diy solar

diy solar

DIY Cell matching process

blutow

Solar Enthusiast
Joined
Dec 20, 2020
Messages
344
I've got 16 lishen 272 cells on order and I'm trying to better understand the matching process and options. I'll be building 4 12v battery packs using 120a overkill BMS's.

My understanding is that I'll want to match cells by capacity as much as possible.

Here is my current plan to figure out capacity:
-Top balance: Group cells into 4 separate 4-cell parallel packs for top balancing. Do this initial grouping/matching based on internal resistance of each cell
-Reconfigure in series: After full charge/top balance, reconfigure each pack in series, add BMS
-Capacity test: Run test on the pack with inverter/shunt, watch cells as they drain and note any outliers dropping voltage faster than others. Compare BMS voltage vs. voltmeter readings to make sure BMS is properly calibrated and all connections are good.
-Gather results: At completion of capacity test, record final voltage of each cell. Santity check that any outliers above are the lowest voltage cells.
-Rinse, repeat for the other 3 packs
-Test individual cells if needed: If I see any dramatic outliers, run capacity test on that individual cell to make sure there wasn't a testing problem, poor connection, BMS calibration issue, etc.
-Match Cells: Review data and sort all cells based on ending voltage of capacity test. Rebuild packs as needed to line up final voltage and call it done.

I've seen some posts from people individually capactiy testing each cell and matching them based on that. I could go that route also, but that would take a lot of time and I'm wondering if the process above could get you pretty close to same results. I've got one of those cheap capacity testers coming, so this is an option, but I don't want to waste a bunch of time if it's not valuable.

Any feedback on the approach described above? Am I missing any important steps? Any opinions on whether it would be worth doing individual cell capacity testing?

I'm also questioning whether capacity matching is even important if all 4 of my packs are going to be charged and discharged at the same time. I guess if I get up against a BMS shutdown limit, it's nice that all the lowest capacity cells are together so it might keep the other 3 packs running longer.
 
I don't see any value grouping cells before the top balance. Internal resistance will vary with SOC, and prior to a top balance the SOC is unknown and will vary.

The best way is to test each cell with one of the 150W dedicated testers. Then group by capacity. If that is unfeasible (or unnecessary!), then just top balance and build the pack.

The odds are your cells won't vary by more than 3% or so.

My take, unless you absolutely need max capacity, just build each pack and test. If they all delivery similar AHs, leave it be. If you have a couple packs with weaker cells holding them back, swap the weak cells into your "leftover" pack. That would let you use a bit more of the total capacity.

At 120A max current, your not going to be seeing a huge effect from internal resistance mismatch.
 
I don't see any value grouping cells before the top balance. Internal resistance will vary with SOC, and prior to a top balance the SOC is unknown and will vary.

Thanks. Yeah, I wasn't sure there was any value to that. I figured it was quick and easy since I'll have the IR numbers for each and it might be better than just grouping them randomly.
 
The way I did it was as follows:

1/ top balance all cells to 3.65V (in parallel).

2/ reconfigure all cells into a single 16S 48V battery.

3/ use capacity tester to drain 48V battery at 150W, monitoring cells hourly (or whatever). 272Ah being drained at ~3A (0.01C) is going to take over 90 hours, so it’s pretty easy to stay on top of cells and identify the weakest cell(s) well before they get anywhere close to the lower knee). If going through an extended period when monitoring cannot be maintained at safe intervals, slow down or even stop discharge entirely until monitoring can be resumed (the capacity testers make this easy).

4/ I actually tracked my weakest cell all the way down to 2.5V monitoring down the knee with a multimeter at which point I stopped the discharge and removed the cell to form a 15S battery, allowing me to continue the discharge and identify the second-weakest cell. Note that this process can be repeated to rank as many cells as worthwhile.

5/ this process is only worth continuing as long as you have cells seemingly several % below the others (meaning 3.5+Ah less or ~70 minutes @ 3A or less in your case). Continue until all remaining cells are clearly over the lower knee (meaning below 3.1V, at least for EVE 280Ah cells) at which point a ranking of the remaining cells based on voltage is likely to be accurate. [see Note below]

6/ with ranking complete, I then reconfigured 48 cells into 2 8S 24V batteries, one with the weakest 8 cells and one with the strongest 8 cells.[or 4 4S 12V batteries]

7/ If the stronger 8S battery has been composed of 8 cells which all reached 3.1V without any yet reaching 2.5V and being pulled from the earlier test, top/balance should be intact and you can check top-balance by charging with an 8S charger (ideally with a BMS in place, but with modest charge current and careful monitoring, you can charge ‘naked’, especially if only getting the cells back into the ‘safe’ SOC zone on 40-50% SOC).

