LiFePO4 Batteries: Discussion Thread for Those Using Them as House Banks

[john61ct]

Originally Posted by CatNewBee

if you follow the recommendations of john and the others with shallow charging and staying at 50% your balancers will fail over time to their jobs properly, one cell will climb to 3.65V while the other is far away, eventually the bms will then engage at 3.65V if set so and you will experience a capacity loss

No idea what you mean by "staying at 50%"?

Nor is the range MaineSail's talking about actually "shallow charging"; wrt to actual usable SoC, very close to the same capacity result as your much higher stop-charge setting.

And following my reco's for well designed balancing gear (if that becomes required), going to such high settings becomes completely optional, perfectly balanced results without the longevity-reducing stress.

[Cpt Pat]

Originally Posted by john61ct

Do not waste time trying to get two different SoC meters to match up at more than one SoC point, ideally near your defined stop-discharge point

which should be nowhere near zero SoC.

None is ever exact, a guesstimate within 4-6% is the best to hope for with today's technology.

Choose to believe the one you think is closer to accurate, or (my choice) the one that causes you to stop (slow down) discharging sooner,

lower average DoD% will greatly extend battery lifetime.

At low current rates, overall LVC or at least alarms should be like 3.2Vpc. At high C-rates maybe 3.1V, I would not go lower than that at rest, treat 3.0V as drop-dead cutoff, try to get recharged up past 3.2V asap.

Sounds like you've already read this:

https://marinehowto.com/programming-a-battery-monitor/

Since you mentioned the link above, I just read it. There's not a word about lithium batteries in the document. Do a text search for "lithium." It's all about lead/acids. The author accurately describes all the lead/acid battery variables that make coulomb counting onerous with lead/acids. Most of those variables don't exist with lithium batteries: ramp-up cycles needed to reach full capacity, significant Peurkert effects, highly variable charge efficiencies, etc. You shouldn't let those issues dissuade you from coulomb counting with LiFePO4 batteries.

BTW: I've stopped shaking my head over all the mentions of: "this voltage equals that SOC." Unless you are referring to the cell voltage with no load and after hours of resting after a charge, where 3.43 volts equals 100% SOC (give or take a few tens of millivolts), you might as well be discussing the measurement of fairies or unicorns.

I've given up on making the point that because, applied long enough, any voltage of 3.4 volts and above per cell can result in overcharging a LiFePO4 battery. Any voltage quoted, without stating the charge current associated with that voltage and stating whether it is a voltage measurement made during charging versus in a resting state -- is meaningless. And a cell measuring 3.65 volts (which is frequently quoted) with a very low charge current invokes fingers-down-a-chalkboard cringes because that is a very badly overcharged cell. At higher charge currents, well that may be OK. So state the current! You guys do own ammeters, right?

I'll defer to this document to explain that in detail: Marine lithium batteries in operation | Nordkyn Design. I am too exasperated to explain it in detail again.

Here's an excerpt, if you don't want to read the whole thing:

"LFP cells simply don’t really charge at voltages up to 3.3V and then fully charge already at 3.4V and upwards. The transition is so abrupt that claiming to control the charging process by adjusting the voltage is purely and simply bound to fail."

There's a serious failure to understand and correlate voltage versus current in this discussion, and in cases where the charge current is continually variable (like for those of us who sail actual sailboats lacking diesel-driven alternators and are using PVs instead for charging), coulomb counting is the only viable option for controlling the charging of lithiums.

[john61ct]

Originally Posted by Cpt Pat

I've given up on making the point that, applied long enough, any voltage of 3.4 volts and up per cell can result in overcharging a LiFePO4 battery.

Do not expect members in general to "remember" points you've repeated even hundreds of time.

Most do not read other current threads much less past posts in such long ones. But where it is important, quite a few of us try to keep educating on the voltage vs SoC and current factor. Don't give up 8-)

WRT SoC meters, yes, some may be more accurate with LFP but not necessarily. Calibrating against precisely timed CC load tests I haven't found many that get closer than 4%.

