Lithium batts and Ah counters

[Col_T]

It is generally held that Ah counters are useless for determining the state of charge of lead acid batteries.

 

is this also true for lithiums?

[WotEver]

1 minute ago, Col_T said:

It is generally held that Ah counters are useless for determining the state of charge of lead acid batteries.

 

is this also true for lithiums?

Not so much, but it’s an irrelevant measurement because you must have a BMS and that BMS will provide all of the info you require. 

[MoominPapa]

1 hour ago, Col_T said:

It is generally held that Ah counters are useless for determining the state of charge of lead acid batteries.

 

is this also true for lithiums?

No. They work much better for Lithium batteries. There are at least two reasons why.

 

1) the Peukert exponent of LiFePo4 cells is essentially zero. Assuming it is zero is much closer to reality than any of the estimated values used with LA batteries.

 

2) Fully charging LA batteries involves running them into a state where some of the electrical energy is not used for the reversible chemical reactions that will produce power on discharge, but instead for splitting hydrogen from water and heating the battery. This electrical energy gets counted going into the battery, but doesn't correspond to electricity that will come out again, adding to the errors. LiFePo4 cells have much higher efficiency and don't lose much energy this way. Plus the BMS has to avoid such "controlled overcharging" because it damages the cells.

 

MP.

[peterboat]

I have noticed they go out of sync, I need to do a reset and tell it how many KWHs it should really have

[nicknorman]

1 hour ago, MoominPapa said:

No. They work much better for Lithium batteries. There are at least two reasons why.

 

1) the Peukert exponent of LiFePo4 cells is essentially zero. Assuming it is zero is much closer to reality than any of the estimated values used with LA batteries.

 

2) Fully charging LA batteries involves running them into a state where some of the electrical energy is not used for the reversible chemical reactions that will produce power on discharge, but instead for splitting hydrogen from water and heating the battery. This electrical energy gets counted going into the battery, but doesn't correspond to electricity that will come out again, adding to the errors. LiFePo4 cells have much higher efficiency and don't lose much energy this way. Plus the BMS has to avoid such "controlled overcharging" because it damages the cells.

 

MP.

Yikes, can I just mention that a Peukert exponent of zero would be catastrophic! Some variable raised to the power zero will always =1. So nothing sensible could come out of such an equation. A Peukert exponent of 1 on the other hand...

 

And can I also mention that Peukert’s exponent doesn’t enter into a calculation about SoC in a properly-designed battery monitor.

2 hours ago, WotEver said:

Not so much, but it’s an irrelevant measurement because you must have a BMS and that BMS will provide all of the info you require. 

I disagree that it is an irrelevant measurement. Whilst as you say it is important to have a BMS, I don’t think all BMSs give you SoC. Their core function is to prevent over-charge or over-discharge, and maybe to balance the cells.

[peterboat]

7 hours ago, nicknorman said:

Yikes, can I just mention that a Peukert exponent of zero would be catastrophic! Some variable raised to the power zero will always =1. So nothing sensible could come out of such an equation. A Peukert exponent of 1 on the other hand...

 

And can I also mention that Peukert’s exponent doesn’t enter into a calculation about SoC in a properly-designed battery monitor.

I disagree that it is an irrelevant measurement. Whilst as you say it is important to have a BMS, I don’t think all BMSs give you SoC. Their core function is to prevent over-charge or over-discharge, and maybe to balance the cells.

My Bms will give a state of change, but as you say that isn't it's primary job

[MoominPapa]

8 hours ago, nicknorman said:

Yikes, can I just mention that a Peukert exponent of zero would be catastrophic! Some variable raised to the power zero will always =1. So nothing sensible could come out of such an equation. A Peukert exponent of 1 on the other hand...

Indeed. It was late when I wrote that.

8 hours ago, nicknorman said:

 

And can I also mention that Peukert’s exponent doesn’t enter into a calculation about SoC in a properly-designed battery monitor.

