← All notes
Engineering note

Phased Arrays: From Antenna Design to Antenna Architecture

Technical deep-dive
September 4, 2026

Gain, bandwidth and scan angle are largely solved. What limits phased array programs now is schedule, cost, integration density and thermal headroom. Yifan Wang, Director of Antenna Engineering, on why the answer is architectural: a compact 3D cavity radiator, surface-mount integration and thermal design solved from the first sketch.

Phased array architecture: radiation and RFIC units, RF distribution network and cooling as independent, upgradable blocks. Visualisation.

Wireless connectivity is scaling faster than the hardware carrying it. Edge compute platforms, autonomous sensors, satellite terminals, radar systems and next-generation remote sensing platforms all need to transmit, receive and steer at the same time, usually inside enclosures with very little room to spare. Phased arrays have done that job for decades, in radar, in 5G infrastructure and in aerospace systems. What is under strain now is the way we build them.

Gain, bandwidth and scan angle are largely solved problems. What limits programs today is schedule, cost, integration density and thermal headroom at power levels the industry has not had to manage before. That pushes phased array work away from element-level optimisation and toward system-level architecture, which is where MILLIBEAM started.

Why the traditional architecture doesn't scale

Conventional phased array development couples everything together. Radiating elements, feed networks, RFICs, multilayer PCB stack-ups, mechanical housings and thermal management sit in one interdependent system. That coupling can deliver excellent performance, but every design decision ripples through the others. A new RFIC forces a new feed network. Higher transmit power forces a new mechanical architecture. Better scan performance forces another optimisation pass on the structures between elements.

The model struggles against a market that now spans communication terminals, industrial equipment, automotive platforms, airborne systems and sensing nodes, each with its own mechanical, thermal and cost envelope. The antenna has to stop being a bespoke subsystem engineered from scratch each time and start behaving like a modular platform. Four problems define that transition.

Development velocity. Programs increasingly need to move from concept to prototype in months rather than years, often against tight board area and power budgets. Standardised, validated building blocks (a proven radiator, an independently optimised distribution network, a selectable RFIC platform, a thermal solution sized to the power level) let an engineering team assemble a system instead of re-deriving one. Schedule and design risk both come down.

Packaging. The radiation unit, RF distribution network, RFIC and thermal components do four different jobs and warrant four different optimisation strategies, yet they are routinely forced into a single monolithic structure. Decouple them into interoperable, independently upgradable modules and a customer can swap an RFIC generation, change a cooling approach or add a frequency band without re-engineering the whole array. That scales in a way a fixed, fully integrated design does not.

Thermal management. Emerging applications are pushing required transmit power to roughly 15 times conventional levels. At that density, thermal behaviour cannot be bolted on once the electromagnetic design is finished. Every RFIC's dissipated heat directly constrains RF performance, reliability and long-term stability, so it has to be in the architecture from the first sketch.

Surface wave suppression. Element-to-element coupling and surface wave propagation remain a persistent source of scan blindness and pattern degradation in wide-scan arrays. The standard fix, continuously optimising the intermediate structures between cells, is expensive and fragile across frequency, scan angle and manufacturing tolerance. The more durable fix is architectural: change the radiation mechanism so the problem is suppressed at source instead of compensated for downstream.

MILLIBEAM's approach

We designed our radiation unit architecture around these constraints from day one rather than retrofitting it. Two characteristics define it.

The first is compactness. The radiating element occupies less than roughly one sixth of a free-space wavelength, which frees up substantial real estate inside the standard 0.5λ × 0.5λ phased array unit cell.

The second is a three-dimensional cavity resonance radiation mechanism, in place of the two-dimensional resonant mode of a conventional patch antenna. That geometry gives the array better surface wave control through the radiating mechanism itself, and it removes an entire category of intermediate-structure optimisation from the development cycle rather than tuning around it.

We have also rethought physical integration. Instead of embedding every radiator into a complex multilayer PCB, the radiation unit surface mounts directly onto the motherboard. The antenna and the feed network become genuinely independent modules, and the system PCB no longer has to serve as the array's electromagnetic backbone. It can be optimised purely for RF distribution, control, power and digital I/O, which is a more cost-effective manufacturing path than a premium multilayer stack-up across the full array area.

Compactness helps a second time here. Because the radiator uses only a fraction of the unit cell, there is room for RFICs and thermal structures on the same board side, while the opposite side stays fully available for additional cooling. The radiator itself can double as a heat spreading path from the RFIC out to the environment, so thermal design no longer competes with the antenna for area.

Building the platform

The next real step in phased arrays will not come from a single gain figure or a smarter beamforming algorithm. It comes from changing the question being asked: how do we design a better antenna system, rather than a better antenna element.

That is the architecture we have built at MILLIBEAM. Radiation, RF distribution, RFIC integration and thermal management are distinct, interoperable building blocks rather than one inseparable structure. What we are aiming at is a phased array platform that adapts rapidly across frequencies, power levels, mechanical envelopes and manufacturing constraints, so customers can move from concept to deployed system on their own schedule.

As wireless connectivity becomes ubiquitous, the antennas behind it need to be just as adaptable. That takes a simplified architecture, modular functions, the physics controlled at source, and electromagnetic and thermal design solved together from the outset.

Yifan Wang is Director of Antenna Engineering at MILLIBEAM.

The silicon behind this note

Designed in silicon. Proven in the field.

The parts discussed here are designed, characterised and shipped by MILLIBEAM. Browse the catalogue, filter by band and package, and request a datasheet.

Acknowledgement of Country

MILLIBEAM acknowledges the Traditional Custodians of the lands throughout Australia where we live and work today. We celebrate the diversity of Aboriginal peoples and their ongoing cultures and connections to land, sea and community. We pay our respects to the Elders both past and present and extend that respect to all Aboriginal and Torres Strait Islander peoples today.