8/ The weaker 8S battery composed of mismatched cells reaching 0% (2.5V) is now effectively bottom-balanced and can be connected in parallel and top-charged/balanced again to 3.65V before being reconfigured into a second 8S 24V battery.

If you have a battery (either the strongest 8 cells or the weakest 8 cells) composed of a mixture of empty and near-empty cells, it’s probably easiest to drain all cells to 2.5V before configuring as an 8S battery and proceeding as in 8/.

Note: For this reason, there is a natural advantage to proceeding with the process in 4/ for the weakest 8 cells [or weakest 4 cells in case of 4S 12V battery target] (which are now bottom-balanced) while leaving the strongest 8 cells [or 12 cells of forming 4 4S 12v batteries] top-balanced so that they can immediately be recharged as an 8S battery [or 3 4S batteries].

The process I outlined above is to characterize 16 cells to create 2 8S 24V battery, but I believe it should be clear enough how to adapt to create 4 4S 12V batteries... [the key point is to decide to stop after fully draining 4, 8, 12 or all 16 cells and treat bottom-balanced 4S packs differently from 4S packs that have remained top-balanced]
 
I'm new to LiFePO4.

My configuration is twelve 3.2v (280 ah each) cells in three banks of 4, for a 12v system. BMS is a Daly unit, 250 amp (4s).

Not all the cells arrived at the same time. When the first eight arrived, I ganged them together as per Will's suggestion charging them all at 3.6v for a couple of weeks. There was negligible current change over two weeks, and I figured they were as "done" as they were going to get. When the last set arrived, I took four off the charger and put the new four on. Over several days I again noticed almost zero change in current, so I figured they were done as well.

When I configured them (all twelve 3.2v cells) for 12v and connected a charger, they registered only about 33% charge on the Daly BMS. They completed charging in about 2 hours or so and the Daly BMS registered 100%. However, one of the sets (#2) showed a higher voltage, I believe it was 3.7 and the rest were 3.3(something). I figured they would even out over time, that's what the balance leads are for right?

Well I have had the battery hooked to my solar charger for several days now and after discharging them down to 75%, I allowed them to charge back up to 100%. For most of that charge cycle groups 2, 3 & 4 were VERY consistent, and sometimes identical in charge voltage, with #1 running just a tad bit lower. Then, as the bank approached 100% I see #2 (again) start to become much higher than the others. Right now, with the BMS reporting 100% charge, #1, #3 and #4 are 3.31, 3.356 and 3.348 respectively... #2 is showing much higher at 3.969.

Can anyone suggest, what is it I am seeing here, is it harmful, and what should I do about it (if anything).

Thanks in advance for any insight.
 
Last edited:
I'm new to LiFePO4.

My configuration is twelve 3.2v (280 ah each) cells in three banks of 4, for a 12v system. BMS is a Daly unit, 250 amp (4s).

Not all the cells arrived at the same time. When the first eight arrived, I ganged them together as per Will's suggestion charging them all at 3.6v for a couple of weeks. There was negligible current change over two weeks, and I figured they were as "done" as they were going to get. When the last set arrived, I took four off the charger and put the new four on. Over several days I again noticed almost zero change in current, so I figured they were done as well.

When I configured them (all twelve 3.2v cells) for 12v and connected a charger, they registered only about 33% charge on the Daly BMS. They completed charging in about 2 hours or so and the Daly BMS registered 100%. However, one of the sets (#2) showed a higher voltage, I believe it was 3.7 and the rest were 3.3(something). I figured they would even out over time, that's what the balance leads are for right?

Well I have had the battery hooked to my solar charger for several days now and after discharging them down to 75%, I allowed them to charge back up to 100%. For most of that charge cycle groups 2, 3 & 4 were VERY consistent, and sometimes identical in charge voltage, with #1 running just a tad bit lower. Then, as the bank approached 100% I see #2 (again) start to become much higher than the others. Right now, with the BMS reporting 100% charge, #1, #3 and #4 are 3.31, 3.356 and 3.348 respectively... #2 is showing much higher at 3.969.

Can anyone suggest, what is it I am seeing here, is it harmful, and what should I do about it (if anything).

Thanks in advance for any insight.
The cells need to be properly top-balanced.

#2 either has more charge than the other cells or has less capacity.

Parallel connect all 4 cells that will be used in a 4S 12V battery, charge to 3.65V until current drops to 0A, then disconnect charger and leave connected for at least several hours if not overnight.

Now reconfigure as a 12V 4S battery and discharge.

if cell #2 hits 2.5V before any other cells, it has less capacity, If another cell approaches 2.5V before cell #2, cell #2s capacity is five (and that other cell probably has the smallest capacity of that bank).

Stop discharging when the first cell reaches 2.5V and then recharge.

