Measuring state of charge in LiFePo...

[TTman]

It seems to be a problem for solar based batteries with a 48V
stack.Manufactureres say it\'s real tricky... temperature/charge
rate/discharge rate blah blah. So why is it so hard to get an accurate
SOC ? 0%= ~40/41V 100% =~57V ... A 12 bit ADC would yield 4mV
resolution. Isn\'t that enough to get an accurate state of charge? My
analogue experience is next to nil..If there is a real problem, how
about 3 ADCs each measuring low middle and high voltage? i.e 40-45;
46-51;52-58 or is that a stupid idea ?

--
This email has been checked for viruses by Avast antivirus software.
www.avast.com

 

[TTman]

On 26/02/2023 17:53, TTman wrote:

It seems to be a problem for solar based batteries with a 48V
stack.Manufactureres say it\'s real tricky... temperature/charge
rate/discharge rate blah blah. So why is it so hard to get an accurate
SOC ? 0%= ~40/41V 100% =~57V ... A 12 bit ADC would yield 4mV
resolution. Isn\'t that enough to get an accurate state of charge? My
analogue experience is next to nil..If there is a real problem, how
about 3 ADCs each measuring low middle and high voltage? i.e 40-45;
46-51;52-58 or is that a stupid idea ?

EG 9.5 kW battery by Givenergy ( UK/China)

--
This email has been checked for viruses by Avast antivirus software.
www.avast.com

 

[Anthony William Sloman]

On Monday, February 27, 2023 at 4:53:43 AM UTC+11, TTman wrote:

It seems to be a problem for solar based batteries with a 48V
stack. Manufacturers say it\'s real tricky... temperature/charge
rate/discharge rate blah blah. So why is it so hard to get an accurate
SOC ? 0%= ~40/41V 100% =~57V ... A 12 bit ADC would yield 4mV
resolution. Isn\'t that enough to get an accurate state of charge? My
analogue experience is next to nil..If there is a real problem, how
about 3 ADCs each measuring low middle and high voltage? i.e 40-45;
46-51;52-58 or is that a stupid idea ?

You have to know the temperature of the battery which gets warmer when you pull current out of it or put current into it.

Ideally, you want a temperature sensor in the middle of the battery, which battery manufacturers don\'t provide. If you know the currents flowing you do know the heat being generated in the core of the battery, and can correct the temperature you measure just out side the cell, but it is tricky.

You need a second ADC to keep track of the temperature and you need to have a reasonable idea of the heat capacities and thermal resistances involved.

It\'s not trivial.

--
Bill Sloman, Sydney

 

[Pete]

On 27/02/2023 4:53 am, TTman wrote:

why is it so hard to get an accurate
SOC ? 0%= ~40/41V 100% =~57V

Because SOC is not reflected in a straight line voltage change between
full and flat.

Your LiFePo battery does not start with full charge at ~57v and linearly
drop down to no charge at ~40/41V

 

[Jan Panteltje]

On a sunny day (Sun, 26 Feb 2023 17:53:36 +0000) it happened TTman
<kraken.sankey@gmail.com> wrote in <ttg6b1$2vjj0$1@dont-email.me>:

It seems to be a problem for solar based batteries with a 48V
stack.Manufactureres say it\'s real tricky... temperature/charge
rate/discharge rate blah blah. So why is it so hard to get an accurate
SOC ? 0%= ~40/41V 100% =~57V ... A 12 bit ADC would yield 4mV
resolution. Isn\'t that enough to get an accurate state of charge? My
analogue experience is next to nil..If there is a real problem, how
about 3 ADCs each measuring low middle and high voltage? i.e 40-45;
46-51;52-58 or is that a stupid idea ?

Charging lifeo4 or any other multi-cell battery configuration involves quite a few things.
For starters you want to know about the status of each individual cell.
In some chargers parallel to each cell there exists a voltage limiter
to prevent over-discharging that specific cell if its full while the rest is still being charged
Cells age and cells differ.
There exist, at least for other battery chemistries, special charger chips
that have several analog inputs just for this purpose,
google is your friend here
https://uk.farnell.com/c/semiconductors-ics/power-management-ics-pmic/battery-management/battery-charger-ics?battery-type=lifepo4

I have considered building a charger that powers each cell individually,
so treats the batteries in the chain each as 1 cell (so without that parallel limiter on each cell).
Takes a bit more electronics... Not a problem when using transformers... should be more efficient,
Have not tried that yet.

