More solar power generated than consummed but battery levels dropped

Ran into something that had me scratching my head for a good few weeks last winter, and I reckon it's worth a proper discussion here.

My system was showing net positive generation — panels pulling in more than the loads were consuming across the day — yet my Fogstar Drift cells were consistently ending each evening a few percent lower than they'd started. Not dramatically, but enough that after a fortnight of "surplus" days I was sitting at a worryingly low state of charge.

Took me an embarrassing amount of time to work out what was happening.

The core issue was where the figures were being measured. My generation numbers came from the Victron MPPT, consumption from a clamp meter on the load side — but neither was accounting for the conversion losses through the inverter, the parasitic draw of the BMS keeping itself alive overnight, or the fact that my battery monitor was slightly miscalibrated after I'd replaced a cell.

Essentially, the maths looked fine on paper but the actual electrons going in and coming out told a different story.

There's also the absorption/float behaviour to consider — if your batteries never quite hit full charge, the monitor's percentage can drift downward even on days that look surplus.

Anyone else noticed this kind of discrepancy? Particularly curious whether those running lithium vs AGM see different patterns here, since the charge acceptance curves are so different. And has anyone found a single monitoring setup that genuinely ties all the figures together without faffing about with multiple devices?

Feel free to share your own setups — the more data points the better on this one.

@MoorHamish sounds familiar! Had almost the exact same head-scratcher with my static caravan setup last autumn.

Worth checking — is your battery SOC being calculated from voltage rather than coulomb counting? Voltage-based SOC is notoriously unreliable, especially in cold weather when resting voltage drops and makes the bank look emptier than it actually is.

Also, what's your BMS reporting versus what your Victron MPPT is showing? Sometimes there's a genuine discrepancy between what the panels claim to generate and what actually reaches the battery after conversion losses.

One thing that caught me out — my inverter was pulling a standing load overnight that I'd completely forgotten to account for. Looked fine on paper but the sums never added up until I actually metered everything individually.

What battery chemistry are you running? That could change the diagnosis quite a bit.

@MoorHamish seen this exact thing with my shepherd's hut setup. Your Victron display is almost certainly showing you instantaneous watts, not accounting for what the battery is actually doing at a chemical level.

Two things worth checking:

1. State of charge drift — if you're using voltage-based SoC rather than a proper Victron BMV shunt, the readings are basically fiction, especially in cold weather
2. Parasitic loads at night — my inverter alone was chewing through more than I'd bargained for in standby

The brutal reality is that net positive on paper means nothing if your charge efficiency is poor. LiFePO4 is miles better than lead-acid here, but a poorly configured BMS can still cause grief by throttling charge acceptance without telling you.

What battery chemistry are you running and have you got a shunt fitted?

@MoorHamish classic case of the battery playing you like a fiddle.

Worth checking your Peukert losses — batteries are sneaky little accountants. They'll cheerfully spend energy at a higher rate than your BMS is logging, especially if you're pulling decent loads whilst simultaneously charging. The maths looks fine on paper but the battery's quietly

Had this on my boat last spring. Turned out my Fogstar Drift cells were sitting at a dodgy state of charge — the BMS was reporting higher than actual because I hadn't done a proper full charge cycle in weeks.

Worth checking if your BMS has had a chance to properly calibrate its SOC reading recently. Mine was way off until I let it hit 100% and hold there for a bit.

Also — what's your battery temperature doing? Cold cells accept charge less efficiently than the figures suggest. Mine were barely absorbing anything useful below about 5°C even though the MPPT was showing decent input.

@MoorHamish this caught me out badly on my narrowboat last winter too. Worth checking whether your inverter is drawing phantom loads overnight — mine was pulling 15-20W in standby constantly, which the solar MPPT wasn't registering as "consumption" in the way the display summarised it.

Also double-check your battery voltage calibration against actual resting voltage with a decent multimeter. My Victron BMV-712 was reading optimistically because the shunt wiring had a dodgy connection — it thought the battery was fuller than it was, so the apparent drop looked inexplicable.

On a narrowboat especially, alternator input can confuse the daily totals if you've been running the engine at all — the figures don't always reconcile cleanly in the app.