Spiral Antenna – Part 2 – Modelling and Simulation with OpenEMS

Please not that at the time of writing this antenna has not been physically built or tested. Simulation and reality can sometimes disagree. I am new to OpenEMS and may easily have made an error.

Electromagnetic simulation software can be very expensive. While some software is cheaper, it often comes with either technical or artificial limitations. There are however a couple of free alternatives.

NEC2 – Originally written by Lawrence Livermore National Labs, NEC2 was open sourced many years ago. Since then a couple of users have created good free or low cost front ends for it. 4NEC2 is my personal favourite. The best feature is it’s speed, especially if you can find a copy of NEC2/MP that makes good use of multiple core moderns processors. The big limitation is with dielectrics. Everything in NEC2 is either wire, free-space or earth.

OpenEMS – A Finite Difference Time Domain (FDTD) tool that uses MATLAB or the excellent free Octave as a front end. It was created by Thortsten Liebig at the General and Theoretical Engineering University in Germany. As suggested in FDTD, OpenEMS uses a time-domain approximation of Maxwells equations to calculate the results. The big advantage of OpenEMS over NEC2 is that it can handle bulk dielectric materials with different EM properties, I will need this to simulate an antenna on FR4 with an Er~=4. While it may not seem it at first, being able to generate geometry mathematically in is another advantage. This is especially true for a spiral antenna. Hence this is the package I decided to learn and use for this project.

Dont Panic!!!

So, lets get this out of the way first. OpenEMS looks pretty intimidating to a new user. It’s really not that bad though. Honestly.

Installation is easy. This is not a tutorial, so do go to OpenEMS and follow the excellent guides there.

On Windows], Octave has a really good installer wizard, just run it, it works fine. Don’t worry about not having MATLAB, for the purpose of using OpenEMS Octave works just fine. OpenEMS is a simple downloaded zip file that you extract to your c: drive(or anywhere else if you want to be different). You only need to add one line to the Octave startup script and type setup at the Octave command line to compile the scripts and you are ready. Paraview is optional. The majority of the plots you will need are generated directly from Octave and may not need to install it at all.

Once installed there are a series of examples and tutorial files included in the OpenEMS/MATLAB/tutorials folder. They are an excellent start and the tutorials on the OpenEMS website will guide you through them really well.

I would guess that the biggest hurdle for many is using Octave/MATLAB. If you haven’t done this before then it is probably worth doing a few tutorials on this. For 99% of tasks Octave works exactly the same as MATLAB and most code will run in either. You will need to understand how Octave generates, manipulates and references vectors.

Give it a try.

My Model and Simulation Results

I have chosen not to attempt to simulate the feed strucutre. The 120ohm tracks(impedance chosen from these simulations) are very fine and would require and exceedingly fine mesh to simulate. Instead I only simulate the arms of the antenna with a lumped feed at the centre that magically creates a input port without and cables or tracks.

Meshing Meshing Meshing!!!

The success of your simulation is really dependent on how your model is meshed. And unlike many of the commercial solvers, it’s up to you to generate it.

The mesh is how OpenEMS breaks down you model into individual unit cells.

Here is mine:-


OpenEMS Mesh

As can be seen I have used a cylindrical mesh. This really seems to make sense for this antenna type. The mesh is naturally finer in the centre where the tracks are thinner and I expect the higher frequencies to be radiated.[edit-i have also tried this with cartesian mesh and it works fine too]

Unlike the simper tutorial antennas this design takes some time to run. Up to an hour on a quad core i7. Really this is down to the required mesh density required for the high frequencies and fine structure of the centre of the antenna and the need to have a large bounding box that needs to be at least 1/2 wavelength from the structure.

Much time later..S11

So here are the simulation results.


Feed Point Impedance

That is a very stable plot of input impedence. The imaginary component stays very close to zero over my 400MHz to 2GHz range and the real impedence is well matched to 120ohms.


Return Loss

With OpenEMS set to use a 120ohms source impedance as derived from the previous plot, I can expect an return loss to look something like this. Anything less than -10dB is really good. These results are amazing!. If I can achieve anything near this in the real world I will be very pleased.