Doesn't it? For Ah counters it must, or how does the battery monitor take into account that 20Ah removed over one hour reduces the SoC more than 20Ah removed over two hours? Or are you saying that a "properly-designed battery monitor" is not an Ah counter? That may be true, but it's off-topic for the original question. Is there an equivalent of the SmartGauge for LiFePo4?

8 hours ago, nicknorman said:

I disagree that it is an irrelevant measurement. Whilst as you say it is important to have a BMS, I don’t think all BMSs give you SoC. Their core function is to prevent over-charge or over-discharge, and maybe to balance the cells.

Having designed and built a BMS, I'd say that determining SoC is necessary for it to perform charge termination, but that's not a universal view..

 

MP.

 

1 hour ago, peterboat said:

My Bms will give a state of change, but as you say that isn't it's primary job

I think it needs to determine SoC to perform its primary job. It's probably doing so, at least in part, by Ah counting.

 

MP.

 

[nicknorman]

19 minutes ago, MoominPapa said:

 

... or how does the battery monitor take into account that 20Ah removed over one hour reduces the SoC more than 20Ah removed over two hours?

 

But it doesn’t. You misunderstand Peukert.

[MoominPapa]

8 minutes ago, nicknorman said:

But it doesn’t. You misunderstand Peukert.

Well. Peukert gives a method of determining the capacity of the battery, which is required to determine the percentage state-of-charge, so it does figure in the calculation. Of course, at a given state-of-charge, the number of amp-hours remaining depends on all of, the capacity of the battery, the number of Ah which have already been withdrawn and the rate at which they were withdrawn, and the rate at which the remaining Ah _will_ be withdrawn. Given the unknowns there, (and the extra ones to do with charge efficiency) it's no surprise that Ah counters are crap on LA batteries.

 

My point is that for LiFePo4, and for fractional-C discharges (ie all of them, for a boat house battery) the calculation becomes independent of the rate of discharge. That makes an Ah counter much easier to implement and much more accurate. The main remaining source of error is small systematic errors in current measurement, which accumulate over time.

 

MP.

 

 

 

[WotEver]

9 hours ago, nicknorman said:

And can I also mention that Peukert’s exponent doesn’t enter into a calculation about SoC in a properly-designed battery monitor.

 

Yeah, sure, if you say so Nick. 

[nicknorman]

28 minutes ago, MoominPapa said:

Well. Peukert gives a method of determining the capacity of the battery, which is required to determine the percentage state-of-charge, so it does figure in the calculation. Of course, at a given state-of-charge, the number of amp-hours remaining depends on all of, the capacity of the battery, the number of Ah which have already been withdrawn and the rate at which they were withdrawn, and the rate at which the remaining Ah _will_ be withdrawn. 

 

MP.

 

Peukert does give a method for determining the capacity of a battery at different discharge rates. But it is important to understand what “capacity” means in this context. A standard capacity measure is to take a fully charged battery and discharge it at the specified rate until the terminal voltage reaches 10.5v. Ah extracted from the battery gives its capacity at that discharge rate.  Peukert helps to predict that at higher discharge rates, the 10.5 terminating voltage is reached having discharged fewer AH than if the discharge rate had been lower.

 

But this has nothing to do with measuring an intermediate SoC when the future discharge rate is unknown (and in the case of a leisure battery, probably on average pretty slow.)

 

Consider how a battery works - by combining some chemicals that, with each molecule of reactant, mobilises one electron (or is it 2, can’t remember). As an electronics person you will know Kirchoff’s first law which is effectively that current is preserved in a circuit. You can’t, for instance “drop current” in the same way that you can “drop voltage” around a circuit. And current is of course the rate of change of charge, which we normally measure in AH. So charge can’t vanish around a circuit. For a fixed amount of chemicals there are a fixed number of electrons which translates into a fixed amount of charge.

