Hello all,
I think I have narrowed down the problem.
The difference between the two installations I have is that one is using the VM-3P75CT and the other one is using the ET340.
The difference between the two meters is their vector registration way of measuring (see documentation).
So while the grid setpoint was 10W, there was a consumer on L3 and thus the solar panels were injecting on L1 to keep the setpoint 10W.
On the VM-3P75CT installation, this energy was recorded as ‘solar-to-direct-use’.
On the ET340 installation on the otherhand, this energy was recorded as ‘grid-to-direct-use’ and ‘solar-to-grid’.
Only consumers on L1 were recorded (as I would expect) as ‘solar-to-direct use’.
I’m 90% confident that this is the issue. This also explains why @cmock has a working system with a ‘slow meter’ as the measuring speed is not the source of the problem.
I will update after the weekend.