Help with Waveports - setting ports properly

[Phanixis]

Help with Waveports

I have run into a problem in my simulations.

I want to simulate the near-field coupling between two adjacent twin-lead lines. To do this, I run two sets of parallel copper lines inside a single vacuum box.

Now I want to define ports, which are defined between each pair of wires on each side of the vacuum box. The problem is that I can only define one waveport per face of the vacuum box, and I need two ports on each side because their are two wires.

Worse still, the waveports can't descriminate between the two sets of wires. If I place the lines far apart so that one wire can't excite the other, and I excite a single wire with port 1, and locate port 2 on the second wire on the opposite end of the box, the simulation will still give me a 100% power transfer for S(1,2).

I have thought about using lumped ports instead, but I don't know the impedance of the transmission lines, and I want to lumped ports to be impedance matched to the lines.

Can anyone help set the ports properly for this simulation? Thanks in advance.

 

[shlomo22]

Re: Help with Waveports

You shouldn't use the whole face of the vacuum box. Draw a rectangle on the surface of the box, taking up part not all of the face, and use that rectangle as the waveport.

By doing so you can make the waveport include one wire but not the other. However, you have to be careful about how big to make the waveport. The fields from any wire extent to infinity but the waveport does not, so a waveport exciting a wire will always be at least slightly imprecise. If your wires are close enough to significantly couple with each other, then the fields are significant throughout the space between the wires - but if you want the waveport to touch only one wire, it cannot enclose all that space, so the results will be significantly inaccurate. This is a dilemma, but I think the way to solve it is to start the wires far apart from each other where there's little coupling, then bring them closer and run them parallel for a while, then separate them again before they reach the waveport(s) at the other end. I suspect the coupling will then vary significantly based on the length over which coupling is significant (i.e. how many times the wavelength) - try it and see. Of course in real life, coupling lengths are finite not infinite, so use the actual length if you know it.

Note: I'm not 100% sure about the previous paragraph, if I'm wrong somebody please point it out.

In addition to lumped ports, and waveports as I described them, HFSS also has a "driven terminal mode" which seems like it might be what you want here.

 

[Phanixis]

Re: Help with Waveports

Ok, I have taken your suggestion and seperated used rectangular faces to separate the waveports from the vaccuum box. Now I have run into a problem where the transmission line poorly guides the waves, and I am suffering losses of 20dB over a 3cm line. Perhaps this is a consequence of the transmission line used, but the dimensions of my line match closely to those of a twin-lead line, and should be able to handle frequencies of several hundred MHz.

I would like to try the driven terminal mode, but I do not know how to set this option.

Thanks for the help.

 

[shlomo22]

Re: Help with Waveports

For driven terminal mode, go to HFSS->Solution Type menu and choose Driven Terminal. Then you have to adjust the port definitions and output variables somewhat, I think.

I forgot to mention: If you want to model a TEM wave then you need at least two conductors. The way I suggested to you, each port has only one conductor. So no wonder the losses are high. You could add another conductor on each port - a wire or a single face of the vacuum box, if the port extends all the way to the edge.