GUIDE to properly Top-Balance and Charge a LFP Battery: Part 1

[DavidKelly]

What defines a grade B cell

At the factory the cells are graded. This is relative to capacity and is based on internal resistance. The can’t capacity test every cell, it would not be economical (The amount of energy!)… So they determine the cell capacity based on the relationship with internal resistance as mentioned above. If the cells internal resistance is outside of the specification on the datasheet it is rejected ( Marked B grade). If it’s real bad. It’s C grade and recycled.

These cells are then batched and sold on the secondary market through tertiary distributors ( nothing in China is wasted). Reputable manufacturers will remove the serial number and mark them B grade (will post a picture).

These cells will not perform the same as A grade or sometimes called HSEV grade ( Electric vehicle grade) cells.

B grade typically find there way into telecoms applications due to low C rates or sold as A grade to fetch the higher value.


This is why buying on Aliexpress must be done carefully. Be careful, there’s a lot of dodgy LFP out there!

 

[G00SE]

At the factory the cells are graded. This is relative to capacity and is based on internal resistance. The can’t capacity test every cell, it would not be economical (The amount of energy!)… So they determine the cell capacity based on the relationship with internal resistance as mentioned above. If the cells internal resistance is outside of the specification on the datasheet it is rejected ( Marked B grade). If it’s real bad. It’s C grade and recycled.

These cells are then batched and sold on the secondary market through tertiary distributors ( nothing in China is wasted). Reputable manufacturers will remove the serial number and mark them B grade (will post a picture).

These cells will not perform the same as A grade or sometimes called HSEV grade ( Electric vehicle grade) cells.

B grade typically find there way into telecoms applications due to low C rates or sold as A grade to fetch the higher value.


This is why buying on Aliexpress must be done carefully. Be careful, there’s a lot of dodgy LFP out there!
View attachment 210882

Do they have data to delineate the cutoff between passing and rejects?
This is the first tangible type info I’ve seen as to HOW these are classified.
Additionally, why would ANY currently available cell offered to regular consumers by 3rd party vendors (18650,imr,ezeal,sfk,docan,gobel, etc) have supplies of grade A available? Wouldn’t those be going into whatever mass scale projects they were built for?

 

[DavidKelly]

Yeah the true safe float voltage is probably in the lower 53's like that.

I'm working on a single voltage charging scheme theory for daily cycled solar systems, because I think a single voltage would greatly simplify things by eliminating concerns about rebulking cycles, eliminating the need for careful charge termination or any termination at all, and allowing balancing in float.

Right now, I'm starting with 3.437 because 55v is a round number, but the best single voltage could be elsewhere.

There are two criticisms my single voltage has to face:

1. Bulk/Absorbing at 3.437 won't be fast enough - This is the big one that I need to explore more in practice. After this summer I will be able to speak more confidently about it. I think that the difference at typical solar C rates is minimal.

2. Floating at 3.437 is too high - Yes, this is definitely true if it was a constant float, like a UPS application. But I think it's ok for two reasons, one that if we absorb at 3.437 in the morning of the solar cycle then we aren't really ramming the charge in, and that should leave a bit of vacancy for the float to keep trickling it in, and two that a solar cycle float is naturally limited to 1-8 hours, depending on how early you acheive the float in the day. The float will always end when the sun goes down, so the float duration is naturally limited.

Think your method is spot on. Might have to play with the numbers a bit depending on the cells you have but should work well. I agree with your reasoning.

I don’t think it will be too high. In many motive applications the charge target voltage is 3,458V.

Your set point will be fine on good quality cells.

Hopefully this is the outcome of this post that 3V-3,5V is good to start and then tailor to your system.

Please keep us updated on how it goes.

 

[DavidKelly]

Do they have data to delineate the cutoff between passing and rejects?

Yes. If you ask them they will share it with you on a batch that you have ordered. Certainly with Eve. They are very professional. We once used it to solve an issue we originally thought was caused by bad cells but data proved it to be a faulty BMS. I digress.

This is the first tangible type info I’ve seen as to HOW these are classified.
Additionally, why would ANY currently available cell offered to regular co

I currently am involved with a lithium ‘rejuvenation’ project for the largest Telecoms company in South Africa. Can’t say who but easy to work out. In South Africa theft and vandalism is very bad. We are then able to disassemble recovered damaged batteries and test cells. Establish the good from the bad based on impeadance testing and other characteristics. Then rebuild new batteries and return to service.