You should now see cell #2 approach ‘full’ (3.65V or 3.6V or 3.5V) coincident with the other cells.
 
Thank you for the reply fafrd -

As I mentioned in my post, I previously DID have eight of these cells hooked up paralell at 3.6v

FOR TWO WEEKS - actually nearly three!

current dropped a very small amount over that two weeks time. At the rate it was going I would have had to leave them on there six months for "current to drop to zero".

Somehow that just doesn't seem right to me... does it to you?

I'm not arguing with you but rather looking for more insight. I couldn't understand why the top balancing seemed to be going nowhere with either group of cells I had hooked up this way.
 
Thank you for the reply fafrd -

As I mentioned in my post, I previously DID have eight of these cells hooked up paralell at 3.6v

FOR TWO WEEKS - actually nearly three!

current dropped a very small amount over that two weeks time. At the rate it was going I would have had to leave them on there six months for "current to drop to zero".

Somehow that just doesn't seem right to me... does it to you?

I'm not arguing with you but rather looking for more insight. I couldn't understand why the top balancing seemed to be going nowhere with either group of cells I had hooked up this way.
Yeah, absolute zero current isn’t critical, but low enough that no real charging is still going on.

For my 280Ah cells, monthly self-discharge is 3% / month or 8.4Ah / cell / month.

For all 12 of your cells connected in parallel, that equates to 101Ah /month or 3.36Ah / day or 140mAh/h, meaning 140mA.

thus is just to provide some scale - once your charge current approaches 140mA for 12 cells or 47mA for 4 cells, you are basically just offsetting self-discharge.

The best thing to do is set the voltage at target of 3.65V or 3.60v and charge until voltage and current stabilizes (which should be under 100mA for 4 cells).

Then disconnect the charger and leave the cells connected in parallel for at least a few more hours if not overnight.

An 0.5mV voltage difference can drive 10A through an 0.05mOhm busbar. Internal resistance is probably higher than that, but even at internal resistance of 0.25 mOhms, an 0.5mV delta between cells will continue to drive a 2A balance current (internally, between cells).

You don’t need to go to this extreme, but the better you complete the top balance, the quicker and easier you will be able to debug your pack.

It’s also a good idea to connect power and ground from opposite ends of the pack (which reduced voltage mismatch between cells along the string).

P.S. I top-balanced (charged) my 16 280Ah cells for over a month...
 
For my 280Ah cells, monthly self-discharge is 3% / month or 8.4Ah / cell / month.

For all 12 of your cells connected in parallel, that equates to 101Ah /month or 3.36Ah / day or 140mAh/h, meaning 140mA.
Your self discharge seems very high, about a full order of magnitude higher than what I'm seeing in my cells. I see about .02a or even less to keep a cell at 3.4v (compressed).
 
Your self discharge seems very high, about a full order of magnitude higher than what I'm seeing in my cells. I see about .02a or even less to keep a cell at 3.4v (compressed).
That was not my measured self-discharge, it is the self-discharge specification on EVE’s 280Ah datasheet (‘<=3%/M’).

3% x 280Ah = 8.4Ah.
8.4Ah / month = 11.7mAh / hour.
So a single 280Ah cell can self-discharge at a rate of up to 11.7mA and be within spec.

On the one hand, that specification is a maximum, so actual self-discharge may be lower.

On the other hand, that specification also stipulates ‘30%~50% SOC’ meaning 3% is the maximum at mid-charge levels and self-discharge at 100% SOC may exceed 3% and still be in spec.

I provided that calculation just to provide some reference of scale for

Once your charge current has dropped to 10-20mA / cell (160-320mA for 8 cells) and voltages are not changing, you are probably just offsetting self-discharge.

0.02A to maintain a 280Ah cell at 3.4V translates to 20mA or 20mAh/hour, which is 0.48Ah/day or 14.4Ah / month.

14.4Ah / month is over 5% of 280Ah, so your self discharge rate (at 100% SOC) is over 5%, not under 3%...
 
I don't think any chemical batteries charge current will ever drop to zero. Lead acid is a bit higher due to the shuttle reaction, but LFP probably will always have some at 3.65V. Now if we are talking 3.55V, the current may taper to a few mA.
 
I'm new to LiFePO4.

My configuration is twelve 3.2v (280 ah each) cells in three banks of 4, for a 12v system. BMS is a Daly unit, 250 amp (4s).
...
Right now, with the BMS reporting 100% charge, #1, #3 and #4 are 3.31, 3.356 and 3.348 respectively... #2 is showing much higher at 3.969.
...
Can anyone suggest, what is it I am seeing here, is it harmful, and what should I do about it (if anything).

LiFePO4 cells must never exceed <2.5v or >3.65v ever.

