1400 Ah Lithium from cells

[nebster]

I agree, with maybe the exception of the "3000 to 5000 happy cycles." Maybe it will be closer to 1000 happy cycles if surprises happen. Personally I think it is still worth it, for the benefits.

[john61ct]

Originally Posted by nebster

I am a skeptic until the data is present.

I am not skeptical that the memory effect is there,

I think waiting until repeated 3.7Vpc with Absorb charges is more harmful than I'd prefer, and

I think it very worthwhile taking reasonable measures to prevent it getting to that point.

My understanding is, it is the result of always stopping at the same place, and the solution involves going a bit higher than usual.

Note it is **just not possible** to use SoC directly as a benchmark to regulate charging, as

A. there is no trustworthy way to directly measure it, and

B. the 100% top point is arbitrarily defined, the manufacturers' spec is already far above what is safe for longevity I consider it worth ignoring.


My (personal) approach (so far) would be to continue using close to 3.45V for day to day, with "just stop", no Absorb, no Float.

With my newish CALBs, that is well over their rated AH capacity, and I believe within 2% of going to full .025C endAmps at 3.6V.

Therefore that is **my** 100% for SoC metering reset purposes.

Continuing to apply any level of voltage after **your** definition of 100% Full has been reached is not the cause of memory effect, and if that is lower than (apparently) 3.6V it is indeed part of the cure.

To prevent memory effect in my case, I will simply vary the end-charge setpoint, both above and below my usual 3.45V-justStop, say at least 1/3 of the bank's cycles.

1. I can hold Absorb until endAmps rather than just stopping at 3.45Vpc.

2. I can adjust the setpoint, up (between 3.5 and 3.6), and down (between 3.3 and 3.4).

3. I can add Absorb time to any of those, more often to the lower points than the higher ones.

Finally, if/when I ever do manual balancing (I don't do anything for that while in use) I may go up to 3.65 - or even 3.7, for short times only and as always, immediately applying a load thereafter.

[nebster]

Originally Posted by john61ct

I am not skeptical that the memory effect is there,

Me either. As I mentioned above, I can see it after just 15 or 20 cycles.

I'm skeptical that it actually matters much, or that the hassles of trying to mitigate it are worth what may be a small return. I'd like to see more data from packs that have suffered the consequences, whatever they may be.

My understanding is, it is the result of always stopping at the same place, and the solution involves going a bit higher than usual.

So, your thinking is that either of the two approaches above (if they were actually possible) would contribute equally to the problem?

[QUOTE]
Note it is **just not possible** to use SoC directly as a benchmark to regulate charging, as

A. there is no trustworthy way to directly measure it, and

B. the 100% top point is arbitrarily defined, the manufacturers' spec is already far above what is safe for longevity I consider it worth ignoring.
[QUOTE]

Yes, that's why I said a theoretical drift-free SOC estimator. It's not the absolute SOC that matters, it's bringing the chemistry back to the same equilibrium potential. A perfect coulomb counter would allow that, and we have to have a baseline to discuss and understand the effect -- or at least to ask the question I did.

With my newish CALBs, that is well over their rated AH capacity, and I believe within 2% of going to full .025C endAmps at 3.6V.

Therefore that is **my** 100% for SoC meteringeset purposes.

Fair enough, but let me throw something out there. You say "newish CALBs," but you don't even say which CALBs. Presumably they vary a little bit.

And whether new or not, my observation across my 112 cells is that I see a lot of variation. In large-series strings, I see significant variation in top-end voltage, being bottom-balanced. If it is a requirement to obtain the "varying stopping voltage" to avoid the memory effect, I'm unwilling to trade low-end discharge safety for it. And I don't see a way otherwise, without active management, to get each cell in a series string to these unique and varying voltages reliably.

So here again, I guess I want to see some extraordinary data before I go to extraordinary lengths to change design.

Continuing to apply any level of voltage after **your** definition of 100% Full has been reached is not the cause of memory effect, and if that is lower than (apparently) 3.6V it is indeed part of the cure.

