tsmc 90nm process TL in HFSS simulation COST too much time

[happsky]

I draw a TL in HFSS using 90nm process,the test structure is simple, but due to there are more than 70 IMD in this process,it cost a long time .After 8 hours' simulation ,the "making initial mesh" is not finished , is there any solution to reduce time?? thank you very much.

 

[snaildr]

I draw a TL in HFSS using 90nm process,the test structure is simple, but due to there are more than 70 IMD in this process,it cost a long time .After 8 hours' simulation ,the "making initial mesh" is not finished , is there any solution to reduce time?? thank you very much.

You can combine those very thin layers in to a thicker equivalent layer, I think that should help make simulations much faster.

 

[kobasa]

I draw a TL in HFSS using 90nm process,the test structure is simple, but due to there are more than 70 IMD in this process,it cost a long time .After 8 hours' simulation ,the "making initial mesh" is not finished , is there any solution to reduce time?? thank you very much.

I agree that the main way to reduce simulation time in this case with HFSS is to combine the dielectrics together. A 3D planar EM software is available from Sonnet that allows you to simulate these types of structures without simplifying the dielectrics because only the metal is meshed in Sonnet and not the dielectrics. A transmission line with 70 dielectrics would only take minutes, if not seconds, to solve with Sonnet.

Visit www.sonnetsoftware.com for further information.

Mike

 

[happsky]

snaildr,kobasa , thank you ,I agreed with you .
And acturely ,I found that using HFSS13 (not HFSS12) will solve "making initial mesh" problem in my case.
thanks again.

 

[kobasa]

snaildr,kobasa , thank you ,I agreed with you .
And acturely ,I found that using HFSS13 (not HFSS12) will solve "making initial mesh" problem in my case.
thanks again.

Ok, that's good....so you don't have to combine the dielectrics to make the initial mesh. However, if it still takes too much memory and simulation time and you do still have to combine the dielectrics, you want to make sure to combine them in parallel. If you need help, please don't hesitate to ask.

Mike