How unbalanced

My packs are out of balance, just not sure what's too bad.
My cheap bms does top end balancing, but I don't charge them enough to get that benefit.

Two strings are around 40-50mv out of balance, idk but I figure that's not too bad.
But the third is currently 70mv apart. I noticed yesterday that it was 100mv different. So I hooked up an imax to just that pack and get it down to like 50mv.
After it rebounded and recharged I'm sitting at .07v difference in that string.

I am about to manually balance it a bit more and see where it's at tomorrow.

But how unbalanced is too much?

Hi, if the packs are recently build may happens, after test them and let rest for some weeks not all cells stay at 4.15 or more. I use the cells that discharge no more than 4.0. So maybe after build the packs may be unbalanced.
Charge the third pack until reachs the others and check after some days, if this pack goes unbalanced again some cells in this pack must be bad and they selfdischarge.

70mV is not too serious yet but starting to climb.

You may need quite a long charge, holding the pack at the top to let a cheap balancer get the cells equal.
Cheap balances only have a small balance current, eg 50mA, so if some cells are low, they only get about 50mA while balancing.
This 50mA might take many hours or even a day or two to balance.

Manual balancing would be faster. Remember to hold the cell at the full voltage for a while until the current has dropped away, that means the cell really is full.

Like jesusangel says above, if some cells drop in voltage much more than others, you might have some bad cells, either self discharger (if voltas drop just sitting) or maybe lower capacity (if pack is used).

Good to hear.
I would guess that pack has higher capacity than the rest. As the lower packs have a voltage closer to the packs in other strings.

I have it manually balanced now by discharging it, not by giving it a full a balance charge.
we'll see what it looks like tomorrow.
I also have those cheap buzzer voltage monitors hooked up now, so I can keep a closer eye on it.

It's not about how much it is unbalanced, but more about how fast it goes unbalanced. I've been without my BMS for about 2 months now (just lazy to put my replacement in since it's cold) but I check on it once in a while. It's off by 50mV now. When I was on the BMS it was off by no more than 25mV.

You need to start with a good set of battery ranges. If you mix 1800 with 2800, you will find they just don't work well as one drains faster than the other then it throws the other's out of whack. I stick mine to a solid range of no more than 300mah difference between every cell. Repackr is a waste of time, tedious output and repacking for something that will eventually deviate if you have a wide range to begin with.

Agree with not2beme, you also need to be watching for drift/self discharge.
That said, if the pack is recently built, you also have to get in into being balanced at first too.

Dmcbudman said:

But how unbalanced is too much?

Click to expand...

Basically, you loose capacity. (by unbalancing)

So your question reconfigures to:
How much capacity can you afford to be lost ?

I have 3 sets of 14s120p (48v) = 42packs (a mix of 6,000 cells mostly from alarmhookup / used but good). Each pack was built different cell counts to keep overall packs at <1% difference between them. I have run (for longer life) at max charge of 56.0v(4.0v/pack) and low (inverter off) at 47.5v(3.4v/pack) and have cycled over 150 times so far.

I recently got my Batrium BMS online, did some balancing around 3.9v/pack but then stopped (set Batrium balance to kick in at 4.10v) because for the most part, 37 out of the 42 stay (up and down) withing 0.06v of each other. I have 5 'bad packs' that I am in process of fixing that are 0.11 outliers - but setting those aside.

My question is.. based on comments above...
Isn't is OK for my packs to be withing 0.06v as they go up/down daily - especially since my operating range is 4.0v to 3.4v so worst case is 4.06 to 3.34v. And that using Batrium balancing to try to keep them perfect (withing 0.2v?) is overkill / wastes power?.

0.2V or 0.02V?

0.2V is too much and any pack that is that far out from the others needs attention. Even 0.1V difference could lead to a major issue.
0.02V is fine, but I'd be worried if the variance starts to increase. 0.4 is when you'd really want to start tracking down the issue.

Also, with the increase in cells in parallel, that variance increases needing attention as the Wh of wasted energy goes up quickly. And this also makes the bms work harder to keep everything in balance.

I'd suggest 0.06V (ie 60mV) is not too bad for variance.
The Batrium cellmons have 10mV resolution.
Top balancing (like the Batrium gear does) should be targeting the same voltage anyway.
In the Batrium s/w, under Menu > Hardware > Cellmon > Bypass Extra Mode have you enabled "Auto Level"?