We are currently processing 13000 units.

by 3rd party vendors (18650,imr,ezeal,sfk,docan,gobel, etc) have supplies of grade A available? Wouldn’t those be going into whatever mass scale projects they were built for?

There’s a lot of us weekend warriors out there hey, Our money is as green as big projects. We also pay way more for small batches. LFP is dirt cheap in bulk - This is how you can buy a 100Ah battery for 400USD. These suppliers have formed a Business model based on buying in bulk and distributing. Their profit is the delta between high volume discounts and low volume markup.

Gobel is fantastic BTW. Ask for Ellie or Jessie.

 

[DavidKelly]

The impedance thresholds are usually on the datasheet.

 

[DavidKelly]

GFB SOC OCV for LF206 for reference. Closer to the 3,5V side at 100% on charge.

 

[G00SE]

Yes. If you ask them they will share it with you on a batch that you have ordered. Certainly with Eve. They are very professional. We once used it to solve an issue we originally thought was caused by bad cells but data proved it to be a faulty BMS. I digress.



I currently am involved with a lithium ‘rejuvenation’ project for the largest Telecoms company in South Africa. Can’t say who but easy to work out. In South Africa theft and vandalism is very bad. We are then able to disassemble recovered damaged batteries and test cells. Establish the good from the bad based on impeadance testing and other characteristics. Then rebuild new batteries and return to service.

We are currently processing 13000 units.



There’s a lot of us weekend warriors out there hey, Our money is as green as big projects. We also pay way more for small batches. LFP is dirt cheap in bulk - This is how you can buy a 100Ah battery for 400USD. These suppliers have formed a Business model based on buying in bulk and distributing. Their profit is the delta between high volume discounts and low volume markup.

Gobel is fantastic BTW. Ask for Ellie or Jessie.

Thanks for the insights

 

[rustyrotors]

GFB SOC OCV for LF206 for reference. Closer to the 3,5V side at 100% on charge.

can you share the soc-ocv data for lf304, or is it basically the same

 

[shvm]

How can it be a function of Chemistry and not have an effect on internal resistance. The FCV will decrease as the cell looses capacity. The increase in internal resistance will affect the FCV.

I can take two identical cells with same FCV. and connect them in parallel as depicted.
The capacity of the combined cell has doubled, The effective IR halved, but the measured FCV or any characteristic voltage for that matter has to be the same. Basic sanity check !!

The literature is clear on this.

Might as well cite one.

Capacity and internal resistance are inversely proportional. A decrease in capacity will see a rise in internal resistance ( and vice Versa).

You need to realise that it is a one way street.
For an aging cell, capacity ↓ will indeed cause IR ↑.
But an IR ↑ doesn't necessarily translates to capacity ↓.

I have said it before.
I am saying again, it is totally common for a cell to have higher internal resistance without affecting measured cell capacity (Ah).



@upnorthandpersonal You reading those ??
This is now reminding me of your thread

I should have known better... (sorry, ranting)

... than to go into discussion with SFK on Andy's YT comment section. For context, the issue is that the JK (new one) resets its values after x amount of time un-powered. This is a bug, obviously, and the reset values of 4.2V per cell are very bad for LFP. So SFK commented under the video...

diysolarforum.com

Are you implying that a resistor placed In series does not incur a voltage drop? The measured voltage will be lower. I suggest you use a resistor with a proportional resistance to that of a cell internal resistance increase.

Not unless current is flowing. But that is not how FCV is determined in the first place.

It does seem we’ve reach agreement on the importance of temperature.

Umm, not at all as far as I'm concerned, when it comes to the charging implementation.

In hindsight, The EVE data is still not enough for determining what voltage to float for LFP.
There is no consideration of the surface charge phenomena and how there is a supercapacitor like behaviour at charge end. There is a depletion of vast majority of mobile charge carriers (Li+ ions) too. Not recommended to stay in this region for long.

This is true on some of the cheap and nasty cell makes a float voltage of 51,8V- 52,5V is not uncommon for a 15S battery.

51.8V-52.5 V comes out to be 3.45 to 3.5 V Cell for 15S battery. That is a dangerous float voltage and is not recommend for even 'grade A' LFP cells due to aforementioned degradation mechanisms.

BTW, I have EV Grade A CALB L160F100 Cells with factory QR.

So to confirm:

• Does your Method calculate the FCV or do you use the SOC OCV curve from manufacturer?

• Does your method compensate for changes in temperature?

