Re: capcitor simulation
Hi midolasi3 -- All the EM tools are the most accurate for one problem or another. No one has a monopoly on accuracy for all problems.
If you have thin dielectric in your capacitor, CST can use their PBA (perfect boundary approximation) that allows up to three different materials to pass through each cell with their boundaries perfectly represented. I think it is likely CST will give a superior answer for such a problem, but would be good to see the problem in detail first.
In general, if you want to quickly check accuracy, look at the current distribution for each EM analysis of a circuit. Current distribution is very sensitive to error. The EM analysis that provides the smoothest current distribution with very strong current on all sharp edges is usually the most accurate. If you need to use HFSS and want high accuracy, I would suggest continuing to refine the mesh until you get a good looking current distribution.
For the result you already have, have you checked the current distribution? It might give you a clue as to where your mesh should be seeded. Seems like there are enough expert HFSS users around, someone could give you a hint on how to seed the mesh for this problem.