Wave Ports are a versatile way of exciting your HFSS designs, which can be very useful for evaluating transmission line geometry, and for making sure that other designs are excited in the most realistic way. In this article, we will explore the capabilities and pitfalls of wave ports that will allow you to get the most out of these excitations.
What they are and how do we use them?
Wave ports are 2D surfaces at the edge of or inside a simulation which through a signal exits or enters the rest of the geometry in the simulation. These surfaces behave as if a transmission line has material properties and cross-sectional geometry incidents to the port extend out infinitely away from the edge of the solution space). Consider the following images from the HFSS help documents – a coaxial bend is pictured containing a few Teflon slugs, excited by a wave port, flat against the edge of the solution space, (the circle marked “port” in the first image).
Though not shown when you assign a wave port, the picture above is electrically equivalent to the picture below. For instance, if a wave were traveling up the coax bend towards the wave port, it would “see” an infinitely long coax line with the same cross-section as the wave port (and keep propagating).
As with any port in HFSS, they are surfaces where incident and absorbed energy is recorded by the simulation, allowing for one to obtain several different electrical properties of the design being simulated.
General Setup
This section details the options available to you when you choose to use a wave port. One can assign a wave port to any sheet or face simply by right-clicking, choosing “Assign Excitation” and then selecting “Wave Port”
After one selects this option, you are greeted with the following dialog, with the most relevant tabs being “General” and “Post Processing”. Through the options available through this dialogue, one has complete control over the number of modes available, their alignment, and polarity. For most user’s applications, however, only a few highlighted parts of the dialogue are relevant and will be covered in the following three sections. For users who would like to use multiple modes and/or higher-order modes of arbitrary alignment, the HFSS help provides an excellent in-depth explanation of how to achieve those ends.
Integration Lines
With the default “Mode Alignment and Polarity” setting checked, the integration line is used to determine the polarity of each mode. In keeping with this, it is primarily important to draw the integration lines in a consistent manner between wave ports of the same transmission line type. The pictures below show the resulting mode of the impact of different directions of the integration line.
For two-conductor t-lines, a good practice is to draw an integration line from the signal pin to the ground pin along the path of the strongest E-field in the anticipated mode joining them. For waveguides, consistent practice is to draw the integration line from the mid-bottom of the open rectangle to the mid-top (similar to the common depiction of the TE10) mode. The “snapping” feature of HFSS drawing aids greatly in this process (triangle when hovering for the midpoint, quarter-circle for circle quadrant, etc.). Visualizations of the placement of these integration lines can be seen in the “Wave Port Best Practices for Common Transmission Lines” section.
De-Embedding
De-embedding is a great way to have a physical excitation in the form of a wave port, without including the response of any of the excess transmission lines in your simulation. It is common to de-embed as much of the transmission line having the same cross-section of the wave port as possible to get only the response of the desired structure.
To demonstrate how de-embedding works with wave ports, consider the case of an open-circuited 5mm long 50-ohm coaxial cable in free space excited by a wave port on one end:
Using the “Get Distance Graphically” option in the “De-embed Settings” pane, one can draw a vector from the center of the near face of the inner conductor to the center of the far face of the center conductor, designating that the entire length of the cable should be de-embedded:
With this setting active, HFSS will find the parameters of the transmission line given by the cross-section of the wave port and remove an equivalent length’s worth of the response of an ideal lossy transmission line with those parameters. Since this is considered “Post Processing”, this will not invalidate your existing solutions. The difference in response is shown in the Smith chart below, with the red line demonstrating the de-embedded response, and the blue line showing the embedded response over a frequency sweep of 0.1 to 10 GHz. One can see that the red line looks very close to a clear open circuit on the right side of the chart (with some end capacitance) and the blue line shows the phase shifting of the 5mm transmission line.
Renormalization
Renormalization simply takes the calculated Z0 of the transmission line and allows you to view the S-Parameters as if they were referenced to another characteristic impedance.
Consider the case of our same 50-ohm coaxial line with a 75-ohm resistor boundary condition at the end of a short length of the line. In this case, the blue dot shows S11 of this line with the calculated Z0 of the wave port (50 ohms), and the red dot shows S11 renormalized to 75 ohms (and thus matched.)
