Disclaimer: This issue is boring. Not only because it has nothing to do with Cyber, AI, machine learning, etc. (nor do most of the things I do and write about). It’s boring because it has to do with only one subject considering 3D electronic/Electromagnetic simulations. That is: Ports – How to connect them, what do they mean, and how they relate to the circuit (schematic) simulations we later run with them.
!Important! Although it is extremely important, port modelling rules are not discussed here. Again, hopefully this is a subject I’ll be able to get to in the future. Sadly, a lot of these subjects require a solver, and these are rather expensive.
So here is a pretty standard flow – You want to simulate an Analog\RF\Mixed signal circuit. You use a 2-3D (sometimes 2.5D is included between them) simulation software. Before starting the simulation you need to connect ports in either of the following fashions:
- Discrete ports (pin to pin, edge to edge, edge to face) for hard to connect parts, replacing discrete parts,
- Waveguide ports – For arbitrary reference planes on transmission lines, or even actual waveguides!
- In antennas simulations, you can infuse all of the space with a plane wave.
The latter point is extremely cool, but it won’t be further discussed here. The first point is easiest to discuss, so let’s start with the second.
Waveguide Ports
This form of port goes through a secondary solver, that is can be referred to as a 2D solver. This type of solver makes an assumption that the structure is continuous perpendicular to the surface.

As you may recall, a 2D problem can be decomposed into Transverse-Electric (TE) or Transverse-Magnetic (TM) polarizations. The naming means that the Electric field or Magnetic field, respectively, have components only tangent to the port surface.

In some cases, especially when there is more than one conductor, the polarization is neither, and both fields have only components tangent to the port surface. This is referred to as Transverse-Electric(-and)-Magnetic, TEM. Another situation is QTEM, or Quasi-TEM. For example, a microstrip line:

In this case we consider the line encased in a metallic shield, but remember that originally it was encapsulated by air or other dielectrics. Basically it is calculated TEM mode, so I won’t elaborate further here.
The last thing you need to remember about this sort of port, is the port is the source. Meaning it has no sources of it’s own. A 2D problem with no sources? How is it solved then? Simple – It’s an eigen-value (self-states of the structure) problem. This is a very interesting problem on it’s own, and hopefully I’ll find time in the future to write a different article about it.
Discrete Ports
In this case, the ports just pushes current, uniformly, from one point\edge to the other. Let’s present the 3 cases this writer knows of:
- Pin to Pin.
- Edge to edge (rectangular face).
- Coax (circular edge to circular edge).
Pin pin is simply a wire current source pushing current from one discrete point to another.

Edge to edge is the same, but the current source is spread across a surface.

Coaxial port is a special case of the Edge2Edge port, as it allows a non-uniform current distribution. The current is more sparse accross the coaxial ground reference, naturally.

On To The Circuit Realm!
Ok, so we finished the 3D simulation, in whatever method\solver we chose. In case we run an S-Parameter simulation, for example, we usually encounter something like this:
As usual, I’m running my circuit simulations using the wonderful QUCS. Props! So there are many questions to answer here. Let’s focus on a few, that will hopefully help you understand ports a little better.
- How do you set how much power flows through the port?
- Waveguide ports have one access point, but discrete ports are connected by two points\edges. How come in the circuit simulator we usually see one connection point?
Power Calculations
Let’s discuss the issue of power infusion shortly. Firstly, Discrete Ports – In this case the answer is simple. The port impedance is usually pre-set by the user (say, 50Ω?). Assuming we are working in the frequency domain, for example, and we want to infuse 1W, the relation is
.
In case this is a wire port, the relation is direct. If this is an edge2edge port, then this amount of current needs to be distributed along the edge. Arbitrarily,

While discussing Waveguide Ports, the calculation is a bit more tricky. As the electric and magnetic fields across the surface are known exactly, the total power flowing through the surface (via Poynting Vector) can be calculated. This figure can be used to normalize the power to any desired quantity.
Single and Double End
The (supposedly) simple case is the waveguide port. The surface is the port, and it translates directly into a single ended port in the circuit solver. So let’s discuss the discrete port case, before I try and explain that nothing is that simple.
The answer here is simple enough: The conversion is done via transformer. Without out a-priori knowledge, the two ends of the port are connected to a 1:1, non inverting transformer, whereas in the secondary coil, the bottom pin is connected to the Ground. In some software (e.g. ADS) the bottom pin is connected to a common pin for the whole S-Parameter block. This allows referencing it to an arbitrary DC voltage or anything else, if desired. Meaning, that if we connect a discrete component like so:

the equivalent circuit is

Hence discrete components can be added to the simulation externally. This is great for matching networks, for example.
Now let’s return for a single to the waveguide ports. Except for rectangular\circular waveguides, most of the waveguide structures have more than one conductor. For me, at least, this causes a bit of confusion regarding the single\double ended ports issue. I find it simple enough to consider the waveguide port as a more elaborate discrete port, with a precise current distribution. What do you think is a more appropriate metaphor? That’s it! Now you have the entire flow panned out. Hopefully in the future I want to elaborate on these two subjects –
- Eigenmode solvers, and how to choose the right port mode.
- Correct port modelling and problems with incorrect port modelling.