How to balance LFP in series? (Updated with cells charged individually)

[Michael9dk]

Background:
20 x EVE LF105 in serie.
They have never been top-balanced.
Discharged op to 60A between charges, for 6 months without issues.

Then one time, it was discharged 74A (3.199V), and I drowe until the BMS shutdown at 83 Amps. I had no other option to get home, and engaged Emergency-mode (well aware that I would fry a cell or more).
One cell was down to 0.7V at light load (and was replaced when I got home).

But what I really noticed, is that about 50% of the cells, always have been out of balance (150-200mV and more), during charge.

My JK-BMS has been balancing with 0.6A for about 48 hours with no significant difference...

I'm waiting for a 2x60V/10A power supply to arrive, and intend to charge the battery with a lower voltage and a very low current.
Would that work if you have the time?

How would you balance a 20-cell series battery WITHOUT disassembling it?

 

[dominik h]

Hook up a charger and let the BMS do its job.
In 48h the BMS can only transfer about 24Ah.
If your battery is completly debalance it will probably take much longer.
But it is also possible that same cells are damaged and have a higher self discharge.

 

[Michael9dk]

My charger turns off when the BMS hit OVP, and need to be unplugged to start again.

Once half the cells reaches 3.65V, the rest is about 3.40V.
After 15 hours rest with balance (>5mV diff. @3.34V) they all settle at 3.333 to 3.337V.

Turning the 18A charger on again, it takes like 10 minutes to hit OVP again with a cell difference of 270mV.

So I'm thinking that floating at 1A/69V (3.45 cell) would let the balancer do its work without hitting OVP instantly?

 

[Zambam]

Sounds like you have more bad cells? To find the bad ones, check each cell's capacity and internal resistance. You can do that w/o disconnecting the series connections. You will need a single cell charger and a battery capacity tester.

edit: Does your EVE LF105 cells have screw terminals. If so, check all connections for corrosion and tightness. That's a big battery! 6.7 KWh! What's it used in? Any photos?

 

[Michael9dk]

I really hope not. They're only 6 months old with ~15 charges. Original EVE from Nkon.

Checked my screenshots of the first charge, and they were off by 220mV, when the charging finished.

My guess is the cells are from 2 different batches. Looking at the deep discharge, SoC were about 20Ah apart (2.66 vs 3.18V at 82Ah discharge).

I'll do the slow charge and see if they will even out, first.

 

[Michael9dk]

Did a slow charge, and can see 10 cells fully charged. The other 10 are really close to each other.
So will have to top those (thankfully 9 are connected in series).

edit: Does your EVE LF105 cells have screw terminals. If so, check all connections for corrosion and tightness. That's a big battery! 6.7 KWh! What's it used in? Any photos?

Yes they're screw terminals. There's no visible corrosion, and no issues with delivering 70 amps.

It's used in a 3-wheel cabin scooter. Living on top of a steep hill, I need some juice.

 

[Michael9dk]

Here its partially assembled.
Used reinforced tape to keep the cells together with light compression. Pack is protected with plexiglass outside.


Almost mounted.

 

[Michael9dk]

Okay,
Lets call them Group A (the 10 good cells) and B (the 10 bad).

Now I've topped all cells to 3.650V, after a full 20s-charge.

Group B nedded about 10-20Ah topping, so they clearly weren't fully charged in 20s.


After 2 days rest with no load:
Group A is resting at 3.60V ±3mV (and one at 3.574).
Group B is resting at 3.34V ±5mV (with three at ~15mV higher).

I haven't tested internal resistance, but expect a difference between group A and B.


Yes buying 10 new cells plus some extras, would be best. But then I might as well get 20, and parallel a good and a bad cell.

But what if added a resistor across group A, while charging?
As in bypassing just enough current, so they reach 3.65 at the same time - like a passive balancer.

 

[Michael9dk]

Prior to topping each cell, I noticed a different voltage drop under light load.

Test case: 8 days since last charge. 20Ah used.
At 12.6A load compared to 0.6A, Group A dropped 20mV, and Group B dropped 30-40mV.

This might be an issue since that load is very light (normal max load is up to 70A/83A). What do you think?


Edit:
Currently collecting charge-data from my 3 unused spare cells. 2mV difference after adding +80Ah juice.
They were down to 3.250V after internal leaking for 12 months in storage. That is quite impressive. Now if all my cells behaved like these, I would be a happy man.

 

[Nix Liou]

Prior to topping each cell, I noticed a different voltage drop under light load.
Test case: 8 days since last charge. 20Ah used.
At 12.6A load compared to 0.6A, Group A dropped 20mV, and Group B dropped 30-40mV.

This might be an issue since that load is very light. What do you think?

I mean it sounds like you simply have some bad cells. If you want to get the most you can out of them anyway, you would want to measure the IR and capacity of them, and mix the bad and good cells to form groups that have the same average. Then you would probably want to get a 5A active balancing board. If the voltage sag under such a small load is double for one group vs the other, then effectively the IR of each group is significantly far off.

However if group A and group B aren't at the same voltage, then different levels of voltage sag doesn't tell you anything except that they are imbalanced. The magnitude of voltage sag is a function of state of charge, and where it is on the discharge voltage graph, basically.

 

[Michael9dk]

That would mean 10 bad cells out of 20. This is not what I expected from grade-A EVE cells from Nkon.nl (maybe they picked from 2 batches).

I really don't like that facts are pointing towards replacing half of them, before they reach 5% cycles.

However if group A and group B aren't at the same voltage, then different levels of voltage sag doesn't tell you anything except that they are imbalanced. The magnitude of voltage sag is a function of state of charge, and where it is on the discharge voltage graph, basically.

In that case they were resting within 39mV, but at up to a estimated 20A different capacity.

Buying a electronic discharger is not an option, since it will take multiple days to test each cell.

I will have to look up on how to measure IR accurately, the DIY style.


What about if added a resistor across group A, while charging?

 

[Nix Liou]

That would mean 10 bad cells out of 20. This is not what I expected from grade-A EVE cells from Nkon.nl (maybe they picked from 2 batches).

I really don't like that facts are pointing towards replacing half of them, before they reach 5% cycles.

In that case they were resting within 39mV, but at up to a estimated 20A different capacity.

Buying a electronic discharger is not an option, since it will take multiple days to test each cell.

I will have to look up on how to measure IR accurately, the DIY style.


What about if added a resistor across group A, while charging?

Yes, you can manually discharge the higher voltage group. You will need to discharge a lot of amps. You should use an active balancer so that you can discharge the higher voltage group and use that energy to charge the lower voltage group.