Choosing the right frequency sweep in hfss

I am designing a dual-band microstrip patch antenna. I would like to know if there are some rules on picking frequency sweep in that case, because when I change the range. For exemple if I first simulate from 800 MHz to 5000 MHz it gives me some return loss results. Then, If i change the ragne and simulate from 800MHz to 1000 MHz, it gives me different return loss result for the frequency of 900 MHz. I would like to know if someone knows why.

thank you

I do not find any logic to simulate for that wide frequency sweep. It is always recommended to sweep close to the region of frequency of interest. What I can comment about getting the different results for same point in different frequency sweep my be due to the interpolation of results and mesh definitions. If you are keeping the number of simulation points same while changing the range of frequency then may be solver is not simulating at the exact frequency of interest and results which you are seeing is interpolation based on the nearest simulation points. Another way to investigate the issue further is to have little more deeper look into convergence profile for a single frequency value and try to refine it or define more strict criterion like increasing the number of converged passes etc. Last and the most important point in defining the sweep is that keep it more and more real that near around the designed frequency point and if it is absolutely necessary to sweep in wide range then spend more time with the mesh and do not trust on default generated mesh.

HFSS (Ansys Electronics Desktop) 18 is out now. It has broadband and multi-frequency meshing.

If you can't get this, a general rule of thumb is that the higher frequency is better -- but it depends on your structure.

The issue is that the solution frequency will determine which mode is dominant when meshing. If your structure only supports one mode over your frequency range then it doesn't matter, but if there are multiple modes present, the solution will set up the mesh for one modal profile. which may be very inaccurate for the excited mode(s) at a different frequency.

The key here is knowing your structure and which modes you expect to see.

I have one question from very useful answer posted by you. Right now I am doing sensitivity analysis from fabrication point of view to determine what is the surface finish we can live for our prototype to have a cost effective prototype without compromising the RF parameters. So mode of interest is one and model is same and variations will be in the range of micron as focus is see the affect of surface finish. What I have done is I have adjusted the initial mesh settings to one micron and then refined further in sub meshing the curved areas. This is my approach the results I see for a sweep of 10 micron change sometimes fail to converge. What is the best way out to have more refined and converged solution. Solution converges for range of values .

It sounds like you just need to keep doing adaptive passes. Make sure that the maximum number is a larger number (e.g., 50), rather than the default.

Which number should be larger ?
I am doing parameter sweep and I have set max number of passes 10 and delta frequency 0.001 keeping the minimum frequency 2.8 GHz while my mode frequency is 2.998 GHz. Point is what should be the considerations while defining the mesh as I have to find the affect of micron level so meshing is very important

The maximum number of passes (10). It's doing ten passes without getting to your convergence criteria, giving that warning (failed to converge), which indicates it's not sufficiently meshed -- so you need to do more passes.

Good Luck.

Things are working fine now but I need bit more explanation about the mesh generation.
I have assigned different mesh setup for different areas of model as it gives me flexibility to define more denser mesh in complex areas from geometry as well as computation point of view. Fine What I am not able to understand properly is what is this Surface Approximation in HFSS mesh as it directly affects the simulation accuracy? Does it controls the mesh on all the surface area of model? It means it over writes the mesh already defined in that area? please little explanation

Assigning the Surface Approximation Mesh Operation forces all of the mesh points in the vicinity of the surface to be located exactly on the surface, thus preserving any complicated surface topology you might have. If you have a seed mesh set up in the same volume, then yes, I believe it prioritizes the surface mesh.