LFP memory effects thread

[AGM]

Originally Posted by OceanSeaSpray

C/3?? You must mean C/30. In all cases, definitely not once the memory effect has deeply set in. I have tested raising the voltage and there is no easy answer in this direction. The battery is too resistive.

Yes, sorry my bad - well spotted.

I'd like to confirm some 'resistivity' which I noticed on a new 4s4p 100Ah LFP battery.
I've no idea if that was the onset of memory, or something else like incomplete formation charging by the factory.

In a nutshell, I saw a noticeable resistance to charge at 3.65Vpc.
But only in one of the 4 cell groups.
This resistance went away after applying 3.75Vpc for a couple of minutes.
It's an almost brand new battery which has only seen a few partial cycles on the test bench.

A single 4p group of cells didn't want to rise much beyond 3.45Vpc using a 2A balance charger whilst the other 3 groups of the 4s string where happily charging to and sitting at 3.65Vpc. They were kept at that level for over an hour by the balance charger with me watching and hoping for the lagging group to catch up but that didn't happen.
I first thought one of the cells is a dud as this looked like increased self discharge in the region beyond 3.45Vpc. Every time the pulsed balance charger pushed the volts up somewhat, it dropped back down again to the same level.

And when taking the affected group to 3.65Vpc with a power supply, the current didn't want to taper off below an amp like on the other groups.
But it did something.
Using the balance charger again, the affected cell group was now hovering around 3.55Vpc, a 0.1V improvement.

Then, encouraged by research paper findings I took the culprit to 3.75Vpc.
Interestingly, I saw a few amps lasting for a minute or two and lo and behold current tapered off fairly quickly after that, well below under an amp, mirroring the behaviour of the other cell groups.
During the following load test the cell groups stayed in perfect balance.

AND best of all, the 2A balance charger (augmented by an ordinary 30A charger set to 14.4V) literally didn't meet any resistance at all taking the 4 groups to 3.65Vpc +/-1mV! in a timely manner.
That was last week and the battery is now resting at ~50% SOC.
I'll give it some more time (looking for potential unequal self discharge) and then do another balance charging session to see if, and how easy the cells reach 3.65Vpc again.

OceanSeaSpray, what charging rate do your cells accept @ 3.75Vpc and do you see any current taper at all at that voltage level?

Next question if you don't mind, how to distinguish between reversible (due to erasable memory) and irreversible (calendar and cycle aged) capacity fade?

[john61ct]

You are assuming the this memory effect cannot cause irrecoverable capacity loss?

I believe that is one hypothesis in question here, and I'm leaning toward no, it can indeed.

Thanks for those charging details, very interesting.

What Ah size is that pack?

And I believe most RC type node - balancing chargers put out 6-10A per cell (or 1S-group), why are you using one that only does 2A?

[AGM]

Originally Posted by john61ct

You are assuming the this memory effect cannot cause irrecoverable capacity loss?

Yes by definition found in the literature, rather than by long term experience.

Originally Posted by john61ct

I believe that is one hypothesis in question here, and I'm leaning toward no, it can indeed.

Then the question arises how to distinguish between memory and calendar/cycle aging?
If we can find a way to erase the reversible loss (the erasable memory), the remaining loss constitutes the irreversible part.
It doesn't really matter if the irreversible part contains some 'irreversible memory' or not.

Well, thinking about it, it does matter and I now understand the essence of your question.
You don't want to be labelled as one of the perpetrators of the slow-death-by-incomplete-charging phenomenon

Originally Posted by john61ct

Thanks for those charging details, very interesting.

What Ah size is that pack?

And I believe most RC type node - balancing chargers put out 6-10A per cell (or 1S-group), why are you using one that only does 2A?

100Ah.
My balance charger can do up to 4A.
But I'm using 2A for the mundane reason my 12VDC mains power supply required for the balance charger is a bit weak.

[john61ct]

I believe the memory effect's causing capacity loss is easily **prevented**.

I think the **recovery** of such losses once they've developed (allowing them to be labeled "temporary")

requires catching it in time, and yes, perhaps after months of precise repetition, some of those losses may prove to be unrecoverable.

I would not personally start charging - during normal cycling - to the absolute maximum "stress ratings" spec'd by the makers, even if I saw the memory effect occuring in my own banks.

That would be like eliminating free speech just because paid-for PR/propaganda has poisoned the internet, where the cure would be more harmful than the disease.

____
Most decent balance chargers require 24V or even 30+V input to deliver rated output.

Serial-connecting a couple server PSUs, or adjustable step-up converters are two very cheap solutions.

