[HFSS]How small a mesh size would be enough?

[sandbo]

Hi,

I have been simulating some CPW waveguides, and with a small dimension like 13um wide of the center lines and 7um gap, then I have a small square hollow within the (blue) CPW center line as 4x10um.
I wanted to simulate the H field being coupled through that square by the red CPW line running perpendicular to to the blue CPW line.

First I ended up with wildly pointing vector H arrows when plotting the H field through that sqaure hollow, such that they point almost horizontally though they were supposed to be vertical.
I then limited the length in mesh operation on selection (for blue CPW line) to around 1um, which is the narrowest structure of the square hollow (the width of the arms), now I am able to get the arrow pointing up straight (the red one, down straight in fact), now it looks like this: (I refined on the blue CPW line to 1um)


However it is taking as long as a few hours to get what could be done originally within 1.

From this point of view, may I know if there is any rule of thumb in achieving the optimal mesh size in terms of a balance in performance and accuracy?
Many thanks for reading.

 

[PlanarMetamaterials]

Hi sandbo,

In working with CPW lines in HFSS, I've found that meshing the air gaps with a maximum length of half the gap width is sufficient for good convergence. I've also found that meshing the air gap seems to be more important than meshing the strip, which it appears what you are doing.

 

[sandbo]

Hi sandbo,

In working with CPW lines in HFSS, I've found that meshing the air gaps with a maximum length of half the gap width is sufficient for good convergence. I've also found that meshing the air gap seems to be more important than meshing the strip, which it appears what you are doing.

Thanks a lot for your advice.
Yesterday I tried to define an extra structure, that is a ring surrounding the square hollow and sized within the strip line,
then I refine the mesh only within that ring, and it worked pretty well in greatly reducing the resource needed while keeping a good accuracy (determined only by the distribution of H field within the square, and the arrow).

I will also apply some mesh to the gap now, thanks again:-D