Untracked consumption in HA - loses on AC/DC conversion?

Been down this rabbit hole with my shepherd's hut setup and it's doing my head in a bit.

Running a Victron Multiplus-II 3000 with a Fogstar Drift 48V battery bank, Cerbo GX feeding everything into HA via the Venus OS integration. My energy dashboard always shows a mysterious "untracked" chunk — somewhere around 8-12% of total throughput just... vanishes into the ether.

My working theory is it's a combination of:

• AC/DC conversion losses on the Multiplus (roughly 4-6% depending on load)
• DC system loads that HA never sees properly (the Cerbo itself, battery BMS comms, fans)
• Rounding/polling gaps in the MQTT sensors — HA isn't sampling at true real-time

The maths never quite stacks up cleanly when you're mixing AC-coupled metering with DC-side figures pulled from the Cerbo. It's like trying to balance your bank statement when half the transactions are in a foreign currency.

Has anyone managed to get genuinely accurate round-trip efficiency figures out of a similar Victron/HA setup? Curious whether adding a Shelly EM on the AC input would plug the gap or just introduce more conflicting numbers to argue with.

Also wondering if this is worse for those on AC-coupled PV versus DC-coupled — intuitively it feels like there should be fewer conversion stages to lose track of with DC-coupled, but maybe I'm wrong.

What are others seeing for untracked losses as a percentage? Is 8-12% just the price of admission with this kind of hybrid setup?

@CamperCarl your Cerbo is showing you battery watts, not wall watts — the Multiplus eats roughly 8-10% in conversion losses before your kettle even knows it's alive.

@BayFrank is right about the conversion losses, and worth noting the Multiplus-II 3000 also draws a standing load just being in inverter mode — I think it's around 20-25W idle even with nothing running through it.

On my garden office setup I noticed a consistent ~15-20W "ghost" load that turned out to be exactly this. The Cerbo doesn't track that separately in the default dashboard view.

Worth checking your VRM portal — under the advanced graphs you can compare AC loads vs battery discharge over time and the gap becomes really obvious. That delta is essentially your conversion inefficiency plus idle draw combined.

If you want to minimise it, switching the Multiplus into charger-only mode when you're not actually using AC loads saves a fair bit over a week.

@LisaHunt88 was probably about to say the standing load is around 20-25W idle on the Multiplus-II 3000 — it's in the datasheet. Adds up over 24hrs on the narrowboat, trust me.

Worth checking your Cerbo's Energy page — there's a section for "AC loads" vs what the battery reports. The gap is basically your conversion losses + that idle draw.

If you're running Victron's VRM portal you can pull historical data and compare DC consumption vs AC metered load over a week — makes the losses pretty obvious.

@CaddyDream finishing @LisaHunt88's sentence is very wholesome, well done both 👏

Anyway @CamperCarl — think of your Multiplus like a bouncer at a club. Every unit of power trying to get from DC battery-land to AC appliance-land pays a small entry fee. Doesn't matter if