You mentioned you have different numbers of cells per pack to get equal capacity. This does mean cells in packs with less cells would get worked slightly harder, but this shouldn't be enough to make a difference.

Part of the question is: "what's the trend?"
The packs you mentioned that are 0.11V (110mV) out, do they
- both get to minimum (go flat first) & maximum (get to full earliest)? Eg the cells have lost capacity.
- never get to full voltage (always lag lower), eg they have 1 (or maybe more) self dischargers or higher capacity than the rest.
Some packs internal resistance may have climbed = uneven charge/discharge vs other packs.

Korishan said:

0.2V or 0.02V?

0.2V is too much and any pack that is that far out from the others needs attention. Even 0.1V difference could lead to a major issue.
0.02V is fine, but I'd be worried if the variance starts to increase. 0.4 is when you'd really want to start tracking down the issue.

Also, with the increase in cells in parallel, that variance increases needing attention as the Wh of wasted energy goes up quickly. And this also makes the bms work harder to keep everything in balance.

Click to expand...

Yes - I ment .02v

---

Redpacket said:

I'd suggest 0.06V (ie 60mV) is not too bad for variance.
The Batrium cellmons have 10mV resolution.
Top balancing (like the Batrium gear does) should be targeting the same voltage anyway.
In the Batrium s/w, under Menu > Hardware > Cellmon > Bypass Extra Mode have you enabled "Auto Level"?

You mentioned you have different numbers of cells per pack to get equal capacity. This does mean cells in packs with less cells would get worked slightly harder, but this shouldn't be enough to make a difference.

Part of the question is: "what's the trend?"
The packs you mentioned that are 0.11V (110mV) out, do they
- both get to minimum (go flat first) & maximum (get to full earliest)? Eg the cells have lost capacity.
- never get to full voltage (always lag lower), eg they have 1 (or maybe more) self dischargers or higher capacity than the rest.
Some packs internal resistance may have climbed = uneven charge/discharge vs other packs.

Click to expand...

>This does mean cells in packs with less cells would get worked slightly harde
My packs are 260ah +/- 1ah each.Some have 2150mah cells (120 +/- a couple) and some 2600mah cells (100 cells +/- a couple). This does not seem to be an issue in voltage variances going up/down.

>have you enabled "Auto Level"?
I experimented with this but at 260ah per pack, it didn't do much when I have 5 'bad' packs with such variance. It seems that letting longmons get hot shortens their life span - not sure how much balancing / shorter life, but makes sense to limit unneeded balancing for longer life. I'll get back to it when I get the bad packs replaced / smoothed out.


>The packs you mentioned that are 0.11V (110mV) out, do they
I have 5 packs (of 42) that go lower and/or higher than all the rest. They happen to be made up of green cell Sony US18650GR-G5 and G7s.

>both get to minimum (go flat first) & maximum (get to full earliest)? Eg the cells have lost capacity.
Yes - these 5 packs to flat first

>never get to full voltage (always lag lower), eg they have 1 (or maybe more) self dischargers or higher capacity than the rest.
They charge faster (go to higher voltage faster) than rest.




>Some packs internal resistance may have climbed = uneven charge/discharge vs other packs.
Not sure if this high voltage variance is Sony G5/G7 characteristic but my main theory is that I damaged them - e.g. they went down to 2.8v or 2.5v or 2.0v over 100 cycles before I got my Batrium.
You suggest that high internal resistance could be the cause of voltage variance... and if I damaged them, it would cause high internal resistance - so the 'damage' theory seems to hold together.

I have cells of same type as good packs to replace these 5 bad packs and will take the bad ones apart and re-runthem thru OPUS and see if i can find bad cells / high resistance. If I do, I'll assume I damaged them. If not - then will look for alternate theory

At what SoC are these imbalances being measured - without any context the figures being mentioned are rather meaningless.

Immediately oncompletion of a balance charge cell voltages should be within the accuracy of your test equipment, or the BMSs ability to properly balance, so practically identical.

Having been correctly balance charged, and left for a period of time(hours, days, weeks) any imbalance is a sign of self discharge - the longer you leave them the greater the imbalance may wellbe.

A battery, having been first correctly balance charged, then fully depleted through discharge,showing an imbalance of a few tenths of a volt isnt anything to be worried about, it's just a sign of mismatched capacity within your cell groups, as is to be expected with second life cells from multiple sources and manufacturers. A battery that is unbalanced when discharged may need very little balancing at high states of charge.New cells can be expected to maintain balance within a smaller window.

