Selecting mesh size in CST Microwave Studio

Hi,
I am trying to simulate a structure with 1 um. as smallest dimension. Does anyone have a clue as to what mesh settings should I take . I am using CST 2010 as the tool and frequency range is 0-40 Ghz.
Thanks,
Emma.

you may start with 1/10 rule (10 cells per wavelength of maximum frequency) but which may not converge to correct result. After initial simulation see what is happening at that minimum dimensional part. if you see major variation of field current near the minimum dimensional part then you may need to increase the mesh density. If total mesh size becomes huge, then you may think of regional manual meshing.
g86

i think 10 cells / lambda is enough
i ask CST application engineer,he said it is enough,if u increase little difference will appear

i think u can take it 20 cell is better this cell size give me better results with measurment

Thanks g86, abdoegg and aaea3 for your help. I will try these out.

i with u
but when we talk about large surface like arrays and parabolic reflectors,i think 10 cells/wavelength is suitable,that what i refers to