I've taken Cpt Pat's warning about too low a C-rate being problematic, especially getting below what should be a reasonable endAmps stop-charge level.

[OceanSeaSpray]

Originally Posted by AGM

OceanSeaSpray, what charging rate do your cells accept @ 3.75Vpc and do you see any current taper at all at that voltage level?

Next question if you don't mind, how to distinguish between reversible (due to erasable memory) and irreversible (calendar and cycle aged) capacity fade?

My cells get into absorption starting at about 0.65C around 60% SOC. Then they stay in absorption for 1.5-2 hours to reach ~75% SOC and by then the current is down to C/40 and voltage at 3.5V/cell. Increasing the voltage only increases the current very marginally (cells too resistive), so I decided it wasn't worth the added stress. I am going to try to erase the memory by continuing charging at this voltage first and I will only increase the voltage later if I have to.

Regarding differentiating from memory effect preventing recharging and permanent capacity loss, see my last post before this one.

[AGM]

Originally Posted by OceanSeaSpray

My cells get into absorption starting at about 0.65C around 60% SOC. Then they stay in absorption for 1.5-2 hours to reach ~75% SOC and by then the current is down to C/40 and voltage at 3.5V/cell. Increasing the voltage only increases the current very marginally (cells too resistive)

Compare this to one of my brand new batteries.
Bulk charging a new battery @ 0.28C to 13.75V (3.438Vpc), current taper duration is 2 hours down to 0.015C.
Then the voltage is increased in steps of 0.16V.
The current change becomes weaker with every step as the SOC approaches 100% (dQ/dV reaches its minimum).

If your cells only respond weakly to a voltage increase it could simply mean the battery is already near full capacity.
At this stage another 25% doesn't seem possible.

On my battery the current taper during the 13 minutes (diagram), and between the voltage steps is still apparent. So there's another indicator the battery isn't fully charged yet.
The slope gives a clue how much more charge the battery is able to absorb.
But I know this particular (new) battery is very close to 100% when I see this behaviour.

It would be interesting to see how your battery responds to the same series of voltage steps. I'm guessing if the incremental capacity (dQ/dV) follows the same downward trend, only a fraction of the observed capacity loss is due to memory.

[OceanSeaSpray]

Originally Posted by AGM

Compare this to one of my brand new batteries.
Bulk charging a new battery @ 0.28C to 13.75V (3.438Vpc), current taper duration is 2 hours down to 0.015C.
Then the voltage is increased in steps of 0.16V.
The current change becomes weaker with every step as the SOC approaches 100% (dQ/dV reaches its minimum).

If your cells only respond weakly to a voltage increase it could simply mean the battery is already near full capacity.
At this stage another 25% doesn't seem possible.

On my battery the current taper during the 13 minutes (diagram), and between the voltage steps is still apparent. So there's another indicator the battery isn't fully charged yet.
The slope gives a clue how much more charge the battery is able to absorb.
But I know this particular (new) battery is very close to 100% when I see this behaviour.

It would be interesting to see how your battery responds to the same series of voltage steps. I'm guessing if the incremental capacity (dQ/dV) follows the same downward trend, only a fraction of the observed capacity loss is due to memory.

A battery is close to 100% when it settles at an OCV also saying so following charging. Mine charges up to ~75% of nominal SOC according to Coulomb counting and the OCV then indicates about the same: it is not full. Furthermore, it is perfectly able to keep charging further, but very slowly, because it has become very resistive. I am getting tired of repeating myself.

I had written just before that "increasing the voltage only increases the current very marginally (cells too resistive)", so you don't need to ask "what would happen if I made little voltage steps". You keep asking for answers that were provided in the previous post. Please read the thread, not just what you are typing.

Furthermore, since your batteries are "brand new", by definition you have nothing to contribute to this thread, which is about clearing the cell memory once it has set in.

Happy to be patient and explain things more than once, but it has its limits.

[newhaul]

Ok so what I am getting is there is actually a need to get an actual full charge on the bank at least once every other month. Preferably monthly to avoid the capacity loss pitfall. Or is the capacity loss that is being experienced a more natural part of the aging process of the bank?
Remember I am just over a year into my bank thankfully not showing capacity loss yet .
Hoping to avoid this if it is possible.

[john61ct]

Originally Posted by newhaul

Ok so what I am getting is there is actually a need to get an actual full charge on the bank at least once every other month.

Well some advocate just using the vendor maximum stress ratings **every** cycle.

But for those who don't, believing a gentler approach avoiding the shoulders extends cycle lifetimes, yes, that is one approach I think would likely work, **if** this memory effect would appear otherwise.

