FNDC SOC irrelevant to LiFePO4 voltage?

[jainsw]

After a few discussions with the SimpliPhi tech and aimlessly looking at my Optics daily charts and event history I've come to the conclusion the FNDC referenced SoC compared to voltage is off by a BIG margin. I submit the following graphics:
This showing the SoC at 46%:
https://www.dropbox.com/scl/fi/xcall0ko ... z8gfm&dl=0
And this showing the same charge with the low voltage at 51.6v
https://www.dropbox.com/scl/fi/8pjnh3f2 ... 2ihvz&dl=0
So, now if you look at the SimpliPhi manual for SoC you see that 51.65v is 90% SoC.
https://www.dropbox.com/scl/fi/m0pdaksj ... 0y7jf&dl=0
And I found on one of raysun's post from 2021 that the FNDC is hard wired to the chemistry specs of FLA (not a direct quote). So, am I to infer that I shouldn't even look at SoC on Optics and just pay attention to the voltage for determining if I should run the generator or not on cloudy days? I've never let my voltage start the generator even though it is set at 2 minutes @ 50.4v. Usually, I was starting the generator manually when the SoC got below 40%, but it appears that's just a waste of propane because according to the SimpliPhi chart, I'm still above 75%SoC on the battery bank. Thoughts?

[chrisski]

For whatever reason, I have also found SOC to be way off for my FNDC and Outback radian and Firyress eFlex batteries.

So much so, I purchased the monitoring device that hooks directly to my batteries, the fortress guardian, so I could get an accurate SOC.

I do not use a generator. I am grid tied.

[raysun]

I'm not here to defend the FNDC. It doesn't have the ability to adjust Peukerts Constant for LFP, as it is "hard wired" for Lead Acid. This may or may not make a difference, depending on battery rate of discharge.
In my case, an FNDC ran in tandem with a Victron BMV-712 battery monitor for 6 years. In that time the "delta" between the two units' reported SoC averaged less than 1.5%. Two factors that probably contributed:
• Low average rate of battery discharge.
450Ah battery has a fairly steady draw of 5A with peaks around 40A (C90 and C11 discharge rates respecively.) At 5A, the Peukerts calclulations are insignificant contributors to derived SoC.
• Regular full charge.
Battery is typically fully charged every day by about 10:30AM. (Later in winter but still before noon most days.) It's a rare event when it doesn't reach Charged Parameters Met. I actually enjoy those days, as it breaks the monotony.
We did have one month during a volcanic eruption where the battery never reached full charge. The span was 28 days. There was no supplimental charging. At the end of that time, the FNDC and BMV-712 SoC readings disagreed by about 5%.

A generator is used for backup, but most of the time it sits idle. It gets run once a month just to get some exercise. Most of those days I have to waste PV charge to give the generator some work to do. "Real" generator production is less than 3% of total energy use.

SoC is a derived value. It is at best an estimate involving either an inadequate method (voltage measurements under load), or a complex method ("coulomb counting" modified by Peukerts Law and estimates for capacity and charge efficiency.) Other than "FYI", SoC isn't particularly reliable as a metric for automating maintenance functions. But then again, neither is voltage, especially for LFP. What does one do when presented with a Hobson's Choice? Pick the least worst under the circumstances.

Is voltage the least worst? Probably. Its a simple metric to measure, however a hard metric to understand, especially for LFP. We talk about LFP's flat charge/discharge voltage curves as one issue, but generally don't grasp the consequences of another, more important issue. The absolute value of the battery voltage is dependent on rate of discharge. (Damn you Peukerts.)
Take a close look at "Voltage v.s. Capacity" tables for LFP and most frequently they will be for open circuit conditions. Tables relating voltage to capacity under load are generally always at high rates. Other than bench testing, who discharges their battery at a C1 rate, or even C5 or C10? Most of us off-grid folks are targeting C20 rates so there's ample runtime until the next charging opportunity. Battery manufacturers list C1 and C5 frequently, but not longer durations. Note the voltage at C1 and that at C5 for capacity estimates are different values.
Also note that for exteremely low rates of discharge an LFP will show very little voltage change until nearly dead.