• Does your method calculate the FCV based on capacity alone?

• Do you still think everyone is doing it wrong?

I don't get your adamancy. There is simply much more to this that needs to be figured out on your end.

 

[DIYrich]

@DavidKelly : any thoughts on compression of cells and cycle life?

 

[DavidKelly]

I can take two identical cells with same FCV. and connect them in parallel as depicted.
The capacity of the combined cell has doubled, The effective IR halved, but the measured FCV or any characteristic voltage for that matter has to be the same. Basic sanity check !!

View attachment 211023

This is very entertaining….

You are proving the above for me. By doubling the capacity you have halved the internal resistance. So capacity and internal resistance are inversely proportional. Yet you demand citations.

I did not say a decrease in internal resistance will change the FCV. An increase will definitively. Something you would see through the addition of a resistor In series. There will be a voltage drop across the resistor be it external or within the battery.

Might as well cite one.

You need to realise that it is a one way street.
For an aging cell, capacity ↓ will indeed cause IR ↑.
But an IR ↑ doesn't necessarily translates to capacity ↓.

Never said it was a two way street. If there existed a technology to restore capacity on a cycled lithium cell. The owner would have more money than Elon.

Think you need to go do some reading. Here is a citation. There are many published works detailing this relationship.

https://www.researchgate.net/publication/286477560_Correlation_between_capacity_and_impedance_of_lithium-ion_cells_during_calendar_and_cycle_life

TY - JOUR
AU - Schuster, Simon
AU - Brand, Martin
AU - Campestrini, Christian
AU - Gleissenberger, Markus
AU - Jossen, Andreas
PY - 2016/02/01
SP - 191
EP - 199
T1 - Correlation between capacity and impedance of lithium-ion cells during calendar and cycle life
VL - 305
DO - 10.1016/j.jpowsour.2015.11.096
JO - Journal of Power Sources
ER -

I have said it before.
I am saying again, it is totally common for a cell to have higher internal resistance without affecting measured cell capacity (Ah).

Within the allotted tolerance by the manufacturer. It’s a sliding slope from there.

Not unless current is flowing. But that is not how FCV is determined in the first place.

A voltmeter with infinite resistance does not exist. In order to measure voltage a small current is passed. If we’re getting technical.

In hindsight, The EVE data is still not enough for determining what voltage to float for LFP.

It is sufficient to make an informed decision for Eve cells. You are trying to make a generalised case for all LFP base solely on chemistry now. At the start of the thread it was Capacity. (regardless of Brand, temperature, internal resistance etc).

You are now deflecting and adding technical jargon to resist answering the simple questions asked of your model.

There is no consideration of the surface charge phenomena and how there is a supercapacitor like behaviour at charge end. There is a depletion of vast majority of mobile charge carriers (Li+ ions) too. Not recommended to stay in this region for long.

Again all interesting and will make for great discussion preferably on its own thread. These arguments were not made in your original post.

51.8V-52.5 V comes out to be 3.45 to 3.5 V Cell for 15S battery. That is a dangerous float voltage and is not recommend for even 'grade A' LFP cells due to aforementioned degradation mechanisms.

Typically occurs in economic applications (where price determined the specification) the product won’t be around long enough so it becomes a non issue. Other than high risk telecommunication sites think about charge controls for RC cars, alarm batteries etc.

Sometimes engineering takes the form of planned obsolescence. Quite frankly most times it does where value engineering produces a bridge that barely stands.

This would make an interesting discussion on another thread.

For DIY guys it’s different as we are vested in max cycle life and passionate about cool tech.

Again the term dangerous is so wrong and hyperbole . LFP will not be dangerous under 3,65V. This is not the message to send to guys starting out. We can argue over cycle life protection But there is no risk of fire and this falls within the operating range in the manufacturers datasheet.

BTW, I have EV Grade A CALB L160F100 Cells with factory QR.
View attachment 211034

Show me your large batches of Eve cells. You have only mentioned these Calb 100Ah cells which is why I say your dataset isn’t large enough to draw such strong conclusions.

I don't get your adamancy. There is simply much more to this that needs to be figured out on your end.

Again deflecting. They are such simple questions that you refuse to answer. Your unwillingness to engage speaks to the lack of confidence in the model. You cannot tell everyone that they are doing everything wrong without being willing to back your argument.

Again. Please answer these simple questions:

So to confirm:

• Does your Method calculate the FCV or do you use the SOC OCV curve from manufacturer?

