I am desiging THz microstrip antennas for Beyond 5G with time domain solver at the CST MWS.
In oder to validate my simultaion results I am trying to simulate my design also with frequency domain solver at CST MWS (FEM solver).
When I am simulating with Frequecy domain solver I am getting the message at the begging :
"The length ratio between the shortest and longest model edges is 1:8571. Short model edges can be high lighted via Mesh:Mesh Control\Check Messh/Check model Edge length...(It may be advatageous to model the metallizations as infinitely"
Please see the attached Fig.
At the end of simulation I am not getting any results.
Can some one help me?.....
Uri Nissanov, Usually the message that you are seeing is considered as a simulation warning, and I am wondering why you can't obtain your simulation results. I am seeing that you have designated thickness for the upper layer and ground layer conductors. The warning states that compared to your structure's whole dimensions, the thickness of conductors is relatively small. Therefore it is better for you to enhance the meshing quality by configuring it manually. Since you are working in Sub THz bands, in this way, you can account for the skin depth effect of the conductors. But remember, enhancing the mesh quality means more simulation time. If you do not care for the skin depth effect of the conductors, you can model the conductors as sheets with zero thickness.
regards,
Hello,
Some resources are given below.
Hope, it will work for research.
https://www.edaboard.com/showthread.php?113273-CST-frequency-domain-solver
https://link.springer.com/article/10.1007/s11277-019-06299-w
https://www.rose-hulman.edu/class/ee/HTML/ECE340/PDFs/MWS_Tutorials.pdf
Thanks,
Dear researhers
Thanks for the cooperation.
I already learned how to validate my CST MWS antenna design with ANSYS HFSS.
This warning occurs only, when tetrahedral meshing is used. The tetrahedral mesh is much more sensitive to the initial mesh quality. The effects of this will be apparent in the adaptive meshing process and solve time (poorly conditioned matrix due to "low quality elements"). To avoid "Low quality elements" we typically recommend modeling thin conductors as zero-thickness sheets. This can easily be accomplished by picking the face of the microstrip where it meets the substrate. Then in the face healing tools, the option to make shape from picked faces will create the sheet. If the thickness is a critical feature (edge coupled microstrip for example), you can set the maximum mesh step by right-clicking on the object and entering the mesh properties dialog. This helps keep the tetrahedrons from having a very long aspect ratio.
Dear Dr. Fatemeh,
Thanks for detail answer.
I already learned how to do validation with ANSY HFSS.
HFSS is better for frequency domain based simulations. (easier)
Good luck.
I'm wondering ,what types of substrate you used at THz band , I think it will be lossy at this band
I used the substrate that is used at mm wave appilcation up to 100GHz and my design was at about 120GHz.
My substrate was Rogers R03003 that the loss tangent and dielectric constant is know and measured up 100GHz.
At my designs and simulations I define this parameter at 100GHz.
with er= 3 , and tand= 0.001 , even though there is rogers 5880 with er=2.23 and tand= 0.004 as I used at my design at 79 GHz.
what about the thickness you used
Uri Nissanov Yes. If you are looking for precise results, you must increase the mesh size for the conductors. I believe that most of the losses at mm-wave frequencies are accounted for conductive copper losses. Therefore, be careful about the thickness and mesh size of the conductors. You can manually increase the mesh size for the conductors only.
I used the substrate height of 0.127 mm, the lowest height that is possible at Rogers R03003 laminate.
It seems warning not error. You must get your desired results. Well the warning shows the thickness of conductor is small as compared to whole design dimension. Furthermore, i would suggest you to use Ansys HFSS for frequency domain simulations.
Dear UriNissanov
I also got the same problem as you. You already knew how to validate my simulation results at CST (Time domain solver) with a frequency-domain solver at CST. Could you share your experience to me? or you can send me email to [email protected].
Thank you very much.
- Badges
- Science topic
- Similar topics
- Phytochemistry
- Extracts