Batteries not charging as expected

[nicknorman]

10 minutes ago, IanD said:

Your last paragraph is what you're doing, not me.

 

I'm saying that LFP charging/termination voltage varies with temperature, and have provided both theory and (other people's) measurements to back this up.

 

You're saying the voltage doesn't vary with temperature, and have provided neither theory nor measurements to back this up.

 

This is nothing to do with charge rates or series resistance, just the inbuilt voltage of the reaction that drives the battery -- does it vary with temperature or not?

 

If charge rates are high so ESR makes a difference then things might change -- but for LFP on boats this is not the case, they're all fractional-C.

You have provided data to show the apparent series resistance varies with temperature, which it does, although not hugely within the normal temperature range likely to be found on a boat (0c to 40c). As you profess to know from ohms law, the voltage difference arising from that is proportional to the current and at 0.05C is very small compared to the cell voltage.

 

You have provided no data to show that the “inbuilt voltage of the reaction that drives the battery”, which I take to mean the open circuit voltage, varies with temperature. As far as I know it doesn’t. Just as it doesn’t with an LA battery. If you do have such data to hand then please do share but a look on the internet doesn’t give any results - probably because the co-efficient is effectively zero and therefore uninteresting.

 

So to be clear, the answer to your question “does it vary with temperature or not?” Is “not”. If you want me to take a different view then you will have to provide data, which so far you haven’t. None of your data was at zero current.

 

And let me be clear, if you can find such data I’ll be pleased, because for me this discussion is not about being right in public for the sake of it, it is about understanding the facts and if I misunderstand then I would prefer to be corrected.

[IanD]

1 minute ago, nicknorman said:

You have provided data to show the apparent series resistance varies with temperature, which it does, although not hugely within the normal temperature range likely to be found on a boat (0c to 40c). As you profess to know from ohms law, the voltage difference arising from that is proportional to the current and at 0.05C is very small compared to the cell voltage.

 

You have provided no data to show that the “inbuilt voltage of the reaction that drives the battery”, which I take to mean the open circuit voltage, varies with temperature. As far as I know it doesn’t. Just as it doesn’t with an LA battery. If you do have such data to hand then please do share but a look on the internet doesn’t give any results - probably because the co-efficient is effectively zero and therefore uninteresting.

 

So to be clear, the answer to your question “does it vary with temperature or not?” Is “not”. If you want me to take a different view then you will have to provide data, which so far you haven’t. None of your data was at zero current.

 

And let me be clear, if you can find such data I’ll be pleased, because for me this discussion is not about being right in public for the sake of it, it is about understanding the facts and if I misunderstand then I would prefer to be corrected.

Good, I'm glad we're starting to agree here 🙂

 

As far as I can see the low ESR means that the voltage difference between 0.05C and 0.5C is small (20mV?) for LFP on both charge and discharge, certainly much smaller than the voltage difference between 0C and 45C (>100mV?).

 

It would be nice to have zero current data but this isn't needed -- all we need to know is the relative changes with current (smaller) amd temperature (larger).

 

I find it annoying that LFP suppliers prefer to ignore these differences and just give simplistic numbers, which I'm pretty sure are hiding what actually happens... 😞

[nicknorman]

4 minutes ago, IanD said:

Good, I'm glad we're starting to agree here 🙂

 

As far as I can see the low ESR means that the voltage difference between 0.05C and 0.5C is small (20mV?) for LFP on both charge and discharge, certainly much smaller than the voltage difference between 0C and 45C (>100mV?).

 

It would be nice to have zero current data but this isn't needed -- all we need to know is the relative changes with current (smaller) amd temperature (larger).

