Apparently we have a different definition of what it means to be balanced. All I care about is that they maintain similar SOC throughout the charge discharge cycle, so that's what I call "balanced". I have never (intentionally) charged a cell to 3.65V.
GUIDE to properly Top-Balance and Charge a LFP Battery: Part 1
- Thread starter shvm
- Start date
Apparently we have a different definition of what it means to be balanced. All I care about is that they maintain similar SOC throughout the charge discharge cycle, so that's what I call "balanced". I have never (intentionally) charged a cell to 3.65V.
Q-Dog
¯\_(ツ)_/¯
If, under normal use, the BMSs do not shut down the batteries I consider them balanced well enough. It really helps to start with well-matched cells.
Good summary. Matches what I also understand about the theory of LFP cells and their charging.
This post is controversial because it goes against the practice of many people here on the forum. It is a tricky issue because let's say we deviate from the manufacturer specs, how can we quantify the damage done to the cells? For example, if I ignore tail current and set my absorb time to one hour, how much will that shorten the life span of my pack? No one really knows. And that's why opinions about optimal charging methods vary so much on the forum. It takes several years to practically test different outcomes.
I think it is essential to at least know there is an optimal charging method, and what it is. Then you can reason about what parts of it to ignore when you tailor your own charging. It is very similar to cell compression. We all know cells should be compressed, because the manufacturer clearly says so. But does it matter? Many of us (including me) look at the spec sheet and deduct that it has very little practical difference for our scenario, so we just ignore it.
It's tricky because LFP has such a long life span, that if you abuse them slightly, you'll only find out many years down the line. So that's why I take the "I've been doing this for years, and it works fine" with a grain of salt, and instead put much more credence in what the manufacturer states about charging.
Of course there are optimal charging characteristics - which are done in a lab. In practice, these are not possible and thus compromises need to be made. You can either get really complex set-ups and figure out tail current (and possible clouds preventing that, thus making provisions for that, etc.) and still never get to the same conditions as in the lab.
The first and foremost killer is temperature: do we all have a climate controlled environment where the temperature is kept at a fixed 25C? Doubt it. So do tail current method at 40C, or any other method, and you'll kill your batteries anyway.
So, instead of making things complex, simplify: stay between 3.0V and 3.55V per cell, no absorb time. Works with practically any equipment out there.
The first and foremost killer is temperature: do we all have a climate controlled environment where the temperature is kept at a fixed 25C? Doubt it. So do tail current method at 40C, or any other method, and you'll kill your batteries anyway.
So, instead of making things complex, simplify: stay between 3.0V and 3.55V per cell, no absorb time. Works with practically any equipment out there.
When the pack is in bulk charge (constant current), it reaches 100% SOC at a certain voltage, which depends on charge current. This can be seen in your diagram. A lower charge current will get the pack to 100% SOC at a lower voltage. Termination should be at that specific voltage, to not over charge the pack.Gosh darn, better get back to my lab and tell all the people there (and our customers), and the reps from the manufactures that work with us, that they're wrong when we put batteries to the test and design systems with them...
If you look at the charge/discharge curve of LFP, you can see that below 3.0V and above 5.5V there is basically no energy left. So with a standard charge, you don't have to go to 3.65V - this is done only for things like capacity testing because that's what is considered, well, the standard for testing. This does not mean that you have to use this in real life - purely because in real life you may never get to the point where cells are at rest (especially when they are powering a house, outside of summer for example).
In reality, the system is dynamic. You don't charge to 100%, then rest, then discharge. You micro-cycle. You don't have to stress the cells by going to 3.65V (or doing a tail current at 3.65V) when you can fully charge even at 3.45V with long enough absorption time. If you stop charging at 3.55V in regular solar applications, even without a tail current, you're going to be at 100%. The reason: suppose a 280Ah battery. The 0.5C standard charge rate is 140A, or 7kW. How many solar arrays of that size are connected to one 15kWh battery? Very few. I have 14kW of solar, connected to 4 (soon at least 6) 280Ah 48V packs, which means the max charge current each battery sees is, what, 70A - at peak production! This means that the voltage during charging also increases less fast. Have a look:
So reaching 3.5V at 0.5C is completely different compared to 3.5V at 0.1C - and these are at constant charge/discharge, which never happens in the real world.
But the chemistry also has an upper voltage limit of around 3.65V. So for higher charge currents, you hit that voltage limit first without fully reaching 100% SOC, so you have to switch to absorption (constant voltage) and wait for the tail current.
In other words: For high current charging, you monitor absorption tail current, which is the topic of this thread. For low current charging, you instead monitor bulk charge voltage. Looking at your diagram, you can see that the the breakpoint seems to be around 0.03C.
Last edited:
You don't need 100% state of charge - 95 or something percent is perfectly fine. For solar applications, you don't reach high charge currents in general. A 280Ah 48V at 0.5C means 140A - that's a 7kW array maxed out! Even at 0.3C, you're talking 84A - 4kW maxed out. Now add two of those packs in parallel, and you're at 42A per pack - 0.15C.
