A way to accurately measure LiFePO4 SOC?

[PNjunction]

The one thing I didn't take into account is real-world use, like aging. Or murdering the battery a few times.

I'm afraid that the only way to accurately judge is by coulomb-counting. Using voltage alone is rough, but as things age, what used to be generally 50% can now easily be 30% in a few years as things age.

So yeah, voltage alone is not really ideal. Like all batteries, we just can't wing it if we're really into them.

[Ampster]

..........
Another advantage of using 15% to 85%, or 20% to 80%, is that small errors are not significant, in terms of harm to the system.

Thanks for the thorough explanation. Yes, it is nice to have a margin of error.

[JohnCFII]

Did you charge your pack to 29.2 volts (3.65 volts per cell) and count the Amphours that went into the pack?

I agree with you that 50% SOC is the flattest part of the curve, but it isn’t totally flat. Accurately measuring to 1/100 of a volt, under reasonably constant conditions seems, with a known accurate meter, after sufficient “rest”, seems to be working for me.

As far as methodology, I derived 26.32 volts by fully charging the pack, in the proper manner, four times. Each time, I then discharged half of the pack’s rated amp hours, while measuring discharged amp-hours simultaneously with 3 different measuring devices. Each time, the rested pack voltage settled very close to 26.32 volts, after discharging half the rated amp hours. I also did one test each discharging to 51% and 49%. Reading pack voltage to 10 millivolts with my Fluke DVM convinced me that 26.32 resting volts is repeatable enough at 50% for my needs. Now, that is always after at least 10 hours of rest. It has been accurate enough that tests of various Charge Efficiency Factors, and Peukert Exponents always gave sensible results.

Of course, as you well know, trying to determine SOC vs voltage, while charging or discharging is many, many times more meaningless.

I also never charge or discharge at any rate other than between about C8 to C14. Perhaps greater extremes of charge and discharge current might introduce wider variation, but I don’t need to go there.

Your point is well taken, but I’m satisfied with this methodology. In a few weeks, I intend to take some time to re-confirm by again charging to 100%, and counting electrons removed to 50.0% of capacity.

Another advantage of using 15% to 85%, or 20% to 80%, is that small errors are not significant, in terms of harm to the system.

[Ampster]

.........

I’ve found that it is workable to, about once every two weeks, discharge my battery pack in the evening down to where the BMV712 indicates a SOC of exactly 50.0%. In the morning, after overnight rest, I then check the battery pack voltage. I’ve confirmed that 26.32 volts corresponds quite well, under this circumstance, to 50% SOC. ........

My experience is a voltage of 26.32 is at the flattest part of the discharge curve and just a 1/10 percent variance in measurement could be as much as a 20% difference in actual SOC.Discharge curves do vary and I am glad that you have found a workable solution.. i rely on the Coulomb counter in both my BMS and my Inverter to give me an accurate measurment about how much my battery has discharged since it was last reset. Like you, I only charge to about 85 or 90 percent but that is when my Coulomb counter resets to 100% and that happens at least several times a week.
I don't know what process you used to confirm your pack SOC from your point of 26.32 volts (3.2875 per cell).Did you charge your pack to 29.2 volts (3.65 volts per cell) and count the Amphours that went into the pack?

My point is that I believe voltage is not a reliable way to estimate SoC at anywhere in the middle of the discharge curve. I have seen too many people receive new cells that all measure exactly the same whether that was 3.28 volts or 3.30 volts. They assume they were all the same SOC and then when the pack was charged one or two were runners hit 3.65 before the rest of the pack. . To me that is evidence that voltage at that part of the curve is not a reliable measure of SOC.I realize in some of those cases it could have been an issue of different capacity but in most of those cases the cells were all close in capacity. Even though the voltages were all the same when they arrived the SOCs were clearly different.

[JohnCFII]

Forgot to say the one thing about amp hour monitors like my victron bmv 712 is you have to get back to 100% full often (maybe every 10cycles) to manually reset & prevent it from becoming inaccurate.

In my experience, this is not totally correct. I chose the Victron BMV712 mostly because it is one of the few units that allows the user to calibrate/reset the indicated SOC to any level. It does not require 100% SOC to calibrate/adjust.

I installed a Victron BMV712 a few months ago for the express purpose of trying to manage SOC of my large DIY lithium pack, generally keeping SOC between 15% or 20% and 80% to 85%.

I’ve found that it is workable to, about once every two weeks, discharge my battery pack in the evening down to where the BMV712 indicates a SOC of exactly 50.0%. In the morning, after overnight rest, I then check the battery pack voltage. I’ve confirmed that 26.32 volts corresponds quite well, under this circumstance, to 50% SOC. If the pack voltage reads higher than 26.32. Volts, I will reset the BMV712 SOC to read somewhat higher than 50% SOC. Converse if the voltage is less than 26.32 volts. The only “problem”, though a happy one, is that the SOC is so repeatable that I haven’t needed to yet make enough adjustments to develop a “table” of adjustments to SOC that corresponds to the morning pack voltage.

I have found that, for my lithium batteries, and my charge/discharge profiles, setting up the BMV712 with Peukert Exponent set to 1.00, and Charge Efficiency Factor set to 0.99, keeps the BMV712 indicated SOC in closest calibration.

The BMV712 is a very nice unit.

[nifty-stuff]

So you want an *at rest* SOC chart for LFP?

Believe it or not, here you go. From Shorai. LFP motorcycle battery manufacturer.