 

The only way electrons are “lost” in the process, is by the by-product of batteries which is the conversion of water into hydrogen gas and oxygen gas. This accounts for the charge inefficiency (ie you need to put more AH back in than you took out). But during discharge, gassing effects are minimal and certainly not what Peukert is describing.

 

Where Peukert comes in is describing the problem of accessing all the chemicals for reaction, when some are buried deep in the plates and take time to make their way to a position where they can react. So if you discharge fast, the less accessible chemicals don’t have the opportunity to react before the terminal voltage reaches 10.5v. But having discharged the battery fast and reached the 10.5 terminating voltage, if you go away for a few hours you would come back to find the voltage significantly recovered and you can carry on extracting current before the terminating voltage is once again reached - this being because the inaccessible chemicals have migrated towards the surface of the plates. So the overall capacity of the battery is what it is, it’s just that you can’t access all of it in a hurry.

 

Certainly in the case of a leisure battery setup, if you discharge your fully charged 100AH batteries at 25 A for 2 hours and then stop on one occasion, and on another occasion at 10A for 5 hours, the remaining amount of chemicals and hence capacity is identical in each case, and if you leave the batteries alone for a while for the chemicals to equalise, the batteries will be in virtually identical states. Peukert is only relevant if you wanted to discharge at 25A constantly until the battery was nominally flat (ie reached 10.5v) and this is why in a battery monitor, the “time to run to flat at the present discharge rate” calculation does use Peukert.

 

Of course none of this is about power / energy efficiency. If you discharge a battery fast, those ah you take out are worth less energy because the terminal voltage is lower, than if you had taken the same AH out slowly. But Peukert doesn’t describe this effect at all and of course doesn’t have any notion of voltage in the equation - it is only about current.

8 minutes ago, WotEver said:

 

Yeah, sure, if you say so Nick. 

I do.

Edited August 28, 2019 by nicknorman

[WotEver]

40 minutes ago, nicknorman said:

Peukert doesn’t describe this effect at all...

Peukert doesn’t describe any effect whatsoever. Peukert merely provides a simple, empirically derived equation that fits the observable phenomenon. That’s the bit that you’ll never accept, and why your understanding of the matter will always fall short of complete. It is not all about diffusion as you continue to maintain. 

[nicknorman]

12 minutes ago, WotEver said:

Peukert doesn’t describe any effect whatsoever. Peukert merely provides a simple, empirically derived equation that fits the observable phenomenon. That’s the bit that you’ll never accept, and why your understanding of the matter will always fall short of complete. It is not all about diffusion as you continue to maintain. 

No I absolutely agree with your first part. It does provide a simple, empirically derived equation that fits the observable phenomena - when taking a fully charged battery and discharging it to its nominal “flat” state, at various fixed discharged rates. Clearly a useful concept for, eg, a UPS (standby power supply) in determining how long the UPS can maintain the load following mains failure, at a known discharge rate.

 

But that scenario is not relevant to a leisure battery SoC gauge. And it is absolutely about diffusion.

 

Anyway, perhaps you would care to explain how you would apply Peukert to a display of SoC when you have no idea what the future discharge rate will be. And how come a fast discharged battery, when discharged to flat, and then left for a few hours, recovers capacity and is no longer flat?

 

Or even provide some specific counter argument, as opposed to just saying I’m wrong?

Edited August 28, 2019 by nicknorman

[mrsmelly]

That bloke Peukert has a hell of a lot to answer for innitt!!

[Sea Dog]

20 minutes ago, mrsmelly said:

That bloke Peukert has a hell of a lot to answer for innitt!!

It does generate an interesting discussion though - unless it degenerates! 

[Dr Bob]

12 hours ago, peterboat said:

I have noticed they go out of sync, I need to do a reset and tell it how many KWHs it should really have

Lots of high tec stuff above but you dont need to know any of that.

Tony mentioned you dont need to measure SoC as your BMS will do that for you???????

