Designing a PCB Antenna

For the device I’m working on, the Particle Electron that operates it may frequently be changed/replaced - as such, I don’t want to always have to change the Particle 3G antenna with it. Furthermore, the device will experience a lot of movement so I need to find a way to properly secure my antenna as well.

I’m looking at the feasibility of simply integrating the antenna into my PCB, and run a connector to it. I have no experience with antenna design though. Is it just a matter of reproducing the copper antenna traces onto my PCB? Is it really that simple?

I don’t have design files or footprints of this particular antenna - my thought was to very accurately measure the traces (length - width - orientation) in real life, and re-producing it and placing it on a PCB. What would the margin of error be? If I’m off by even a few % for a dimension, would that screw up the entire antenna?

I would be hesitant to design an antenna myself even if I’m just reproducing a reference design. When you get into RF design, I believe the PCB traces form a waveguide with the ground plane instead of just conducting electricity such as a GPIO. There are also a bunch of inductors that come into play to tune the antenna. And you have to put your product through FCC testing for certification. I don’t have a solution for you, just realize it’s pretty complicated.


Yeah - after reading some guides I was a bit put off with the tuning especially, and realize that it’s going to become very costly, time-intensive and not going to be fail-proof to produce hundreds of PCB antennas myself.

I think I will just put clips on my PCB that can hold the Taoglas antenna in place. Will have the same effect in practice, and is much simpler to do.

Are you saying that the Taoglas antenna will “travel” with its Electron when the latter is swapped out? I ask because the u.FL connector has a pretty low # of connects/disconnects spec’d - something like 40 or so.


The Taoglas will be soldered onto the uFL connector. Not using any solder for this connection will be impossible - have tested it, and due to the motion forces of the device it will disconnect on its own regularly.

So since it’s soldered directly onto it, any replacement/removal of the Electron means the Taoglas has to travel with it.

My plan now is to use some kind of clip to hold it in place (hard to find a component designed to hold something of this shape in place), or to just use 3M tape and re-apply it each time the Taoglas is put back.

Have you considered SMA connectors/antennas? They should have more cycles, and allow you to switch electrons without removing the antenna.

But the connection that is made at the uFL connector on the Electron still needs to be soldered IMO. The uFL connection by default is too flimsy and it doesn’t take much to disconnect it. So if this connection is soldered, IMO it doesn’t make much difference what antenna is used at that point, as the Taoglas PC104 already is pretty good. The Electron (important) also needs to be easily removable by other people - just taking it out or popping it into the PCB, with no removal of solder etc. That’s why something like a uFL to SMA that’s soldered onto the PCB won’t work. PCBs won’t swap, but Electrons will.

For reference, my devices will be out of reach for me (or anyone else) for months at a time, and they experience some very significant motion forces on a daily basis. Even a failure rate of 5% over the course of 6 months would be unacceptable.

There are plenty of options for PCB mount antennas both from Taoglas and other vendors (eg Molex, etc). You solder these down to your PCB, and run a 50 ohm trace from a u.FL to them, then use a u.FL to u.FL jumper cable to connect the RF.

As this cable is very small and light, it’s unlikely to be badly affected by motion.

Note that you should very carefully read the antenna application note and take all the recommendations. Often there are minimum groundplane sizes required for proper operation.


A fairly simple patch antenna is quite easy to calculate.

The dielectric constant for FR4 is around 4.4, depends on your frequency, thickness etc etc.
Ideally you would use a spectrum analyzer to verify the calculations after producing it but not everyone has such an expensive device.

Maybe something like this would also help you out:

Or a simple dipole (not etched on the pcb) :