But then, the current systems seem to work.
(the one that came with my 3 cell system broke down for an other battery chemistry
so that is how I know about this as I went looking for the chip).
For lifepo4 above 90% charge the voltage rises fast, so 90% as turn off point may work:
https://cleversolarpower.com/lifepo4-voltage-chart/

I have several lifepo4 batteries.
The big 250 AH box has a build in charge controller
and came with a charger for 230V 50 Hz,
The small ones I charge on the lab supply,
using some Lifepo4 charge graph I got from the web
and monitor time, voltage and set the current.
Monitoring battery temperature is important, as the flat part of the charge is strongly
dependent on that,
When charging below 1C , say .3C the battery temperature will not rise that fast,
But in general it is not so critical
I do look at the amount of power send into the cell when charging
and that, if it was empty use about a bit more than the advertised capacity to refill...
I know its a hack but seems to work
Anyways..

 

[Ed Lee]

On Monday, February 27, 2023 at 12:49:32 AM UTC-8, Jan Panteltje wrote:

On a sunny day (Sun, 26 Feb 2023 17:53:36 +0000) it happened TTman
kraken...@gmail.com> wrote in <ttg6b1$2vjj0$1...@dont-email.me>:
It seems to be a problem for solar based batteries with a 48V
stack.Manufactureres say it\'s real tricky... temperature/charge
rate/discharge rate blah blah. So why is it so hard to get an accurate
SOC ? 0%= ~40/41V 100% =~57V ... A 12 bit ADC would yield 4mV
resolution. Isn\'t that enough to get an accurate state of charge? My
analogue experience is next to nil..If there is a real problem, how
about 3 ADCs each measuring low middle and high voltage? i.e 40-45;
46-51;52-58 or is that a stupid idea ?

I use multiple of 12V. My maximum is 12.6V (50.4V for 48V system). 57V will kill the cells.

Charging lifeo4 or any other multi-cell battery configuration involves quite a few things.
For starters you want to know about the status of each individual cell.
In some chargers parallel to each cell there exists a voltage limiter
to prevent over-discharging that specific cell if its full while the rest is still being charged
Cells age and cells differ.
There exist, at least for other battery chemistries, special charger chips
that have several analog inputs just for this purpose,
google is your friend here
https://uk.farnell.com/c/semiconductors-ics/power-management-ics-pmic/battery-management/battery-charger-ics?battery-type=lifepo4

I have considered building a charger that powers each cell individually,
so treats the batteries in the chain each as 1 cell (so without that parallel limiter on each cell).
Takes a bit more electronics... Not a problem when using transformers... should be more efficient,
Have not tried that yet.

I built a 32x array. It worked for a while, but a short somewhere blew up half of them. So, i have not tried it again. It scared some of my friends badly, since it was next to the batteries. I said: \"relax, it\'s just charger fire, not battery fire\".

I was planning to charge them with 4 to 8 generators. So, running an octo engine EV.

 

[TTman]

But then, the current systems seem to work.
(the one that came with my 3 cell system broke down for an other battery chemistry
so that is how I know about this as I went looking for the chip).
For lifepo4 above 90% charge the voltage rises fast, so 90% as turn off point may work:
https://cleversolarpower.com/lifepo4-voltage-chart/

I have several lifepo4 batteries.
The big 250 AH box has a build in charge controller
and came with a charger for 230V 50 Hz,
The small ones I charge on the lab supply,
using some Lifepo4 charge graph I got from the web
and monitor time, voltage and set the current.
Monitoring battery temperature is important, as the flat part of the charge is strongly
dependent on that,
When charging below 1C , say .3C the battery temperature will not rise that fast,
But in general it is not so critical
I do look at the amount of power send into the cell when charging
and that, if it was empty use about a bit more than the advertised capacity to refill...
I know its a hack but seems to work
Anyways..

Ah... I see the problem now 10% to 90% is just 200mV.Add some variables-
charge/discharge current v temp and it must get realy tricky. My 9.5kW
186 AH battery can charge/discharge at 3.6kW from my inverter ( 80A
max.) Problem with this manufacturer is that the SOC software falls over
with anything other that full charge then full discharge DOD=0%. You can
watch the charge and at some point ( i.e. 75%) the SOC software suddenly
decides the battery is at 100% and so stops charging.


--
This email has been checked for viruses by Avast antivirus software.
www.avast.com

 

[Anthony William Sloman]

On Monday, February 27, 2023 at 3:05:18 PM UTC+11, Anthony William Sloman wrote:

On Monday, February 27, 2023 at 4:53:43 AM UTC+11, TTman wrote:
It seems to be a problem for solar based batteries with a 48V
stack. Manufacturers say it\'s real tricky... temperature/charge
rate/discharge rate blah blah. So why is it so hard to get an accurate
SOC ? 0%= ~40/41V 100% =~57V ... A 12 bit ADC would yield 4mV
resolution. Isn\'t that enough to get an accurate state of charge? My
analogue experience is next to nil..If there is a real problem, how
about 3 ADCs each measuring low middle and high voltage? i.e 40-45;
46-51;52-58 or is that a stupid idea ?
You have to know the temperature of the battery which gets warmer when you pull current out of it or put current into it.

