BLS 4x 3.2V 200Ah LiFePO4 - voltage difference between cells

[cyanogenic]

Newbie here, following everything from Will's Youtube videos

So my setup:
1. 4x 3.2V 200Ah LiFePo4 cells (from BLS, Aliexpress)
2. ISDT 8S Battery Go balancer
3. 100A Daly BMS with Separate Port
4. Victron battery protect
4. 1500W inverter

My issue:
These 4 cells were balanced to within 5mV at 3.2V when I got them. I've been discharging them slowly via inverter and an LED light. Got them to about 3.1V and then one of the cells would start dipping down a LOT more than the others, e.g. 3.08V, 3.08V, 3.08V, 2.98V.
This cell would continue to drop like a tonne of bricks when discharging further, e.g. 3.0V, 3.0V, 3.0V, 2.75V

These aren't the exact voltages (doing them off memory), but that is the general idea. The cells stay within 20mV of each other from 3.1V - 3.4/3.5V , but when it gets below 3.1V the cells start deviating quite a bit.

I'm trying to do a bottom balance now with them around 3V, but it is taking a LOOONG time due to the difference in voltages.

Could there be an issue with one of these cells?

 

[gehowi]

cyanogenic,
Do these cells have 6mm hex bolts with nuts or 4mm bolts with washer? Type ETC NAH3L0 ?

If 6mm then you received old stock. ETC does not produce this anymore.
I ve got 16 of these in my first order, with 1 self discharging cell immediately , 4 expanded cells replaced after 1 month (on my own expense!) and now again (3 months) 4 cells with too large deviation.

BLS (ETC) Cells with 4mm bolts should be ok. I have 20 of them performing normal with equal voltages - for now.

Check cell flatness/expansion: Put all the cells side by side together. If you get gaps between cells more than 1mm, cells may be expanded in the middle due to discharging too far=permanently degraded. This expansion grows rapidly under 2.5V, even after 1 discharge to 2V and it never comes flat again. New cells in this condition means you got old stock which has been discharged too far. I got 4 of these.

Don't trust the Daly BMS as well. It should turn off the output when any cell goes under 2.5V. In my own cell datalog I discovered that my 200A Daly sometimes turned off when one cell went below 2.3V already which certainly degrades those weak cells further. In my experience never go lower than 2.7V or higher than 3.5V/cell. (10.8V-14V)

Normally all cells in a string must match very close. If not, the total available capacity will be (much) lower because the BMS will turn off faster and the capacity in the good cells remains unused.
It is hardly possible to keep cells with a capacity deviation balanced. Standard BMS as the Daly do not transfer energy but put a load on the cells with the highest voltage to bring them down. In case of 16 cells with one low voltage cell, this may be 35mA*15 = 0.5A or 25W wasted power for as long as there is imbalance.

At the end of charge/discharge, voltage can rise/decrease very quickly for the weakest cells. That's a normal behavior if the other cells are +- in the same SOC.

Measuring cell voltages every 30 sec and writing data in a log file is the best way to analyze cell problems. For 12V it can be done by resistor dividers and an ESP32. This MCU has several analog inputs and an internal SPIFFS file system which can keep a lot of data. Since it is an MCU with wifi, it's possible to monitor anything from a browser.

In such 24 h graph for 16 cells you can see that 4 cells are degraded. It would cost a lot of energy and time to get this information from charging/discharging 16 cells manually.

Try to charge the low V cells separately and several times to the same level or higher than the other cells. You may use the 3.3V from an old PC power supply for this, then you can't overcharge and most can deliver >10A.
If the imbalance comes back after time, your only option is replacing cells or live with reduced capacity...

 

[electric]

you should distinguish initial balance with ongoing balance. Even perfectly healthy cells need initial balance because they are shipped at various SOC levels. Even 1-2 Ah difference would take many hours to balance with small balance current in the BMS.
Also, you are not balancing anything at 3.2V, you are just wasting your time. You must balance at the upper or lower knees, below 3.0V or above 3.5V. In the middle the voltage is too close even when SOC difference is large.
Your cells maybe good or bad, but you can't tell until you do initial balance and then some use cycles to see if balance stays decent enough.
Another good test for bad cell is to charge fully and then let it rest for a few days with completely open circuit. Good cells will settle to about 3.38-3.4 and will stay there, while crappy cells will keep falling to 3.3-3.35 within 1-2 days.
What you described above is just lack of initial balance, not proving bad cells yet.

 

[cyanogenic]

Do these cells have 6mm hex bolts with nuts or 4mm bolts with washer?

4mm hex bolts with washer

Check cell flatness/expansion:

All flat

Measuring cell voltages every 30 sec and writing data in a log file is the best way to analyze cell problems. For 12V it can be done by resistor dividers and an ESP32.