The above is assuming a reasonable size battery, let's say 200Ah with an end of discharge cell voltage of around 3.5v

Sean said:

At what SoC are these imbalances being measured - without any context the figures being mentioned are rather meaningless.

Immediately oncompletion of a balance charge cell voltages should be within the accuracy of your test equipment, or the BMSs ability to properly balance, so practically identical.

Having been correctly balance charged, and left for a period of time(hours, days, weeks) any imbalance is a sign of self discharge - the longer you leave them the greater the imbalance may wellbe.

A battery, having been first correctly balance charged, then fully depleted through discharge,showing an imbalance of a few tenths of a volt isnt anything to be worried about, it's just a sign of mismatched capacity within your cell groups, as is to be expected with second life cells from multiple sources and manufacturers. A battery that is unbalanced when discharged may need very little balancing at high states of charge.New cells can be expected to maintain balance within a smaller window.

The above is assuming a reasonable size battery, let's say 200Ah with an end of discharge cell voltage of around 3.5v

Click to expand...

>Having been correctly balance
The packs were balanced by Batrium at3.7v about 40 cycles ago but have been daily cycle. Longest 'no sun cycle' has been 3 days. Have not observedsignificant self-discharge (e.g. significant difference in lows/highs) over the last 50 days.

>reasonable size battery, let's say 200Ah
Its a780ah@48v battery (made up of 42packs at 260ah each). All cells are used but at tested90% or betteroriginal capacity and within 2-8yrs old per date codes.

>At what SoC are these imbalances being measured
The cycle is
1)inverter on at52v (3.71v/pack)- 59% Soc
2) high (float)at 56.4v (4.03v/pack) - 88% Soc Doesn't get that high this time of year - Averaging 52v high.
3) inverter off at 47.5v(3.40v/pack) -29% Soc

Inverter draws average of 100ma per cell - with high of 200ma per cell for short bursts. The PV array charges at 360mah per cell max - but averages more around 100mah per cell this time of year. These values should not be stressing the indiviual cells that much as the worst ones have a nominal discharge of 500mah capability.

In the example (graph)in earlier post:
-Highwas Batrium battery bank51.1v (3.65v/pack nominal) - bad packs went to 3.75v, 3.73v, and 3.70v - about 0.1 higher than average of good packs.
- Low was 47.5 (3.40v/pack nominal) - bad packs went to 3.26v, 2.29v, 3.30v - about 0.1 lower than average good packs.

So bad packs are swinging .1 higher and .1 lower than group of good packs in between - which are 0.06v withing each other.

Note:The Midnite Classic Aux1 (Inverter) Onis the Classic's 'battery charging voltage' which is higher than actual battery voltage as soon as inverter draws power. This is reasy for 52 vs 51.1 above.

You are balancing at 3.7v ?

What cells are you using ? Edit, ignore that question, I see you are charging to 4.03 (occasionally)

You'll have to explain why you aren't top balancing please and why the conservative 4.03 top target

---

offgridinthecity said:

- Low was 47.5 (3.40v/pack nominal) - bad packs went to 3.26v, 2.29v, 3.30v - about 0.1 lower than average good packs.

Click to expand...

Just to clarify this comment - are you using control logic to cease discharge based just on total pack voltage, rather than pack and any individual cell voltage ?

Sean said:

You are balancing at 3.7v ?

What cells are you using ?

Click to expand...

These are all18650 cells of several brands from alarmhookup and power2spare.

When I got myBatrium BMS (recently)I manually set it to3.7 to do an initial balance of the pack because this was where the pack was at the time. Being December / rainy, there is little sun and didn't want to wait several days to go up to 4.0 to see how it worked.

Continuing to balance (auto and 3.7 top) over the next few days- it was mainly the 5 bad packs that did all the balancingbecause they were trying to swing way high (.1 +)and this just made them go way low as in .15v lower than rest. So I stopped balancing asgood 38 packs are stable (within 0.6v thru full cycle)and let the5 packs swing higher.1 and lower .1.

Its my plan to replace the bad packs and then revisit balancing to see if it tightens things up with all good packs.
Then re test the cells in thebad packs and perhapsfindsignificantly higher internal resistance orless capacity that when originally tested (each cell is marked) as the explanation for the wild voltage swings / loss of capacity compared to other packs.

If you are able to, grid charge the entire pack such that itfullybalances around 4.1v/cell - set a end of discharge pack voltage, and cell voltage that equates to about 3.5v/cell and run a few complete discharge/charge cycles - you'll get a lot clearer view of the relative pack capacities.