My battery monitor (Victron BMV) is part of the BMS although in its BMS role it is only monitoring voltage.

I never looked at my BMV for SoC with the LA's and I never look at it for the Lithiums. Why would you?

I estimate my state of charge of my lithiums in my head based on firstly voltage at rest (having ploted a graph of voltage vs amp hours used) and secondly the amp hours used figure.

The biggest issue I have is that as I am only using the range 30% to 80% SoC and only take it to 100% after 2-3 months. In this time, the BMV amp hours used figure can drift. For example, I reckon I was around 20Ahrs out after the last 12 week period between excursions up to 100% - similar to what peter said. I just paid more attention to the voltage, making sure it was above 12.8V (at rest) and the charge sources disconnecting at 13.6V.

It really isnt an issue though as you dont need to know the SoC accurately if you are between 30 and 80% charged.

To get to 100%, I use voltage and tail current (ie charging from a 30A charger),  I aim for a voltage of 13.9V and a tail current dropping from 30A to 20A. That is 98% full and I can reset the meter.

Edited August 28, 2019 by Dr Bob

[WotEver]

1 hour ago, mrsmelly said:

That bloke Peukert has a hell of a lot to answer for innitt!!

That’s the problem though innit... he didn’t have any answers, only observation. He was as puzzled as anyone else. 

37 minutes ago, Dr Bob said:

Tony mentioned you dont need to measure SoC as your BMS will do that for you???????

No he didn’t. Read it again.

 

He said that the BMS will provide all of the information you require, as you yourself confirm later in your post. 

[peterboat]

2 hours ago, Dr Bob said:

Lots of high tec stuff above but you dont need to know any of that.

Tony mentioned you dont need to measure SoC as your BMS will do that for you???????

My battery monitor (Victron BMV) is part of the BMS although in its BMS role it is only monitoring voltage.

I never looked at my BMV for SoC with the LA's and I never look at it for the Lithiums. Why would you?

I estimate my state of charge of my lithiums in my head based on firstly voltage at rest (having ploted a graph of voltage vs amp hours used) and secondly the amp hours used figure.

The biggest issue I have is that as I am only using the range 30% to 80% SoC and only take it to 100% after 2-3 months. In this time, the BMV amp hours used figure can drift. For example, I reckon I was around 20Ahrs out after the last 12 week period between excursions up to 100% - similar to what peter said. I just paid more attention to the voltage, making sure it was above 12.8V (at rest) and the charge sources disconnecting at 13.6V.

It really isnt an issue though as you dont need to know the SoC accurately if you are between 30 and 80% charged.

To get to 100%, I use voltage and tail current (ie charging from a 30A charger),  I aim for a voltage of 13.9V and a tail current dropping from 30A to 20A. That is 98% full and I can reset the meter.

At the moment my bank is sitting at 65.6 volts it thinks it has 50.94 AH remaining  which it thinks is 13.59KWH ! All this equates to it thinks that I have 22% left in my batteries, I suspect its wrong so I will fully charge them and reset the metter, like you I tend to believe the voltage rather than anything else

[MoominPapa]

5 hours ago, nicknorman said:

Peukert does give a method for determining the capacity of a battery at different discharge rates. But it is important to understand what “capacity” means in this context. A standard capacity measure is to take a fully charged battery and discharge it at the specified rate until the terminal voltage reaches 10.5v. Ah extracted from the battery gives its capacity at that discharge rate.  Peukert helps to predict that at higher discharge rates, the 10.5 terminating voltage is reached having discharged fewer AH than if the discharge rate had been lower.

 

But this has nothing to do with measuring an intermediate SoC when the future discharge rate is unknown (and in the case of a leisure battery, probably on average pretty slow.)