I'll look into these

Try to charge the low V cells separately

I'll have a dig around to see what I have if the cells continue to be out of balance

you are not balancing anything at 3.2V, you are just wasting your time.

What you described above is just lack of initial balance, not proving bad cells yet.

Gotcha, will wait and see what they're like after the bottom balance finally finishes and do a few cycles.

 

[gehowi]

I think your M4 batteries may be ok but it is possible one cell has less capacity. I needed 20 M4 cells to create a perfect match for a 16S bank, which does not mean the other 4 are bad.

Now I am checking the real capacity of the faulty 200Ah M6 batteries from BLS (3months old):
Charging to 3.4V until current is less than 0.5A.
Rest for 1 hour: cell V= 3.32V.
Then discharging with 10A constant current to 0A: Measured max 54Ah in 5h33m.
That's 25% of the advertised 200Ah!

May be you can do a separate capacity test on 4 cells but of course don't discharge lower than 2.5V...

 

[cyanogenic]

I think your M4 batteries may be ok but it is possible one cell has less capacity. I needed 20 M4 cells to create a perfect match for a 16S bank, which does not mean the other 4 are bad.

Now I am checking the real capacity of the faulty 200Ah M6 batteries from BLS (3months old):
Charging to 3.4V until current is less than 0.5A.
Rest for 1 hour: cell V= 3.32V.
Then discharging with 10A constant current to 0A: Measured max 54Ah in 5h33m.
That's 25% of the advertised 200Ah!

May be you can do a separate capacity test on 4 cells but of course don't discharge lower than 2.5V...

View attachment 1152

Wait your 200Ah battery only had 54Ah?! That's terrible

 

[electric]

Now I am checking the real capacity of the faulty 200Ah M6 batteries from BLS (3months old):
Charging to 3.4V until current is less than 0.5A.

Why only to 3.4? That is way too low if you are testing a cell. Why not to 3.65 which is actual correct max charge voltage?

 

[gehowi]

You need a supply of 3.65 V or higher to get enough current in the cell. But as soon as you connect a power source, voltage will drop to the cell voltage due to low resistance. The current will be limited by the max current the supply can deliver, the wires and the rising internal cell resistance.

I stop charging when the cell reaches 3.4V. Charging to 3.65V will hardly put 10% more energy in a cell but will shorten the cell life drastically. A good BMS will turn off when any cell reaches +-3.5V. The best cells in the string may not even reach 3.4V at that point.

In space, batteries are charged / discharged far less than these limits to get max lifetime.
In smartphones the limits are used to get max autonomy. So they last not much longer than 3-5 years.
For expensive 200Ah cells I prefer less capacity and longer life.

The 200Ah on the label is based on max charge to max discharge in labo conditions and without lifetime concerns. In real life 100-150Ah is more realistic - if you don't want to replace cells after a few years. Anyway, it's always wise to divide Chinese specs by 2...

The usable capacity for the tested cells with M6 screw is indeed only 54Ah, may be 65 if I charge to 3.6. I will try it, I have spare cells. (takes loooong time..)
More recent cells with M4 screws are certainly much better, but I did not do capacity tests because I don't want to create imbalance in my bank...

For my 16S banks I always turn off charge when the bank reaches 54.5V or 3.4V/cell. Any further charge make some cells rising quickly (in 10 min) to unhealthy voltages above 3.6.


As electric said, it's best to check imbalance at lower voltages. Next 24h charts shows this.
At 3.2-3.4V there seems to be no imbalance:


At lower charge the cell quality is obvious:
The upper four are M4 cells; the other 12 are M6.
It's clear the next cell to replace is the yellow C3 with a difference of 0.3V. This cannot be balanced anymore.
Actually all 12 M6 cells will need to be replaced. All cells are 200Ah and new but do not match at all after 3 months.

 

[electric]

@gehowi you are not wrong and you seem to understand concepts, but your design optimization is way too conservative for commercial applications. I do this for a living, so if I design for 60/40 optimization, I'd get fired. Even a napkin design should be 80/20, commercial product at least 90/10. Of course optimization factors are different for space and cell phone and everything in between. You may have picked one factor out of 3 which effects battery life, optimized way too conservative for that one factor, but remaining 2 factors could still ruin your results.
With all due respect, I wish I could spend an hour talking to you face to face, but I already got sucked in too much into yet another online forum full of nonsense, so I need to pull back, turn off my notifications and get back to work. No offense. I know I sound like an ass, I'm just too frustrated with forums and social media.