 

I find it annoying that LFP suppliers prefer to ignore these differences and just give simplistic numbers, which I'm pretty sure are hiding what actually happens... 😞

No you misconstrue the data and its analysis. If the voltage difference between 0 and 45c is 100mV at 0.5C, then at 0.05C it is 10mv. So if your point is that the correct full charge termination varies between 3.645 and 3.655 ( 14.58v and 14.62v for a 12v system) then we can agree. However in practical terms this difference is insignificant, especially when you bear in mind that “fully charged” is not at all well defined for a Li battery and boat electrical systems are unlikely to be accurate to 40mV.

As to you penultimate sentence, you persist with saying that changes with temperature are very significant whereas changes with current are not, which demonstrates that you don’t understand the process and are immune to any rational input.

[DShK]

I'd like to chime in with - 

 

My understanding is that you want to avoid overcharging. What is overcharging in a lithium battery? My understanding is that it's when the charge it can accept gets so low it is unable to deal with the ripple in the incoming voltage. So this ripple begins damaging the cells.

 

I don't see why voltage would be a better indicator of this than the current the battery will accept.

Edited August 4, 2023 by DShK

[IanD]

31 minutes ago, nicknorman said:

No you misconstrue the data and its analysis. If the voltage difference between 0 and 45c is 100mV at 0.5C, then at 0.05C it is 10mv. So if your point is that the correct full charge termination varies between 3.645 and 3.655 ( 14.58v and 14.62v for a 12v system) then we can agree. However in practical terms this difference is insignificant, especially when you bear in mind that “fully charged” is not at all well defined for a Li battery and boat electrical systems are unlikely to be accurate to 40mV.

As to you penultimate sentence, you persist with saying that changes with temperature are very significant whereas changes with current are not, which demonstrates that you don’t understand the process and are immune to any rational input.

 

Excapt that's only correct if the voltage difference is due to internal resistance -- and it can't be, otherwise at 2C it would be 400mV, and it's considerably less than half of that (<200mV) looking at the charge and discharge curves.

 

Which means that at C/2 -- what I used for the calculations, and very high for LFP on boats! -- the drop due to internal resistance would be something less than 50mV, or 20mV at 0.2C which is a more typical upper limit in real life (100% to 0% SoC in 5 hours).

 

In turn this means the voltage difference due to temperature is still around 80mV-100mV (I rounded it down). Which is a *very* significant SoC change, bigger than the difference between 20% SoC and 80% SoC, and a *lot* bigger than the voltage drop due to internal resistance at boat-level currents -- double even at C/2, 5x bigger at a more typical C/5.

 

That's rational, it's all read straight off the data sheet... 😉

[rusty69]

I'm glad i am too ignorant to understand any of this. Ignorance is bliss.

[IanD]

29 minutes ago, DShK said:

I'd like to chime in with - 

 

My understanding is that you want to avoid overcharging. What is overcharging in a lithium battery? My understanding is that it's when the charge it can accept gets so low it is unable to deal with the ripple in the incoming voltage. So this ripple begins damaging the cells.

 

I don't see why voltage would be a better indicator of this than the current the battery will accept.

Because an LFP battery will carry on accepting current (e.g at 5% of C) until the voltage gets well over 4V/cell (and carries on rising) and it starts to destroy itself. It's not the same mechanism as LA where excess current when it's fully charged just vents hydrogen and oxygen at a roughly constant voltage until it dries out.

 

It's why voltage is the key parameter to watch with LFP batteries, and why they should never be used with a conventional alternator...

[nicknorman]

25 minutes ago, DShK said:

I'd like to chime in with - 

 

My understanding is that you want to avoid overcharging. What is overcharging in a lithium battery? My understanding is that it's when the charge it can accept gets so low it is unable to deal with the ripple in the incoming voltage. So this ripple begins damaging the cells.

 

I don't see why voltage would be a better indicator of this than the current the battery will accept.

Voltage ripple is an interesting issue. My BMS takes 1.6 microseconds (ie virtually instantaneous) to sample each cell voltage so I did wonder if I was going to see a lot of ripple from the alternator. However I don’t, and I suspect this is due to the smoothing effect of the rather large input capacitors built into the Combi. So in my case at least, there is no significant ripple.

