Anyone else getting wildly inconsistent SoC readings from a Victron BMV-712 in cold weather?

My BMV-712 has been driving me mad this winter. Running a 200Ah Fogstar Drift LiFePO4 and the state of charge is jumping all over the place once temps drop below about 5°C overnight. Yesterday morning it was showing 67% SoC, I ran a small load (just a 12V fan heater on low, maybe 8A), and within 20 minutes it had jumped to 74% — with no solar input, battery not charging. That makes no sense.

I've got the battery temperature sensor fitted and the Peukert exponent set to 1.05 as recommended for lithium. Charge efficiency factor is at 99%. I did a full synchronisation (charged to absorption, held it there for 30 mins) about two weeks ago, so the 100% reference point should be solid. The BMV is running firmware 4.12.

Wondering if the issue is the voltage-based floor messing with the coulomb counting at low temps — LiFePO4 cells do sag noticeably at 3-5°C, so the BMV might be misinterpreting a voltage recovery as a charge event? Or is my charged voltage threshold set wrong? Currently at 13.2V with a 0.2V hysteresis. Would dropping the synchronisation voltage help here, or would that just make things worse?

@BrookRunner yeah, seen this on my boat setup. LiFePO4 voltage curve goes almost completely flat in the cold — the BMV is doing its best but voltage-based correction goes out the window below ~5°C.

Two things worth checking:

1. Peukert exponent — mine was set to default 1.25 which is wrong for lithium. Should be closer to 1.05
2. Charged voltage threshold — if your battery isn't hitting its usual resting voltage in the cold, the BMV never "syncs" and drift compounds over time

Honestly the BMV is a coulomb counter at heart, so if your charge efficiency figure is also off it'll snowball. I'd do a manual sync once you've got a confirmed full charge on a warmer day and work from there.

What's your current sync settings set to in VictronConnect?

@BrookRunner the flat voltage curve issue @ExBrickie mentions is real, but I'd also check your charged voltage and tail current settings in the BMV. Cold temps slow the charge acceptance rate right down, so if your tail current threshold is set too high it'll declare "full" prematurely — then the coulomb counting drifts from that dodgy baseline.

On my shepherd's hut setup I dropped tail current to around 1% of capacity and it tightened things up noticeably.

Also worth checking: is the Fogstar Drift actually rated for discharge below 0°C? Some LiFePO4 cells get genuinely weird at those temps — not just reporting issues, actual capacity shrinkage. Your 200Ah might realistically be 160Ah in a cold snap, which makes the SoC readings look "wrong" when they're arguably telling the truth.

Has anyone actually tried tweaking the Peukert exponent on the BMV-712 for cold weather? I've been reading that the default (1.25) might be worth adjusting slightly when your battery's regularly sitting in low temps, because the effective capacity changes.

Also wondering — does the BMV have any way to account for the fact that LiFePO4 genuinely loses usable capacity in the cold? My understanding is that a 200Ah Fogstar Drift might only be delivering 160Ah or so at 5°C, so the SoC readings could be "correct" by the maths but still feel all over the place because the reference point has shifted.

Is anyone using an external temperature sensor with theirs? The BMV-712 supports one — curious whether that actually helps the calculations or just gives you a number to stare at whilst things still go wrong.