How do I calculate a PCB antenna trace width?

I'm totally stuck here trying to find a single source that describes how to determine a PCB antenna trace width. I understand how to determine the length based on the targeted wavelength, but the width is lost on me.

To preface, I'm very new to this and am trying to design a half-wave curved dipole PCB trace antenna. The guide I'm using as a reference is from Texas Instruments. Click design guide at this link.

In this TI DN041 guide it says the following in section 4 Designing a Half Wae Dipole Antenna on a PCB:

The width of the trace does have an impact on the antenna performance, particularly the bandwidth. In this design note we will stick to a width of 300 um (12 mils), and refer the reader to more thorough papers on antenna theory if they wish to probe deeper into this particular property of the dipole.

So I went on a hunt to try to find these "thorough papers on antenna theory" and came up short. I did, however, find a few resources that pointed me to a broken link for an online "microstrip" calculator. I did some digging and found a new link to it here (click "Microstrip Trace").

Now I'm not sure that a PCB trace antenna counts as a microstrip trace because PCB trace antennas are not supposed to have anything on other layers above or below it so that the antenna performs well. Yet a microstrip has a groundplane underneath it as illustrated below:

Then I did some digging and found that this is used to calculate transmission lines and that "PCB antennas are not transmission line."

Someone on this very site tried to use this calculator for that purpose and it was mentioned that their use of the microstrip calculator was incorrect due to the ground plane. So using a microstrip calculator seems to be incorrect (even though some people used it as such). So then how do I calculate it?

So to summarize into a single question, how do I determine the width of a PCB trace antenna?

• Notice that at your last link ("someone on this very site"), there is no ground plane beneath the trace. Therefore this is not a microstrip structure. If you really want to optimize your antenna trace width I suspect you need to do a numerical analysis with an EM simulator like CST or HFSS. Feb 28, 2020 at 18:36
• And the answer to that question did include "you don't need a 50 ohm transmission line to make such an antenna in the first place. You don't want that ground underneath, because you want radiation of the monopole" so it's not true nobody mentioned it was incorrect. Feb 28, 2020 at 18:37
• @ThePhoton yes, hence the reason why I reference it. People appear to be using the microstrip calculator for antenna traces, yet antenna traces have no ground plane below and microstrips do. Feb 28, 2020 at 18:43
• @ThePhoton I rephrased that line. I was more referencing how no one mentioned that their use of the microstrip calculator was incorrect for determineing the PCB trace width. Feb 28, 2020 at 18:44
• @Davbog fix your question as per what the photon says. That answer clearly states you should not use a micro strip calculator when trying to design an antenna. Feb 28, 2020 at 19:27

The TI reference design should include a layout. My advice is to copy that layout exactly for your first iteration and test it. Pay special attention to preserving the spacing between the antenna poles and the reference planes alongside (but not beneath!) them.

If it works well enough, go with it (it likely will.) If not, hope is not lost. You can adjust things a bit (spacing and width) to get things right, as the TI appnote said. To do this in a structured way you will need to acquire a antenna design tool and develop the skillset to run it, or talk to the FAE at the chip vendor.

Worst case? Hire a consultant. Which might be your best case if you've never done this before and you have bosses breathing down your neck for schedule.

• I'll agree... people make PhD & master degree thesis about how to design an microstip antenna for a particular type of communication. I doubt that a neophyte could manage to build an antenna by himself in a reasonable amount of time. Jul 1, 2020 at 19:25

Cypress has a great tutorial.

It will take some time to study it. Its a great one for begginers, but you will need something like a minivna to get your hands on it. Theory is different from real life.

For example after designing an antenna for 2.4Ghz, my minivna shows me this when i use it to see which frequency is my antenna build for:

(The lower the graph, the better the signal)

You see as points 1 and 2, is where my antenna best functions. Around 2.4Ghz, but not exactly 2.4Ghz.

Also At point 3 it has a weaker signal.

My point is, it will take time to get there. Study as many sources as you can, gather information and keep testing.