The SOC-OCV curve of LFP battery is non monotonic

[toms]

Can you elaborate? I found 1Amp active balancing effective on my 42 kWh pack compared to the 200 mAmps my traditional shunt based BMS was able to do.

The active balancing keeps the voltage equal, but this doesn’t mean it’s keeping the SOC equal.

This often results in energy being both taken out and added to the same cell in a single charge cycle.

If you don’t have the means to reduce current at the correct voltage for balancing then active is a balancing measure that works at the expense of extra cell activity.

 

[robby]

This relationship is part of the reason active balancing is often a waste of time.

Active balancing is so much better than passive.
The time it takes the cells to balance is just a few minutes vs an hour with passive balancing.

 

[curiouscarbon]

The active balancing keeps the voltage equal, but this doesn’t mean it’s keeping the SOC equal.

This often results in energy being both taken out and added to the same cell in a single charge cycle.

If you don’t have the means to reduce current at the correct voltage for balancing then active is a balancing measure that works at the expense of extra cell activity.

the OP's graph goes up to 3.3280 V

balancing only when multiple cells are above 3.40 V seems to be better than at the range shown in original post.

this makes me want to gather some data too..

3.4 V area would seem to be more monotonic than the 3.3 V area

 

[Substrate]

It may assist others to explain why these findings bother you. It's a DIY forum full of people eager to learn and absorb new information.

I'm going to channel my inner Professor Jeff Dahn :

Op has vendor supplied battery and bms
Also has vendor supplied black-box charge/discharger with special sensor.
Has his own multimeter (possibly high end that tracks similar to special vendor supplied sensor)

No reason given (other than tripping the bms) on this VERY minute non-monotonic response. I don't know of any bms that would trigger on monotonicity. Special circuit? Obsession with monotonicity?

Purpose? Non-monotonic response might be being used to try and determine cell SOH, nevermind the bms trip or SOC. This SOH may be dependent on chemical degradation like SEI growth and plating. Small amounts of non-monoticity seen as minor voltage-diffusion variations when intercalating through the SEI layer might be used as an indicator.

These "levels of non-monoticity" to determine SOH are not known. At least to us.

The real problem presented here is that the vendor-supplied battery/bms combo is tripping on it. Like me.

 

[curiouscarbon]

non-monotonicity makes sense to me especially when considering SOC estimation. any voltage region with low local monotonicity would need to rely on coulomb counting moreso than say 3.5v or 2.8v extremes

higher local monotonicity -> easier to estimate SOC from voltage

 

[Ampster]

The active balancing keeps the voltage equal, but this doesn’t mean it’s keeping the SOC equal

I have never expected the SOC to be equal. My limited knowledge suggests that unless one has perfectly matched cells equal voltage in the knee would not mean equal SOC.

 

[toms]

I have never expected the SOC to be equal. My limited knowledge suggests that unless one has perfectly matched cells equal voltage in the knee would not mean equal SOC.

Assuming a battery that is top balanced - the top balance voltage is then 100% SOC for each cell (actual useable capacity may differ).

As you discharge and return to that SOC setpoint (or 80-90%), the voltage of each individual cell will not be the same throughout the cycle.

If you allow the battery to cycle back to it’s original balanced state with no intervention, you will minimise the cell ion activity which will maximise lifespan.

With active balancing you are transferring energy between cells throughout your balancing range - you have no idea if the voltages will naturally converge or not. Experience has shown that it is normal for voltages to diverge and then converge as the cell cycles. To minimise cell activity you need to only balance when the first cell reaches its maximum voltage. If you do this with active balancing it is fine, but that doesn’t work with many BMS, so people set the balancing to start early which increases cell activity.

 

[Substrate]

@Leon_Xp C'mon man, don't leave us battery nerds hanging.

Why is this such a concern? Another similar thread suggested noise. This thread suggest an SOC reading tripping a bms, cause by non-monotonic response.

