Skip to content

A Practical Guide for Antenna Matching Circuits

This post will explain a practical way to create an antenna matching circuit using open source and commercial tools for PCB antennas.

Whiteboarding circuit schematics for engineering design

Making sure an antenna is well matched is good practice when it comes to a maintaining a solid radio link and ensuring a high radiation efficiency for the longest links. Matching an antenna should not be complicated. Let’s start with why antenna matching is important.

Why is antenna matching important?

What is an antenna but a transition from a waveguide to free space. Depending on the antenna, the bandwidth could be quite small or the resonance just outside of the ideal location. Matching networks can help correct these performance issues.

A matching network can also sacrifice antenna quality factor and a bit of efficiency (by adding loss) to increase overall bandwidth.

Feed network block diagram architecture

Lumped element components are idea for lower frequency antennas. It is best not use them over 6 GHz otherwise other factors become more significant. It’s also a good rule of thumb to use 0402 components or smaller whenever possible.

For higher frequency antennas, waveguide is highly recommended to be used to match the antenna. In this case, the matching strategy changes depending on the type of waveguide selected.

Whether you have measured or simulated an antenna, that data can be used to create the matching network. A VNA is used to measure the scattering parameters of the antenna. Since it’s typically a 1 port device, the measurement would be simply the return loss or VSWR. This measurement is generally saved as a touchstone file for simulation, which will be used in our model.

The touchstone file contains both the magnitude and phase information of the antenna return loss. As an important note, the waveguide length between the matching network and the antenna interface is important so be sure to take that into account. If possible, simply place the matching network as close as possible to the antenna.

There’s a couple ways to optimize the performance of a low frequency antenna (ie: less than 6GHz). In this post, the open-source tool Qucs and the commercial electromagnetic solver HFSS is used to chose the lumped element components to match an antenna.

Optimizing Lumped Element Component Values with Qucs

Qucs has a built in impedance matching tool, but it is our suggestion to use the optimizer for this task since the tool is limited. Using an S-parameter system component, some capacitor/inductor elements, and a Power Source a matching network circuit is built. The image shows a PI network using parallel capacitors and a series inductor. A T network could also be used and the lumped elements could be changed. The match will depend on the location of the center frequency on the smith chart.

Qucs feed network using lumped elements to match an antenna

Using either a Pi- or T-network, the entire smith chart is nearly covered so it’s possible just to build a few iterations and use a gradient or min/max optimizer to find the best answer. If no solution is found, a particle swarm or genetic algorithms can be a better way to find the global minimum.

Keep in mind that each element added will add losses, which ultimately lowers the antenna efficiency and gain. Once the network is determined, an optimizer block can then be used to tune the parameters resulting in a matched antenna.

Another strategy would be to use a tuning table. This table would use a selection of existing components that can be used to optimize for discrete solutions at frequencies around the center to realizable lumped element components.

If your device is difficult to tune after assembly such as embedded antennas in some dielectric materials, it’s probably best to create a tuning table and select-on-test as an optimization strategy.

If you’re wondering, select-on-test just means exactly what it implies: manufacture a bunch of tuned variations, and select the one that performs the best.

The lumped element values can be used rounded to the nearest realizable value or converted to your waveguide of choice like microstrip for trace antennas and implemented without the need for components. This would be an advanced topic and would require a more sophisticated solver like HFSS, FEKO, or CST in order to simulate and predict performance.

HFSS makes it really simple and fast to simulate antenna performance using lumped element components. Since gradient or genetic optimization of the lumped element component values is relatively straight forward, here are some simple steps to create a (not so obvious) tuning table for a select-on-test:

  1. Add the matching network pads to your antenna model
  2. Add the linking pads between components and set their boundary condition to Lumped RLC
  3. Create variables for each of the lumped element components added and set the RLC boundary to each variable (such as C1, L1, or C2…)
  4. Create an excel table of existing components and import them into a sweep analysis while being sure to have both ‘Save Fields and Mesh’ and ‘Copy geometrically equivalent meshes’ selected
  5. Run the sweep analysis and select the tuned values for your select-on-test
aem antennas patch antenna on a flex pcb

It’s important to note that if you are matching a measured antenna, any cables that are not apart of the antenna should be de-embedded from the circuit model. Also, any adapters used during calibration or microstrip line around the components should be added to keep the electrical lengths representative and provide a more accurate match.

If you liked this article, follow us on LinkedIn.

We hope you found this article useful. Let us know how we did by getting in contact with us!

Leave a Reply