Problem with grounding passive circuits with two ports in HFSS

When I design an inductor or other passive circuits with two ports at different sides, I notice that HFSS will always difficult to find the ground and adding an idea inductor between the two ports.

what makes the simulation result difficult to understand and needing deembed.

Is there anybody have similar experience as me?

Are you working with lossy silicon? Or are you working with low-loss dielectrics?

I usually prefer to use a planar simulator for such structures; they determine the characteristic impedance through port de-embedding techniques that work much better for lossy dielectrics. Unless there is something that really requires full 3D volume meshing, it is a lot more efficient and usually more accurate to use one of the planar EM simulators for such a problem, especially now that most of them will handle thick metal.

--Max

maybe I will try Sonnet later