 

[Supervstech]

@gehowi you are not wrong and you seem to understand concepts, but your design optimization is way too conservative for commercial applications. I do this for a living, so if I design for 60/40 optimization, I'd get fired. Even a napkin design should be 80/20, commercial product at least 90/10. Of course optimization factors are different for space and cell phone and everything in between. You may have picked one factor out of 3 which effects battery life, optimized way too conservative for that one factor, but remaining 2 factors could still ruin your results.
With all due respect, I wish I could spend an hour talking to you face to face, but I already got sucked in too much into yet another online forum full of nonsense, so I need to pull back, turn off my notifications and get back to work. No offense. I know I sound like an ass, I'm just too frustrated with forums and social media.

I hope you stay on here. Your experience should really help out!
In commercial settings, batteries get replaced on a schedule, and money isnt as tight as most on a DIY solar budget! I know you are totally correct, and would love to be in that arena, but alas, moth filled wallet, and ramen meals are the norm...

 

[gehowi]

electric, I fully agree with your opinion from your point of view. Commercial applications usually need to deliver maximum capacity and who cares if all cells have to be replaced after 5 years.
I just wanted to give some useful advise to cyanogenic to protect his private investment and because I have the same cells.
I appreciate your knowledge and I am sure I could learn a lot more from you.

 

[electric]

You can get 10 years from cells charged to 3.5V and even to 3.55V, while using 80%-90% of their actual capacity, but you have to realize that temperature effects aging just as much as voltage and also that calendar aging exists even if you don't use them at all. All 3 factors matter, so take everything into account. Planning for more than 10 years is silly. To me it's not reasonable to pay for 40% extra capacity and not use it, but 20% extra is much more reasonable. 3.4V is a resting open circuit voltage of a fully charged cell, but it is too low for a charge voltage. In my experience 3.5V is the minimum charge voltage, even for very long cell life. Sure you can live with 3.4V charge, but you are fooling yourself if you think it's better than 3.5V.
However, all of this assumes healthy decent quality cells from the factory. If your cells are junk, nothing else matters. Swollen cells or cells with self discharge are garbage. Good cells have no measurable self discharge, you can put them on a shelf for 1-2 years and they keep the same SOC.
Also, for initial pack balance I use this rule of thumb - when highest cell gets to 3.6V, the lowest cell must be at least 3.5V. If not, then keep balancing further. Conversely, when lowest cell gets to 3.5V, highest cell better be below 3.65V or you are in trouble (not immediately, but in the long run). I absolutely cannot stand the idea of bottom balance, it is a huge waste of time. Top balance is much easier and faster, with same end results. Good cells should have less than 2% capacity deviation and even 1AH of mismatch would cause overcharge or over-discharge of high/low cell, so the idea that you can bottom or top balance and not use the BMS is silly. You overpay for unused capacity just to save on BMS, makes no sense. Just top balance and use the BMS, so many BMS choices now days, it's not like 2009 when we had to make our own DIY BMS boards.

P.S. These were just random points in no particular order and not all of them in response to this thread, just something that I see all the time. Just a brain dump, take it or leave it.

P.P.S. If you ask me to prove anything, I won't bother, most of my sources are under NDA's or private, backed by 10 years of experience and 1000s of batteries built.

 

[gehowi]

Thank you for your extra advice! You don't have to prove something. I knew from your first question already, you know very well what you are talking about.
You are right, temperature is important. In my selfmade BMS I monitor individual cell temp as well and I keep them all in a very stable 20° room. Must say that the cells do not vary more than 2° in temperature, even they are degraded and the charge exceeds 50A sometimes.
I think that calendar aging is my trouble. The chinese seller sold me old stock in my first order which I did not knew. So I got 16 garbage cells and all the trouble that comes with it.
I had no experience with large capacity Lifepo4. I had only 8X 110Ah deep cycle Lead acid before. So I thought it was my fault.

To analyze the problems and to get valid prove for the seller I built my own BMS system from scratch 2 months ago which emails me every day all cell, charge and discharge data. So I see the rapid degradation of the crap cells while the other 48V bank works perfect.
It also detects solar panel problems like shadowing from trees and automatically disable the shadowed group before cell damage occurs.


I had no other option than buying new cells (with some price compensation) to get replacement cells or I had no working bank at all. Then I saw the different connection type (other bolts). With the 16 new cells I had no problem at all to balance.
I am still negotiating with BLS battery to get all bad cells replaced but this is China you know.. If they stops communication, there is nothing
I can do. I have to prove everything I say.

I know I can get more energy from my batteries but I keep it safe until all trouble is solved. I (should!) have +- 30KW power from LA and Liion and for now that's enough to bridge 3 days of cloudy weather without charging from mains.
I would not start such investment again nor advice DIY to start one. I think I spent at least the equivalent of 10 years of electricity bills.
May be that's why I want my cells to survive 10 years

 

[electric]

@gehowi your C1 is overcharged at 3.7V. Did you top or bottom balance? You are certainly torturing your C1 right now. Need to bleed it off relative to other cells.

