Over the past eight years of my career, I’ve been fortunate enough to work on several different cutting-edge projects here at Plextek including a number of mm-wave designs. These have included both research programmes and product development, covering radar sensors, communication systems, and medical applications. This has given me a great depth of experience into different approaches to meet our customer’s specific requirements. Over this time, I have found that one of the most challenging aspects of working in this frequency range is the integration of mm-wave front-end electronics with an optimised antenna solution, to provide an efficient, and controlled radiation pattern.

Whilst there are many different approaches to a mm-wave antenna design, in my experience, by far the most popular choice for mm-wave applications is some variation of the patch antenna. To me, this is because they are both cheap to produce and surprisingly efficient, albeit over a narrow bandwidth. Whilst there are techniques to improve the bandwidth of the patch antenna, I have personally found that they do not necessarily offer the best performance when compared to alternative approaches.

That said, like all decisions in engineering, the best approach depends on what your priorities are and what you are trying to achieve. For highly cost-sensitive applications, it is difficult to justify a more complex and ultimately more expensive approach if an edge-fed patch antenna will meet your requirements.

A rectangular edge-fed patch antenna, as its name implies, is excited with a signal launched from a microstrip transmission line to one edge of the rectangular element (the element being the specific part of the antenna where radiation occurs). The rectangular element, like the microstrip transmission line, is also over a large ground plane with often a thin microwave dielectric between (a thin microwave dielectric is usually required for the mm-wave electronics to provide a low inductance path to ground). This causes the element to resonate at a specific frequency due to its length, dielectric thickness, and dielectric permittivity. Whilst this thin dielectric enables efficient radiation, this only occurs over a narrow range of frequencies resulting in the element bandwidth being a particularly narrow band. To improve the element match, an inset feed can also be used as shown in the pair of patch elements below.

To increase the gain of the antenna, multiple elements can then be added by feeding the next element in series with a short length of microstrip from the far edge of the first, and so on. This works well providing all the elements radiate in phase with respect to each other. To ensure all the elements radiate in phase, a specific length of microstrip should be used for the required frequency. This causes the series-fed array to also provide a narrow array bandwidth as it relies on a specific wavelength which, over frequency, will inevitably change.

To improve the element bandwidth, a stacked patch is a popular choice. Here, stacking a second patch above the first effectively increases the depth of microwave laminate between the top patch and the ground plane, trading a small amount of efficiency for bandwidth. This approach does, however, have a financial cost associated with it, as multiple layers of expensive microwave dielectric are now required.

To improve the array bandwidth, a feed network which provides an equal electrical line length to each element is often required. This can be achieved through a corporate feed network where the transmission line is split as it feeds all the elements with the same length from the source. An example of this is shown in the 4-element via-fed stacked patch array below.

It is often desirable to increase the gain of an antenna generally to improve the system range, whether it be for a communication or radar system. However, a high-gain antenna is not always advantageous as it essentially directs all the energy in a specific direction, but less in others. To use an analogy it’s like squeezing a balloon, a high-gain antenna can be squeezed such that there is one main lobe where most of the signal (or air in the case of the balloon) is going, but the amount of signal in other directions is reduced. This is fine if this is the intended direction, and for fixed point-to-point communication links this is usually the case. However, in a dynamic system where the direction of radiation needs to change rapidly, an electronically steerable solution is usually required.

Electronically steerable antennas are usually realised in the form of an array of separate elements, like the examples discussed above, but by adjusting the phase of the signal radiating from each can cause the direction of peak gain to change. The challenge, particularly at mm-wave frequencies, is the ability to provide the required phase to each element whilst managing the interaction between nearby elements. This is where, in my opinion, achieving good performance with patch elements becomes challenging as, without careful design, the tightly spaced elements and their feed network can interact. This causes the relative phase between elements and the current distribution across the array to be affected, resulting in a less-than-ideal radiation pattern.

An alternative to the patch element is a radiating slot. A slot in an infinite conductor is considered as the complement to a dipole in free space. A patch antenna is often modeled as two slots side-by-side radiating in phase; this results in an increased element gain for the patch over a thin slot element, but reduced beamwidth. Whilst this allows a patch array to achieve a higher total gain for an array with the same number of elements when compared to a slot array, the reduced beamwidth means a patch array cannot steer over as wider a scan angle.

Another advantage I see with using slot elements in an array is the ability to feed them from waveguide rather than microstrip, or similar transmission lines. The signal is constrained within the waveguide which helps to lower coupling between adjacent transmission lines used to feed nearby elements. This inherent isolation of the feed network helps to achieve a more controlled radiation pattern over wide bandwidths and scan angles. Naturally, however, there is a higher cost associated with a waveguide fed slot array antenna as it tends to require a more complex design and manufacturing process. An example of this can be seen below where an 8 x 10-element slot array (with an additional single 10-element array to one side) is fed from substrate-integrated-waveguide. This has been designed directly onto a multilayer PCB which can support the associated RF front-end electronics.

Designing the antenna directly onto the same PCB as the electronics are usually the most cost-effective approach. Not least does this prevent the need for an additional separate antenna but simplifies the assembly process. However, achieving efficient, wideband antenna performance on a PCB which also contains mm-wave electronics can be very challenging. This is because the electronics tend to require contrasting PCB requirements to the antenna, particularly with respect to the thickness of the PCB dielectric as discussed earlier. Our conference paper and presentation at EuCAP 2020 aims to address this fundamental issue for use in substrate-integrated-waveguide (SIW) fed antenna arrays. The approach presented demonstrated a significant reduction in the insertion loss of substrate-integrated-waveguide using a novel feed approach for a multilayer PCB design.

If further improvement in antenna efficiency and bandwidth is required, particularly for high-gain antennas, the use of a sectoral horn could be considered. Horn antennas are a popular choice for microwave and mm-wave applications because they offer good, predictable performance and their operation can be accurately calculated through well-documented equations. The challenge with using a horn antenna for mm-wave applications is generally feeding the signal into each sectoral horn element. An example where Plextek has implemented a sectoral horn array for an mm-wave radar sensor can be seen on Texas Instrument’s website. This short-range sensor was optimised to achieve a wide coverage using TI’s IWR6843 radar-on-chip device.

Whilst the sectoral horn antenna approach can offer improved bandwidth and efficiency for high-gain elements, it does come at a cost both to manufacture, as it requires an additional component, as well as its larger size and weight. The example below is of a 64-element sectoral horn array.

There are many other solutions to mm-wave antennas such as parabolic reflectors, as used in the development of the FOD radar, but the examples explored in this blog are probably the most relevant to consumer electronics both for communication or short-range radar systems.

I hope that this is both interesting and informative, as to me, the antenna choice is vitally important to any mm-wave design as it dictates a considerable amount of the system around it.