 

Consider how a battery works - by combining some chemicals that, with each molecule of reactant, mobilises one electron (or is it 2, can’t remember). As an electronics person you will know Kirchoff’s first law which is effectively that current is preserved in a circuit. You can’t, for instance “drop current” in the same way that you can “drop voltage” around a circuit. And current is of course the rate of change of charge, which we normally measure in AH. So charge can’t vanish around a circuit. For a fixed amount of chemicals there are a fixed number of electrons which translates into a fixed amount of charge.

 

The only way electrons are “lost” in the process, is by the by-product of batteries which is the conversion of water into hydrogen gas and oxygen gas. This accounts for the charge inefficiency (ie you need to put more AH back in than you took out). But during discharge, gassing effects are minimal and certainly not what Peukert is describing.

 

Where Peukert comes in is describing the problem of accessing all the chemicals for reaction, when some are buried deep in the plates and take time to make their way to a position where they can react. So if you discharge fast, the less accessible chemicals don’t have the opportunity to react before the terminal voltage reaches 10.5v. But having discharged the battery fast and reached the 10.5 terminating voltage, if you go away for a few hours you would come back to find the voltage significantly recovered and you can carry on extracting current before the terminating voltage is once again reached - this being because the inaccessible chemicals have migrated towards the surface of the plates. So the overall capacity of the battery is what it is, it’s just that you can’t access all of it in a hurry.

 

Certainly in the case of a leisure battery setup, if you discharge your fully charged 100AH batteries at 25 A for 2 hours and then stop on one occasion, and on another occasion at 10A for 5 hours, the remaining amount of chemicals and hence capacity is identical in each case, and if you leave the batteries alone for a while for the chemicals to equalise, the batteries will be in virtually identical states. Peukert is only relevant if you wanted to discharge at 25A constantly until the battery was nominally flat (ie reached 10.5v) and this is why in a battery monitor, the “time to run to flat at the present discharge rate” calculation does use Peukert.

 

Of course none of this is about power / energy efficiency. If you discharge a battery fast, those ah you take out are worth less energy because the terminal voltage is lower, than if you had taken the same AH out slowly. But Peukert doesn’t describe this effect at all and of course doesn’t have any notion of voltage in the equation - it is only about current.

Interesting: I'd always wondered where the extra charge goes. Though I suspect it's not as simple as you describe - there must be ways by which electrons can travel between the electrodes without passing through the external circuit, and those paths may be favoured at high discharge rates. 

 

HOWEVER. None of this is relevant to the original question, as none of this behaviour is apparent in LiFePo4 batteries at fractional-C discharges. If you start with a "full" LLiFePo4 battery where full is an arbitrary state, discharge it by x coulombs and then recharge it by x coulombs, it will be in the same "full" state to an accuracy which is determined more by how well you can measure charge flow than anything else. If you define another arbitrary state of the battery as "empty", where a "full " battery becomes "empty" after discharging by C Coulombs,  then the current percentage SoC of the battery is given by 100*(C-x)/C and that number is a useful measure of the SoC of the battery.. It allows me to know things like: I usually use between 25% and 30% per day, the current SoC is 45%, so I can stay 'till this time tomorrow, but probably not another night after that.

 

MP.

[Lily Rose]

My head hurts!

[nicknorman]

27 minutes ago, MoominPapa said:

Interesting: I'd always wondered where the extra charge goes. Though I suspect it's not as simple as you describe - there must be ways by which electrons can travel between the electrodes without passing through the external circuit, and those paths may be favoured at high discharge rates. 

 

HOWEVER. None of this is relevant to the original question, as none of this behaviour is apparent in LiFePo4 batteries at fractional-C discharges. If you start with a "full" LLiFePo4 battery where full is an arbitrary state, discharge it by x coulombs and then recharge it by x coulombs, it will be in the same "full" state to an accuracy which is determined more by how well you can measure charge flow than anything else. If you define another arbitrary state of the battery as "empty", where a "full " battery becomes "empty" after discharging by C Coulombs,  then the current percentage SoC of the battery is given by 100*(C-x)/C and that number is a useful measure of the SoC of the battery.. It allows me to know things like: I usually use between 25% and 30% per day, the current SoC is 45%, so I can stay 'till this time tomorrow, but probably not another night after that.

