W-band planar antennas offer a compact alternative to conventional horn-based structures for systems operating from 75 to 110 GHz. At these frequencies, however, short wavelengths make geometry, materials, feeding and measurement especially consequential. A research programme at Maynooth University explored these challenges through simulation, fabrication and testing of planar antennas and small dielectric lenses intended for focal-plane pixels.
Although the work was motivated by astronomical instruments, its engineering process is relevant to technical teams developing high-frequency components: model the electromagnetic behaviour, manufacture prototypes, measure them under controlled conditions, and examine how individual elements interact when assembled into an array.
Why develop W-band planar antennas?
Planar antennas are formed on dielectric materials such as printed circuit boards. Patch and slot antennas are examples. Their compact form makes them candidates for arrays in which each antenna is paired with a small lens, or lenslet. The challenge is that structures and feed connections become difficult to design and manufacture as wavelength decreases.
The project focused on W-band detector pixels as an alternative geometry to traditional horn antennas. Its broader scientific context was measurement of the cosmic microwave background (CMB), including its weak polarization signal. In an array intended to detect such signals, unwanted coupling between neighbouring elements matters: crosstalk can contribute an additional signal that must be understood and accounted for.
Simulation and prototype development
Several antenna geometries were modelled and optimised in CST Studio Suite to predict performance around 100 GHz. The work concentrated on patch antennas, including circular and rectangular designs, and investigated both back-fed and side-fed configurations. Alternative designs, including slot antennas, were also considered.
Simulation provided a way to examine electromagnetic behaviour before committing to hardware. But predicted performance alone was not enough. Prototypes were manufactured in-house and measured using a vector network analyser (VNA) setup operating in W band. Comparing measurements with simulations helped assess the designs and exposed practical considerations, including the influence of dielectric thickness and the challenge of coupling a feed efficiently to the test source.
Feed configuration and return loss
The feed determines how energy is delivered to the antenna, making it a central part of the design rather than a minor connection detail. The research compared back-fed and side-fed patch approaches. In the side-fed work, an inset structure on the patch was examined as a way to reduce return losses. This design process also considered how the antenna could be connected and measured using the available test arrangement.
For engineering teams, the practical takeaway is to treat the antenna, feed, substrate and measurement interface as a connected system. A simulated radiator may not behave as intended if its feed or test transition is not represented adequately in the design and evaluation process.
Dielectric lenslets and array crosstalk
Small lenses can be placed over planar antennas to form lenslet pixels. The project examined lenslets made from high-density polyethylene (HDPE), including hemispherical, cylindrical and plano-convex shapes. A key design concern was whether the lenses could support the intended array arrangement while limiting coupling between neighbouring pixels.
A particular focus was a truncated plano-convex lens. Truncating, or cleaving, its sides changes the lens geometry and was investigated as a means of reducing crosstalk. Plano-convex lenses with cleaved sides were manufactured and tested, and crosstalk between neighbouring pixels was considered through both modelling and measurement.
This illustrates why evaluating one antenna in isolation may be insufficient for array development. The behaviour of adjacent elements, their lenses and their spacing can influence the system-level result. Array-level analysis helps identify interactions that a single-element model would not capture.
Measurement beyond the planar antenna
The work also applied Gaussian Beam Mode Analysis (GBMA) to a multi-moded pyramidal horn operating at 2.7–5 THz. In the experimental configuration, a cryostat window truncated the propagating beam. The analysis was extended to examine that truncation and verify whether the beam could pass through the window without loss of signal.
This separate study reinforces a broader point in high-frequency engineering: boundaries and apertures in the measurement or operating environment can affect electromagnetic fields. Modelling those constraints can help engineers interpret far-field patterns and assess whether the test configuration is affecting the observed result.
Practical implications for high-frequency design
The W-band development process demonstrates a useful sequence for advanced antenna work:
- Model the complete structure: include the antenna geometry, substrate and feed arrangement when evaluating performance.
- Build and measure prototypes: use VNA measurements to compare physical behaviour with simulation predictions.
- Assess the array, not just the element: investigate crosstalk and other interactions between neighbouring pixels.
- Account for test conditions: consider how coupling structures, apertures and windows may influence measurements.
W-band planar antennas can support compact focal-plane designs, but successful development depends on integrating electromagnetic modelling with practical fabrication and measurement. For engineers and technical decision-makers, the central lesson is to validate the full configuration—including feeds, dielectric lenslets and neighbouring elements—rather than relying on an isolated antenna design. That system-level approach provides a clearer basis for evaluating performance and managing crosstalk in high-frequency arrays.