[Cpt Pat]

(session timed out)

I agree with the stop-discharge voltages you quoted above. Those voltages should yield a stop-discharge at about 20% SOC - a good conservative value that won't get you in trouble if there's moderate cell imbalance at the bottom of the curve. You can use cell voltages to determine state-of-discharge -- but you can't use voltage alone to determine state-of-charge.

My contention is the misapplication of the article linked in your post vis-à-vis lithium batteries. Coulomb counting is much simpler and more reliable with lithium batteries versus lead/acids.

> Calibrating against precisely timed CC load tests I haven't found many that get closer than 4%.

That's my expectation too. When I recalibrate coulomb counting, I get about +/- 5% accuracies with the actual 100% SOC. That's one reason why I stop-charge at an 80% coulomb-counted (indicated) SOC. But in my operational environment, all other options are less accurate than coulomb counting. There's just too much variation in charge current harvested from PVs.

[alctel]

Chiming in here to say I also use the lower voltages for my cells - I think my alt regulator disconnects at 3.45 (to protect diodes), the rest of the charge sources turns off at 3.475, followed by the charge bus relay at 3.5 and the main 'oh ****' relay at 3.55

[Cpt Pat]

Originally Posted by alctel

Chiming in here to say I also use the lower voltages for my cells - I think my alt regulator disconnects at 3.45 (to protect diodes), the rest of the charge sources turns off at 3.475, followed by the charge bus relay at 3.5 and the main 'oh ****' relay at 3.55

You can still overcharge lithiums at 3.45 volts per cell -- if it's applied long enough. That's why you should never "trickle charge" lithiums.

The battery manufacturers specifications are written for a specific charge current (not less or more) charging the cells to a specific voltage -- with all charging stopped once that voltage at that specific current is attained.

Those conditions are really not achievable in an operational environment, and that's where all the correlative voltage-versus-current complications come in. I've solved that issue by counting amp/hours taken out and put back into the battery (coulomb counting). It's imperfect, but close enough if given wide tolerances, like 20% to 80% operational SOC ranges.

[alctel]

Yeah, I have turned off 'float' for everything (by setting it really low)

Solar is the only thing that is always 'on' anyway, my alternator and tow generator are normally turned off unless I'm at around 50% battery

[john61ct]

Originally Posted by Cpt Pat

you should never "trickle charge" lithiums.

The battery manufacturers specifications are written for a specific charge current (not less or more) charging the cells to a specific voltage -- with all charging stopped once that voltage at that specific current is attained.

Those conditions are really not achievable in an operational environment

I find that statement odd.

In your solar-only case your current rate is under 0.03C correct, you use Ah counting to terminate charge since your Bulk rate is lower than endAmps.

However that is a very unusual circumstance.

If one is charging at say 0.3-4C, there is no problem doing CC-only charge termination as you describe.

In fact it then becomes OK to go to a much higher stop-voltage 3.55 - 3.60V, if you calibrate that point to a gentle ending SoC%.

since there is no Absorb//CV holding time at all, and once current input stops, the cell voltage drops down to the usual resting say 3.34Vpc, or even lower if loads are active.

How is any of that "not achievable in an operational environment"?

[Cpt Pat]

Originally Posted by john61ct

I find that statement odd.

In your solar-only case your current rate is under 0.03C correct, you use Ah counting to terminate charge since your Bulk rate is lower than endAmps.

However that is a very unusual circumstance.

If one is charging at say 0.3-4C, there is no problem doing CC-only charge termination as you describe.

In fact it then becomes OK to go to a much higher stop-voltage 3.55 - 3.60V, if you calibrate that point to a gentle ending SoC%.

since there is no Absorb//CV holding time at all, and once current input stops, the cell voltage drops down to the usual resting say 3.34Vpc, or even lower if loads are active.

How is any of that "not achievable in an operational environment"?