Is this a case of:

1) Stump the readers
2) Use of a special vendor monotonic echo-response-network sensor that you have duplicated results with a simple multimeter where the SOH (health) is directly related to the OCV model?

3) Forcing a freak like me to channel Jeff Dahn, and make up stuff based on my own lack of chemical engineering?

Come clean, bro. It's killing me.

 

[Ampster]

Forcing a freak like me to channel Jeff Dahn, and make up stuff based on my own lack of chemical engineering?

At least Jeff Daun can explain what is important and what is not.

 

[Ampster]

With active balancing you are transferring energy between cells throughout your balancing range - you have no idea if the voltages will naturally converge or not.

That may be true but mine seem to stay withing .001V of each other from resting to 3.45 CV stage. I do have three LF280 in parallel in a 3P16S pack.

 

[toms]

That may be true but mine seem to stay withing .001V of each other from resting to 3.45 CV stage. I do have three LF280 in parallel in a 3P16S pack.

If that’s the case you don’t need balancing at all. My pack often goes for years between the times i activate the balancing function.

 

[Substrate]

At least Jeff Daun can explain what is important and what is not.

Oh yeah, total respect for the guy. His classic "Why lithium batteries die and what to do about it" video is unforgettable.

I'm probably reading into the thread topic more than I should, like there is some critical detail that is being held back.

Or it's a case of "Does your bank suffer from monotonicity? My monotonic healer circuit is now available!"

 

[Ampster]

If that’s the case you don’t need balancing at all.

I have disabled balancing on my Orion BMS but might also disable the JK active balancer as well and see if there is much drift. I do think three LF280 cells in parallel probably does reduce the individual cell deviations to an average amongst the three in parallel.

 

[toms]

At least Jeff Daun can explain what is important and what is not.

I often struggle to explain myself well. This topic i would place in the “minor detail” column of importance. It is interesting to discuss nonetheless.

The voltage / SOC thing peaked my interest as i was recently showing a friend how far apart my cells were when the first cell reached 3.40V (at low C rate), then how they all ended up at 3.45V together with no interference.

If i’m charging at a higher C rate they stay much closer in voltage as they travel the SOC range,

 

[Substrate]

I often struggle to explain myself well. This topic i would place in the “minor detail” column of importance. It is interesting to discuss nonetheless.

I understand. I feel the same way. I have a tendency to go energizer bunny - on and on and on. I'm my own worst critic, believe me.

As a stress relief from all the precision, I just went back to the EV days with a 4S pack I built with external lvd, and no bms. Fused of course. Top balanced the cells, and charger CV is set so that I'm always between about 3.45 and 3.55 between cells in series use. 100mv or under balance, I'm good! Heretical I know. Non-critical application and I'll keep my eye on it.

Hah, that's a "danger zone" project category now.

 

[Leon_Xp]

@Leon_Xp C'mon man, don't leave us battery nerds hanging.

Why is this such a concern? Another similar thread suggested noise. This thread suggest an SOC reading tripping a bms, cause by non-monotonic response.

Is this a case of:

1) Stump the readers
2) Use of a special vendor monotonic echo-response-network sensor that you have duplicated results with a simple multimeter where the SOH (health) is directly related to the OCV model?

3) Forcing a freak like me to channel Jeff Dahn, and make up stuff based on my own lack of chemical engineering?

Come clean, bro. It's killing me.

I think this is the result of a complex electrochemical reaction inside the cell, and I will ask someone with in-depth research in this field, such as Dahn.

 

[DavidKelly]

When I tested OCV on a 100Ah LFP cell, I found that its SOC-OCV curve was non monotonic at 70% - 99% SOC. Is this a problem of cell production process?View attachment 115866

Very interesting.

I’ve seen this before on SOC OCV curves from some manufacturers (even Eve). Thought it was a typo but maybe it is true. Never had to privilege to try it.

Thanks for sharing.