I described the issue also several months ago to their german support representative without any solution.Thank you @kirkdis
I have described the problem to them and they sent me this video showcasing that there is not any actual problem... But i think this is 113/114 firmware on the video.
Regarding the video, here are my thoughts: the inverter in the video is not a SPF5000ES. The only thing which seems common to the SPF is the same display unit.
the issue with the unreliable switch back to grid happened for me in 2 out of 10 cases (the dry contact seemed everytime to switch correctly!). that means in 10 days where the inverter needed to switch back to the grid 2 days where not successful. My assumption is that the phase sync from the inverter drifts away over time from the grid phase. Only when this is the case the inverter is not able to switch back to grid or hard switches back to grid, this depends on the firmware as there are different behaviours in each one. This video is not proofing that their crappy firmware is working fine. (test condition and only one switch back)
Under test conditions I can switch 100 times back an forth to grid in SBU without any issues. I can not check what they programmed in their firmware but for my experience with the SPF inverters the longer the inverter is connected constantly without interruption to the grid the more often it happened here. In SBU the inverter is usually constantly connected and this seems to benefit the cause.
You can sort the Growatt support engineers in two groups: group one doesn´t have any clue about the technology, group two is not able to communicate with the customers properly and misunderstands most of the details.
And exactly this is the biggest issue Growatt is facing and leads to these weird explantions in the provided video...
As I use my inverters not anymore with a constant connected grid I assume this is and was the solution to solve the isse with the switch back.
Due to the fact I use uti as working mode as soon the grid is available the inverter makes the phase sync and switches over to energy provider power. This is for me the most reasonable explanation why the external switch works great and the internal switch is unreliable.