There is no need (apart from a top/bottom balance, or capacity test) to go anywhere near those values. Do you fill your petrol tank until gas is pouring out on your shoes, or run the tank empty until you need to call a tow truck ?

A typical BMS does not have the current ability to balance 280Ah cells.

Your BMS is the last line of defence against cell damage, you should configure it to never let a cell exceed 3.65v or go below 2.5v (but for day-to-day usage your BMS cell HVP/LVP should be well inside this range)

Your charger/inverter is the first line of defence, but this is only reading the pack voltage, so it won't prevent a single cell from over/under voltage.

Go back and re-read the top balancing guide in the resources section.

Disassemble your problematic 4s pack and repeat the top balance, more is not better, you should NOT leave the cells at 3.6+ volts for days/weeks. Personally I would recommend that you keep the cells connected in parallel at the end of your top balance, and let them sink down to 3.5x over a few days before you remove the busbars.
 
So, if you know the capacity of your cells and you know the the IR at 50% SOC what's the best way to pair up a 2P8S pack. Sort (pair) by Capacity first then IR, or IR first then Capacity?

I've attached a small spreadsheet illustrating the two pairing methods, which is better Method A or Method B? What's the rationale?
 

Attachments

  • Screen Shot 2022-12-05 at 8.21.20 PM.png
    Screen Shot 2022-12-05 at 8.21.20 PM.png
    85.3 KB · Views: 12
Depends on your current. High current applications should match against IR, so each cell in the block shares current equally. Lower current applications should pair cells so that each block/group has the same capacity.

Your IR values are close enough that contact resistance variance between cells will exceed that IRs effect on current sharing, though for low current applications, it probably doesn't matter.
 
Thanks Luthj. So you like Method A. Capacity first for a .1C typical current system. That's how I was leaning.

It's for a trailer I'm building. The biggest draw will be the electric incinerating toilet and the split system air conditioner. When they are both on it will pull about 60 amps or .11C. The bank is actually 2P16S, I simplified it to 2P8S for my question.

If I did make a crazy mismatch pair, let's say I paired up a 300AH cell with a 100AH cell in a 2P8S config, what would happen? Let's assume they were top balanced, identical IR at 50% SOC and the pair has only one lead to the BMS. As the parallel pair discharged would the 100AH Cell reach 2.5v about 3 times faster than the 300AH cell then continue to approach zero Volts by the time the BMS read 2.5V? In how many ways is this a terrible idea?
 
I may not have been clear. You want each cell block (logical cell group), to have the same capacity. Otherwise the pack will be limited by the 2 lowest capacity cells (matched together).

So you want to pair cells in such a way that each pair has as close to the same capacity as reasonable.

If I did make a crazy mismatch pair, let's say I paired up a 300AH cell with a 100AH cell in a 2P8S config, what would happen?
It would be unusual for both to have the same IR, but its possible if the construction is different enough. Ideally the 100AH cell would have 3x the internal resistance of the 300AH one. In that case, they would share current proportionally, and should match each other for SOC, even at higher currents. If the IR is the same, the 300AH cell would lag behind the 100AH cell in charge and discharge slightly. There could be a bit of cross charge/discharge if the current was high enough (during resting periods).

It is possible to use some advanced software to model this, but I don't have access to it (cost about 10k per license!). Besides, it would require a full characterization table for the cell types in question.

The real question is what do the other cell pairs in the battery look like in this scenario? If the rest are 300 and 300 (600AH total), then the pack will be limited to ~400AH If the rest of the cell pairs are around 400 (say 150 and 250), then the pack will still function just fine.

The real issue would be a 280AH nominal cell with 100AH, that indicates a major issue, and that cell could have very unusual characteristics. High self discharge, local heating, or even contamination/breakdown of the electrolyte. In which case it could ruin that cell pair.
 
I went through a similar thing with my 32x cells. I did a capacity test on all of them and then matched the highest to lowest to average out their capacity. Mine went something similar to this:

283Ah -> 263Ah = 546Ah
281 -> 265 = 546
280 ->267 = 547
etc
etc
etc
273 ->272 = 545

So at the end, I had a pack that averaged ~546Ah.

I'm not sure if this is the correct way to do it, but they stay pretty well in balance. Again though, like @Luthj said, I am using relatively low C rates, usually well under .1C. This makes internal resistance less important, but still important.
 
@Luthj and @Lt.Dan, Thank you. I'm happy to realize my 2 weakest cells are not what will define my pack capacity. I've attached my updated theoretical pairing chart. I get an (theoretical) extra full hour of air conditioning this way! It's my first 2P pack. Thanks again for the upgrade.
 

Attachments

  • Screen Shot 2022-12-07 at 9.33.28 PM.png
    Screen Shot 2022-12-07 at 9.33.28 PM.png
    76.9 KB · Views: 6
Back
Top