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

Had this exact headache last winter in my static caravan setup. Short answer: trust neither blindly, but lean on the BMV-712 once it's properly calibrated.

Here's my thinking — your BMS is watching cell voltages and doing rough maths, whereas the Victron is doing proper coulomb counting (tracking actual amps in/out over time). The BMV wins on accuracy if you've set your battery capacity, Peukert exponent, and charge efficiency factor correctly in the app.

Common culprits for the discrepancy:

• Tail current not set right — the BMV might not be syncing to 100% when your battery actually hits full
• Capacity figure wrong — especially if you've got Fogstar or similar cells that perform better than rated
• BMS is just... optimistic — some cheaper units basically just read resting voltage and call it a day

What I'd do: run a full charge to 100%, let it sync the BMV manually, then do a controlled discharge and see which one tracks better against real-world usage.

Also worth checking if your shunt is on the right side of the system — caught me out once, had a load bypassing it entirely. Felt very silly.

What BMS are you running? Makes a difference to how much I'd trust it. Some of the JK units are actually decent. Others... less so. 🙈

Anyone else had this? Curious whether others just ignore one and crack on.

Reply by Ben1989:

Good shout @DustyCaptain, and I'd add one thing worth checking — the BMV-712 is only as good as its shunt installation. If anything is bypassing that shunt (common culprit is a direct earth connection somewhere on the battery negative side), your coulomb counting will be off from the start and the calibration never quite catches up.

Had this exact issue on my narrowboat setup. Spent weeks convinced my LiFePO4 pack was degrading before I spotted a rogue earth wire going straight to the hull without passing through the shunt.

Once the wiring's clean, the BMV with proper charge efficiency factor and Peukert settings tends to give a more usable day-to-day picture than most BMS SOC readings, which can jump about quite erratically depending on the brand.

Yeah this drove me mental on my narrowboat for months tbh. What finally sorted it for me was making sure the BMV had a proper full charge sync — basically let the bank hit absorption voltage and hold it, then the BMV resets to 100% and everything lines up proper.

Also worth checking your Peukert exponent setting in the BMV config. Most people leave it default but if you're on LiFePO4 (Fogstar or similar) you want it closer to 1.05 rather than the lead-acid default. Makes a noticeable difference to accuracy under load.

My BMS readout and BMV are now within 2-3% of each other pretty much constantly. Once you nail the calibration the BMV becomes genuinely reliable.

Reply by Compo55:

Something nobody's mentioned yet — have a look at your charge efficiency factor in the BMV settings. It defaults to 99% which is fine for lithium, but if you're running AGM or gel it should be more like 95%, sometimes lower on older batteries. Get that wrong and the BMV's coulomb counting drifts further out with every cycle, even if your shunt is wired perfectly.

Also worth checking whether your BMS is reporting cell-level SOC based on voltage curves rather than actual amp-hours in and out. That's why you'll often see them disagree most noticeably at partial states of charge where the voltage curve is relatively flat. They're measuring different things, effectively.

@ExChippie94 I'd be curious what your sync point ended up being on the narrowboat — 100% at absorption end or full tail current?

Good points all round. One thing I'd add from experience with both my shepherd's hut and the boat — synchronisation events are your friend here.

The BMV-712 drifts over time if it rarely sees a proper 100% sync point. Make sure your absorption/float voltage thresholds are actually being reached consistently, otherwise the BMV never resets its reference and the error compounds across cycles.

On the BMS side, most budget units (anything not Daly Smart or JK-grade and above) report SOC purely from voltage curves, which is notoriously unreliable under load due to voltage sag.

My workflow: set the BMV as primary readout, configure the tail current threshold properly for your specific cells (LFP wants something around 1-2% of capacity), and let the BMS handle protection duties rather than SOC reporting. Two different jobs.