If your battery manufacturer specs call for exactly 0.3C or 0.4C charging (most specs that I've seen are "0.5C"), and if you can maintain exactly that charge current continuously, then it is achievable. You'll need very sophisticated regulation on your alternator with sensing and feedback to compensate for other variable house loads during charging, and maybe a governor controlling your engine RPM.

In my case, I now have 325 amp/hours of LiFePO4 batteries, and there is no way my PVs will produce exactly 0.3C (30%) of that current, and do so constantly. (That pesky Sun keeps moving and setting, and I can't afford gyro stabilizers to keep my PVs aimed exactly at the Sun.) I will be very impressed if anyone's PVs can achieve that! 98 amps at a nominal 14 volts? Very impressive! That's a minimum of 1400 watts of PVs!

Adjusting for the actual harvest efficiency, I'd only need about a 60 foot boat to have enough area for the PVs - covering every square inch of deck.

I'm not a Fanboy of lithium batteries. (https://www.urbandictionary.com/define.php?term=Fanboy) I'm an engineer, and so practicality is paramount. Coulomb counting is practical in my implementation. And simple.

[newhaul]

Originally Posted by Cpt Pat

I will be very impressed if anyone's PVs can achieve that! 98 amps at a nominal 14 volts? Very impressive! That's a minimum of 1400 watts of PVs!

I'd only need about a 60 foot boat to have enough area for the PVs - covering every square inch of deck.

Or a lagoon 440

[john61ct]

No need to be precise, ballpark's close enough.

0.4C is just an example, even 0.1C is well above endAmps, can still do CC-only stop-charge, or go to holding some CV if you have a reason to.

And not talking solar either, whatever the source, bank don't care.

My point is your stop-charging based on coulomb-counting is a very unusual choice,

of course it's fine, excellent for your use case, not saying otherwise, great solution, dictated by a very unusually low C-rate.

I also am no LFP fanboy, in fact believe they just aren't needed, nor suitable at all, for most owners

[JmanC]

Originally Posted by john61ct

Yes those cell level voltages are very far apart and indicate the BMS is faulty.

Use another BMS or dedicated balancing gear to do the job, many options there.

The factors to look at are:

"start balance" voltage should be well below **your preferred** Absorb voltage setpoint, and that really should be adjustable IMO. Some balancers will balance at any voltage, and need IMO to be controlled to only do so as SoC is nearing the top at the end of the charge cycle

"balance current" should be high enough that your cells - in their current condition **and** as they wear over the next few years - get to fully balanced before the point **where you prefer** to end the charge cycle. No such thing as too high, but note that most give a high spec, actual current may depends on the voltage delta between cells.

______
I consider 3.65Vpc to be a very high target / stop-charge / Absorb voltage, especially for normal cycling usage, and most especially if needs to artificially be held for a long time just to get balancing finished.

And yes such poor design of balancing circuitry is the norm for most cheap BMS, and even many that cost more than say $100.

If you do convert your balancing process to another BMS you could get a "non-protective" version with a better design

and keep your existing BMS - after thorough testing - for its protective functions - just keep your normal cycling charge voltage setpoint lower than its start-balance setpoint, in effect disabling its balance function.

Or use dedicated balancing gear, only connected when needed.

Some even adapt RC hobby chargers for that purpose.

Just do not let multiple balancing devices work at the same time on the same set of cells.

Thanks John, appreciate the help. My turn off with bare cells is building the housing and all the gear needed. being a minimalist and all.

[JmanC]

My capacity test is running a generic load on battery until somewhere down near 11.5V and counting AH consumed with monitor.
My understanding is between even 12V under load & 10.5V there is roughly only 2% capacity.

From my tests on my newish battery when doing this capacity test there is no diff between 0.1 & 0.3C loads, they both yield the same AH consumed with very small differences.
A few full discharges at the same rate always gives slightly different capacity, varying usually by 0.5-1AH

So, What about for an older lfp battery that's aged & worn? Do I need a high rate of .3C? Or is 0.1C good enough for obtaining a number for the battery monitor?