So I would say that overcharging is when the voltage is above the maximum specified by the manufacturer (which is down to the chemistry) and the current has decreased to a fairly low value (depending on the excess voltage). And the longer you hold that condition for, the worse it is.

 

Or to put it more usefully, the way to avoid overcharge is to charge to the specified voltage, wait for the current to decrease to 5%, then stop charging. Did I mention this before?

[IanD]

1 minute ago, nicknorman said:

Voltage ripple is an interesting issue. My BMS takes 1.6 microseconds (ie virtually instantaneous) to sample each cell voltage so I did wonder if I was going to see a lot of ripple from the alternator. However I don’t, and I suspect this is due to the smoothing effect of the rather large input capacitors built into the Combi. So in my case at least, there is no significant ripple.

So I would say that overcharging is when the voltage is above the maximum specified by the manufacturer (which is down to the chemistry) and the current has decreased to a fairly low value (depending on the excess voltage). And the longer you hold that condition for, the worse it is.

 

Or to put it more usefully, the way to avoid overcharge is to charge to the specified voltage, wait for the current to decrease to 5%, then stop charging. Did I mention this before?

 

The combis like Victron have massive input capacitors (a hundred thousand microfarads or so?) which do act as an effective ripple filter -- it's also why there can be problems when connecting them up to a system if these aren't precharged to the DC input voltage.

 

Yes you did, and you're absolutely correct 🙂

 

Now, about what the voltage should be, especially at different temperatures... 😉

Edited August 4, 2023 by IanD

[nicknorman]

7 minutes ago, IanD said:

 

Excapt that's only correct if the voltage difference is due to internal resistance -- and it can't be, otherwise at 2C it would be 400mV, and it's considerably less than half of that (<200mV) looking at the charge and discharge curves.

 

Which means that at C/2 -- what I used for the calculations, and very high for LFP on boats! -- the drop due to internal resistance would be something less than 50mV, or 20mV at 0.2C which is a more typical upper limit in real life (100% to 0% SoC in 5 hours).

 

In turn this means the voltage difference due to temperature is still around 80mV-100mV (I rounded it down). Which is a *very* significant SoC change, bigger than the difference between 20% SoC and 80% SoC, and a *lot* bigger than the voltage drop due to internal resistance at boat-level currents -- double even at C/2, 5x bigger at a more typical C/5.

 

That's rational, it's all read straight off the data sheet... 😉

I’ll call it a night because clearly you’ve had plenty of gin already.

Just now, IanD said:

 

Yes you did, and you're absolutely correct 🙂

 

Now, about what the voltage should be, especially at different temperatures... 😉

3.65v per cell for LiFePO4 regardless of temperature.

[IanD]

4 minutes ago, nicknorman said:

I’ll call it a night because clearly you’ve had plenty of gin already.

3.65v per cell for LiFePO4 regardless of temperature.

Not what the numbers I just posted said. Did you actually look at them or just poo-pooh them?

 

Or are you doing a TWC and using your own version of Ohm's Law? 😉

 

P.S. Too much gin -- shome mishtake, shurely?

Edited August 4, 2023 by IanD

[nicknorman]

10 minutes ago, IanD said:

Not what the numbers I just posted said. Did you actually look at them or just poo-pooh them?

 

Or are you doing a TWC and using your own version of Ohm's Law? 😉

 

P.S. Too much gin -- shome mishtake, shurely?

I didn’t much look at your numbers but I did look at the data.

[IanD]

10 minutes ago, nicknorman said:

I didn’t much look at your numbers but I did look at the data.

The numbers are straight from the data...

 

Change in voltage due to current at typical boat rates (0.2C) ~20mV

Change in voltage due to temperature (0-45C) *excluding* current 80mV~100mV

Change in voltage due to SoC (20%-80%) ~100mV

 

Conclusion: charge/discharge rate doesn't make much difference to SoC (so a constant voltage is fine), but temperature makes a big difference (so it isn't).

