Had exactly this with my shepherd's hut setup last summer. Turned out my absorption voltage was set a touch low — the Victron was hitting its target voltage too quickly under light load and calling it done. Bumped absorption up to 14.2V and extended the absorption time, made a huge difference.
A few things worth checking:
- Battery voltage vs controller voltage — are they actually the same? Volt drop across your cable run can fool the MPPT into thinking the battery is fuller than it is. Proper cable sizing matters more than people realise.
- Tail current setting — this is the one most people miss. If your tail current threshold is too high, the controller exits absorption early even if the battery isn't actually full. With a 280Ah bank you want that set quite low, maybe 2-3% of capacity.
- BMS communication — if you're running a Fogstar or similar with a BMS that's talking to the Victron via VE.Direct or similar, the BMS might actually be requesting the float transition. Worth checking what the BMS is telling it to do in VictronConnect.
What lithium cells are you running? Some of the budget LiFePO4 packs have quite flat charge curves and the MPPT can genuinely struggle to tell where you are in the state of charge. A proper battery monitor like a BMV-712 alongside the MPPT makes diagnosing this so much easier — you can actually see what SoC you're hitting when it drops to float.
What does your VictronConnect history graph look like? That'll tell us a lot.