Results do not Converge
Hello,
I'm performing a convergence study on the attached model. The idea is to run multiple simulations for the same model using different mesh sizes, and then determine the appropriate mesh size for this model to guarantee accurate S-parameter results.
The simulations were performed using custom meshing for 3.5 mm, 2.5 mm, 1.5 mm, 1.0 mm, 0.5 mm in all of these trials the S-parameter plots were different.
Thank you.
Answers
-
Hi Mr
The problem is with the connections (or not) at the 'squares' (barely) touching at corners only.
For an ohmic connection, I would rather model this as seen in the image below. An ohmic connection can only occur across triangle edges
Also consider how the model would be physically constructed/etched: there would be some sort of finite size metallic connection there.
As for the mesh size, I don't think 0.5 mm is that fine. The squares are electrically very small themselves, then having a couple of triangles across the width of each square is recommended. This will also help to accurately model the 'inter-square' coupling between adjacent squares.
0 -
Thank you for your response Mel, but I do not think that's the reason. I tried to add small squares to the corners and that didn't seem to solve the problem.
I have also created a new model (see attached) where I made sure that no two square 'islands' are only attached by corners and I was still not able to obtain convergence.
could there be any other issues causing this? I realize that 0.5 mm might still be not an appropriate mesh size, but I am worried that having the mesh any smaller would cause inaccuracies as the wire feed radius is 0.3 mm.
<?xml version="1.0" encoding="UTF-8"?>
0 -
Have you tested how the results change with changing radius?
An option instead of wires is to use a cylinder meshed into triangles to feed the model.
See e.g. a monopole made from triangles only. But the model contains fine geometric detail, so finer meshing is most likely going to be required.
0