Nonetheless, voltage is likely better than SoC as a process trigger.

[raysun]

Good, bad, or indifferent as the FNDC is, at least its not (always) batshit insane like this Sol Ark all in one. Note the deeply discharged battery (48V @ 130Ah LFP) being charged to 34% SoC at 53.9V, then deciding "All good, leap to 100%".
Full charge voltage is 54.4V according to mfgr specs.

[raysun]

The trouble with voltage. Battery at 70% SoC and charging.

1st photo shows 53.85V @ 36A charge current
2nd photo shows 54.07V @ 72A charge current

What's the "real" battery voltage in this case?

SimpliPhi specifies the "Optimal operation parameter" (whatever that means) of 51.4V @ 75%. Most likely this is open circuit.

[fcwlp]

Note the deeply discharged battery (48V @ 130Ah LFP) being charged to 34% SoC at 53.9V, then deciding "All good, leap to 100%".

Most of us with experience with more than one OB system, have also seen an OB FNDC do the magic jump in SOC even on lead acid. What is useful with modern systems is that they actually talk to each, closing the loop and coordinating the charge. My guess is that this graph represents an open loop system.

[raysun]

Yes, open loop.
I "kind of" understand the OB FNDC "Force Float" mechanism for lead acid batteries. However, I don't know what drives the Sol Ark "leap of logic".
Do you?

[dtich]

Is that sharp hockey stick SOC curve at near freezing temps? I find the FNDC SOC calculations near freezing completely worthless. As one might expect actually.

Because of this and other reasons I absolutely use voltage triggers and not SOC for AGS.

Of course voltage will vary with load, lower with higher discharge current, and higher with higher charge current.

Discharging: voltage drops with more current;
Charging: voltage rises with more current;
Both are caused by internal resistance and battery chemistry and,
Cold amplifies both effects.

No current in or out, resting state, is the only way to accurately measure a battery's true SOC via voltage, and still one must account for temp. the effective capacity of lithium batteries is substantially derated in cold, so the FNDC does its math wrong.

So, you can have an SOC of 20% w a per cell voltage of 3.30vdc, and an 80% SOC with per cell at 3.32, hardly a difference in voltage, but a huge difference in charge. voltage derived SOC measurements (estimates) vary tremendously with load and temp. the FNDC does use coulomb counting largely, however, as i understand it and seems to be from empirical experience, will fall back on voltage-derived when it senses quick changes, eg. a large jump up to absorb voltage - as happens in cold-weather charging, which can fool the FNDC into thinking the battery has hit full charge when it very much hasn't. low temps are just not friendly to electricity management. until you reach super conductor territory, lol.

[raysun]

Is that sharp hockey stick SOC curve at near freezing temps? I find the FNDC SOC calculations near freezing completely worthless. As one might expect actually.

In Hawaii, no freezing temps.

[dtich]

ah. verrry strange then.

[jainsw]

During this time of year, October-Februray, in the mountains of Colorado, I generally gnash my teeth, pull hair, and kick snow drifts about why my energy storage needs to be supplemented by my generator. Could be snow on the panels when I'm gone, or a snow event that comes and goes taking 48 hours, or just those days when the sun comes up nice and strong over the mountains to the east, then clouds move in mid day, then the sun comes out before sitting below the mountains to the west. I think my batteries are good and sufficient for my application but the jest of it is, I likely need more panels for this time of year and an inverter/ccs that will charge faster. .

[raysun]

What is the model / specs of the generator?