I'm having trouble understanding this sentence, but it might just*be me.

Let's say my definition of "full" is "85%" (whatever that means), and I cannot even get to 3.6v to achieve it -- because my maximum charge rate is low enough relative to pack size that I stay in CC the whole time -- then something in your statement has to give. Right?

Cheers!

[OceanSeaSpray]

Originally Posted by nebster

Well, it sounds plausible, but we don't know about the mysterious 100+ RVs gentleman, since he's not here to ask more questions. We also don't hear about this mysterious "catastrophic" knee-shortening in other installs, and I am a skeptic until the data is present.

(If, in fact, the "force charging" every three years is necessary, by the way, I am not sure if that constitutes a catastrophe.)

Not in the business of naming and finger pointing. It is unfortunate that he took his report down. When your bank is so stuffed that it has no capacity left and complete dismantling and several days offline are needed to "recover" it, it is a fiasco as far as I am concerned. A good lithium system is one you can forget.

Originally Posted by nebster

Let me revisit this by posing a hypothetical pair of charging approaches:

Approach 1: charge at 3.55V (or higher) using a *theoretically* perfect, drift-free coulumb counter, stopping charge at 80% SOC -- or pick any SOC, really -- every time, using a current that is low enough that we stay in CC.

And... how do you propose to do that? What if your cells go over your 80% figure without asking for permission long before the voltage is 3.55V? You don't understand charging properly. There is no relation between voltage and SOC. Time, C-rate, voltage and the condition of the cells all feed into what SOC is reached.

Originally Posted by nebster

Approach 2: charge at a lower voltage, say, 3.40V, with a CC/CV regime, but with a taper and current- or time-based termination that also arrives at the same 80% SOC. (Or again, some other SOC meaningfully less than 100% and the same as above, for the sake of this thought experiment.)

That you could do with fresh cells, but it would take a long time because you haven't got enough charging voltage to be efficient.

Originally Posted by nebster

Question: do both batteries end up with the same "memory hump" and knee-lengthening over time? Put another way, is it the act of always stopping at a certain SOC, or is it the "trickle-charging" at the end of a slow charge that does it? Do we know?

They would both end up with the same memory effect. It stems from having failed to displace all the charge carriers.

The effect of trickle charging is overcharging and depositing lithium where it should not be, like on the edges of the plates, and can't be recovered afterwards ever. Trickle charging will erase the memory first if there is one, and then damage the cells.

[nebster]

Originally Posted by OceanSeaSpray

And... how do you propose to do that? What if your cells go over your 80% figure without asking for permission long before the voltage is 3.55V? You don't understand charging properly. There is no relation between voltage and SOC. Time, C-rate, voltage and the condition of the cells all feed into what SOC is reached.

Charge CC and then to a CV max 3.55V.

I could have put it more precisely, but it gets really tedious.

I can't believe you would say "there is no relation between voltage and SOC." Of course there is, in steady-state open circuit?

That you could do with fresh cells, but it would take a long time because you haven't got enough charging voltage to be efficient.

I guess we're getting to the heart of it here: what is suitably "fresh," then. My cells are plenty fresh enough. Will they still be, at 500 cycles? 1000? That's exactly what I'm asking, but no one seems to have anything to show us.

They would both end up with the same memory effect. It stems from having failed to displace all the charge carriers.

That's good to hear. It means we can adopt a charge strategy that makes the most sense, orthogonal to any memory effect creation it may have.

Trickle charging will erase the memory first if there is one, and then damage the cells.

So... how fast does it erase the memory? Can we just "overcharge" in tiny amounts and then observe the results? Does it take one cycle? Or are the number of cycles required proportional to the size of the effect that has built up? IOW, is this actually a big deal, or is it a maintenance step?

[CatNewBee]

Have you all read the same article?

