What about CEL NEC NExxxx in plastic package and ATFxxxx?
It looks that for NEC's S parameters given from chip center, not from leads. I examined parallel feedback oscillator with microstrip resonator, and it seems that feedback microstrip loop lenght = exactly inverse of S21 angle (more or less 1deg) at frequency of oscillation
So if S21 angle is -100deg
then feedback loop microstrip length is N*360+100 deg.
If i calculate line lenght from FET leads, it gives wrong phase relation: 14deg differs from S2P data.