[jainsw]

During this time of year, October-Februray, in the mountains of Colorado, I generally gnash my teeth, pull hair, and kick snow drifts about why my energy storage needs to be supplemented by my generator. Could be snow on the panels when I'm gone, or a snow event that comes and goes taking 48 hours, or just those days when the sun comes up nice and strong over the mountains to the east, then clouds move in mid day, then the sun comes out before sitting below the mountains to the west. I think my batteries are good and sufficient for my application but the jest of it is, I likely need more panels for this time of year and an inverter/ccs that will charge faster. .

Honestly, what you described is so real Winter in the mountains is just brutal for solar. The issue usually isn’t the batteries, it’s that your winter solar production just can’t keep up. That’s super common in places like the Colorado mountains. You’re thinking about it the right way. The real fix usually isn’t more battery, it’s more panels and a system that can charge faster when the sun does show up. Your setup actually sounds pretty solid , it’s just that Colorado winter is out there playing dirty with your system

Indeed winter and solar don't mix well. Today was another day of limping by with clouds most of the day. I did set my FM60 to reboot every 90 minutes so that may have helped a bit since I discovered it was "tapping out" and only logging about 1/2 the capacity it has shown to produce on a sunny day. I'm looking at changing out the old 190 watt Evergreen solar panels that have been standing for 17 years and replacing them with some 390s. That should put us over the goal line. Now, I just have to figure out what to do with the FM60 if it is necessary to replace it. I may just go to a different inverter with built in charge controllers.

[raysun]

I may just go to a different inverter with built in charge controllers.

I'm working with a neighbor who has one of the top-of-the-line All-in-One units. I'm a bit shocked at how poorly it performs, and that's after I helped them through their issues and had the company tech support folks review and confirm its optimally configured.

[raysun]

the SoC calculation can be way off when paired with a flat-voltage-curve chemistry like SimpliPhi. That’s why you're seeing 46% SoC while the battery voltage clearly indicates you’re still around 90%.

This is not entirely true. The FNDC, and every "coulomb counting" monitor does not derive SoC from battery voltage, so the different voltage curves have little impact on the derivation. The monitors track current and rate of charge / discharge as the driving metric.
That's not to say voltage is disregarded entirely. It is used as a factor in the Peukerts calculation, as rate of discharge causes a voltage drop in both lead acid and lithium ferro phosphate. The voltage behavior is embodied in Peukerts Constant, typically 1.25-1.50 for LA, and 1.05 for LFP.
The source of potential inaccuracy in the FNDC SoC calculation is due to its fixed Peukerts Constant set to LA value, rather than being adjustable for LFP chemistry. How large the error depends on how large the rate of discharge. For smaller discharge rates (e.g. 20 hour and longer) the error is very low, on the order of <2% by my measurements over the past six+ years. At higher rates of discharge (e.g. 1 hour) the error would be greater.

In this kind of setup, relying on FNDC-reported SoC will almost always lead to early generator starts and wasted fuel. Voltage is actually a much more reliable indicator for LiFePO4 — especially during low loads and morning checks — because the chemistry holds voltage very steadily until it gets close to empty.

If SoC is a poor choice of triggers, then voltage is even worse, given the very attribute described above: flat voltage curve during discharge. At low rates of discharge, the voltage between 80% SoC and 20% SoC can be as little as 0.1V With such a small voltage "delta" how can one set a reliable trigger? The pragmatic answer is to observe the system under "typical" discharge rates and note the voltage while in the "neighborhood" of the desired battery capacity at which charging should commence.
Note this will be no less "fuel wasting" than using a charging trigger based on SoC. The "saving grace" in the ambiguity of both SoC and Voltage start with LFP is charging below virtually 100% full is constant current with the battery accepting the full rate of charge. Also LFP is perfectly happy at partial state of charge, so if charging is terminated below 100% SoC, no harm done. If there is a long series of partial charges, its suggested to execute an occasional full charge to allow the BMS to "top balance" the cells and the monitor to sync @ 100%.

