Dear George Dishman .

This is worth considering: the disc on the left is a circular patch antenna with four feed points, 90o hybrids and 180o hybrids are small off the shelf items that can also be implemented in printed cct. format. The hybrid on far right has two inputs, one is terminated the other active simply by reversing these you switch from LHCP to RHCP. Somewhat surprisingly patch antennas can have gains as high as 9dBi There fore a 3*3 array of these will give you 18.5dBi of gain. Off axis it may not be quite circular (axial ratio). Quadrifilar log periodic spiral can be used for more bandwidth and quadrifilar helicals for more gain, all using the aforementioned feed. The helical's and spirals give more options re: axial ratio. Google or Wiki: "Butler Matrix" for novel beam steering technique using 90o hybrids. You will like it no need for phase shifters you have fixed switchable multiple independent fixed beams beams. all of the phase shifting is fixed and integrated into the feed their is sharing of the phase shift elements (hybrids and delay lines) It is brilliant!

https://www.microwaves101.com/encyclopedias/circularly-polarized-antenna-feed

Barry

Dear George Dishman

To answer all your concerns. We should open a table conversation containing 4 to 5 PhD holders and try to simplify it one by one. However, I will try to answer my field of interest, "Planar antennas with Reflectors", if you want to add a reflector behind your circular polarised antenna, I think Frequency Selective Surface (FSS) as a reflector will be a good choice. Utilising FSS can get the advantage of its filter behaviour to create a band-stop filter that can help you prevent interference and prevent radiation in the opposite direction. Besides, using FSS as a reflector leads to obtaining a directional radiation pattern. The antenna gain will be at its maximum due to the reflected waves ongoing from the antenna and the Reflector. It will create a constructive interference, which leads to high directivity and gain.

Thank you both for your helpful answers. Barry, your suggestion of a patch antenna and the link showing how it is fed with passive splitters is interesting but has the problem of being a narrow bandwidth solution. Digging around, I've seen typical figures of 3% of the centre frequency.

For the quadrifilar design, if the wavelength is comparable to the circumference of the helix then it gives circular polarisation in the axial direction, "end launch", which would be ideal although not a planar implementation. The problem again would be that over a wide bandwidth, when the wavelength is larger than the helix, it switches to isotropic lateral radiation with linear polarisation.

The Butler Matrix would be a good way to generate the feeds but it would need to be 2D like figure 4 in this document:

https://www.semanticscholar.org/paper/Research-of-multi-beams-antenna-array-using-butler-Zheng-Hua/4c4f729d5c388bee94a27bffb6d5c097360b3954/figure/3

However my array would be a large "honeycomb" of hexagonal elements so the details would need to be adjusted.

I instead think that a design based on this would be a better starting point:

https://hexandflex.com/2018/09/29/spiral-antenna-part-1/

Barry, you might be interested in the author's modelling, he used a tool that can take a MATLAB physical design and also handles materials with varied properties, not just metals. That's covered in this page:

https://hexandflex.com/2018/10/05/spiral-antenna-part-2-modelling-and-simulation-with-openems/

Ahmed, you are on exactly the track I was thinking about. However, I need the solution to be broadband. The rear reflector would work but the distance needs to relate to the wavelength. The signal fed to the antenna is mostly radiated at a point around the spiral that also relates to the wavelength so my idea is that the reflector distance behind the surface tracks should increase as the track radius increases, resulting in a helical strip that is wound round a cone.

Do you think that would be effective?

Having seen nothing on my idea for months, I've just found a paper that describes exactly what I've been thinking about. They say there's "nothing new under the Sun" and that applies here.

http://old.ursi.org/proceedings/procGA02/papers/p1274.pdf

A lot more detail is available here:

https://apps.dtic.mil/sti/pdfs/ADA426107.pdf

Dear George Dishman "Barry, you might be interested in the author's modelling, he used a tool that can take a MATLAB physical design and also handles materials with varied properties, not just metals. That's covered in this page:"

Thanks, I would be interested in "OpenEMS".

Patch antennas are generally used on copper backed dielectric low loss pcb material. This means the radiating patch is closely coupled with its image in the ground plane. This is why the bandwidth is so low. By eliminating the dielectric board material and increasing the patch height above the ground plane bandwidths as high as 30% are achievable gain is reduced but beam width increases approaching hemispherical element patterns from a planar element.

At 160gHz patch height would be 3 or 4 mm, patches can also be mounted on grounded posts in place of pcb material, providing the post is in the centre of the patch at a zero voltage node. this also acts as lightening protection and low frequency suppression filter in recieve mode Hybrids used in Butler matrices can also be broad band.

Barry

George, I worked on a somewhat similar problem some time ago. This conference paper Conference Paper Multispectral very wide-view sensing concept