[CatNewBee]

Originally Posted by Cpt Pat

Since you mentioned the link above, I just read it. There's not a word about lithium batteries in the document. Do a text search for "lithium." It's all about lead/acids. The author accurately describes all the lead/acid battery variables that make coulomb counting onerous with lead/acids. Most of those variables don't exist with lithium batteries: ramp-up cycles needed to reach full capacity, significant Peurkert effects, highly variable charge efficiencies, etc. You shouldn't let those issues dissuade you from coulomb counting with LiFePO4 batteries.

BTW: I've stopped shaking my head over all the mentions of: "this voltage equals that SOC." Unless you are referring to the cell voltage with no load and after hours of resting after a charge, where 3.43 volts equals 100% SOC (give or take a few tens of millivolts), you might as well be discussing the measurement of fairies or unicorns.

I've given up on making the point that because, applied long enough, any voltage of 3.4 volts and above per cell can result in overcharging a LiFePO4 battery. Any voltage quoted, without stating the charge current associated with that voltage and stating whether it is a voltage measurement made during charging versus in a resting state -- is meaningless. And a cell measuring 3.65 volts (which is frequently quoted) with a very low charge current invokes fingers-down-a-chalkboard cringes because that is a very badly overcharged cell. At higher charge currents, well that may be OK. So state the current! You guys do own ammeters, right?

I'll defer to this document to explain that in detail: Marine lithium batteries in operation | Nordkyn Design. I am too exasperated to explain it in detail again.

Here's an excerpt, if you don't want to read the whole thing:

"LFP cells simply don’t really charge at voltages up to 3.3V and then fully charge already at 3.4V and upwards. The transition is so abrupt that claiming to control the charging process by adjusting the voltage is purely and simply bound to fail."

There's a serious failure to understand and correlate voltage versus current in this discussion, and in cases where the charge current is continually variable (like for those of us who sail actual sailboats lacking diesel-driven alternators and are using PVs instead for charging), coulomb counting is the only viable option for controlling the charging of lithiums.

In theory a full cell has a resting voltage around 3.4...3.43V, but you barely will charge a large cell to 100% at this voltage in a timely matter. BTW, it translates to 13.6...13.72V pack voltage, above the recommended 13.5V float.

I f you have infinite time you may charge a cell full, but not in a normal cycle with real world off the shelf chargers.

3.65V is the 100% SOC charging voltage for LiFeYPO4 / LiFePO4 cells and 4.00V is the highest safe voltage before a cell starts to degrade.

You can argue for longevity and advocate lower setpoints, and you can do what you want with your battery, but I barely would seriously consider advice from someone who does not know the difference of Ah and Amp/hours.

Sorry for this statement, but we had discussed it 10 times or more in this thread alone. IT IS AMP * HOURS nor AMP / HOURS.

1 Amp is the flow of 1C/s or in writing 1 Coulomb per second. 1 Amp / h would result in 3600 C / (h*h) , what would be an infinite acceleration and not a capacity of a storage of 3600 Coulombs, what a Ah represents.

[Cpt Pat]

Originally Posted by CatNewBee

You can argue for longevity and advocate lower setpoints, and you can do what you want with your battery, but I barely would seriously consider advice from someone who does not know the difference of Ah and Amp/hours.

Sorry for this statement, but we had discussed it 10 times or more in this thread alone. IT IS AMP * HOURS nor AMP / HOURS.

Ah, yes. The ad hominem attack. The last refuge of the defeated debater.

See https://acronyms.thefreedictionary.com/A%2FH

"A/H" is an accepted typographic convention -- not a mathematical formula. You say to-MAY-doe and I say to-MA-doe. Shall we argue about how many angels can dance on the head of pin? Because this isn't an argument -- it's just contradiction.

https://www.youtube.com/watch?v=ohDB5gbtaEQ