AIO islanding not seamless when charge rate 0 or battery charge paused

I have recently noticed that even when solar is below 6kw and SOC is below 80% my islanding isn't seamless if the battery charge is paused or the charge rate is set to 0 (to export generation).
Although logically I get why, as the solar has theoretically nowhere to go I dont believe this was always the behaviour.
Does anyone have any experience of this issue and can shed any light on a solution?
Firmwares are:-
Gateway A0.013, Dongle 206
All-In-One D0.616-A0.616, Dongle 299, BMS 12

Hopeful Is the SoC actually accurate is the first question with all GE kit?

I suspect yours isn't and is under-reporting the real SoC. When was the battery last run through a couple of charge/discharge cycles?

Vestas Thanks for your query, yes the SoC is pretty accurate as I dump to 15% and charge to 100% every day.
I did a series of tests and islanding was fine as long as CR was up and Pause Charge was disabled. When either was not the case the islanding failed to be seamless. In the past I have had Earth leakage trips on occasions when switching but this is not the problem at this time.

Hopeful Ah well its maybe one for @hoggy

I was always under the impression that if the inverter was paused or set to 0W it wouldnt be able to achieve the seamless changeover - the inverter would be in a powered down / idle state and need time to wake up and and output something?

I guess that theory would be proven if you instead set charge rate to the lowest possible none 0 value (which would keep all the internal DC bus 'warmed' and ready to step in)

If it used to work I guess you could load that older firmware into it - they are all on GitHub.

hoggy Thanks, I find with 0 CR it still charges the battery at about 500w so hence the pause charge. With Eco on the AIO can still supply demand if it goes above the solar and I thought if solar was higher than the battery could consume it would wind the solar back with a frequency hike and then sort its live out.
I dont think I have ever seen a set of rules surrounding seamless islanding, only things people have deduced. There is a newer AIO firmware set on the portal but I am loath to try loading that given the current support...or lack of....and the missing change logs

Thanks to Paul at PredSupport I have resolved this issue by manually loading D0.620-A0.620 which has specific fixes for Islanding switching timing where it overrides the pause and charge settings to consume the spare solar. On grid restore the original settings continue.
Note I did lose my serial number during the upgrade, which Paul quickly fixed. I dont believe this always happens, from what I have read, but its something to be prepared for.

Hopeful I wonder if the speed fixes were the ones in D617
https://github.com/DJBenson/giv-firmware/blob/main/Firmware/AIO/Gen1/Inverter/Release%20Notes.pdf

I just raise that as some may already be on the previous release like me of D619 and am scared to death of doing an update now free support is gone.

So for you, going from D616 to 620 you will get those speed optimisations. Me going from D619 to D620 I am more likely to just get a heart attack and a declaration from the wife along the lines of "What did you do!!!????"

Guybrush-Threepwood Yes Paul called out 617 as the important one. The only other one of interest is in 619 so you are safe there too

A620-D620 27 Sept 2025

1. Bug Fix. Add missing hardware type codes for the 3.6 and 5kw version of AIO

A619-D619 19 Aug 2025

1. Bug Fix. Multiple grid loss EPS fix.

A618-D618 - 29 Apr 2025

1. BMS Upgrade optimisations
2. Add new function of ultra low battery Re-Activation logic (Ready for BAAA0013). Battery can be force charged from as low as 60v. Previous 130v. For Service Desk use only.

A617-D617 - 07 Apr 2025

1. Add GW controlled Import limitation for AIO with GW, inc parallel
2. Increase Import limitation range to 1-255A for EU
3. Reserve and cut-off optimisations
   4. Off Grid switching speed optimisations