DESS does not follow negative AcPowerSetpoint on 3-phase Multi RS, manual ESS export works (BUG)

Bit of a niche one but genuinely losing my mind here — anyone else found that DESS refuses to honour a negative AcPowerSetpoint on a multi-phase setup?

Running a three-phase arrangement in the van workshop (yes, I went full mad scientist) with Victron kit throughout — Cerbos, Venus OS on the latest, the lot. Manual ESS export works an absolute treat when I kick it off myself, but the moment I let DESS take the wheel it just... ignores the export command entirely. Sits there like a golden retriever that's heard "walkies" but decided the sofa is better.

I've got the VM-3P75CT wired in proper over Ethernet, not the dodgy WiFi lottery, and it's reading the grid perfectly. The assistant sees the data, it just refuses to act on it when conditions should trigger export.

Wondering if this is a Venus OS 3.75 quirk or something deeper in how DESS calculates setpoints across phases — does it average them? Prioritise one phase? Nobody seems to know and the Victron docs are characteristically... sparse.

Has anyone on here with a three-phase Multi RS setup actually got DESS export behaving properly, or is this a known firmware wobble we're all just quietly suffering through? Fogstar cells doing their job perfectly, this is very much a software drama rather than anything in the battery stack.

Drop your configs below — curious whether it's reproducible or if my install is just haunted. 🔧

@RussHunt82 seen something similar on a single-phase setup with my Multiplus-II so potentially not purely a 3-phase issue. When you say manual ESS export works, are you testing that via the dbus command directly or through VRM? Because if dbus works but DESS ignores it, that points squarely at the DESS algorithm overriding the setpoint rather than a hardware/comms fault.

Worth checking whether your grid feed-in limit in the ESS assistant config is actually set to allow export — DESS won't push negative setpoints past whatever hard limit is baked in there, regardless of what you tell it to do. Victron's documentation on this is characteristically vague.

Also — firmware version? There were some odd ESS scheduling behaviours reported around v3.10-3.14 on the CCGX/Cerbo side.

@RussHunt82 classic Victron left hand/right hand situation — DESS overrides the AcPowerSetpoint because it's doing its own grid-point calculation and essentially ignores manual ESS registers when it decides to take the wheel 🚗

Worth checking your DESS target setpoint in the Venus GX settings isn't fighting your manual value; they'll arm-wrestle each other all day.

Known workaround: disable DESS temporarily, confirm manual export works (sounds like yours does), then re-enable and check whether your grid meter phase compensation setting is set to individual phase rather than total — on 3-phase this causes absolute chaos with negative setpoints.

Victron's GitHub has an open issue on this; @ExBrickie your single-phase weirdness might be a separate DVCC interaction rather than the same root cause.

Interesting one — had a related headache when I was fine-tuning my garden office setup with a Multiplus-II. DESS does seem to treat its own scheduling logic as gospel and won't yield to manual AcPowerSetpoint overrides properly.

Worth checking if ESS Mode 2 vs Mode 3 is contributing here — DESS essentially assumes control authority that conflicts with external setpoints. Some people have had luck writing directly to the MQTT topics and bypassing the DESS layer entirely, though that's a bit of a faff.

Have you raised it on the Victron Community forum? The devs actually engage there. @BoatPaddy is right that it smells like an architectural conflict rather than a straight bug, but that doesn't mean it should behave this way on a 3-phase installation — feels like it warrants a proper ticket rather than a workaround.

@RussHunt82 just to clarify something pedantic but I think it matters here — are you certain the negative AcPowerSetpoint is actually being written to the correct path on the DBus, or is DESS potentially overwriting it on the next scheduling cycle before the inverter acts on it?

In my van setup (single-phase, I know) I found there's a timing race between the DESS optimiser loop and any manual setpoint writes. Worth logging the actual DBus value at ~1 second intervals to confirm it's persisting.

Also — what firmware version on the Multi RS? There were some DESS-related changes in recent VenusOS releases that specifically affected setpoint handling. The Victron changelog might be worth cross-referencing against when this behaviour started.

Ran into something adjacent on my boat install last winter — not three-phase but DESS was absolutely bulldozing my manual setpoints whenever it decided it knew better about grid interaction. Turned out the culprit was the ESS mode setting itself; DESS essentially takes the wheel completely and treats AcPowerSetpoint as advisory at best.

Worth checking your ESS Assistant version too — there was a point where certain firmware combos on the Multi RS specifically had this split personality where manual export worked because it bypassed the DESS logic entirely, which is probably why manual works for you but DESS doesn't.

@FormerMariner1 raises a fair point — negative setpoint meaning "export this much" versus "draw from grid" can get confused depending on which layer of the stack is interpreting it. What firmware are you actually on?