 

I don't see what other conclusion can be drawn from this -- do you?

Edited August 4, 2023 by IanD

[DShK]

26 minutes ago, IanD said:

Because an LFP battery will carry on accepting current (e.g at 5% of C) until the voltage gets well over 4V/cell (and carries on rising) and it starts to destroy itself. It's not the same mechanism as LA where excess current when it's fully charged just vents hydrogen and oxygen at a roughly constant voltage until it dries out.

 

It's why voltage is the key parameter to watch with LFP batteries, and why they should never be used with a conventional alternator...

 

Surely the voltage only continues to rise if you program the charging device to be allowed to go higher?  As the batteries can't exactly force the charging device to provide higher voltage. If you program everything to charge up to 15V then it's another story...

 

LFP batteries will fully charge at quite low voltages. They will very readily accept charge, so there is no need to allow the voltage to get that high.

[IanD]

1 minute ago, DShK said:

 

Surely the voltage only continues to rise if you program the charging device to be allowed to go higher?  As the batteries can't exactly force the charging device to provide higher voltage. If you program everything to charge up to 15V then it's another story...

 

LFP batteries will fully charge at quite low voltages. They will very readily accept charge, so there is no need to allow the voltage to get that high.

Correct, but you asked why just current couldn't be used -- the answer is exactly what you said, you need a voltage limit as well.

 

LFP are very easy to get charge into, but are also much less good at protecting themselves from destruction than lead-acid -- you need a much more intelligent charging system/BMS than with LA (i.e. not just an alternator!) to keep them charged and not send them to an early grave...

[nicknorman]

4 minutes ago, IanD said:

The numbers are straight from the data...

 

Change in voltage due to current at typical boat rates (0.2C) ~20mV

Change in voltage due to temperature (0-45C) *excluding* current 80mV~100mV

Change in voltage due to SoC (20%-80%) ~100mV

 

Conclusion: charge/discharge rate doesn't make much difference to SoC (so a constant voltage is fine), but temperature makes a big difference (so it isn't).

 

I don't see what other conclusion can be drawn from this -- do you?

Yes I do. If you charge the battery by charging to the specified voltage, holding to 0.05C, then stopping, you will get to 100% SoC within a very small margin.

Your first “Change  in” not relevant because it is at 0.2C whereas the relevant condition is 0.05C

Your second “Change in” is not relevant because it occurs at 0.5C - when you say *excluding* you are attempting to deceive because although current isn’t varying, current is 0.5C and the voltage difference you refer to is down to the varying reaction rate (modelled by varying internal resistance).

Your third “Change in” is not relevant because we are talking about fully charging, not some SoC between 20 and 80% and anyway that is the zero current voltage.

[IanD]

2 minutes ago, nicknorman said:

Yes I do. If you charge the battery by charging to the specified voltage, holding to 0.05C, then stopping, you will get to 100% SoC within a very small margin.

Your first “Change  in” not relevant because it is at 0.2C whereas the relevant condition is 0.05C

Your second “Change in” is not relevant because it occurs at 0.5C - when you say *excluding* you are attempting to deceive because although current isn’t varying, current is 0.5C and the voltage difference you refer to is down to the varying reaction rate (modelled by varying internal resistance).

Your third “Change in” is not relevant because we are talking about fully charging, not some SoC between 20 and 80% and anyway that is the zero current voltage.

I'm not trying to decieve, you said you wanted this explaining -- but when I tried, you don't seem to want to listen.

 

I'm going to have one more go (using the Winston curves, because these have more data) and then give up...

 

1. Internal resistance causes the voltage to vary with current on both charge and discharge.

2. From the plots, the difference on voltage between 0.5C and C is about 50mV on both charge and discharge (internal resistance is similar).

