simulate a microstrip in HFSS

[zhul3]

Just a simple microstrip...

1. a rectangular ground box (pec)
2. a rectangular substrate box (e.g. ε=2.2)
3. a rectangular conductor box (pec)
4. two rectangular (no box) wave ports (not assigned material), the same width as conductor, the same height as ground+substrate+conductor

After simulation, S21 and S11 are both very bad.....

I understand HFSS sees all faces of the model which are exposed to the outerspace as an E boundary. So I am thinking to add an air box outside the microstrip structure. Since Perfect E boundaries are also on the air box, I think the air box should be much bigger than the microstrip, otherwise it will like a suspended substrate stripline structure.

But the problem is how I can set the wave port into a big air box? HFSS doesn't allow any port set in the material on both sides.

Instead of using air box, I also thought of using "radiation boundaries" on the microstrip itself, but the result is even more ridiculous.

Could any friend here clarify how I can solve this problem? Thanks

just a simple microstrip.....

 

[rotmanlens]

well brother i see that both the pecs are boxes.which to me sounds a bit corny.pecs(ground and copper) are always been rectangles.and radiation boxes dont let waveports inside of the box.so lump ports has always been the excitation for microstrip.

i hope it helps.since u said ur s11 and s12 are ridiculous i am guessing you dotn have a proper matching network.please look into.

i dont have acess to HFSS now but when i get to school i will look into your design.and post more comments and suggestions.

regards
sajid mohammed.

 

[zhul3]

Hi Sajid,

Thin Rectangular box should be fine for the GND and conductor. Otherwise you cannot assign materials for just rectangular. I am trying to use lumped port and see whether it will be better.

The dimensions of the microstrip is set as a "50 ohm" one, so even it is not that accurate, it should not be away that far...


Lee

well brother i see that both the pecs are boxes.which to me sounds a bit corny.pecs(ground and copper) are always been rectangles.and radiation boxes dont let waveports inside of the box.so lump ports has always been the excitation for microstrip.

i hope it helps.since u said ur s11 and s12 are ridiculous i am guessing you dotn have a proper matching network.please look into.

i dont have acess to HFSS now but when i get to school i will look into your design.and post more comments and suggestions.

regards
sajid mohammed.

ΩΩΩΩΩ

 

[rotmanlens]

hi lee.

what is the frequency at which you are operating and what is the width of microstrip.

and i am guessin both the sides have lump ports.

 

[zhul3]

Frequency is around 20 GHz and the width of conductor is 1.465mm, which I calcuated for a 50 ohm impendance.

hi lee.

what is the frequency at which you are operating and what is the width of microstrip.

and i am guessin both the sides have lump ports.

 

[rotmanlens]

everything seems alrite.just cant figure out whats the matter with the whole messed up s parameters when i did the same simulation with a infinite ground plane with 20 mil thickness of epsilon 2.2 s11 was way below -30dB and s 12 close to 0dB on desinger.

let me look into it more my friend.if u have the solution let me know what was wrong with it.
sajid mohammed.

Added after 51 minutes:

i think i figured the problem.let me upload the file so that u can check with it alrite.

sajid mohammed.

Added after 54 minutes:

i actually redid the whole simulation from the beginning.all the numbers are good. but to my suspicion the duplicating of the rectangles for the lump ports could be a reason for improper result and in the project tree i saw a move command which looked suspicous.i have the file which i have simulated i didnt bother for the length because it is meaningless.

if it really helped .dont forget to add "it helped ".

thanks a lot again.

sajid mohammed.

 

[zhul3]

i actually redid the whole simulation from the beginning.all the numbers are good. but to my suspicion the duplicating of the rectangles for the lump ports could be a reason for improper result and in the project tree i saw a move command which looked suspicous.i have the file which i have simulated i didnt bother for the length because it is meaningless.

if it really helped .dont forget to add "it helped ".

thanks a lot again.

sajid mohammed.

Hey sajid,

I changed to lumped ports and set a big air box outside, now the result is making sense.

Thank you.

Lee