Instead, charge to 3.5V, than float at 3.37 or below once voltage reached. No issues.
Instead, charge to 3.5V, than float at 3.37 or below once voltage reached. No issues.
Last edited:
I see no major contradiction between you and @shvm. He describes the underlying theory of what optimal charging is, which is good to know about. You explain what a practical solution is, i.e. how to get as close as practically possible to optimal charging.
I also meant to say: You have probably come up with the 3.5V value experimentally, correct? There is also a theoretical way to arrive at the same voltage, by looking at the pack capacity and charge current, and that what the OP is about. Two ways to arrive at the same conclusion.
It's more the way it's worded in the first post:
Essentially: "I'm here to tell you you're all wrong. Here is the correct way. There is no other way, they are all wrong". I'm showing that there is a world of difference between the things that work in a lab, and something that works out in the field - and I have experience with both. I'm currently in the process of setting up a new lab here (replacing the old one) specifically for battery testing, for all kinds of purposes (not just theoretical stuff, but for real world applications such as grid storage etc.) to support the battery manufacturers and industry using them here in the area.
The 3.5V comes by choice. You can make it 3.55V or 3.48V - I just like the round number, and I know most equipment can't measure to 0.01V accurately anyway, so it gives a safety margin as well.
To add, it basically boils down to: this is a DIY forum with people building their own batteries, with all kinds of skill level. Is it better to:
a) Tell them to get equipment in order to implement tail current, somehow, depending on a voltage table this will be different, so make sure you pick the right one. Or failing that:
Oh, and balance the pack every few months, somehow, and top balance properly before with this method, not the other one, etc.
or
b) Set your cut-off at 3.0V per cell on discharge and 3.5V for charge. Don't absorb. Set your float at 3.35V and the balancer start in your BMS (or active balancer) at 3.45V. Done. You don't even need to balance the cells if you have an active balancer. It may take some time, but it's fully hands-off and you can perfectly use the battery in the mean time.
a) Tell them to get equipment in order to implement tail current, somehow, depending on a voltage table this will be different, so make sure you pick the right one. Or failing that:
Oh, and balance the pack every few months, somehow, and top balance properly before with this method, not the other one, etc.
or
b) Set your cut-off at 3.0V per cell on discharge and 3.5V for charge. Don't absorb. Set your float at 3.35V and the balancer start in your BMS (or active balancer) at 3.45V. Done. You don't even need to balance the cells if you have an active balancer. It may take some time, but it's fully hands-off and you can perfectly use the battery in the mean time.
Partimewages
Solar Addict
@BjornM
I agree with @upnorthandpersonal. This member has multiple posts where their the smartest person in the room. My age tells me those people are the ones to watch out for.
We have a multitude of experience here, just pay attention and you will learn. ?
I agree with @upnorthandpersonal. This member has multiple posts where their the smartest person in the room. My age tells me those people are the ones to watch out for.
We have a multitude of experience here, just pay attention and you will learn. ?
I agree that humility isn't his strong suit. But he is correct in how he describes the theory of charging LFP. That is a necessary foundation of knowledge before you look at a practical solution, even if reality never matches a lab perfectly. For example, if I know that my charge current is low and fluctuates, but is normally within a certain range, then I can estimate what a matching bulk voltage would be, instead of just picking some random number between 3.40 and 3.60. Ok, we will still not know how much this will actually extend the lifespan of the pack, but at least we have tried our best.
You (and others as well) should read up on "Systems Thinking" (recommended book: Thinking in Systems, by Donella H. Meadows).
As a user, you don't need to know the theoretical details of charging LFP. It is not a necessary foundation. You're building a system powering a house, or van, or... - you need to look at that as a system. Getting the charge details 'perfect' is not what is important - being able to have a system that is not complex, reliable and free of failures and needless complexity is.
It's the same as all the people going for 100% state of charge. Why? Why does your system need the battery to get to 100%? Does it fail when you only get to 95%? If yes, your system is designed wrong and you should add capacity - not stress to battery going to 100% with tail currents or whatever. You're focusing on the wrong thing.
Tulex
Solar Wizard
This is like listening to a race car driver talk to a commuter. Most of us don't have race cars for batteries, nor do we need that type of performance.
We just need to know best practice to get the most out of our batteries. The rest is just confusing.
We just need to know best practice to get the most out of our batteries. The rest is just confusing.
Agreed. The charge methods that you and other senior members advocate here are totally fine. I have no doubt that people who follow your advice will have great packs for many years. The OP fails to mention this, and is unnecessarily dismissive of your methods. Hence the push back, which could have been avoided by writing in a more humble way. However, the OP is laying the theoretical foundation for how to optimize charging further, which is valuable and highly interesting to a few of us, including me.
Yep - and by the way, context matters. If this thread would have been titled "Ideal charge parameters discussion for a LFP battery" or similar, where the discussion was about the nitty gritty details, I'd be all for it. I have a bunch of nitty gritty details that haven't even come up here yet. But that was not the point of the thread as was made very clear in the first paragraph written in the post.
