Radio frequency (and general) convergence question

I have a model which COMSOL is having a very difficult time finding a stationary solution for - the solver gets a fair way, but then tends to oscillate and never settles down to a converged solution. Is there any way for me to visualize or get more information about how the error in the computation is distributed across domains? It would be nice to know where exactly I need to improve meshing, or if certain parts of the problem are particularly difficult for the solver to arrive at a solution for.

Thanks very much for any guidance you have.