The memory effect I read about is resulting of power recuperation while breaking and acceleration afterwards in a e-mobility use case with high currents. The effect is a higher cell voltage in relation to capacity than expected for the SOC due to the mentioned "memory effect". It is not about normal charge cycles and no-charge periods.

The memory effect was reversed not by charging - it stops because of the high cell voltage - but either by time with no charge, that cures the memory effect, the cell balances itself over time if there is no charge source or quicker by a deep cycle - discharge of the cell, so all voltage cluster get consumed and drained.

On a vessel / house power wall this effect will not be an major problem. There are no QUICK charge/discharge cycles with no rest for the cells. The cells will follow a daily use pattern, with a slow recharge with varying current by solar (way less energy going into the bank at a time) and some heavy loads around breakfast, lunch and dinner followed by light discharge / rest during the night hours that allow a healing of the memory effect - if any - during the day use.

There is no quick start-stop heavy load/recuperation at all.

Regarding the RV installation. There is no data available about the BMS, chargers, use pattern, depth of cycle and heat involved, nor about the way the cells were installed (vertical vs. horizontal), wired, monitored and balanced, nothing about alarms low voltage or over-charge, nothing about capacity and currents (A/C-heavy use, heat in the battery compartments...).

There can be various issues with the install and the use / abuse of the bank - I would not blame the memory effect on LFP failures - especially if the fleet was used as rental RV's chances are the customer do not care much about battery health.

If you really care about memory effect AND want to mitigate it by charging - the option is to keep the charger at charging voltage and let the cell absorb current until it is really full - that is called FLOAT - If you do not want FLOAT and insist on turning off charger your second option to give sufficient time to charge to a higher SOC is to use ABSORPTION phase for a longer time by an adequate voltage instead of dropping the charger when absorption level reached...

Otherwise you will not mitigate the memory effect, but amplify it.

Anyway, I am pretty sure this effect is a minor issue in a house bank, there are no quick high-current subsequent cycles with a overall drain of the battery like in automotive use.

[OceanSeaSpray]

Originally Posted by nebster

Charge CC and then to a CV max 3.55V.

I could have put it more precisely, but it gets really tedious.

I can't believe you would say "there is no relation between voltage and SOC." Of course there is, in steady-state open circuit?

In most instances, your 80% SOC will be reached before the voltage has had a chance to climb up to 3.55V, unless the C-rate is very high. Unless the voltage is so low that it can't charge a cell, there is no relation between voltage and SOC! There are very important things here about charging that you haven't understood.

Originally Posted by nebster

I guess we're getting to the heart of it here: what is suitably "fresh," then. My cells are plenty fresh enough. Will they still be, at 500 cycles? 1000? That's exactly what I'm asking, but no one seems to have anything to show us.

Fresh just means "not messed up by repeated partial cycles and systematic inadequate charging". The cells must behave normally.

Originally Posted by nebster

That's good to hear. It means we can adopt a charge strategy that makes the most sense, orthogonal to any memory effect creation it may have.

Memory effect is a part of the reality of operation with fractional cycles and opportunistic recharging that often fails to complete. We have to live with the fact that it happens. What matters is being able to overcome it and recover from it. This is what rules out the "low voltage" weak charging strategies that sound great at cycle #1 with new cells. It only seems to take 3.5V/cell to sustain healthy operation on the long run, even when every cycle doesn't get there. In winter here, we can go through a few months with very few or no full recharge cycles. In the spring, when the solar picks up, the battery health comes back to what it should be with only 10 minutes of absorption at 3.5V/cell. If the first cycle doesn't fully succeed, the next ones gradually do the job and we don't stress anything.

Originally Posted by nebster

So... how fast does it erase the memory? Can we just "overcharge" in tiny amounts and then observe the results? Does it take one cycle? Or are the number of cycles required proportional to the size of the effect that has built up? IOW, is this actually a big deal, or is it a maintenance step?