Wave Port Best Practices for Common Transmission Lines
Some best practices and proper modes for common transmission lines are given below. Notice the vector E-field plot of a solved wave port by clicking “Port Field Display”> “PortName”>” ModeName”:
Coax
Typical Integration Line and Mode Field Display:
How to Define Wave Port Geometry:
- Assign port directly to an annular side of the dielectric by pressing “F” while the modeling pane is in focus, then clicking the side in question:
Waveguide
Typical Integration Line and Mode Field Display:
How to Define Wave Port Geometry:
- Assign port directly to a side of the dielectric/air subtract object by pressing “F” while the modeling pane is in focus, then clicking the side in question:
Microstrip
Typical Integration Line and Mode Field Display:
How to Define Wave Port Geometry:
- Draw rectangle centered on signal, bordering ground conductor, about 8 line-widths wide and 10 substrate heights tall:
- It is strongly recommended to place incident to the edge of the solution space with a radiation boundary active, and under your simulation setup to check “Advanced”>” Use Radiation Boundary on Ports”
Grounded Co-Planar Waveguide
Typical Integration Line and Mode Field Display:
How to Define Wave Port Geometry:
- Draw rectangle centered on signal, bordering ground conductor, about 8 line-widths wide and 10 substrate heights tall:
- It is strongly recommended to place incident to the edge of the solution space with a radiation boundary active, and under your simulation setup to check “Advanced”>” Use Radiation Boundary on Ports”
Wave Port Placement
There are a few important points to take under consideration when placing a wave port.
Leaving Between Wave Port and a Discontinuity:
It is important to leave a length of transmission line with the same cross-section as a given wave port for HFSS to solve the wave port properly. If this is not done, fields that are part of a transition incident on the wave port may cause additional non-propagating modes to be present at the wave port. Pictured below is a transition from a rectangular waveguide to a circular waveguide – as pictured on the left, the changing cross-section close or incident to the wave port would incur this problem. On the right, a length of the line has been added to mitigate this issue. Remember that the response of this length of the line can be removed through de-embedding detailed earlier.
The exact amount of space to leave between a discontinuity and a wave port unfortunately is a bit involved to determine. From the help:
“To precisely determine a distance you can solve the 'port only' with one additional mode (for a correct port definition this mode will be evanescent) and extract the decay length of this mode from its complex gamma. Then, as a rule of thumb, place the first discontinuity at least three times this decay length.”
In the author’s experience, 2-3 guided wavelengths at the highest frequency of operation is a good place to start. Pay special attention to the warnings or errors in your message manager pertaining to evanescent or additional modes.
Placing Wave Ports inside of the Solve Area:
All the wave ports described so far have been incident to the edge of the solution area, but one can also add wave ports inside of the solution area by simply adding a PEC backing. Pictured below is a section of waveguide with a wave port completely internal to the solution area. The blue rectangular prism was drawn, covering the entire extent of the wave port, and then its material was set to “pec”. Simulation executes without error with this configuration.
Using “Solve Ports Only” to Get T-Line Parameters and Applications
Though wave ports are most often used as an excitation in order to find the response of a 3D structure, HFSS also offers the option of solving only for certain properties of the ports themselves. This is quite useful when one would like to investigate an existing transmission line geometry, or when one wants to create a new transmission line for which there are no design resources available.
In order to designate an analysis in the design to “Solve Ports Only”, simply double-click your solution, and check the “Solve Ports Only” checkbox in the general tab, then enter the frequency you’d like to solve the wave port mode at:
From this ports-only analysis and the frequency it’s solved at, the solver is able to provide the following quantities about the transmission line that the wave port is a cross-section of:
These can be quite handy for performing all kinds of applications that would otherwise require the time to set up and run a full-wave simulation. One quick calculation would be to using the real component of the complex propagation constant to estimate the loss of a given length of a transmission line. Another could be to use the Z0 of the wave port to optimize for dimensions/material properties of an unconventional transmission line that doesn’t have clean empirical design equations. Both applications are explored in the subsections below.
Application: Using Gamma to Estimate Loss and Phase Shift of a Transmission Line
In the following section, it is demonstrated how one can use the “Gamma” port result property to predict the loss and phase shift of a transmission line with the same cross-section as a solved port. This becomes practical when you’d like to find a good approximation of the response of a line that is too long to simulate, but for which you’d like to use the real material properties to account for loss.