• Does your method compensate for changes in temperature?

• Does your method calculate the FCV based on capacity alone?

• Do you still think everyone is doing it wrong?

 

[DavidKelly]

@DavidKelly : any thoughts on compression of cells and cycle life?

This is the age old debate hey. Might even elicit a more passionate response than the top end voltages.

Technically the Aluminium casing are supposed to provide sufficient resistance (This is advertised on some datasheets). If you’ve ever had a pouch cell go bad and swell you’d realise how much the casing does hold back. Prismatic cells are just pouch cells in an aluminium casing.

That being said. My view on it is that the operation of the cell is heavily dependent on geometry (Think anode to cathode distance, sEI layer thickness, Tesla cylindrical cell performance vs standard etc). All this being necessary for maintaining cycle life. They do swell and so we compress to prevent it. It’s very apparent on first charge.

In telecoms it’s rarely done as the batteries are so cheap and value engineered that it’s not cost effective. Their expected lifespan doesn’t warrant it. For example Narada batteries use a plastic retainer at the top and the bottom of the cells with a sizeable air gap in between and hence no compression. The C rates are also fairly low although that’s changing with 5G.

More premium network operators pay more and then the batteries have some form of compression. Their ROI periods are much longer and there sites are more secure so the batteries tend to stick around. (The telecom battery theft issue in SA is a big problem).

When it comes to bigger batteries like the forklift battery’s we manufacture it is essential. Large per unit cost , Large temperature changes, large C rates, Demand to meet cycle life. vibration etc it just makes sense. Not to mention the extra weight adds ballast for the forklift. I don’t believe in squeezing the cell. Just gently preventing expansion. It’s also very important not to obstruct the vents.

For home Gamers I dont think it’s a terrible idea but promise we won’t burn you at the stake if you decide not to

I’ll post some pics of the various types used. Very blessed at the moment to have access to a wide variety of batteries.

 

[Partimewages]

Sometimes the ignore button is useful LOL!

 

[shvm]

In order to measure voltage a small current is passed. If we’re getting technical.

Smart!
Take your best Voltmeter, capable of highest precision and measure the FCV of a cell, with and without a resistor in series.

Entertain us on a video for everyone to see. Let's face the physical reality together.

So capacity and internal resistance are inversely proportional.

Proportionality only applies to a two-way street.
Add a resistor in series to those two cells in parallel and run a capacity test. Verify this on your own end. For science!

Again all interesting and will make for great discussion preferably on its own thread. These arguments were not made in your original post.

The primary reason for this entire thread to exist is this, cell life concerns due to time spent at high voltages without any current in LA charging models, and cell debalancing.

Again. Please answer these simple questions:

So to confirm:

Let's attend to your very pertinent 'simple' questions.

So to confirm:

• Does your Method calculate the FCV or do you use the SOC OCV curve from manufacturer?

FCV is never being calculated in the first place.

• Does your method compensate for changes in temperature?

No. And it never will be. basis a data "accurate" to 4 decimal places of voltage. That is an immediate red flag.

• Does your method calculate the FCV based on capacity alone?

FCV is never being calculated in the first place.
Just like the 3.65 V number is common to LFP regardless of capacity. I'm not calculating 3.65 V either, rather taking it for granted.
This guide is about battery charging models, not calculating FCV.
Nobody prevents you to put your own numbers and come with something else.

Read here, the same model being employed for NCM charging (which isn't LFP).

esphome-jk-bms-can/documents/README/Cut-Off_Charging_Logic.md at main · Sleeper85/esphome-jk-bms-can

Contribute to Sleeper85/esphome-jk-bms-can development by creating an account on GitHub.

github.com

Guess next, you should feel the need to correct me on NCM FCV calculations and how I need to account for temperature.
This is basic strawmanning.

• Do you still think everyone is doing it wrong?

This is a rhetorical question.

Look, I will be clear. I'm not entertaining this further unless you actually put your claims on paper, (on a video) and show us where the wrong lies. There are things beyond just decimal point precisions. Otherwise you are just adding to the page count of this thread.

 

[DavidKelly]

FCV is never being calculated in the first place.

From your statement “Then it is a simple matter of applying linear regression to find the desired (voltage, current) pair connecting these two extremes. (FCV, 0 C) and (3.65 V, 0.05 C). There is only one unknown in this statement of which can be solved by simple linear regression…But we’re not calculating it? Or the unknown value lies between FCV ,an unknown, and 3,65V.