I suspect, based on you being confident in the quality of the cells, that you aren't fully charging all of the packs, this needs to be done at quite a slow rate unless you can force cool the mons.

Balancing in the middle Will cause this scenario... Either top or bottom and since you have batrium its top.

The problem is that the Soc vs voltage is all over in the middle. I would say just running the bank 5 min can cause O.05v difference again. Especially when you mix number of cells like you did.
Same number of cells types and capacity on each cell in each pack is essential IF you want to Keep the balance just somewhat equal during a cycle.

+1, you need to be balancing at about 4.1V
Otherwise you'll never have the cells properly balanced where the reactions in the cell have really got to "full" and "absorbed" for a while.
You'll need to hold them at 4.1V until the current drops away ("absorbing").
Given the packs are unbalanced now, you'll have to take care not to overcharge some of them while bringing them all to 4.1V.

Like daromer said "balancing" voltages in the middle of the range is not the same as having balanced SoC. Ideally we'd like the pack voltages to track each other, but it sounds like you have concentrated different types of cells together instead of distributing them evenly around the packs. This is likely to mean differences in SoC when voltage = same.
The pack made of Sony green cells highlights this!

Back to your graphs, even if you started with apparently equal capacities, they don't look like they are the same now (because of going flat first + getting to full first behaviour) . Some cells (Sony green) have probably aged badly &/or been damaged as you're suggesting.

Thing is that optimal is to balance in middle if you want to run it from top to bottom at its average voltage. But IF you do that you can not stare at the numbers since they will change instantly when you start to use them. 50 or even 100mV apart is not unusual on mixed packs.
Balancing in top will ensure that next time you charge up they will be equal at top. If they start to drift apart like yours do when discharging or charging that is due to that you have a mixed pack with uneven capacity.

You say you have them equally balanced? How did you test that? With an Opus or equal that easy is 5-10% off?
Only way to know if they are equal is to run iR and capacity test after they are built. Capacity test also need to be done during several different currents to ensure that the plotted graphs allign. Then you see what I talk about.

So my recommendation is
Top charge and balance.
Test them down to bottom and see how many Ah/kWh you get out of the bank untill first pack drops out. Thats the available capacity you have. If you want an more even end you need to add capacity to the low packs to counter the lower capacity in them.

Very cool, much appreciate info/discussion here. I'll take some steps in the next few months and post back the results.
For the record:
1) Yes - OPUS to test individual cells and added individual to get 260ah +/1 1ah per pack. Did not do pack level tests.
2) Yes - I did mix cells because many purchases over 1.5yrs.. and building the packs one at a time over that 1.5yrs.
3) I am in the process of replacing bad packs / homogenize the 14s string they are in:
There are 3 rows of 14s packs in parallel:
Row1 (14packs) - is stable, made of ebike 36v 4.4ah cells (avg 2150mah). Its 50% LG MF1(s) and 50% Samsung ICR18650-22Ps
Row3 (14packs) - is stable, 100% LGDAS31865 (avg 2200mah). *made big purchase to get all same cells

Row2 (14packs) - is the problem. The 5 bad packs are 100% Sony G5/G7s. The 9 stable packs are 50% Panasonic NCR18650A (avg 2800mah) and 50% LGABB41865 (avg 2600mah).

THE PLAN:
1) Replace Sony G5/G7 packs to homogenize the 14s row of packs. I purchased Panasonic NCR18650A and LGABB41865 to replace the Sony G5/G7s with same 50/50 construction as the other 9 packs. Hopefully, this will stabalize 'Row2' with 14 similar packs.

However, going forward, the best strategy I can do is go target each 14s 'battery' to be same cell/chemistry. So I hope that if I have each 14s 'battery' consistent it will reduce problems even though the 3 14s batteries are hooked in parallel.

2) After #1 - do some explicit 4.1v top balancing and tests you advise and see how it goes.
Assuming the packs run within 0.06 (or even 0.10) of each other up and down along with perhaps minor auto-level/top balancing - I'll relax and go with flow

3) Retest the Sony G5/G7s and see if they are damaged - e.g. retest and look for significantly different capacity/internal resistance.

4) See about getting setup to test the packs as a whole... and perhaps do that in future.

Again, thank you very much for the discussion/info. This project is more than a hobby with big $$ spent and I hope to run this battery bank for many years, add to it, and get my money back eventually. Current projection for overall solar expenditures recovery is 21yrs.