RS PV inverter zero feed-in

Been experimenting with zero feed-in on my RS Smart Solar setup paired with a Cerbo GX, and I think this deserves more discussion here because the behaviour is genuinely interesting once you dig into it.

For anyone unfamiliar: when you disable AC-coupled PV feed-in within the ESS settings, the GX actively throttles your AC-coupled inverter rather than just ignoring excess generation. It's doing real-time frequency shifting to signal the PV inverter to back off. Clever stuff.

My situation is a static caravan with a grid-tied SolarEdge unit on the AC output side — yes, I know, slightly unconventional — and getting zero feed-in dialled in properly took some patience. The Cerbo was occasionally hunting; the PV would ramp down, battery would dip slightly, PV would ramp back up. Bit of an oscillation loop until I adjusted the ESS minimum SOC and smoothed out the response.

A few things worth discussing:

• Does anyone have experience with this on a fully off-grid system where there's no actual grid connection at all? I'm curious whether the frequency-shift mechanism still behaves predictably when the Victron inverter is the only voltage reference.
• What PV inverters have people found respond cleanly to the frequency signal? SolarEdge seems reasonably well-behaved but I've heard mixed reports about some cheaper units.
• For those running Fogstar or similar lithium banks — does your BMS communicate fast enough to prevent over-charge events during the transition period before the GX catches up?

This feels like one of those settings that's powerful but not entirely obvious in its consequences. Worth sharing experiences before people just tick the box and wonder why their system is behaving oddly.

@VanGill agreed, the RS's grid code compliance layer makes zero feed-in behave differently to what you'd expect coming from a standard Multiplus setup.

Worth noting for anyone following along — the feed-in limiting on the RS runs through the ESS assistant logic, so your CT clamp placement is absolutely critical. If you're measuring grid at the wrong point relative to your EV charger, the system can't distinguish between "exporting" and "powering the car," and you'll see erratic throttling on the inverter output.

My setup runs a 22kW Type 2 charger behind the Cerbo's grid metering point deliberately — means the ESS sees EV load as a house load and responds properly rather than fighting itself.

Also check your DVCC settings; shared current sense needs enabling or the RS won't honour the feed-in limit accurately under dynamic load conditions like EV charging ramp-up.

@VanGill @QJ_Builds — worth flagging that the zero feed-in response time varies quite a bit depending on whether you've got the CT clamp on the grid side or the load side. Victron's own documentation is... optimistically vague about this. In my setup I initially had the sensor positioned wrong and was seeing ~2-3 second lag before the RS throttled back, which absolutely resulted in a few watts trickling out. Repositioned correctly and it tightened up considerably.

Also, the Cerbo's MQTT feed introduces its own latency into the control loop — tiny, but measurable if you're paranoid enough to be logging at 1-second intervals (guilty). The RS does have onboard processing that responds faster than anything routed through the Cerbo, so understanding which control path is actually active in your specific config matters enormously here.

@TracyAllen that response time point is key — I spent ages chasing what I thought was a CT clamp wiring issue on my boat install before realising the AC sensor polling interval was the actual bottleneck. Default ESS assistant settings weren't helping either.

Worth checking your ESS minimum SOC and feed-in excess solar flags in VEConfig — they interact with zero feed-in in ways the manual glosses over.

Also, if you're on a weak grid connection (common on marina shore power), the RS can get twitchy about its voltage reference for the feed-in limiter. Switching the AC input type in the Cerbo fixed mine immediately.

@CaddyProject that's a really useful point about the CT clamp red herring — I've been second-guessing my own wiring for exactly the same reason on my setup.

Quick question for the thread: does the zero feed-in accuracy degrade noticeably when loads are very small? I'm finding that at low consumption (under ~200W) there seems to be a small but consistent export of maybe 20-30W that I can't eliminate. Is that just within the acceptable tolerance of the ET340 meter, or is there a tuning parameter I'm missing in VRM/Venus OS? @VanGill or @QJ_Builds — have either of you managed to tighten this up, or is it essentially a hardware limitation at that power level?

@TrevorRoberts75 the CT clamp self-doubt spiral is a rite of passage — I've personally rewired mine three times convinced I'd got the orientation wrong, only to eventually realise the Cerbo just needed a firmware update and a stern talking-to.

One thing worth adding to this thread that hasn't been mentioned yet: the RS Smart Solar