How to Design a PCB Trace Antenna with Limited Datasheet Info?

I’m working on placing a PCB trace antenna, but the only information available in the datasheet is the antenna type and gain — no physical dimensions or layout details were provided.

After some exploration, I found that manufacturers often provide a .dxf file for the antenna layout, which can be imported into the PCB project. I tried that, but it didn’t go well. I’m also confused:

  • Does simply importing a .dxf file ensure optimal performance?
  • Do we need to consider dielectric thickness, substrate properties, etc.?

In my case, the datasheet doesn’t mention the substrate material, so I’m unsure how to proceed.

Is it possible for designers to create a custom trace antenna (like a coplanar waveguide) based on the design needs?

From what I understand, EM waves resonate with the trace length — typically around λ/4. So, in theory, can we calculate and design a PCB antenna ourselves?

Also, is there any calculator or tool available to assist with designing PCB trace antennas like CPWG or microstrip lines?

Any guidance would be truly appreciated!

2 Likes

It looks so simple when you see a PCB antenna on a board, but it is not simple at all when you have to make it part of your board. For things like this you really need to talk to the company whose part you are designing in.

Importing a .DXF of the antenna is not a key to instant success. You do need to know the dielectric thickness. You need to know where the associated GND layer is (more on that layer later). You need to know where the planes on your board need to stop. And more.

It is very likely, though not certain, but the assumption is that the substrate materials will be standard FR4 (nothing fancy like Megtron 6 or Rogers 4350 etc.) for reasons of cost and simplicity.

A coplanar waveguide is not meant to be an antenna but a means of transferring the RF energy between the antenna and the RF chip. The trace becomes the antenna when the GND plane underneath stops.

The length does determine the resonant frequency, but don’t forget that the Dk of the PCB material changes the signal velocity, so you might think that the tuned length is wrong.

We might be able to work out a few things from what we can see in the picture. The trace thickness of the antenna is 0.5mm. We can see that the trace to GND copper on the top layer looks like it is about 1.5mm. So if we know that 0.5mm is a 50 Ohms trace (because it is the same width as the 50 Ohms feed in the image) then the plane underneath must be about 0.25mm away. For the digital design work that might also be needed elsewhere on this board, you would want a clearance of more like 0.05mm to 0.10mm between top layer and the GND underneath. When this figure is used, the coplanar waveguide impedance drops to between 15 Ohms to 27 Ohms respectively. Clearly an impedance mismatch that is so bad is not a good plan.

When you look at your layer stack, you realise that you can repurpose part of a layer that is 0.3mm down into the board to be a local GND if you clear away anything on layers in between. You also workout that this means the PCB trace must now be 0.6mm to get it to look like a 50 Ohms coplanar waveguide, so this is good enough. But when you get to the edge of the region where there is a plane underneath, to the point where there is nothing but the FR4 material present, there you want it to be 0.5mm so you can neck-down the track as you leave the plane boundary and approach the inductor (or whatever). Now you are into the antenna structure, but how thick does the FR4 material need to be here? It could be expected that it is 0.25mm to match the separation of the coplanar waveguide - but that’s not very convenient for the rest of the design. What if you made it 1.6mm. It might work, but does the extra thickness of FR4 detune the structure?

In the end, antenna design is anything but simple when you are trying to integrate it into a larger design. Unless this is your forte, the best (and safest) plan is to go back to the manufacturer and request their assistance. At the very least, make sure you get one of their evaluation boards and measure it all carefully.

One final remark. In the image taken from the data sheet, it is interesting to read “If end customer wants to use other type of antenna, customer need [sic] to take care of certification”. This is true but a surprising remark to make because there is not enough data given to ensure that their customers could “copy” what might be assumed to count as a reference design and expect that they will all behave the same. Normally this is why most RF modules have an antenna socket (say a U.FL) on the board so that a co-ax cable can take the RF to the “standard” antenna outside the box the unit is housed in. Now you can buy an antenna and know that it meets the specifications and so you can easily claim “type certification” from the module manufacturer.

When it comes to certification, it is usually the case that if the customer designed (copied) the antenna into their product, then they are not able to claim type certification and a respectable lab should reject any such claim. This concern remains, whether or not it is a licensed RF band that you are using. There are still rules to obey and organisations to enforce compliance.

While this can’t answer all the concerns, it hopefully gives some pointers as to what to ask/expect from the manufacturer.

2 Likes