Ideally, you want a temperature sensor in the middle of the battery, which battery manufacturers don\'t provide. If you know the currents flowing you do know the heat being generated in the core of the battery, and can correct the temperature you measure just out side the cell, but it is tricky.

You need a second ADC to keep track of the temperature and you need to have a reasonable idea of the heat capacities and thermal resistances involved.

It\'s not trivial.

If you want more detail

Electrochimica Acta 60 (2012) 269– 273 can tell you more. the paper title is

Effects of temperature on charge/discharge behaviors of LiFePO4 cathode for Li-ion batteries

Authors are

Lixia Liaoa,b, Pengjian Zuoa, Yulin Maa, XinQun Chena, Yongxin Ana, Yunzhi Gaoa,∗, Geping Yina,∗∗
a School of Chemical Engineering and Technology, Harbin Institute of Technology, No.92, West Da-Zhi Street, Harbin 150001, China
b College of Science, Northeast Forest University, No.26, Hexing Road, Harbin 150040, China

Figure 2 is informative - not as informative as a more intelligible exposition would have been, but it gives you some idea of what happens.

The Arrhenius equation shows up just once in the article and not in a useful place.

https://shop.gwl.eu/blog/LiFePO4/What-is-the-behavior-of-a-LiFePO4-battery-at-low-winter-temperatures.html

is a bit more explicit but it confuses internal resistance with capacity, which isn\'t helpful.

--
Bill Sloman, Sydney

 

[Anthony William Sloman]

On Tuesday, February 28, 2023 at 12:12:13 AM UTC+11, TTman wrote:

But then, the current systems seem to work.
(the one that came with my 3 cell system broke down for an other battery chemistry
so that is how I know about this as I went looking for the chip).
For lifepo4 above 90% charge the voltage rises fast, so 90% as turn off point may work:
https://cleversolarpower.com/lifepo4-voltage-chart/

I have several lifepo4 batteries.
The big 250 AH box has a build in charge controller
and came with a charger for 230V 50 Hz,
The small ones I charge on the lab supply,
using some Lifepo4 charge graph I got from the web
and monitor time, voltage and set the current.
Monitoring battery temperature is important, as the flat part of the charge is strongly
dependent on that,
When charging below 1C , say .3C the battery temperature will not rise that fast,
But in general it is not so critical
I do look at the amount of power send into the cell when charging
and that, if it was empty use about a bit more than the advertised capacity to refill...
I know its a hack but seems to work
Anyways..

Ah... I see the problem now 10% to 90% is just 200mV.

And where that 200mV range is does depend on the battery temperature (if not all that much), If there is current moving in or out of the cell its internal resistance changes the voltage you see at the terminals. You can beat that by measuring the voltage when you aren\'t moving current into or out of the battery, but that means high current switches, which cost money. Modulating the current moving in and out lets you detect the internal resistance from the consequent modulation in the battery voltage, but that tend to end up in the \"much too hard\" basket.

Add some variables- charge/discharge current v temp and it must get really tricky. My 9.5kW
186 AH battery can charge/discharge at 3.6kW from my inverter ( 80A
max.) Problem with this manufacturer is that the SOC software falls over
with anything other that full charge then full discharge DOD=0%. You can
watch the charge and at some point ( i.e. 75%) the SOC software suddenly
decides the battery is at 100% and so stops charging.

Internal resistance tends to change as the battery ages. If somebody has made a bad assumption ...

--
Bill Sloman, Sydney

 

[legg]

On Sun, 26 Feb 2023 17:53:36 +0000, TTman <kraken.sankey@gmail.com>
wrote:

It seems to be a problem for solar based batteries with a 48V
stack.Manufactureres say it\'s real tricky... temperature/charge
rate/discharge rate blah blah. So why is it so hard to get an accurate
SOC ? 0%= ~40/41V 100% =~57V ... A 12 bit ADC would yield 4mV
resolution. Isn\'t that enough to get an accurate state of charge? My
analogue experience is next to nil..If there is a real problem, how
about 3 ADCs each measuring low middle and high voltage? i.e 40-45;
46-51;52-58 or is that a stupid idea ?

Make sure the cells are actively balanced. This removes single-cell
dominance in the string, allowing generalizations to work in
capacity estimation for varying discharge rates.

RL