much charge as it needs and give as much power as it has. And they'll balance each other automatically. As soon as there's one cell giving up, it won't hold as much power as before and will start sucking the power from all other cells connected parallel to it. Depending on how many cells in parallel you have, you'll notice that sooner or later - with few cells sooner, with many cells later - by a faster voltage drop or lower max. voltage than the other parallel blocks in your series.

So from my point of view, it's highly recommended to monitor the voltage of all your parallel blocks. And as soon as you notice a higher voltage difference in one pack, check each cell in it and find the bad boy.

daromer said:
Leave them sitting for a week or 3. If they dont drop then they may be good but do a 2nd charge to see if you get them bumped up. They generally sit down to 4.12-4.14 rather fast.


Note that with 2nd hand cells there are no "right" ways. its merely better ways :)
Click to expand...

that's a good way to find out if a cell is still good or not. I noticed the same thing, some fully charged cells have less than 4.1V, some have even more than 4.2V... I always messure the voltage ~1 hour after they were fully charged and taken out of the charger, then leave them alone for abotu 14 days and meassure the voltage again. In my packs, I only use cells with a max. voltage drop of 0.3V between these two messurements.
 
You must log in or register to reply here.
Back
Top