EM Simulation taking a long time on ADS

[arwen16]

Hello,

I am simulating an overall 60mm long structure consisting of two non-uniform SIW structure of 25mm each with a gap between them, from DC to 10 GHz (Npts=50). I used 6 GHz as the mesh frequency and 80 cells/wavelength. The EM simulation of the structure took more than a day. Initially, I did the EM simulation of each structure individually, each taking about 10 hrs. The results were not accurate as the gap was not considered. So, I decided to simulate the overall structure with the gap but the simulation is taking more than a day.

Could someone please tell me where am I going wrong? I wish to reduce the simulation time considerably.

 

[BigBoss]

You should provide these information first.
-Which EM simulator you've been using ?
-What is your substrate and how it's complex ?
-How is your layout complex ?
-Which mode you've been using in ADS Momentum (if it was) ? RF Mode or Microwave Mode
-Why 80 cells/wavelength ?? Is there any forced circumstance ?
-What are your computing resources ? (Processor, how many cores, memory etc.)

 

[arwen16]

- I am using ADS Momentum Microwave.
- I am using an FR-4 substrate of 1.6mm thickness
- A non-uniform curved SIW structure
- I read somewhere that if you have a large bandwidth, it is better to reduce the mesh frequency and increase the number of cells/wavelength.
- 32 GB RAM (* cores)

 

[BigBoss]

Post your project file as archived format. (Do not zip/rar whole project directory, make it archive format).
Let me try on my computer.

 

[volker@muehlhaus]

- I read somewhere that if you have a large bandwidth, it is better to reduce the mesh frequency and increase the number of cells/wavelength.

Technically, that makes no sense. The maximum cell size for mesh is calculated from cells/wavelength using wavelength at mesh frequency. So if you use half the frequency and double the cells/wavelength, you get the exact same mesh as before.

In your case, go with 10 GHz as mesh frequency and 20 cells/wavelength. Some selective edge mesh might be useful, that depends on the layout and conductor width & gap width.

To help you, we need to see more details of your model. One possible simplification might be to use thin (sheet) metal for the planes instead of thick metal.


Best regards
Volker // Keysight Certified Expert EDA