I suppose in a grid-down situation, since my two SB 6kw inverters are downstream of the grid they would just reboot after seeing the grid from the SI is up.... if it's too much juice I could always manually flip the main power switch on one or both if the SI isn't controlling their output very well but of course that brings dumb manual control into something that should be automatic and seamless.
Honestly, I wish SMA would show some circuit diagrams and instructions for connecting odds-ends of their hardware.... such as connecting older models of SB with newest SI, and vice versa.
I suppose pretty much any gridtie inverters downstream of the SI would function no matter what the brand as long as the SI puts the frequency over the shut down of the inverters to cut power if not needed for the battery.
SB do typically go offline by the time SI recognizes the failure and establishes local grid. Loads may see a 30 millisecond glitch and should remain online. SI can carry the load until SB return (up to 6 kW per SI, a bit more for a short time. 6.7 kW is the GT PV limit for grid-backup of US models. European model can work with twice that and would need SB to return real fast if 12 kW of PV was supporting loads)
Any basic UL1741 grid-tie inverter can work downstream of SI during grid failures, but only so long as 100% of its output is used either for AC loads or battery charging. If power output is excessive, SI increases frequency and knocks it offline. Loads are carried on battery, and if SI has frequency within range then after 5 minutes GT inverter comes back online. As soon as battery charge current tapers off enough that there is excessive PV generation it is knocked offline again.
Default settings of SI let it drop below 59.5 Hz and GT inverter won't connect. That setting would need to be changed for basic UL-1741 GT.
It may be that GT inverters with Rule 21 watts-Hz function would work well. Sunny Boy with compatible "backup" function do work well.
Oldest/newest SI. For US market, pretty much the same (except for really old SI4248). The SI5048 and SI6048 appear to differ only efficiency or thermal performance (different continuous wattage) and the size of lugs for battery cables.
The RS485 piggy back instructions for SI and for SB give some help. It appears the daughter card is same for most if not all, but the kit has various part numbers and comes with different sleeves etc. for wiring of various SI and SB.
The RS485 cabling plan shows wiring for the SB which have an adapter (all except newer -40 and -41; those get AC wiring only, no data cable)
What that cabling plan assumes is that you always have a Sunny Web box, Gateway, Sunny Data Control, or other device with pullup/pulldown for RS-485. Without any of those, you need to place jumpers enabling the pullup/pulldown in one Sunny Boy. At least for my short cables there was no problem leaving pullups connected in SB whether Sunny Data Control was connected or not. I put a bulkhead RJ45 socket in SI, and either plug in an RS485 terminator or else cable to my Sunny Data Control.
It will be automatic and seamless if total SB capacity is within pass-through current limits and SB is in the "backup" or "Island" mode.
I had bought 10000TLUS-12 because it is supposed to be supported, but discovered in "backup" it tolerates wider frequency but doesn't reduce power output. It remained online up to 64.5 Hz at full power, disconnected, and reconnected about when frequency dropped to about 62 Hz. SMA reviewed their firmware and acknowledged the bug. They gave me the workaround of configuring for "island" when used with Sunny Island between SB and the grid. They just updated list of supported inverters, showing this series is OK for off-grid but not backup. I've since bought SB 5000US to replace it.
It appears -40 and -41 if configured for Rule 21 should work. I assume those have watts-Hz function.