Erasing the memory requires a full recharge, period. Overcharging, i.e. raising the voltage when the current won't flow, is always harmful. It is important to prevent the memory effect to become deeply rooted, otherwise it is going to take a lot of voltage and time and cycles to wipe it out and the cells won't thank you for it.
Having a correct charging regime means that the memory effect can clear naturally through cycling and doesn't become a serious issue.

[arsenelupiga]

read thru this stuff. HUH!

Dont mind learning all sort of knots even colregs, but this electrical stuff is over the top. Rather use very little electricity.

i am going to replace my 480 AH GEL with GEL again when time comes

[nebster]

Originally Posted by OceanSeaSpray

In most instances, your 80% SOC will be reached before the voltage has had a chance to climb up to 3.55V, unless the C-rate is very high.

It was an example chosen to illustrate contrast between Approach 1 and 2. Please don't get hung up on it. Later readers, feel free to substitute "CC/CV to n volts" if it makes it simpler to focus on the distinction.

Unless the voltage is so low that it can't charge a cell, there is no relation between voltage and SOC! There are very important things here about charging that you haven't understood.

I can assure you there are things that I don't understand about charging, but this is not one of them. Your sentence is only true if you put the adjective "charge" in front of "voltage." That's where you lost me.

Fresh just means "not messed up by repeated partial cycles and systematic inadequate charging". The cells must behave normally.

Yeah, there's that cyclic reasoning again. If they behave normally (as opposed to catastrophic knee-shortening, as you described) for a long time, then maybe this is not that interesting?

My cells show a bump in their charge curve that doesn't seem to affect the end result at all. If that's all we have, this is a second-order concern that I think we can mostly not worry about.

Memory effect is a part of the reality of operation with fractional cycles and opportunistic recharging that often fails to complete. We have to live with the fact that it happens.

You write as if there's a "full cycle" coming at some point, but for those of us who choose to end our cycle at a partial SOC, there is no such thing. Unless we manually override. I'm willing to do that, once in a blue moon, if it makes a dramatic difference.

What matters is being able to overcome it and recover from it. This is what rules out the "low voltage" weak charging strategies that sound great at cycle #1 with new cells. It only seems to take 3.5V/cell to sustain healthy operation on the long run, even when every cycle doesn't get there. In winter here, we can go through a few months with very few or no full recharge cycles. In the spring, when the solar picks up, the battery health comes back to what it should be with only 10 minutes of absorption at 3.5V/cell. If the first cycle doesn't fully succeed, the next ones gradually do the job and we don't stress anything.

With a passively (un)managed, high-series-count pack, I don't see it being viable to get all the cells to 3.5V reliably. Unless they're miraculously consistent out of the factory, or top-balanced, or actively balanced, getting ALL cells to 3.5V will push SOME cell or cells way too high. That's what my data suggest, anyway.

Erasing the memory requires a full recharge, period. Overcharging, i.e. raising the voltage when the current won't flow, is always harmful. It is important to prevent the memory effect to become deeply rooted, otherwise it is going to take a lot of voltage and time and cycles to wipe it out and the cells won't thank you for it.

Is it safe to conclude that your first two sentences here are unrelated to the remaining ones?

Regarding the remaining ones, what constitutes "deeply rooted," what did it take to get to that level, and how does the pack behave in that case? These are the data that are absent. Do we have an example of a disaster-case with a real LFP pack in a real boat or mobile application in a low-C ESS performance envelope?

It's great to hear that you see a tiny bump when your batteries sit all winter, one that you can erase with one or two charges. Those of us who are going to operate hundreds of cycles to a stopping SOC that is meaningfully under 100% want to know what the practical consequences may be. Things look awfully benign after ~30 charge cycles, over here in the "lab." When does this memory problem start getting Real? No one seems to know, and you're unfortunately the only one attempting to teach us by talking about it.

[CatNewBee]

I am not buying many of the concerns brought up here. some are relevant, some are not so - but in the end the main goal is to leverage and use the power of the LFP and not to garden the cells all the time and fear the loss of a single cycle or an Ah in life expectancy.

