Continue to Site

Welcome to

Welcome to our site! is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

HFSS deembed waveport for coax

Not open for further replies.


Newbie level 5
Oct 12, 2009
Reaction score
Trophy points
Activity points
Hi all,

If I have a coax (whether it be inside the airbox or outside) that connects to my structure (coax probe connects inner coax PEC conductor to structure), and the waveport is defined on the farthest face away from my structure, what is the length that I need to deembed?

Should i deembed from my waveport through the coax towards my structure, with a deembed distance of coax length? Or coax length + probe? I want to find S11 at the input to my structure, and I am guessing that deembeding should do the trick (since the waveport is defined all the way on the other side of the coax).

Also I thought that if I have any coax length and deembed the whole length of the coax, my results should be the same. However my S11 changes as I vary the coax length, even though I deembed the whole coax. Why is this happening?


It sounds like you want the s-parameters of the structure by itself, in which case you would de-embed to the edge of the probe.

As far as different lengths of coax giving you different S11, I think that your coax might not be long enough in the first place. I'm thinking that you used a perfect coax to go from the boundary wall to your structure. This is typically how you need to simulate in HFSS. So, if S11 depends on your coax length, try making it longer and then adjust the length and see if S11 depends on the length of the coax.

Basically, you want to let all the fields that can couple to the boundary wall to die down before it hits your structure. So, in my experience, I've needed a longer input line and then de-embed it to get the s-par of the structure without any affect from that feedline.

Hope this helped any. Good luck!


    Points: 2
    Helpful Answer Positive Rating
Hi tomtom,

Your comment about making the coax longer makes a lot of sense. I have been using a short coax (because currently I was trying to figure out how to make the simulation work with the coax inside the radiation box. Of course I used a PEC cap drawn on the waveport) I will instead extend my coax all the way to one of my radiation box's walls and remove the cap.

But there are two setups that I see possible, which one is better or perhaps outright wrong?

1) My coax extends from my structure all the way to the radiation box wall. The very long coax is totally within the box, and I define a waveport on the face of the coax that touches the wall

2) I have a long coax OUTSIDE my radiation box, and when the coax reaches a box wall I have a probe (same size as my inner conductor) inside the box going all the way to the input of my structure.

I'm thinking perhaps 1) is the better way to go, although the HFSS book and many examples have the coax drawn outside the radiation box with only the probe inside the box. In either case, I would de-embed the length of coax+probe?

Thanks a bunch

Hmm ... The way I would do it is to have the coax start on the side of the radiation box and place the waveguide port on that face. So make the radiation box big enough such that the coax is in the middle of the face of the box. Then make the coax long enough and de-embed to the pin.

Also, I would make the bottom of my radiation box be a the bottom of my structure and make sure that it has metal boundary condition because you need everything in your simulation to have the same ground and this will define the same ground for everything.

I think if you're still confussed, it might be best to post a couple pics of your simulation and I might be able to help you further if possible.

Hi tomtom,

thanks for your reply. you raise a point about a ground, which is what I have been wondering about actually.

My structure is very simple to explain in words. Its just a single loop that I created by sweeping a circle. However the loop is not closed (swept only like 355 degrees) so the loop ends are not joined together. My goal is to excite this loop with a 50ohm coax, so I can find inductance from the mismatch of the 50ohm coax and the impedance of my loop inductor. Specifically, L=im(Z11)/w.

I'm not sure where I would put a ground such as a plane with PEC boundary condition with infinite ground checked. In real life, my loop would just exist by itself with no physical ground plane near by (unlike structures like CPW's that have an explicit ground plane in fabrication).

Currently, my setup is as follows: I have only one coax for this simulation, and it extends all the way to a radiation box wall like you suggested. I have one loop end attached to my PEC coax probe, and the other loop end attached to the PEC coax outer conductor. The reason I did this is because I view coax stimulation as the input field going through the coax inner conductor into one of the loop ends. Then the other loop end is connected to the outer conductor, which I deem "ground" to close the current path. But I don't think this is quite right. I've also tried using two coaxes, one for each end so I have two waveports. But for this I don't know how to have any "ground" in my design.

I appreciate your suggestions

Not open for further replies.

Part and Inventory Search

Welcome to