Accusations of Humility aside,
The main goal of this guide was always the pointless nature of theoretical deliberations on 'real world compromise' charging parameter which can be readily derived from standard charging model.
Like people literally going into all kinds of rabbit hole regarding ideal absorption time, absorption voltages, balancing and most importantly, floating etc.
I Mean just take a look at this:
The timestamp at 19:55 even reads: "The perfect charge cycle"
No it isn't. Why you would even use a BMS perfectly capable of closed loop comms to implement something like this is beyond me.
I will admit it, I don't know much and I'm not an expert by any stretch.
Bur, at the same time I know there's only one way someone can implement 'compromised LFP charging' with Hardware that is perfectly capable of implementing Standard Model using closed loop communication. (ESP32 + CAN/RS485 )
And the only way to do that is by being not aware of Standard Charging model for LFP. I fail to get otherwise why would anyone do this.
The post is there to prevent someone getting into so deep of a rabbit hole (and influencing others) so as to start thinking of compromised LFP charging itself as some sort of ideal/perfect charging for LFP when it really isn't.
The main goal of this guide was always the pointless nature of theoretical deliberations on 'real world compromise' charging parameter which can be readily derived from standard charging model.
Like people literally going into all kinds of rabbit hole regarding ideal absorption time, absorption voltages, balancing and most importantly, floating etc.
I Mean just take a look at this:
The timestamp at 19:55 even reads: "The perfect charge cycle"
No it isn't. Why you would even use a BMS perfectly capable of closed loop comms to implement something like this is beyond me.
I will admit it, I don't know much and I'm not an expert by any stretch.
Bur, at the same time I know there's only one way someone can implement 'compromised LFP charging' with Hardware that is perfectly capable of implementing Standard Model using closed loop communication. (ESP32 + CAN/RS485 )
And the only way to do that is by being not aware of Standard Charging model for LFP. I fail to get otherwise why would anyone do this.
The post is there to prevent someone getting into so deep of a rabbit hole (and influencing others) so as to start thinking of compromised LFP charging itself as some sort of ideal/perfect charging for LFP when it really isn't.
You're fooling yourself if you think anyone knows the "optimal" way of charging LFP batteries. Given the number of variables, the best anyone can do is extensively test and report what works best in their test. That's true of any engineering discipline. You define the boundaries outside of which there's an unacceptable risk of failure. Inside those boundaries is an infinite space of varying degrees of success.
TBH, I don't even think that what you're referencing in your OP is in any way "optimal". The datasheet says "standard", which I read to mean that it's what they used in their tests. I have a hard time believing that charging at such a high current to max voltage is even good for the cells. Both of those are contradictory to what I've read.
If you have real-world test results showing that your "optimal" charging method will significantly extend the life over Andy's "compromised" method, then I would be interested in seeing it.
TBH, I don't even think that what you're referencing in your OP is in any way "optimal". The datasheet says "standard", which I read to mean that it's what they used in their tests. I have a hard time believing that charging at such a high current to max voltage is even good for the cells. Both of those are contradictory to what I've read.
If you have real-world test results showing that your "optimal" charging method will significantly extend the life over Andy's "compromised" method, then I would be interested in seeing it.
Well after all of this I adjusted my battery to 3.35v per cell across all 16 cells. So how long can I leave it like that? Does it do any good cycling it once a month or something or can it sit at 3.35v per cell indefinitely without damage?
3.35V charge voltage, or storage? What are you trying to achieve?
Battery is used for power outages and that is all. I want it ready to run my home when the power goes out.
That's probably reasonable. From what I've seen, you just don't want to leave it too low due to self discharge, but I don't know much beyond that.
In the early days of LiPo batteries, it was suggested to store them with a relatively low charge, but I've heard different for LiFePo4, and having empty batteries would be pretty useless for backup anyways.
What you have done is to present a theory of how to charge the battery when you follow the standard model in the manufacturer's spec sheet. This involves keeping track of tail current, etc. This is all good and interesting as a starting point for discussing charging methods. Keep in mind though that this is not necessarily the optimal model, its just a model the industry people have agreed on because it is easy to understand and works reasonable well for EVs. Given the decades of experience they have with manufacturing, this model should be respected, but not taken as gospel for all situations. The reason I say this is because there are no real world tests on what lifespan improvements you will get from it for a solar installation. This needs to be quantified. Is it 20% longer life? 1%? Optimization for its own sake is pointless. This kind of optimization might involve a DIY MCU board that speaks to the BMS and charger, and is way over most people's heads.
You chose to go against the grain here at the forum and tell why "everyone is charging the wrong way", when you instead could have worded it "the theoretical background to why most people's charging methods work reasonably well, and how we might improve it further". That's why you get so much pushback. What it comes down to in the end is that the traditional charging methods work fairly well, even from a theoretical point of view. For example the @upnorthandpersonal method of bulk to 3.50 V, no absorption, float at 3.37 V.
That is the first thing i do. And yes it is the most important.
Similar threads