Things break, thing need sometimes fixing or replacements. I consider lead acid installs more troublesome then LFP.

Many cruisers not having an technical background are scared away from LFP when reading this concerns and contradicting recommendations. Just take them with a grain of salt.

There are hundreds of successful DIY LFP installs around in the RV and cruising community, that just work for years.

Look a the Yanmar engine. Impeller may break, engine may overheat, there is a memory effect on wear and tear if the engine overheats, if you not tune the vents properly it can brake. If you conclude it is unsafe to use a diesel engine, because there may be problems, well, you will row over the ocean.

Same for LFP. The discussion is interesting on a academic level and some learning is possible, but in a real long distance cruising this all is set and forget stuff. A lead acid bank can also fail while on the way.

Also battery technology evolves, the replacement will be more powerful, smaller, lighter and safer. By then he current available LFP technology will help power my travels in comfort.

[arsenelupiga]

Originally Posted by CatNewBee

I am not buying many of the concerns brought up here. some are relevant, some are not so - but in the end the main goal is to leverage and use the power of the LFP and not to garden the cells all the time and fear the loss of a single cycle or an Ah in life expectancy.

Things break, thing need sometimes fixing or replacements. I consider lead acid installs more troublesome then LFP.

Many cruisers not having an technical background are scared away from LFP when reading this concerns and contradicting recommendations. Just take them with a grain of salt.

There are hundreds of successful DIY LFP installs around in the RV and cruising community, that just work for years.

Look a the Yanmar engine. Impeller may break, engine may overheat, there is a memory effect on wear and tear if the engine overheats, if you not tune the vents properly it can brake. If you conclude it is unsafe to use a diesel engine, because there may be problems, well, you will row over the ocean.

Same for LFP. The discussion is interesting on a academic level and some learning is possible, but in a real long distance cruising this all is set and forget stuff. A lead acid bank can also fail while on the way.

Also battery technology evolves, the replacement will be more powerful, smaller, lighter and safer. By then he current available LFP technology will help power my travels in comfort.

just in case i change my mind...

where would you recommend talented but electrically semi-literate would start to raise knowledge to the level that Lithium DIY is possible ?

[CatNewBee]

I cannot answer this in general.

My first sources have been various local forums in Germany - there is a vibrant LiFeYPO4 community in the RV area and renewable energy forums with a lot of installations, but this are German speaking sites, so probably not the right source for you.

I am sure there is a community around your location too that can provide help and information. You can read about many stuff around, I see more negative posts in the CF forums than usual, while other sites are way more positive about the technology. I assume this is because some system integrators are more keen to provide out of the box solutions in consulting and installation of ready to use systems from Victron, Mastevolt and all the other LFP manufacturer and trying to scare away DIY approaches to protect their market.

Also in the sailing community there seem to be more people willing to spend lots of money to the refitting industry for turn key solutions. In the RV world there are usually less customization aftermarket provider, there are no ports with services around for the cars everywhere, and people tend to do more technical projects themselves regarding solar, electricity, batteries etc.

Maybe you'll find there some ideas - also renewable energy sites have large communities around with knowledge and experience.

But in the end, you need to sort out what you learn from this sources. There are very professional people, inexperienced folks and trolls as everywhere in social media.

My learning curve was over years monitoring, watching, reading and seeing installs, talking to people around and looking how long this setups really survive. The technology seemed to me too expensive in the first place and potentially troubling, but meanwhile I am quite confident it is the right step forward.

In the end you learn most when trying it / doing it yourself and doing some measurements / proof of concept.

At some point you have to jump in the water and start swimming instead of reading about flotation and different styles other people swim and float.

The good news is, you are not alone. You always can ask questions and get help when needed, discuss your project and get some advice. It is not rocket science either.

In general you need 3 main components:

1.) 4 cells connected in series to achieve a 12V system
2.) a BMS that watches the cells and does the balancing and system protection to prevent over / undervoltage of a single cell
3.) some sort of relay capable to switch off the system controlled by the BMS.

