Help on HFSS simulation problem

[lhasayak]

Attached please find the HFSS file. It is single ended microstrip line that goes from top to bottom layer through via. The lumped port is added on two ends. However, the simulation result doesnt seem correct. I couldnt find the problem with it. could anyone take a look? Thanks

 

[dingjingfeng]

this is a ground problem of HFSS
the ground you think is not a true ground

 

[lhasayak]

sorry. i didnt get it. though two segments of microtrip line reference two different layers of ground, the return displacement current can still go across between two ground layers.
if i use waveport instead, the result looks all right. How come it doesnt work for lumped port?

 

[jallem]

Your model does not have a DC continuity.
The second graphics is a modified version which
I added two vias for continuity. At RF is better
to think in terms of transmission lines.

 

[lhasayak]

jallem and other gurus,
please see the attached file for waveport usage on this example. even without via shorting two separate grounds, the S11 and S12 look all right sincce as i said at relatively higher frequency the two ground planes can couple to each other.
My question is that the result using lumped (without adding shorting via) looks obviously wrong. is this the limitation of lumped port?

 

[dingjingfeng]

In my opnion, the main difference between the luped and wave port design is

the Two Ground in Waveport is already connected together with outer boundry.
but the ground of the lumped port design are floated

 

[lhasayak]

Thx for replying this.
The whole structure is enclosed by air, to which i assigned rad boundary. so i dont understand how and where two grounds get shorted in the case of waveport. as far as i know, the default boundary of outer is PEC. However, it is not the case here.

 

[dingjingfeng]

you can just shrink the ground plan and tell us the result

maybe I had a mis-understanding

 

[jallem]

Reducing the ground plane will give you the correct
answer, see below:

 

[dingjingfeng]

the Lumped port is seemed as an internal port by the simulator, it's better to have a short return pass between them.

I don't have much experience for using the Waveport.

as to Transmission line, if it is matched, the s11 shoud be very samll and the insertion loss S12 should be large(near zero)

I cannot understand why the S12 is lower than -5dB at low frequency, they are short at nero zero frequency.

 

[lhasayak]

I can verify it by extending air and corresponding rad boundary. That is interesting finding for me. I still dont understand for waveport application, how come the two grounds get shorted? The rad boundary doesnt short the two planes. The port doesnt short them either. what gives the DC continuity?