[jainsw]

After a few discussions with the SimpliPhi tech and aimlessly looking at my Optics daily charts and event history I've come to the conclusion the FNDC referenced SoC compared to voltage is off by a BIG margin. I submit the following graphics:
This showing the SoC at 46%:
https://www.dropbox.com/scl/fi/xcall0ko ... z8gfm&dl=0
And this showing the same charge with the low voltage at 51.6v
https://www.dropbox.com/scl/fi/8pjnh3f2 ... 2ihvz&dl=0
So, now if you look at the SimpliPhi manual for SoC you see that 51.65v is 90% SoC.
https://www.dropbox.com/scl/fi/m0pdaksj ... 0y7jf&dl=0
And I found on one of raysun's post from 2021 that the FNDC is hard wired to the chemistry specs of FLA (not a direct quote). So, am I to infer that I shouldn't even look at SoC on Optics and just pay attention to the voltage for determining if I should run the generator or not on cloudy days? I've never let my voltage start the generator even though it is set at 2 minutes @ 50.4v. Usually, I was starting the generator manually when the SoC got below 40%, but it appears that's just a waste of propane because according to the SimpliPhi chart, I'm still above 75%SoC on the battery bank. Thoughts?

You're not wrong , what you're seeing is a very common issue when using the FNDC with LiFePO4 batteries. The FNDC was originally designed around lead-acid behavior, so the SoC calculation can be way off when paired with a flat-voltage-curve chemistry like SimpliPhi. That’s why you're seeing 46% SoC while the battery voltage clearly indicates you’re still around 90%.

In this kind of setup, relying on FNDC-reported SoC will almost always lead to early generator starts and wasted fuel. Voltage is actually a much more reliable indicator for LiFePO4 — especially during low loads and morning checks — because the chemistry holds voltage very steadily until it gets close to empty.

So yes, on cloudy days, using voltage as your reference for generator decisions is the better approach. And from the numbers you shared, it really does look like you've been starting the generator much earlier than needed.

Thank you for that confirmation! I thought I would look back over the past 60 days and the worse day was November 24th in which my lowest voltage was 50.9v and the SoC measured it at 28%. According to the SimpliPhi chart the actual SoC for 51v is 50% so I was just shy of that for a bit. Interestingly, the hour prior to that my lowest voltage was 51.4v but I had a load (hot water kettle to make coffee) that came on and pulled the voltage down to 50.9v (the average that hour was 51.03v and the highest voltage for that hour was 51.1v). Prior to those two hours (5am and 6am), it looks like I was loosing .1v per hour using about 3-4kWh per hour. The voltage at 6am that dropped my bank down from 51.4v to 50.9v was 7kWh. That kettle takes a lot of juice but you gotta have coffee! Sorry for such a long reply.

[llucklinn]

I found this discussion interesting because the difference between the FNDC’s reported SoC and the battery voltage could have a major impact on generator decisions. Basket Random

[fcwlp]

On a system that I acquired the maintenance of the reported SOC and voltage were significantly different. It is a 24V LFP bank that originally had 33 24V 50AH batteries. Batteries were configured with 2 strings of 6 and 3 strings of 7 connected to a bus bar with multiple different lengths of 4/0 ranging from 5' to 15' in length. It now has 5 sets of 6 batteries in parallel with equal length cables and a name plate capacity of 1,500 AH. Batteries are about 3 years old. A couple of weeks ago there was a temperature excursion in the room due to running the gen without the covers on, resulting in a room temp of 160+F. Only reason we know that is that at 5:30 in the morning, the FM80s reported over temp faults.
.
What I have found over the past couple of weeks is that the actual battery capacity is closer to 800AH. By adjusting the FNDC down to 800AH bank size, the SOC is getting closer to matching what is actually being consumed overnight. So, if your SOC is way off take a look at what you are actually consuming overnight and how that relates to battery voltage.
.
For open loop LFP systems, I always use voltage for AGS.