solve inside hfss

I'm a new HFSS user, I'm simulating a coupled Coaxial resonator and the inner and outer conductor are assigned to be of copper material, when I simulate I get this warning message

HFSSModel1
[warning] Solving inside a solid with high conductivity may require a large mesh

and I can't find the problem, but when I remove the 'solve inside' option from the material properties the warning message goes, but the results are wrong so if anyone knows what to do please tell me Thanks

Hi,
could you please upload the file the way I can have a look at it and try to fix the problem? You have to improve mesh refinement inside the copper using mesh operations (select the object and right-click on it-->mesh operations). Then plot mesh to verify your mesh improvement. Another solution that may come helpful could be the use of finite conductivity boundaries on the metal surfaces,resulting in a reduced time for the whole simulation. But I need to view the file to say what's the best way to simulate it.

Bye

Ivan

Hey,

Thanks a lot, probably you going to find a lot of things that are not right please if you can advise me about anything that I can do to improve my simulation please tell me thanks a lot

thank you

Hi,
I have HFSSv11. So, I hope you can use 11 version as well to open the file I'm gonna modify.
I saw meshing is very poor. Moreover, there's no need to define mode vector at port since TEM mode is enforced by the structure geometry and boundaries. I have a couple of questions : what's the point of that air gap between the two inner conductors? Is it part of the model or do I have to remove it and make a unique inner conductor? And what's the exact operating frequency you want to view fields at? Let me know so I can go ahead with the simulation.

Ivan

Hi,

Thanks a lot for your help I really appreciate what u r doing, the gap thing is a capacitive coupling, the resonator is a capacitive coupling resonator and this is how the model should be, and the operating frequency is 1.0351 GHz.
please if u can list all the modifications u have done thanks.

Husam

ok, next week I will fix meshing and will send you the corrected file.
Bye

Hi Husam,
I suggest to remove copper and use PEC. This shouldn't affect resonant frequency. Using copper may result in a complex frequency, the real part indicating the resonance (very sligthly different from the one calculated with PEC) and the imaginary part linked to ohmic and dielectric losses. But you can avoid solving inside the copper, with a significant reduction in total simulation time. Anyway, a meshing improvement must be set for dielectric material. For setting the mesh refinement:
1) Select the object;
2) Right-click on it and click 'assign mesh operations' -->'inside selection'-->'length based';
3) Reduce maximum length of element (The default value is set to 20% of the maximum edge lengths of the bounding boxes of each selected object?s faces): try very low fraction of wavelength, sometimes I use 1000 or more fractions of wavelength to have a very high resolution, but it depends on the computational resources you are allowed to use. If you want to limit number of thetaedra fix a max limit (i.e. 3000).
4)Go to setup-->right-click-->apply mesh operations. Then plot mesh, after object selection and right-click on it: you will see the meshing refinement of the object;
5) If you are not satisfied of the refinement (too accurate or still too poor, repeat previous steps).

Another meshing improvement can be realized improving number of adaptive passes and percent mesh refinement per pass (default 30%, if you want try with a 50% or more).
Naturally, this operations cause a longer simulation time.
There's no need of setting port mode because it is forced by boundaries and geometry.
If you want to keep copper, you gotta apply the same operations inside the material itself, but meshing improvement must be extremely high in good conductors such as copper. This implies very high time consume. Another solution could be the use of finite conductivity boundaries: apply these conditions at copper object surfaces (not to be solved inside) and copper conductivity is set as default (be careful of using an interpolated analysis, fixing the maximum frequency of the range as solution frequency, if you are going to use either non-zero thickness or/and roughness).

Actually, If I were you, I'd have used another solution type for resonators: eigenmode solution. There's a really explicative example in HFSS to learn how to use this kinda solution. I seem it is called 'cavity resonator'. Try to run it, it's not difficult to understand. Moreover, there should be a .pdf tutorial in HFSS directory.

Hope this helps.

Ivan

Thanks a lot :) , I have tried to do the finite conductivity boundaries thing the results were not fine, but I saw in couple of HFSS tutorials that they don?t use the outer conductor (for coaxial cables) so I removed it and I increased the length of the small feed in inner conductor and I got much better results I?ll upload the file for you and I?ll upload the results for different mesh refinements please if you can check them because I think the first HFSS model is wrong so see this one and if u please tell me if there is anything wrong with my model thanks a lot you have been a great help.
One last question is it possible to assign the outer conductor as a ground because I want to check if I can do the model with the outer conductor (of the coaxial cable) and compare the results.
And other thing for all the simulations I still get the warning message (the large mesh one) but I think it?s not that important.

Thanks For everything
Regards

Husam Barjawi

The message appears because meshing inside the conductor is still poor and meshing must be exremely high i conductors. I think you can avoid to solve inside the conductor. Solving inside will give you more information about skin effect and ohmic losses inside it. But I think you are more interested in resonances. As for the outer conductor, just try to apply a PEC along the external surface of the dielectric. Have you tried with the eigenmode solution?
Bye.

I.

Added after 3 minutes:

PS-->By the way, the port mode is not well-defined in your last design. Just set no integr. line and put the outer conduct. using a PEC to enforce field reference at port. What was wrong with the use of finite cond. bound.? Have u tried with a zero thick boundary (avoiding to solve inside the copper of course)?

No i have not tried the eigenmode i was planning to try it tonight, but i tried the PEC instead of copper and i had the same warning msg when i assign solve inside, and when in general i simulate the model without solving inside the results i get are really wrong I think it affects everything (I'm not an expert but I tried many things the best thing till now is when i remove the outer conductor) I want to ask u, when there is no outer cond. does it means that its connected to ground?

With pec, there's no need to solve inside. As you know, skin depth is zero if the surface is a perfect conductor, so there's no need of solving inside and no field is inside the inner material in this case. The wrong results you get without solving inside may depend on the way the excitation and boundaries are defined.
The absence of the outer conductor for the modal soution at hand causes no reference for the dominant mode enforced by the wave port. Try with eigen mode solver, leave the only dielectric cylinder removing all metal cilinder and PECs. For a better understanding have a look at the HFSS example called 'cavity'. Define the number of modes you want to calculate the resonances for. Define the lowest frequency to let resonances calculation to start from, min and max number of adaptive passes (I suggest a number of passes sufficiently high, min 8 and max 15), and verify the convergence. If not converged, increase the number of max adaptive passes. The more you set convergence constrains (low deltaS) and number of passes, the better your accuracy will be. The default value of deltaS, anyway, should be enough. After the structure is simulated, go to setup1, righ-click,then select eigenmode data: a list of N modes with their relative resonances and Q factors will appear. Frequencies are complex, to keep account for dielectric losses. Get the real part. That is a commonly used way to simulate resonators. Let me know.

Ivan.

a useful topic for me

hey Ivan

Sorry for not replying for a long time but i had an accident and it was not good, I have tried the eigen mode thing and its fine but it would not help me, because i'm investigating the gap effect on resonance and some different end Tips for the resonator, but thanks for everything man, I was doning somthing wrong before (finite cond.) but now it worked fine thanx again man, everything is fine now.

Hello guys,

I am a beginner for HFSS, I am trying to simulate a tungeston via with very high conductivity, and look at the S-paramters, and I already have the same warning
"solving inside a solid with high conductivity may require large mesh", I got the s-paramters results and it didnt make sense to me, so I thought the mesh refinement is the cause, any help form your side is apprreciated.

Thanks a lot
Shadi

i m havin warning message like....THE BACKGROUND DOES NOT FULLY TOUCHED THE FOLLOWING BOUNDARIES :PER E1 ( GND PLANE)