SOC estimation using coulomb counting

[Jonndeka]

Hi everyone,

I got stuck in the concept of SOC estimation using Coulomb counting and SOC-OCV curve look up table.

Actually, one of the main issues which arise for me is when the EV is switch on and then you switched it off. After one minute you are going to switch it on and now EV is under operating. So how this method can use OCV-SOC lookup table while the batteries have not reach the steady-state (I mean voltage equilibrium).

We just can use the look up table when the battery have long rest and the battery voltage is a steady-state mode.

Do you have any comments about it?

Thanks,
Jooni

 

[fechter]

If you have a coulometer, you don't need to worry about measuring the voltage. I've seen SOC meters that work only off the voltage and it's hard to get an accurate reading when the load is changing. Probably need to use a microcontroller and some software to track the peaks and calculate from there.

 

[Hillhater]

Voltage measurement can be very difficult with the different discharge curves for various Li chemistries.
LiFePo in particular has a very shallow voltage curve until it is almost empty !

 

[Jonndeka]

If you have a coulometer, you don't need to worry about measuring the voltage. I've seen SOC meters that work only off the voltage and it's hard to get an accurate reading when the load is changing. Probably need to use a microcontroller and some software to track the peaks and calculate from there.

I guess coulometer is used for accumulating Ah overtime and it's not applicable for SOC estimation for a cycle.
Could you please clarify how can I use coulometer for SOC estimation?
If EV is switch off and when we switch on, how coulometer can estimate the initial SOC?

 

[Jonndeka]

Voltage measurement can be very difficult with the different discharge curves for various Li chemistries.
LiFePo in particular has a very shallow voltage curve until it is almost empty !

I agree!
So what's the solution?

 

[docw009]

Could you please clarify how can I use coulometer for SOC estimation?

Start with a full charged battery. Put it under load and monitor the Ah or WH or whatever your "coulumb counter" measures along with the voltage. Take data points. When the battery finally shuts dowm, you will have the discharge curve data, and also know the capacity of your battery. The measured capacity will depend on what load you choose. so you should pick something typical of operation.

The above measurements may be unwieldy for a powerful EV, For my ebike batteries, it's easy to do with inexpensive gear, Nonetheless, it's not really necessary. If you look at data sheets for 4.2V li-ion cells. they are all around 40-50% at a nominal 3.6V, I want to be on my way home when I'm that low.

 

[fechter]

If you know the capacity of your battery and reset the coulometer after a full charge, you'll know the SOC by subtracting the Ahr used from the Ahr capacity. Divide this Ahr remaining by the capacity and you have % remaining. I have a cheap coulometer that has programmable voltage set points that will automatically reset when the voltage gets to the full charge point and gives a display of % remaining.

 

[Jonndeka]

Start with a full charged battery. Put it under load and monitor the Ah or WH or whatever your "coulumb counter" measures along with the voltage. Take data points. When the battery finally shuts dowm, you will have the discharge curve data, and also know the capacity of your battery. The measured capacity will depend on what load you choose. so you should pick something typical of operation.

The above measurements may be unwieldy for a powerful EV, For my ebike batteries, it's easy to do with inexpensive gear, Nonetheless, it's not really necessary. If you look at data sheets for 4.2V li-ion cells. they are all around 40-50% at a nominal 3.6V, I want to be on my way home when I'm that low.

Actually your proposed method is suitable for constant current load or low duty application. I want to use for heavy-duty application and our discharge current is very variable.

 

[Jonndeka]

If you know the capacity of your battery and reset the coulometer after a full charge, you'll know the SOC by subtracting the Ahr used from the Ahr capacity. Divide this Ahr remaining by the capacity and you have % remaining. I have a cheap coulometer that has programmable voltage set points that will automatically reset when the voltage gets to the full charge point and gives a display of % remaining.

In real-word application we always don't have full charge battery and in reality maybe the EV or other electric application might be charged partially. So we can not use fully charged method for SOC estimation. Right?

 

[fechter]

If the coulometer is calibrated to a full charge in the beginning, it can be used for a while, but over time will tend to lose accuracy over time due to errors in the current measurement and losses in the battery. Having both coulomb counting and voltage sensing to correct the measurement periodically is probably the best, but would require some pretty smart software to stay accurate.

 

[lnanek]

They do require some sort of memory or write to flash capability AFAIK - unless you want to start from empty then charge all the way before using, or initialize it with a voltage table lookup and known capacity or something. That's why you see features like "power off memory" for many:

https://a.aliexpress.com/_mMazV1q

Also why things like the Bosch ebike batteries have a way to reset them, holding the button down for ten seconds. It's not a big storage requirement, though. Just like a tiny coin cell keeps PC motherboards from forgetting their BIOS settings.

 

[Jonndeka]

If the coulometer is calibrated to a full charge in the beginning, it can be used for a while, but over time will tend to lose accuracy over time due to errors in the current measurement and losses in the battery. Having both coulomb counting and voltage sensing to correct the measurement periodically is probably the best, but would require some pretty smart software to stay accurate.

Thanks for your reply.
So if you mean "voltage sensing" is estimate SOC through OCV-SOC look-up table for SOC correction or estimate initial SOC, this is my first question.
I'm saying when the battery has not passed the relaxation time, we can not use OCV-SOC look-up table. Because battery voltage didn't reach to steady-state.

Did you mean anything else?

 

[Jonndeka]

My big concern is how can we identify the initial SOC while the battery didn't reach equilibrium voltage (steady-state). In this case we can not use OCV-SOC look-up table.

 

[fechter]

If the coulometer is reasonably accurate, it should give an accurate SOC for at least one charge/discharge cycle. The OCV-SOC lookup can correct the reading whenever the voltage is steady for long enough, like when the battery is idle or reaches either full charge or full discharge voltage.

 

[Jonndeka]

whenever the voltage is steady for long enough

So how long is enough for a LFP cell to reach steady-state voltage?
As you know for some application we don't have constant current and so that the voltage is variable.
I just would like to know how many second/minutes does need the battery reach steady-state and coulometer corrects the new SOC?
This issue is important for identifying of initial SOC when we can not wait long time for battery steady-state.

 

[fechter]

It takes about a minute at zero current. You can test with a voltmeter after applying a heavy load.

 

[Jonndeka]

If the coulometer is reasonably accurate, it should give an accurate SOC for at least one charge/discharge cycle. The OCV-SOC lookup can correct the reading whenever the voltage is steady for long enough, like when the battery is idle or reaches either full charge or full discharge voltage.

We use the current transducer for sensing current.

https://www.lem.com/sites/default/files/products_datasheets/dhab_s_137.pdf

I would like to know coulometer is more accurate than that?

 

[fechter]

Any coulometer uses a current sensor and software to calculate the coulombs. The accuracy of the current sensor determines the accuracy of the coulomb counter. That sensor looks good for very high currents.