 

[Keith_PDX]

@gehowi
When did you purchase these cells from BLS? I just purchased 8 of the 150Ah from them and the listing states August 2019 date of manufacture. I really appreciate you sharing the M4 vs M6 details. My cells get delivered in two days so I'm excited to see which shows up.

Second question, do you have a bill of materials and source code you'd be willing to share for your esp32 voltage logger? I'd really love to build one but don't have any experience with Arduino or similar.

Thanks for your contributions

 

[gehowi]

@electric, this snapshot was an old situation only to show that the cell monitor shows over/undervoltage or deviation from the average voltage.
3.7 V would be indeed not very good. The BMS turns off the solar inputs one by one, depending on the load, if cells go higher than 3.4 or whatever setting I choose.
I did top balance many times already for the crap cells but after 5 cycles it's the same problem again because the cell capacity is completely mismatched. For the other bank I even never needed to do any balance. At my second order I told BLS battery that I would dispute any battery that would deviate more than 0.05V or had any expansion on arrival. They checked very well this time and it's a perfect match.

I have a question for you: what should be the powerfactor for such 200Ah Lifepo4 cells? If I charge 100A, how many A should I get out? For Lead acid it's 140 in and only 100 out.

 

[gehowi]

@Keith_PDX
I purchased the first (bad) 16 cell set in july 2019. The next 20 cells in august 2019. But I did not get any listing of manufacturing dates and there is no date information at the labels. So they can tell whatever they like.
Based on the picture of the top label, the manufacturer (Wuhu ETC Battery Limited ) cannot even say when those batteries are manufactured and writes: " we think that you purchased un-qualified batteries from China " . I am still in contact with ETC. They don't know BLS Battery. It's just a trader, they say.
However, BLS sends very fast in good package and gives more support than other Chinese sellers. But when it comes to (expensive) replacements under warranty, you will need very good arguments and proves to get something done. I can even understand that because it's easy to ruin your batteries in very short time. So, don't count on any warranty after 6 months.

For the datalogger: it's a work in progress, made for my situation, not industrial and without decent PCB for now.
For 8 batteries you need 8 analog inputs. Since we need to measure hundreds of a volt, the ADC's of an ESP32 are not stable enough and not linear. You can use 2 X I2C 4 channel ADC's like ADS1015 with a separate very stable voltage regulator. Even then you may suffer a lot of interference from large currents in the near cables, especially when higher than 20A.
If you have no experience with such hardware or Arduino, ESP and C++ , it will be hard to get it work. And I don't have the time to support it...

 

[electric]

@electric
I have a question for you: what should be the powerfactor for such 200Ah Lifepo4 cells? If I charge 100A, how many A should I get out? For Lead acid it's 140 in and only 100 out.

When speaking in terms of AH in vs AH out, this is called Coulombic efficiency and LFP chemistry is essentially 100% Coulomb efficient, in other words it's Peukert value is 1.0. Before people point out that nothing in life is 100% efficient, let's clarify that Coulombic efficiency does not account for difference between charge voltage and discharge voltage. So, if you speak in terms of WH in vs. WH out, which accounts for both Amps and Volts, then efficiency becomes largely dependent on your C rates of charge/discharge. Lower C rate means less voltage sag ( during charge sag goes in the other direction, i.e. you must apply higher voltage to push more current in ). So, higher C rate means more WH losses, which is exhibited in cells heating up more.
As an example, if you take a low C rate application, say C/5 ( 5 hours charge + 5 hours discharge ), average Vdiff would be 3.4V-3.2V=0.2V per cell, which means a 200AH cell will lose 40WH in heat out of total nominal 640WH capacity, which corresponds to 93.75% round trip power efficiency.
There are many approximations here, so end result is not precise and depends on multiple factors. Since I used per cell voltages, efficiency will be the same regardless of battery voltage.
Now you can see that it's much easier to operate in Amps and AH and just ignore the Wattage losses. During bulk stage most chargers are in constant current mode, so you are operating in Amps and AH anyway, understanding that heat losses in the system are due to power conversion inefficiency in both the charger and the battery.
Note that approx. 95% power efficiency in the battery is incredible compared to lead acid, which is one of many reasons why Lithium is winning the energy storage market.
Sorry for long winded answer...hope it helps.

 

[gehowi]

@electric,
I had to read your answer twice to fully understand but everything you say is 100% correct. You are a genius.
I am speechless, I have only questions for you

 

[Keith_PDX]

@gehowi
From your experience are these the new or old ETC cells?