Sounds like your mystery is solved
.
And none of us are idiots (hopefully) but none of us are perfect.
546.7W / 680.52W = 80.34% efficiency (which is still low).
First, if you check voltage at the input wire into the SCC (or what it reports), you will see it is lower than the voltage coming out of your combiner box (losses of ~3% are not unusual).
Second, when you state ‘from controller to battery’, is that measured at controller output or battery input? If you measured at battery input, there are more losses in those cables and as measurement directly at SCC output will give you a more accurate read on it’s actual efficiency.
Then depending on your SCC you can see what ‘max efficiency’ it specifies and discover how much worse the reality is than the specification.
Even if you measured voltages at combiner output and battery input and you have 3% losses in both sets of cables, that would still translate to 563.61W / 660.10W = 85.4% efficiency from your SCC.
That’s still pretty piss-poor and I’d still be concerned that 96.5W would hear it up over time…
As a reality-check, my 1000W GTILs are only 80% efficient and when they are taking in 1000W of battery energy to put out 800W of 120V AC, that 200W of internal heat causes the cooling fans to kick-in.