Another approach is to just **vary** the stop-charge setpoints a bit, sometimes go a point or two higher or lower in voltage, sometimes charge CC-only, otherd put in some Absorb time, etc.

Personally I think just doing regular **monitoring** of the charge cycle, capacity tests etc, is enough, so you will detect any memory effect, catching it early is the key.

[newhaul]

If possible could someone that is actually experiencing this issue please weigh in on what seems to be working to mitigate the issue.

Seems to be the last several pages of just repeating but with more technical data points .

[AGM]

Originally Posted by OceanSeaSpray

A battery is close to 100% when it settles at an OCV also saying so following charging. Mine charges up to ~75% of nominal SOC according to Coulomb counting and the OCV then indicates about the same: it is not full. Furthermore, it is perfectly able to keep charging further, but very slowly, because it has become very resistive. I am getting tired of repeating myself.

Ok, I think a brief description of top memory is in order.
Reasearch has shown there's a developing 'voltage hump' starting near the (lowish) voltage target, if charging is terminated before reaching 100% SOC.
This hump becomes more pronounced with increasing cycle count.
You describe this as 'cells have become very resistive' which is in line with the above.
And here's the important bit:
The voltage hump actually looks more like a layer extending all the way up.

The process of erasing memory restores some of the lost capacity by overcoming this voltage layer so that current can flow more freely again. It's like developing a pathway to the temporarily locked section of total capacity.
An icebreaker is a good analogy.
Of course it takes energy to do so, part of which is the required voltage increase.

And that's what led me to do the stepped voltage experiment.
Energy (As*V) is transferred instantly to the battery in every step.
In a battery with no memory, there's no extra voltage layer to be overcome, no ice to be broken.
Therefore an SOC of close to 100% can be readily achieved with a relatively low charging voltage, like 3.44Vpc.
Very little additional charge can be delivered by stepping the voltage higher as can be seen in the diminishing current peaks.

I'm guessing, had there been memory the observed current peaks would grow on every voltage step, as would the baseline current.

Unfortunately I cannot repeat the experiment with a memory affected battery.
And that's why I'm here, hoping someone with a suitable charger (ideally capable of closed loop charging) is willing to join in.

[john61ct]

Another factor is, just because coulombs are leaving your charger and **appear** to be entering the battery

does not mean any actually usable energy is being stored in the battery.

Especially as you go past the voltage shoulder, even more so in CV stage at low current rates (under say .03C)

higher and higher percentages of inbound coulombs are actually dissipated as wasted chemical activity and rising internal temperature.

This can be verified with precise load testing - the only truly accurate measure of SoC between 0 and 100%. Less accurately with precise & accurate measurement of isolated resting voltage, but that's almost as tedious.

My main point is, you may think the SoC gap between stop-charge profiles X and Y is 5-10%, when in fact, only counting **actually usable energy** the gap is well under 2%.

So yes, go higher voltage maybe higher current to "break through" try to recover capacity. Just don't overestimate how much of that energy is actually getting stored as opposed to just adding surface charge.

[tanglewood]

Originally Posted by OceanSeaSpray

It is quite clear for me. When I have charged up to 76%, the voltage is up to target and the current so far down that it is taking forever to continue, so classic termination has been reached. The OCV then settles down to something also indicating the same SOC.
The OCV is an indication of the distribution of the charge carriers in the cell and it says there are plenty more left to displace.

If the cell was suffering from permanent capacity loss due to loss of lithium, the OCV would indicate a high SOC following the end of charge. Furthermore, no matter what I do, the current won't taper down to zero. The cells keep charging gently at their own pace. It looks like it is a matter of continuing until they really reach the top... so far it hasn't been practical to do it and the memory effect still stands.

OK, very interesting.


You are defining/detecting this memory loss as when SOC as determined by charge voltage and return current is higher than SOC as determined by post-charge OCV. Correct?

[rgleason]

I'm guessing if the incremental capacity (dQ/dV) follows the same downward trend, only a fraction of the observed capacity loss is due to memory.

AGM - Is one of the things you look for to determine if a battery is nearing 90-95%, that "dQ/dV"? In other words, when you increase voltage in a small jump, the spike in Current becomes less and less responsive. The spike is less and the drop from that spike becomes more gradual because the battery resistance has started to go up a little?


I am having difficulty with OCV, what is it?

[john61ct]

Voltage measured when the battery is isolated.

Colloquially often used interchangeably with "resting" voltage, but that has an additional time element since dis/charging was last active.