See the chart in this FAQ, in the "Can I use a lead-acid battery charger or tender" section for the chart. Uh, yeah, don't use a wall-wart motorcycle charger on your 400ah bank - but the voltages are right anyway:

https://shoraipower.com/faq

Divide the 12v chart by 4 for individual cell voltages, and apply to your 48v bank. (the 12v batts have 4 cells inside, right so divide by 4...)

As valid for a 2.5ah LFP motorcycle batt as it is for a 400ah bank.

So - if an individual cell is sitting at rest (4-12 hours no load) 3.585v, then it is 100% full. Don't let it sit like that for long!
And, we can see that once a cell has reached 3.21v, *at rest*, it has reached the 20% knee you don't want to slide past.

So if you want to go by *RESTING* voltage alone, and take your own monitoring into account, this is handy rough estimation for SOC.

Yes, the Shorai's have a slightly different LFP chemical makeup than yours, but at least this is *something*.

I signed-up just to come "like" this post and to say how very helpful (and direct) it is!

Not only am I moving (SLOWLY) up the food-chain (from fun little AA NIMH & solar / LED projects, to small LifePo4 batteries and an inverter), I'm also trying to wrap my head around all the complexities that come with larger batteries, voltages, etc.

... then add the complexity of trying to gauge capacity on LifePo4 batteries with super flat discharge cures. UGH!

Thanks again @pnjuntion for the great reply / info!

[openplanet]

Something else I've been meaning to ask. The current draw during the night on my four, 100ah, 48v modules is at most 1 amp each. So my morning (before sun!) for all intents and purposes can they legitimately be considered to have been resting? In the case, of course, the purpose is to determine SOC from voltage! THanks.

[openplanet]

I have a large 16s LFP pack I use daily for my house power. Consequently, it is always under some load or charge.

The slope of LFP is not very large, but it is there, and it is quite linear between 20% and 80% SOC. Moreover, at 16s, it is much easier for regular meters to read it with enough precision to be useful.

Under typical house loads, as long as the demand has been pretty stable for a little while, I can tell you the SOC of my battery just by looking at the pack voltage readout, to within 5% or so.

Many of the other comments here are also good ones, but don't let the conventional wisdom get you too worried. With common sense, and some time invested up front, you too can become a human battery SOC meter.

Nebster, would you be willing to share a list of Voltage-SOC pairs, maybe in 5% SOC increments? That would be a great help to those of us trying to zero in on values that can work for us. THanks!!!

[Jman]

Forgot to say the one thing about amp hour monitors like my victron bmv 712 is you have to get back to 100% full often (maybe every 10cycles) to manually reset & prevent it from becoming inaccurate.

[nebster]

Another issue is when charging off solar the rate is not solid like AC charging & so at lower rates I have found it too difficult to know where 80% is, you can easily skip past it & be at 90%+. The battery always absorbs it's way to full quite easy even with lower voltages & rates as I understand it.

Yeah, it's pretty tricky to use voltage-based settings to stop at 80% or lower when at a low average rate like you might have with solar. You need a big solar array or a small battery to hit the CV threshold "hard" enough, and that can tough (or impossible) to do. Using a counter and/or a BMS that has something similar built in is the way to go then, for sure.

(And it's also probably the way to go even if you do have a big charger and a very reliable CC/CV regime you can follow.)

[Jman]

I have been trying to monitor my dropins staying between 20-80% with voltage alone & completely gave up. It varies too much with loads so not worth it imo. This was with an accurate DMM at terminals. I found the 20-35% region to be more flatter than the rest of the flat curve.
Another issue is when charging off solar the rate is not solid like AC charging & so at lower rates I have found it too difficult to know where 80% is, you can easily skip past it & be at 90%+. The battery always absorbs it's way to full quite easy even with lower voltages & rates as I understand it.
So I just use an AH counting monitor & program it well. I figure at worst it will be like a cars fuel gauge, not super accurate but gives a good idea.

I found charging at lower voltages just means the battery takes longer to fill up & doesn't just stop at some magic region. 13.4V float voltage seems to be enough to fill the battery up (high 90%) with enough hours.

[nebster]

I have a large 16s LFP pack I use daily for my house power. Consequently, it is always under some load or charge.

The slope of LFP is not very large, but it is there, and it is quite linear between 20% and 80% SOC. Moreover, at 16s, it is much easier for regular meters to read it with enough precision to be useful.

Under typical house loads, as long as the demand has been pretty stable for a little while, I can tell you the SOC of my battery just by looking at the pack voltage readout, to within 5% or so.

Many of the other comments here are also good ones, but don't let the conventional wisdom get you too worried. With common sense, and some time invested up front, you too can become a human battery SOC meter.

[PNjunction]

It is a little bit high due to a slightly different chemistry. But close.

If you split the difference between the 100% and the 90% voltage, one ends up very near 3.45v resting - that would be my maximum at rest. But again, you don't want to be sitting at that for very long.

Sure, they make LFP voltage based soc meters. The Tecmate-Optimate TS-127 meter mainly targeted at 12v motorcycle users gets one in the ballpark with LFP / agm / flooded. I compared it against my Fluke, and it's close enough for shade-tree observations as the led floats around the outer edge of the meter.

Still, at the end of the day, voltage alone, be it ANY battery chemistry is not always the best for determining true SOC (or health), but there ya go. Better than nothing.

[openplanet]

I just looked at the chart you provided a link to. Must say I'm surprised that 57.36v resting is considered 100%. That seems high.

[openplanet]

Thanks very much for this info.
I have Midnite Classics and a Whizbang Jr. so I'll be monitoring charge in/out very diligently.
Seems like whoever comes up with an accurate SOC-o-Meter for LiFePO4s will do well...