But you don’t calculate the value, don’t use manufacturers charts, tell everyone anything over 3,37V is over charging and dangerous.

Everyone is wrong but you won’t say what the correct value is.

On your Calb100 cells. What is the FCV you are using?

FCV is never being calculated in the first place.

But your argument contained a single FCV across all LFP based on chemistry alone. What is that value?

Just like the 3.65 V number is common to LFP regardless of capacity. I'm not calculating 3.65 V either, rather taking it for granted.This guide is about battery charging models, not calculating FCV.

Your advice was to only charge to FCV. What is the value that should be targeted if it’s not calculated?

Nobody prevents you to put your own numbers and come with something else.

I provided the manufacturers SOC OCV curve and stressed the importance of temperature in determining the value. A good BMS would include this information in its calculations.

I also endorse the 3V to 3,5V model for beginners as it would deliver results without all the technical fuss. It is not dangerous to operate the cell under 3,65V and we could have another conversation about protecting cycle life by not going that high. I might add that the BMS operating limits should be broader than the charge and discharge settings to avoid disconnection.

Read here, the same model being employed for NCM charging (which isn't LFP).

esphome-jk-bms-can/documents/README/Cut-Off_Charging_Logic.md at main · Sleeper85/esphome-jk-bms-can

Contribute to Sleeper85/esphome-jk-bms-can development by creating an account on GitHub.

github.com

Start a new thread, happy to discuss…

This is a rhetorical question.

Look, I will be clear. I'm not entertaining this further unless you actually put your claims on paper, (on a video) and show us where the wrong lies. There are things beyond just decimal point precisions. Otherwise you are just adding to the page count of this thread.

As Oprah says, if it ain’t bringing you joy…

So you still think everyone is wrong but won’t tell us why we are wrong ?

 

[shvm]

Your advice was to only charge to FCV

Any voltage between [3.65 V to FCV] (which I'm taking as 3.37V) is acceptable target bulk voltage in the model.
I personally use 3.525 V/Cell

Are you really sure you read 5 pages deep into the thread??

I am repeating this again. Unless you go and put the effort to understand what is being done,
you will keep mistakenly attributing things to me.

Start a new thread, happy to discuss…

But you're the one discussing things that aren't part of this thread.

You can start a new thread along the line of "LFP cell FCV and its dependence on temperature".

 

[DavidKelly]

Any voltage between [3.65 V to FCV] (which I'm taking as 3.37V) is acceptable target bulk voltage in the model.
I personally use 3.525 V/Cell

So target 3,525V and float at same voltage or float at 3,37V?

Merely trying to understand your logic.

 

[DavidKelly]

This is the age old debate hey. Might even elicit a more passionate response than the top end voltages.

Technically the Aluminium casing are supposed to provide sufficient resistance (This is advertised on some datasheets). If you’ve ever had a pouch cell go bad and swell you’d realise how much the casing does hold back. Prismatic cells are just pouch cells in an aluminium casing.

That being said. My view on it is that the operation of the cell is heavily dependent on geometry (Think anode to cathode distance, sEI layer thickness, Tesla cylindrical cell performance vs standard etc). All this being necessary for maintaining cycle life. They do swell and so we compress to prevent it. It’s very apparent on first charge.

In telecoms it’s rarely done as the batteries are so cheap and value engineered that it’s not cost effective. Their expected lifespan doesn’t warrant it. For example Narada batteries use a plastic retainer at the top and the bottom of the cells with a sizeable air gap in between and hence no compression. The C rates are also fairly low although that’s changing with 5G.

More premium network operators pay more and then the batteries have some form of compression. Their ROI periods are much longer and there sites are more secure so the batteries tend to stick around. (The telecom battery theft issue in SA is a big problem).

When it comes to bigger batteries like the forklift battery’s we manufacture it is essential. Large per unit cost , Large temperature changes, large C rates, Demand to meet cycle life. vibration etc it just makes sense. Not to mention the extra weight adds ballast for the forklift. I don’t believe in squeezing the cell. Just gently preventing expansion. It’s also very important not to obstruct the vents.

For home Gamers I dont think it’s a terrible idea but promise we won’t burn you at the stake if you decide not to

I’ll post some pics of the various types used. Very blessed at the moment to have access to a wide variety of batteries.

Hauwei ESM48100AhB1V1 batteries with compression.

 

[Helles]

sorry, I cant delete this invalid message