Radiation Pattern and Gain

RHCP Directivity

RHCP Directivity 900MHz

This plot shows the Right-Handed polarisation. The peak gain is heading directly upward with very little back lobe. The LHCP plot is just the mirror image of this with peak gain straight downward. The simulation predicts about 4.7dB of directivity.


OpenEMS is a powerful tool for simulating antenna and other RF circuits. I was able to design and simulate an Equianglular Spiral Antenna that predicts good results. I will shortly be receiving manufactured PCBs that I will conduct initial tests on and see how closely simulation and measurement match.

Posted in Antennas, Uncategorized

Spiral Antenna – Part 1


My original Palm Tree Vivaldi Antenna has proved amazingly popular. Much more so than I originally intended. The design aim for this was primarily for it to be small and portable.

When deciding what to do next I decided to create a survey and see what other people were interested in.


R&D Survey Results

The overwhelming response was that people wanted a bigger version that would cover lower frequencies. I can understand this, there are loads of interesting signals below 800MHz.

Read more ›

Posted in Uncategorized

Reconfigurable Ultra-Wideband Antenna Array Kit – part 2 (Spacing)

Why Spacing Matters

Firstly, the antenna array I am designing is a fixed in-phase array. Hence we assume that the total path length from each antenna to the single input/output connection are the same length. This makes the design and analysis much simpler, especially for an ultra-wide-band design.

In my design signals that arrive at right angle to the antenna will add up co-herently (in-phase)

spacing boresight

Signals arriving from boresight add up in phase

Read more ›

Posted in Uncategorized

Ultra-Wideband Antenna Lab Measurements


Antenna being tested in anechoic chamber

My Vivaldi Antenna was recently tested by Antenna Test Lab who offer an excellent professional antenna testing facility. Antenna Test Lab are able to provide customers with full 2D/3D antenna measurement using their anechoic antenna testing chamber. Read more ›

Tagged with: , , ,
Posted in Antennas

UWB Antenna R&D Survey

I am conducting a quick survey to help discover what developments people would most like to take my UWB antenna forward. I would be really grateful if you could take a few minutes to have your say.

Posted in Uncategorized

Reconfigurable Ultra-Wideband Antenna Array Kit – part 1

My next project after successfully building my Ultra-Wideband Vivaldi antenna is to use it to create a UWB antenna array. I will assume that most readers are at least slightly familiar with the concept of an array, I will give a quick introduction, but for those that want some background reading here is a good site to look at.

In essence we create when we sum the output of multiple antennas to increase the gain and directivity. Of key importance when we do this is to control the phase of the incoming signal that we are adding. I will be building one of the simplest forms of array, where all antennas are summed with an equal phase offset. This will have the effect of increasing the gain on the antenna boresight (and reducing the gain at other angles (I will get to grating lobes in another post)).

array basic

All antennas fed with equal phase offset.

The simplest way to do this is with a ‘corporate’ feed that consists solely of 2 way Read more ›

Tagged with: , , ,
Posted in Uncategorized

Making an Ultra-Wideband Antenna – Part 3 (performance)

Update – this antenna was professionally measured at AntennaTetLab results here

Follow this link to see more info on the original creation of the Palm Tree Vivaldi Antenna developed by Dr. Alexandre at the Laboratory Maxwell in Brazil.

Final Antenna (The Palm Tree Vivaldi Antenna)

Final Antenna

Here is the final antenna. I am still waiting on some SMA connectors for the 1mm PCB. For now I have bodged a 1.6mm SMA connector onto one of the boards for testing.

Antenna with temporary 1.6mm SMA

The antenna measures about 90mm x150mm x 1mm so is extremely portable.

S11/Return Loss/SWR

Return loss or SWR(Standing Wave Ratio) is a measure of how much of the power that you send to an antenna is reflected back to the input port. This may also be called anĀ S11 measurement. Return loss and SWR are basically 2 ways of representing the same thing and it is fairly easy to convert between measurements if required. SWR is given as a ratio while Return Loss is normally(but not universally) quoted in decibels.

Read more ›

Tagged with: , , ,
Posted in Antennas