Designing a Streamlined 8 x 8 Element Waveguide Array for mmWave Usage

Author : Marcus Walden, David Eliston and James Henderson, Plextek

04 October 2023

Figure 1: Model of the 8 x 8 element waveguide array and feed network - a) XZ plane (transparent) b) YZ plane (transparent) c) XY plane (transparent) d) Isometric view (solid)
Figure 1: Model of the 8 x 8 element waveguide array and feed network - a) XZ plane (transparent) b) YZ plane (transparent) c) XY plane (transparent) d) Isometric view (solid)

Waveguide arrays provide efficient, directional antennas for RF signals. With new prospective applications opening up all the time, waveguides addressing mmWave frequencies (i.e. those above 26GHz) are becoming increasingly important. 

Designs can often be conveniently realised as a stack of multiple machined layers, such as a Ku-band corporate-fed waveguide array. In the mmWave band, designs often use 2 x 2 element subarrays to achieve low profiles (and thereby save valuable space). Here the design, simulation and realisation of an 8 x 8 element mmWave waveguide array with a complex internal corporate-feed network is discussed. 

The challenge faced here was that the required array would form part of a larger mechanical structure. This precluded the use of machined layers, meaning that aluminium selective laser sintering (SLS) based additive manufacturing techniques needed to be relied upon. 

Design parameters 
The required antenna would operate at 78.5GHz, with an 8 x 8 array of uniformly weighted open-waveguide apertures that had 2.5mm x 1.5mm dimensions. It would have a 3.3mm pitch and the waveguide cut-off frequency would be 59.96GHz. Grating lobes were expected at large angles from the array boresight, owing to the greater than half-wavelength element spacing - though these were reduced by the element pattern. 

Figure 2: Simulated (red trace) and measured (black trace with markers) reflection coefficient of 8 x 8 element waveguide array
Figure 2: Simulated (red trace) and measured (black trace with markers) reflection coefficient of 8 x 8 element waveguide array

SLS was used to 3D print the waveguide array in aluminium. It was manufactured in layers from the waveguide apertures up to the WR12 input port. SLS manufacturing rules include a minimum allowable wall thickness of about 0.8mm, which would have a bearing on the aperture size and element pitch. Overhanging horizontal sections were not feasible, but chamfers up to 45° angles would be possible, as well as cylindrical or spherical domes with maximum radii of 2.5mm. 

The RMS surface roughness was kept within 10µm. To comply with 3D-printing process rules, sections of the rectangular waveguide were rotated by 45°. A combination of E-plane and H-plane T junctions were used to maintain phase alignment across the array. To minimise the overall antenna volume, the corporate-feed network had to be folded back on itself. The overall length of the corporate-feed network from the WR12 input port to the radiating aperture came to 41mm. 
 
Simulation 
The SEMCAD X 3D simulation platform was used to produce a simulated rendering of the waveguide array in free space. Figure 1 describes the model from various different perspectives. For modelling convenience, a cuboid structure with dimensions of 26.4mm x 26.4mm x 13.0mm was employed. An effective conductivity of 3.5 x 105S/m was used to account for the metal surface roughness, based on simulation and measurement of a 50mm straight waveguide fabricated via the same process. Surface roughness also affects the transmission line phase constant, not just the attenuation constant, but was deemed to be negligible enough to ignore in this particular analysis work. 

Figure 3: . Simulated 3D radiation pattern at 78.5GHz (50dB scale)
Figure 3: . Simulated 3D radiation pattern at 78.5GHz (50dB scale)

The simulated reflection coefficient, as illustrated in Figure 2, is less than -10dB over the range 76.25GHz to 80.85GHz. Owing to acute time constraints, parametric optimisation was not used, but would be expected to yield wider impedance bandwidths. 

The simulated 3D radiation pattern at 78.5GHz is shown in Figure 3. The half-power beamwidth was about 7.5°. Peak gain was +25.4dBi versus a directivity of +27.4dBi. The 2.0dB loss (63% efficiency) proved consistent with the attenuation constant for the model’s effective conductivity and corporate-feed length. The peak close-in sidelobe level was about -13dB, consistent with uniform aperture weighting. Grating lobes appeared at ±70° from boresight at -19dB and -26dB in the XZ and YZ planes respectively.

Measurement data
Photographs of the fabricated waveguide array from the front and rear are shown in Figure 4. While the simulation model has a cuboid structure, the fabricated design would have excess metal removed at the rear in order to reduce weight. Figure 2 also shows the measured reflection coefficient between 75GHz and 84GHz. Return loss was generally better than 6dB (VSWR =3) over approximately 77.2GHz to 82.8 GHz frequencies. This upward shift in frequency for low VSWR may indicate the 3D-printed waveguide channels were narrower than actually designed. Peak measured boresight gain at 78.5GHz came out at +25.8dBi, consistent with the simulated peak gain (+25.4dBi). 

Figure 4: Fabricated 8 x 8 element waveguide array - left) Front view showing the 2.5 x 1.5 mm waveguide apertures, right) Rear view showing the application-specific WR12 flange
Figure 4: Fabricated 8 x 8 element waveguide array - left) Front view showing the 2.5 x 1.5 mm waveguide apertures, right) Rear view showing the application-specific WR12 flange

Next stages
Post-processing of the 3D-printed waveguide array is expected to reduce the surface roughness and improve antenna efficiency, through use of a sophisticated micromachined process. Use of electroforming instead of SLS may help to lower surface roughness. Corporate-feed losses could be reduced by extending WR12 dimensions further into the design from the input port. Parametric optimisation during the design phase would yield a wider impedance bandwidth, thereby making the design more tolerant to potential manufacturing variations. Finally, amplitude weighting across the array aperture would reduce sidelobe levels.


Contact Details and Archive...

Print this page | E-mail this page






This website uses cookies primarily for visitor analytics. Certain pages will ask you to fill in contact details to receive additional information. On these pages you have the option of having the site log your details for future visits. Indicating you want the site to remember your details will place a cookie on your device. To view our full cookie policy, please click here. You can also view it at any time by going to our Contact Us page.