3. From Ohm's law, the difference between 0C and 0.5C should also be about 50mV (so 20mV for 0.2C).

4. The voltage difference at 0.5C between 0C and 55C on discharge (above 90% SoC) is about 150mV (+50mV at 55C, -100mV at 0C).

5. This compares to 20mV change with current at 0.2C, the normal maximum charge current likely to be seen on a boat.

6. The ~20mV change due to current (for a boat) has a *much* smaller effect on cell voltage than the ~150mV change due to temperature.

 

If you disagree with any of this, please point out where -- because as far as I can see these are all simple calculations straight from the data, based on good old Ohm's and Kirchoff's laws and linear superposition 🙂

 

(and the "pure LFP" curves are very similar to the Winston ones, as you'd expect apart from a small change in absolute voltage)

 

Now look at the charge/discharge curves attached. If you choose 3.65V charging/termination voltage (100% SoC) at 25C, this is equivalent to about 3.6V at 55C (undercharged a bit) and 3.75V at 0C (overcharged quite a lot) -- the same 150mV range as in point 4.

 

Everything published about LFP says that a 150mV variation in Vterm is *much* too high to be able to guarantee reaching 100% SoC at high temperature (highest Vcell) without +150mV overcharging at low temperature (lowest Vcell).

 

Night night 🙂

[nicknorman]

As they say, assumption is the mother of all cockups. You have made several false assumptions which are causing you confusion.

 

False assumption 1: The battery can be accurately modelled by means of a voltage source and a resistance.
No, over a wide range of current there is not a “resistor” effect, ie there is a non-linear (non-ohmic) change in voltage with current. This is because whilst there is some ohmic resistance from the terminals to the business end, there is also some “resistance” arising from pushing the reaction fast. It is a similar concept to pushing water through a hose pipe. At low pressure and flow rates, if you double the pressure you more or less double the flow. But at high pressure and flow rates, if you double the pressure you get nothing like double the flow.

 

False assumption 2: The charge and discharge functions are symmetrical - identical characteristics except for the sign of the current.

No, we all know that whilst discharging is allowed down to -20c or so, charging is disallowed below 0c or at least must be severely curtailed. Discharging at low temperatures is normal, charging at low temperatures plates the electrode with metallic lithium. So using data from discharge and applying it to charge is an erroneous strategy.

 

False assumption 3: 100%SoC is precisely defined.

No, even your graphs show that. They go to 120%! So if the battery can take more charge than 100%, then it wasn’t really 100% was it! Why? Because there is a compromise to be stuck between stuffing more charge into the battery vs damage and shortening life. You can charge beyond the nominal 100% but in doing so, you hit a steep part of the cycle life curve. Charging to 100% ie 3.65v/cell until current falls to 5%, is the generally accepted best balance between capacity and life. But a few mV or 10s of mV either way isn’t going to make a significant difference.

 

False assumption 4: where temperatures are quoted, these are the internal temperatures where the reaction is taking place. No, they are ambient temperatures. Which at low current flows will be very close to internal temperatures. But at high current flows such as 0.5C,  the internal temperature will rise significantly.

 

False assumption 5: You need to charge at exactly the specified maximum voltage to get to 100%SoC.

 

No, even if you charge at say 3.5v per cell and hold it until the current has subsided further, you will get to the same SoC. It will just take longer.

 

With so many false assumptions in your work, I really can’t be bothered to work out which is causing you the problem. Maybe all of them.

 

On the earlier subject of open circuit voltage vs temperature, I found a couple of pieces of research. Here is one of the result graphs. It shows virtually no OCV temperature co-efficient at high SoC, which is what I said all along.

 

 

Compare all that with the fact that if you charge to 3.65v/cell at 1C you will be at about 80% SoC, whereas if you charge to 3.65v/cell at 0.05C you will be at 100% SoC, and everyone can see that your claim that monitoring current is not necessary, whereas adjusting charge voltage for temperature is, is incorrect. To put it politely.