Voltage as indicator of LFP charge level

[tanglewood]

A bunch of people will probably have a fit over this, but here goes. After a few weeks running my house on LFP, it looks to me like bank voltage is a perfectly good measure of SOC, just as it is with LA. Not perfect, but close enough to have a good sense where you are.


As with LA, you can only pay attention to voltage when there is a modest, steady load. Voltage sags with large loads, and take a little time to recover. But nothing like the hours that people profess. I've never seen it take more than maybe a minute or two for voltage to stabilize.


Also required is descent voltage metering. The voltage swing between full and empty isn't huge, but it's easily detectable with a good metering system.


On my LA systems, I stopped using SOC meters well over 10 years ago, and never looked back. On my 48V house system, fully charged the batteries are 50.4V (25.2V), and 48.0V (24.0V) when ready to recharge at about 50% SOC. That's a spread of 2.4V (1.2V) that pretty accurately correlates with SOC.


On my LFP system, there is actually MORE spread in voltage between my max charge point, and my generator-start recharge point. After a full charge bank voltage is 53.6 (26.8V), and at recharge to it's at 50.4V (25.2)V. That's a spread of 3.2V across the SOC range, and from what I'm seeing, it drops in a way that reasonably correlates with SOC.



So from what I'm seeing so far, voltage is a plenty accurate SOC reading, at least for my needs. Maybe this is because I don't really care what the SOC is down to a few percent, or even 10%. Around 20% is more than enough. Really, what matters the most is knowing when they are full, and when they are getting close to, or in need of recharge by the generator.

[poiu]

I agree, but decide charge level based on the lowest cell when discharging and the highest when charging.

[tanglewood]

Originally Posted by poiu

I agree, but decide charge level based on the lowest cell when discharging and the highest when charging.

My warnings and disconnect protections are based on per-cell voltages. But charger limits and generator auto start are based on bank voltage. I'll get ample warning if any cells are straying.

[rgleason]

Top: (26.8V) or 3.35vpc

Bottom: (25.2V) or 3.15vpc



This is a .2v difference at cell level and at 48vdc it would be .2v x 16 = 3.2v difference.

[tanglewood]

Originally Posted by rgleason

Top: (26.8V) or 3.35vpc

Bottom: (25.2V) or 3.15vpc



This is a .2v difference at cell level and at 48vdc it would be .2v x 16 = 3.2v difference.

Looks right. Are you just restating, or is there a question in there?

[nebster]

Yes, if you go back in one of those other monster threads from a bit ago, I mentioned that I have gotten to where I can guesstimate SOC of my battery to within 5 or 10% if you give me voltage and current. It is very linear in the 20-90% range. (And much easier to see at 16s versus 4s.)

I still prefer the coulomb counter, though, because with the frequent syncs we get it is quite accurate, and it is a lot more user-friendly for the rest of the family on board.

[tanglewood]

Originally Posted by nebster

Yes, if you go back in one of those other monster threads from a bit ago, I mentioned that I have gotten to where I can guesstimate SOC of my battery to within 5 or 10% if you give me voltage and current. It is very linear in the 20-90% range. (And much easier to see at 16s versus 4s.)

I still prefer the coulomb counter, though, because with the frequent syncs we get it is quite accurate, and it is a lot more user-friendly for the rest of the family on board.

Glad someone else sees the same phenomenon.


I have really become spoiled by auto gen start. With it, battery monitoring couldn't be simpler, because there is none. You just never look or care what the SOC is. If it gets too low, the generator starts and runs for a couple of hours, then shuts off. Meanwhile, life carries on.


Ok, I exaggerate a bit. I do look at the bank voltage because I'm a geek. But nobody else in my family even knows how to check, let alone how to interpret. And the best part is that it doesn't matter. Day to day operation totally passes the "wife and kids" test.

[nebster]

Agreed ... AGS is great!

I just completed my ~one-year checkup on my cells. All of them are still as stable as day 1. I have about 100 "cycles" on them so far, so they are still babies.

[hellosailor]

Tanlewood-
I'd agree with you. Battery voltage CAN be used to determine SOC, even if it is not always exact or the best way to do it. The problem with lithium being that the voltage has to be very accurately measured, which means at the least you should be using a calibrated voltmeter, and you'd want a back-up or calibration check annually to make sure it stays that way.
And of course, to stay far enough away from the danger zones (high or low) in the battery and the cells, so that you didn't have to worry about them.

Long time ago I tried to recalibrate a vintage battery meter on a boat. I could get it to read correctly at 14.4 when the alternator was on, or at 12.3 when the battery was down, but either way, if I got it calibrated right for one, it would be a half volt off for the other. I think that kind of cheap analog meter problem is why a lot of folks don't want to hear about voltage being used to read SOC. That, and the float that is normal in DC meters these days.
Personally...I prefer to to screw around with hydrometers and acid, even when that option is available. Voltage can work, if you're "doing it right".

[tanglewood]

Part of my point/realization is that voltage as proxy for SOC is EASIER to measure on LFP than LA, not harder. The voltage swing from, in practice, full to empty, is larger for LFP than for LA.


The only place where LFP requires more accurate voltage measurement is in monitoring per-cell limits. With LA, it really doesn't matter too much because everything gets re-aligned when you go through a good full charge. So voltage drift between cells doesn't matter too much. Where with LFP it does, both because there is no natural re-balancing, and because running out of range leads much more quickly to irreversible damage.


Now this doesn't mean that existing meters are accurate. One thing I have found is that they totally suck. My inverter, MPPT, and monitoring system is all Schneider commercial solar equipment, and the metering in it is terrible. And I only noticed when I started paying close attention to in for my LFP bank. It's so bad, in fact, that I think the equipment is generally unsuitable for use with LFP. Different devices are metering the battery as much as .7v off. So I have had to program offset charge values into each device, based on it's metering error, to get charging to work reasonably well. It's really pathetic, and is just another thing in a long list of reasons to never recommend or use their equipment again.


Fortunately my BMS is dead nuts on. I'm very pleased with how that has worked out.

[senormechanico]

Some, but not all meters have both slope and gain calibrations.
They may have adjustability inside, but probably not.
Gain adjusts the numbers read on the screen from zero to x, while slope does not affect gain, but moves everything up and down.
Zero will move with no voltage applied.


Just because the meter is "digital" does not guarantee accuracy !!!
That being said, I have $1.77 digital meters on both my home's grid tie backup and the boat's Lifepo4's as well.

Every one of them has been personally calibrated by me.
Oh, and they sometimes drift a bit, but what do you expect for $1.77 and free shipping?
It keeps me busy ! I check calibration every month or so with my Fluke 77 which gets occasional recal's from the factory with inside connections.

lol.