How to simulate the complex antenna structure over 50 GHz?

I am now simulating antenna over 50 GHz with package signal traces, vias and solderballs. The structure is very complex (5 layers). I am simulating it in HFSS. the initial mesh costs even 1 G RAM. 1st round mesh can not be finished with message of " insufficient memory'. Can anyone provide some idea how to decrease the calculation load in HFSS? or suggests me using other softwares for such complex structure?

Try using CST 2006B version with a new subgridding function! The mesh cell can be significatly reduced thereby for the occupying memory!

Try to find a symmetry in your structure (if you're lucky) and use symmetry planes to reduce the mesh size.

Exclude very small details from your structure which don't have great impact to your structure properties at your highest frequency (50GHZ?).

Otherwise add more ram and switch to 64-bit OS + HFSS.

EDIT: I would like to see your model. If there are no patents or any other secrets, could you upload your structure?

hi,

you can try semcad x. for evaluation, you can get full version with one month license (www.semcad.com)...

cheers,

Thanks. I am simulating the structure in CST2006 now. I am new to it. For the same complex structure, which can not be done by HFSS, it can work in GST2006 and get the final result although it costs almost 12 hours. It shows advantage than HFSS at this point. It is true that I still need to learn mesh and port set up to make the full simulation run.

CST 2006 uses the PBA technique according to its mannual? Could you explain what is it and also something about subgridding technique? BTW, how can i get the CST2006b? I am using 2006A, so CST2006b can improve my simulation better?[/img]

[quote="oflee"]hi,

you can try semcad x. for evaluation, you can get full version with one month license (www.semcad.com)...

cheers,

Thanks, I have asked for a trial version. It is based on FDTD right? Will follow up once i get the result. But I once use the XFDTD, it is okey for the same complex problem at 5 GHz. I need to check the problem in 50 GHz

Added after 3 minutes:

Thanks. AT 50 GHz, the small details maybe have influence to performance when they are compable with wavelength. I hope get more ideas but can not upload the model owing to the project patents.