To demonstrate one such scenario, a 5 cm long length of Z0 = 50-ohm, the coaxial transmission line was drawn in HFSS, with the center and outer conductors being made of stainless steel, and the dielectric being made of Teflon. Stainless steel and Teflon were chosen as they should provide for conductor and dielectric losses, the stock bulk conductivity being 1.1 MS/m and the TanD of Teflon being 0.001. This should make for a bit of insertion loss, for which a lossless transmission line would be a poor approximation. The length of stainless-steel coax is excited by two-wave ports on either end, with the integration lines as shown:
Two simulations were performed on this geometry, a “Solve Ports Only” setup as described at the beginning of this section, and a normal HFSS simulation that directly finds the S-Parameters of the transmission line. As also stated in Table 1 under the top heading of this section, gamma is the complex propagation constant. Gamma describes the S-parameters of an ideal lossy transmission line matched to the reference impedance as follows:
This is a more general version of a more familiar formulation of the S parameters of a lossless transmission line, which just describes the phase shift of a transmission line at a given frequency as it varies with the length of the transmission line:
Since gamma has an extra real part α being exponentiated, this acts as an attenuation per unit length which additionally captures the conductor and dielectric losses of the transmission line along with the phase shift.
Similar to how one might understand β being a function of frequency (β = 2 π/ λ), α is also a function of frequency. The real and imaginary components of the complex propagation constant solved for Port 1 as part of the “Solve Ports Only” simulation is plotted below (remember that β = im(γ), and α = re(γ)). Both simulation setups were performed over a frequency range of 0.01 – 10 GHz.
After the “Solve Ports Only” simulation setup was solved, a normal modal HFSS simulation setup was run, which found the FEM-simulated S-parameters of the transmission line. The 3D solved vs.. predicted insertion loss (S21) is compared in the plot below. The red lines corresponding to the left Y-axis show the magnitude of the insertion loss, while the blue lines corresponding with the right Y-axis show the phase in degrees. Symbols show the 3D solved S-parameters, while the solid lines show the response predicted from the “Solve Ports Only” 2D sim finding gamma, which is then plugged into the ideal lossy transmission line S-parameter equation, S21 = e-γl.
As you can see, the 3D-solved insertion loss matches almost exactly the predicted response only using the 2D solved gamma, with only some minor fluctuations in magnitude near the top of the frequency range not being captured. Though for this simulation, the 3D-solved S-parameters took less than a minute to simulate, one could imagine the utility of the 2D solved gamma approach to finding a response of a very long transmission line that takes into account the real lossy material parameters, while not requiring the time and compute resources to solve such a transmission line.
Application: Using “Solve Ports Only” to find Z0 = 100 ohms for an Unconventional T-Line
Another very useful result made available by a “Solve Ports Only” simulation is the Z0 of the transmission line which the wave port comprises a cross-section.
Consider the following atypical transmission line, an air-filled “coaxial” line, with a square rather than a circular outer conductor. As one might imagine, there are no easily available design equations for this geometry to produce a desired reference impedance. The geometry has been parametrized with the variables shown in the picture below, and the variables have started with arbitrary values shown in the table below the picture:
At either side of the transmission line, wave ports were added with integration lines as shown in the picture below:
Adding a “Solve Ports Only” solution setup at 10 GHz, the port E field distribution was plotted. (as shown in the “Common Problems and Troubleshooting” > “Field Display” section earlier in this article). The field distribution seemed to be physical, looking to be similar to the mode of a normal circular coaxial cable but with the E fields veering off near the inside of the outer conductor in order to make the tangential component of the E-field at the surface of a conductor zero.
After an initial simulation, an optimization was set up, in which a goal was set to tune the characteristic impedance of the transmission line at 10 GHz to be 100 ohms. The optimizer was allowed to change the radius of the inner conductor, and the height and width of the outer conductor. The optimization setup is outside of the scope of this document, but the results of the optimization are compiled in the table and graphs below:
In summary, the optimization quickly converged and produced a geometry with the desired reference impedance. This approach could be extended to many different types of transmission lines, from unconventional microstrip and stripline transmission lines to waveguides, and countless other transmission lines of arbitrary geometry not mentioned.