Electronics

Electronics

Topics related to HFSS, Maxwell, SIwave, Icepak, Electronics Enterprise and more.

Boundary setting for characteristic mode analysis

    • fparveen
      Subscriber

      Hello, 

      I am trying to find the characteristic modes of a vivaldi antenna in HFSS 2023 R2. 

      I have set lumped port excitation to the port. 

      I have created a lambda/4 free space all around. I usually use 'Radiation boundary' when using HFSS (Modal Network) solution type to find S11. But in the Characteristic Mode solution type, the option 'Radiaion boundary' is not available. Can anyone suggest how should I setup the boundary ?

       

      Thanks

    • Faezeh Ladani
      Ansys Employee

      Hi, 

      You don't need an airbox and Rad boundary in Characteristic Mode solution setup. This solution is based on integral equations. 

       

       

       

       

      • fparveen
        Subscriber

        Hi,

        Can I use 'distributed memory matrix solver' using two computers (16GB RAM on each) ?

        To specify my problem,  my antenna (characteristic mode solution type) simulation cannot complete due to out of memory problem. I have 12 GB RAM. It requires almost all the memory (~11.9 GB) for adaptve pass 1. But it stops when it starts adaptive pass 2.

        As, the simulation can't be completed at the solution frequency, it doesn't even start the sweep simulation. 

         

        Please suggest if I can combinedly use two or four computers each having 16 GB RAM. If yes, please kindly suggest how should I combine them and how to setup HFSS for that. 

         

        Thanks 

    • fparveen
      Subscriber

      Thanks. I removed the excitation and airbox. Now simulation runs with only the antenna. 

       

      But the simulation cannot complete due to out of memory problem. I have 12 GB RAM. It requires almost all the memory for adaptve pass 1. But it stops when it starts adaptive pass 2. 

       

      My setting is solution frequency = 4 GHz 

      I set a linear step discrete frequency sweep from 1 to 4 GHz. (BTW it cannot start sweep simulation, as it fails on the 2nd adaptive pass during the simulation at the solution frequency.)

       

      The mesh size in adaptive pass 1 is almost 5000, I cannot know the mesh size for adaptive pass 2, as it could not finish the pass 2. But it should not exceed 65,000 in pass 2. (I am using student version.) Also, the error is not saying that it stopped due to student version limit. It says about out of memory issue. (So, I think I can do it using student version. I just need to accomodate the task within 12 GB RAM. Or do with bigger RAM.)

       

      I didn't change any mesh settings. Can you please suggest how I can finish the simulation within 12 GB RAM? Is there any way that I can get acceptable accuracy by reducing the mesh or by doing anything else? 

       

Viewing 2 reply threads
  • The topic ‘Boundary setting for characteristic mode analysis’ is closed to new replies.