That's all, everything else is the same as with lead acid.

To improve and tune your system you may re-program your charge sources to lower levels to extend the live expectancy of your new battery - but this is not mandatory. You can use GEL settings (the lowest available in lead acid world) or even the storage mode settings without recond program as your bulk charge (13.5V). You may need to disable temp sensors too. because they do not make sense with LFP. If you have already programmable charge sources (like the Victron controllers), you can use the LFP settings or program your own limits. This may even extend the life of your battery. But even if not doing it and going with GEL settings while cruising and turning off everything when not on board your battery will survive many years. LFP's are much more robust than lead acid batteries as long as the cell voltage remains in a state above 3.0V and below 3.5V.

[john61ct]

Originally Posted by nebster

So, your thinking is that either of the two approaches above (if they were actually possible) would contribute equally to the problem?

Sorry I'm confused. Which two "approaches"? To me that implies the attempted solutions rather than causes.

> CALBs. Presumably they vary a little bit.

Not that I'm aware, chemistry is chemistry, all the large LFP prismatics should have the same behaviour afaic. Winston/Thundersky adding Yttrium doping may change things a bit, I believe they can charge at a lower temp.

> I'm unwilling to trade low-end discharge safety for it.
Not sure what you mean.

How low do you let it go, in voltage (light vs heavy loading) and estimated SoC?

I would think in that situation just vary the top/stop setpoints using whole bank or string/pack/block aggregate-average.

> So here again, I guess I want to see some extraordinary data before I go to extraordinary lengths to change design.

Are you trying to keep everything automated? Are these setpoints difficult to adjust?

[john61ct]

Originally Posted by nebster

I'm having trouble understanding this sentence, but it might just*be me.

You had asked if going past 100% was the cause of the memory effect. I stated no, it is caused by repeatedly stopping charge at a "too low" voltage X, and (more so) at the same point every time without variation.

Originally Posted by nebster

Let's say my definition of "full" is "85%"

My definition of 100% Full is "charge to 3.45V and stop". Right now that is higher than the AH rating, call it 100AH, so for now drawing down 15AH after resetting the BM is 85%.

At some point, hopefully many years, the AH capacity will be lower, so maybe drawing down 13AH will then be 85%.

My point is that the SoC% are not objective, the user defines them, so saying "your Full is at 85%" makes no sense, your Full is by definition your 100%, there is in practice no "real 85%", SoC can't be used to define anything, it is the measure that each, user must decide how to define for herself.

> I cannot even get to 3.6v to achieve it -- because my maximum charge rate is low enough relative to pack size that I stay in CC the whole time

So stay in CC until Vpc climbs above 3.5 once in a while. Even at low charge rates, you just keep charging longer right? Are you trying to do this on solar, saying the sun goes down too soon?

All maintenance protocols, like equalizing a lead bank, have their requirements. If the user wants to do it they need enough energy and the right gear. To me that means getting back to sufficient shore power and a user-adjustable charger/power supply.

I am always only in CC, IMO LFP in normal cycling use should not get any Absorb time.

The exception is when running periodic maintenance protocols, e.g. for the sake of top balancing or (maybe, now) preventing memory effect.

[CatNewBee]

Originally Posted by arsenelupiga

just in case i change my mind...

where would you recommend talented but electrically semi-literate would start to raise knowledge to the level that Lithium DIY is possible ?

Zhe BMS I use

REC d.o.o. Manufacturer web site

Installation Video:

http://www.rec-bms.com/images/A_BMS/ABMS_connection.mp4

Configuring settings:

https://www.youtube.com/watch?v=sRr_5vBKupw

Uploading custom Firmware - if you need one:

https://www.youtube.com/watch?v=2di-Dpv2Nf8

Here a large install using many cells and the big brother of the ABMS:

https://www.youtube.com/watch?v=j8dWyNSwyqk