 

MP.

Have a read of this if you are interested, although as you say, nothing to do with lithiums:

 

http://www.bogartengineering.com/wp-content/uploads/docs/PeukertsComments.pdf

 

[Detling]

18 hours ago, Col_T said:

It is generally held that Ah counters are useless for determining the state of charge of lead acid batteries.

 

is this also true for lithiums?

It is not the Ah that is usually wrong it is the Soc percentage as this needs the capacity to be accurate and as that drifts with age, temperature, and a hundred other things it renders percentages meaningless if not frequently reset. Smartgauge uses a clever algorithm from measured voltage over time and is surprisingly accurate usually. Lead acid change from 12.8 ish to 12 volts from 100% to 0%, quite a range. Lithium change voltage a lot less and then only at the ends of their capacity is 95% is almost the same as 5% voltage wise, the volt change being outside those elbows. This is where the BMS cuts off the charge or the load to protect the battery. To get a decent idea of the capacity left then Ah counting is the only thing going to get near, the much commented on Peukert does not apply to Lithium so set it to 1 and the charge efficiency is about 98% not all amp hour counters use either of those parameters and some don't allow you to set them. Trying to read SoC from voltage on Lithium really is a non starter as the temperature variation over a 20 degree range can exceed the voltage variation between 25% and 75% Soc.

[Dr Bob]

13 minutes ago, Detling said:

To get a decent idea of the capacity left then Ah counting is the only thing going to get near, the much commented on Peukert does not apply to Lithium so set it to 1 and the charge efficiency is about 98% not all amp hour counters use either of those parameters and some don't allow you to set them. Trying to read SoC from voltage on Lithium really is a non starter as the temperature variation over a 20 degree range can exceed the voltage variation between 25% and 75% Soc.

Incorrect.

Voltage is a great way to estimate SoC.

I have been doing this for the last 5 months.

Counting amp hours is a problem when you don't sync the meter.

Have you got lithium's?

[cuthound]

32 minutes ago, Detling said:

It is not the Ah that is usually wrong it is the Soc percentage as this needs the capacity to be accurate and as that drifts with age, temperature, and a hundred other things it renders percentages meaningless if not frequently reset. Smartgauge uses a clever algorithm from measured voltage over time and is surprisingly accurate usually. Lead acid change from 12.8 ish to 12 volts from 100% to 0%, quite a range. Lithium change voltage a lot less and then only at the ends of their capacity is 95% is almost the same as 5% voltage wise, the volt change being outside those elbows. This is where the BMS cuts off the charge or the load to protect the battery. To get a decent idea of the capacity left then Ah counting is the only thing going to get near, the much commented on Peukert does not apply to Lithium so set it to 1 and the charge efficiency is about 98% not all amp hour counters use either of those parameters and some don't allow you to set them. Trying to read SoC from voltage on Lithium really is a non starter as the temperature variation over a 20 degree range can exceed the voltage variation between 25% and 75% Soc.

 

12.7 to 10.5 volts is the generally accpeted range for wet type lead acid batteries.

[MoominPapa]

26 minutes ago, Dr Bob said:

Incorrect.

Voltage is a great way to estimate SoC.

I have been doing this for the last 5 months.

Counting amp hours is a problem when you don't sync the meter.

Have you got lithium's?

Below are three graphs, First is Soc (%) for the last four days, battery current, and battery voltage: you can see the Dr Bob is right.

 

Second is the same, but with an estimate of open-circuit voltage by linear regression, rather than actual terminal voltage.

 

Third show the discharge in Ah by Ah counting, an estimate based on voltage, and the best estimate output from a Kalman filter which is using both the other sources of data,