I am designing a microwave lumped element resonator which is contained in a metal box. My simulations consists of the resonator with PEC BCs simulating the metal of the resonator as well as the limits of the simulation box. My issue is that when I simulate the design with no loss (no material losses or scattering BCs) I get a divide by zero error related to emw.etaRef (not sure what this is anyway), but as soon as I include any loss (in my case I add a tiny patch of scattering BC in the simulation box) the eigenfrequency search goes without a hitch.
Any ideas concerning what's going on would be appreciated.
-Steven