Victron BMV-712 showing different SOC to my BMS — which one do I trust?

Had exactly this with my shepherd's hut build last winter. Two readings, two different stories — maddening at first.

Short answer: trust your BMS for protection, trust your BMV-712 for day-to-day monitoring — but only once it's properly calibrated.

The BMV works by coulomb counting, so if it starts from a dodgy reference point it just drifts further and further off. Mine was reading 78% SOC when the BMS was screaming low-voltage cutoff. Turned out I'd never let the bank fully charge to 100% after installation, so the BMV had never had a clean sync point to work from.

What fixed it for me:

• Charged the Fogstar Drift LiFePO4 cells slowly up to absorption, held it there until current tapered right down
• The BMV then synced to 100% and from that point the two readings tracked within 3–4% of each other
• Also made sure the Peukert exponent and charge efficiency factor in the BMV settings matched LiFePO4 (not the default lead-acid values — an easy mistake)

The BMS SOC estimate is often just voltage-based anyway, which on LiFePO4 is nearly useless across that long flat discharge curve. So ironically the BMV, once calibrated, is probably more accurate for actual state of charge.

Worth checking what algorithm your BMS is actually using before putting too much faith in it.

Anyone else found the two diverge again over time? Curious whether periodic full charges keep things aligned or whether there's still drift.

@LynnCrane same setup in my hut — been through exactly this.

The BMV wins for accuracy day-to-day once you've nailed the peukert exponent and charge efficiency factor. Most BMS units just count voltage steps which is pretty crude.

Key thing people miss: synchronise them both by doing a proper full charge and letting the BMV reset to 100%. Do that regularly and the drift stays minimal.

One caveat — if your BMS is cutting out before the BMV hits your low SOC alarm, listen to the BMS every time. It's seeing cell-level data the BMV can't touch.

Had mine disagree by about 12% last January during a cold snap. Turns out low temps were tanking my actual capacity — BMV was right, BMS was protecting against a weak cell. Both were doing their job.

Really glad this thread came up — I spent ages confused by this exact thing in my narrowboat setup last year.

One thing worth adding that nobody's mentioned yet: the divergence between the two readings often gets worse if your BMV's Peukert exponent isn't set correctly for your battery chemistry. LiFePO4 is much flatter than lead-acid, so if you've migrated from older settings or used defaults, your BMV can drift noticeably over time.

@LynnCrane and @VoltJohn — did either of you do a proper synchronisation point after a full charge cycle? That reset made a massive difference for me. Once the BMV resyncs at 100%, the two readings usually come back much closer together within a day or two of normal use.

The BMS catching cell-level anomalies that the BMV simply can't see is the other crucial bit people overlook.

Worth adding — the BMV-712's coulomb counting is only as good as your calibration. If your charged voltage, tail current, and Peukert exponent aren't dialled in for your specific cells, it'll drift over time and the two readings diverge further.

I run Fogstar Drift LiFePO4 in my cabin and spent a weekend logging both readings against actual cell voltages to get the BMV settings right. Once sorted, it tracks beautifully.

The BMS figure is usually just a rough voltage-based lookup anyway — not true coulomb counting — which is why it can jump around, especially mid-SOC where the LiFePO4 discharge curve is so flat.

So really it's less about which to trust and more about whether your BMV is properly configured in the first place.

Great thread. I'll add something from the marine side that might help @LynnCrane and others.

On my boat I run both, and the divergence usually tells me something useful rather than being just noise. If the BMV reads higher than the BMS — typically means your charge current tail isn't quite right and the BMV thinks you're fuller than you are. If the BMS reads higher, you've often got a cell imbalance skewing its calculation.

The fix that sorted mine: synchronise them after a proper full charge — let the bank hit absorption, hold it until tail current drops to around 2% of capacity, then let the BMV auto-sync to 100%. Do that monthly and the two readings stay within a few percent of each other.

@Inverter_Pro's point on calibration is key — garbage settings in, garbage readings out, regardless of how good the hardware is.

Good points from @Inverter_Pro on calibration — that's where most people go wrong.

One thing I'd add: battery chemistry matters here. LiFePO4 has an almost flat voltage curve, so the BMV-712 struggles to infer SOC from voltage alone. It's almost entirely reliant on coulomb counting, which drifts over time.

My fix: set a solid synchronisation point. I let mine sync at 100% only when the Victron MPPT hits float and tail current drops below 1% of capacity. Took a bit of fiddling with the BMV settings but it's spot on now.

Also worth checking your Peukert exponent — Fogstar Drift cells behave differently to a cheap no-name pack. Wrong value there will throw your readings off gradually.

BMS still wins for protection though. Non-negotiable.