ESS in Grid Loss Mode works fine when all 3 grid phases are off - but not if L1 is still present

Been doing some deliberate grid-loss testing on my Victron Quattro setup after reading about the grid instability incidents on the continent recently. Got me thinking — how robust is my own ESS configuration when things go partially wrong rather than completely wrong?

Here's what I've found, and I'd be curious whether others have replicated this:

Full grid loss (all phases drop simultaneously) — the Quattro handles the transition cleanly. UPS-style switchover, barely a blip on the critical loads.

Partial phase loss — specifically where L1 stays present but L2/L3 drop — this is where it gets messy. The system seems confused about whether it's actually in a grid-loss condition. My understanding is that ESS uses L1 as its primary reference for grid presence detection, so if L1 remains energised the inverter doesn't fully commit to islanded operation. Result: the loads on the dropped phases aren't picked up properly, and I'm seeing some odd behaviour from the MPPT controllers as well.

My workaround currently is to set the AC input current limit aggressively low on L1 during testing, which forces the ESS into a different operating mode, but that's a hack rather than a solution.

Has anyone properly configured the grid code settings or adjusted the phase compensation parameters in VEConfigure to address this? I'm running firmware 498 on the Quattro and using a Cerbo GX.

Also wondering whether a neutral contactor arrangement might be the cleaner engineering solution here — physically isolating when partial phase conditions are detected.

Anyone in the UK had real-world partial phase loss events rather than just test scenarios? Particularly interested if you're on a rural single-phase DNO supply that occasionally does weird things.

Really interesting edge case @SueThompson — this one caught me out too during testing on my narrowboat setup.

What I found is that the Quattro's grid detection logic on L1 effectively acts as the "heartbeat" for the whole ESS grid-loss response. If L1 is still alive, the system doesn't declare a full grid outage even if L2 and L3 have gone. Stays in grid-connected mode, which in a real partial-grid scenario can cause some genuinely odd behaviour with loads.

Worth checking your grid loss detection settings in VEConfigure — specifically the voltage and frequency window tolerances on L1. Tightening those helped me get more reliable transitions.

Also, make sure your ESS Assistant is on the latest version. There were some improvements to multi-phase loss handling in recent releases that aren't widely talked about.

@SueThompson this is a well-documented Victron quirk that trips people up. When L1 is still present the Quattro doesn't fully enter island mode — it continues attempting grid synchronisation on that phase, which can cause erratic inverter behaviour or even a brief output dropout.

The fix is in your ESS assistant settings: ensure "Grid lost" conditions are configured to trigger on any phase loss, not just total grid absence. You'll find this under the AC input current limits section in VEConfigure.

Also worth checking your grid code settings — some codes (particularly older UK G98/G99 configurations) have phase-specific detection thresholds that won't flag a partial loss correctly.

Test with a proper current clamp on each phase during your simulated outages. I did exactly this on my static caravan install before trusting it for winter. Don't assume symmetrical behaviour just because full loss works correctly.

@SueThompson mine did the same — turns out the Quattro treats L1 as the "master" phase so if it's still sniffing grid it won't fully commit to island mode, which is exactly the wrong behaviour when you want a clean failover on the narrowboat.

Worth checking your grid code settings in VE.Config — there's a minimum voltage/frequency window that determines when it declares "grid lost." Tighten that window and it'll trip faster when L1 goes wobbly rather than hovering in this awkward half-connected limbo.

Also check your AC input current limit isn't fighting the ESS assistant logic at the same time — had two things arguing at once on mine before I sorted it.

Great thread @SueThompson. To add a bit more detail to what @ExFarmer and @DevonDweller are getting at — worth checking your Grid Lost Detection settings in VEConfigure. Specifically the AC input current limit and the voltage/frequency thresholds on L1. If L1 is limping along even at degraded voltage, the Quattro can still see it as "grid present" and won't fully commit to inverter mode.

I'd also look at your ESS Assistant version — older versions were notoriously slow to respond to partial phase loss scenarios. Updating the assistant firmware sorted a similar issue for me.

One practical test: use a proper contactor to isolate individual phases deliberately and monitor the CCGX/Cerbo logs in real time. The transition behaviour is all captured there and makes it much easier to diagnose exactly where the hesitation is occurring.