
What Is Phased Array Antenna Design?
Phased array antenna design is the process of arranging multiple radiating elements on a common aperture and controlling the phase (and often amplitude) of each element’s signal to steer or shape the radiated beam electronically—without mechanical rotation. The technique enables beam steering in microseconds, supports multiple simultaneous beams, and scales from sub-6 GHz 5G base stations to 77 GHz automotive radar modules. Getting it right on a PCB demands coordinated decisions across element geometry, inter-element spacing, feed-network topology, phase shifter integration, and laminate selection.
Element Selection and Geometry

Every phased array starts with the unit cell—the individual radiating element repeated across the aperture. The element type determines bandwidth, polarization purity, and how tightly you can pack the array.
Common Element Types
| Element | Typical Bandwidth | Polarization | Best Fit |
|---|---|---|---|
| Rectangular patch | 3–8 % | Linear or dual-linear | Narrow-band radar, 5G FR2 |
| Stacked patch | 15–25 % | Dual-linear / circular | Wideband SATCOM, LEO terminals |
| Vivaldi / tapered slot | > 3:1 | Linear | EW, UWB imaging |
| SIW slot | 10–15 % | Linear | mmWave arrays above 60 GHz |
For substrate-integrated waveguide approaches at mmWave frequencies, SIW antenna design techniques offer low radiation loss and straightforward integration with planar feed networks. Vivaldi elements suit wideband electronic-warfare arrays but demand more board area per element, which conflicts with tight lattice spacing at higher frequencies.
Substrate Thickness Trade-Off
A thicker substrate increases patch bandwidth (roughly proportional to h/λ₀) but excites surface waves that raise mutual coupling and degrade scan performance. For a rectangular patch at 28 GHz on Rogers RO4350B (Dk 3.48, Df 0.0037), a substrate height of 0.254 mm (10 mil) gives about 5 % impedance bandwidth, while 0.508 mm (20 mil) pushes bandwidth past 8 % at the cost of stronger surface-wave coupling beyond ±50° scan. The choice between thick and thin antenna substrates must be made with the scan-volume requirement in hand.
Inter-Element Spacing and Grating Lobes
Element spacing is the single most consequential layout parameter. If the spacing d is too large, grating lobes appear inside visible space; too small, and mutual coupling degrades active impedance and increases scan blindness risk.
The Spacing Rule
For a linear array scanned to a maximum angle θmax from broadside, the grating-lobe-free condition is:
d < λ₀ / (1 + sin θmax)
At 28 GHz (λ₀ ≈ 10.71 mm) with ±60° scan, this gives d < 5.73 mm. Most designers target 0.5 λ₀ (≈ 5.36 mm) as a safe default. At 77 GHz (λ₀ ≈ 3.90 mm), the spacing drops below 2 mm, which is why antenna-in-package architectures become attractive—routing hundreds of phase-controlled paths on a standard PCB at that pitch is extremely challenging.
Mutual Coupling Budget
Aim for element-to-element coupling below −15 dB across the operating band. Higher coupling shifts active impedance as the beam scans, causing scan-dependent mismatch loss. Full-wave simulation of at least a 4 × 4 finite sub-array (with periodic boundary conditions on the interior cells) is the minimum fidelity needed to capture these effects. An antenna simulation workflow in HFSS or CST that exports embedded-element patterns and S-parameter matrices feeds directly into the system-level beamformer model.
Phase Shifter Architecture
The phase shifter converts a control word into a progressive phase taper across the array, steering the beam. Three dominant implementation styles exist on PCB-level phased arrays:
- Analog (varactor-loaded or ferrite): Continuous phase control, simple bias, but amplitude varies with phase state. Common in S- and C-band arrays.
- Digital MMIC (4- to 6-bit): Discrete phase steps (e.g., 5.6° LSB for 6-bit), excellent repeatability, dominant at X-band through Ka-band. Devices such as the Analog Devices ADAR1000 integrate four channels of phase and gain control in a single package.
- True-time-delay (TTD): Eliminates beam squint in wideband arrays by applying a group-delay offset rather than a fixed phase shift. Essential when fractional bandwidth exceeds roughly 10 % and scan angles exceed ±30°.
For a 64-element array at 28 GHz using 6-bit digital phase shifters, the peak quantization sidelobe level is approximately −19.5 dB (≈ −6 × number of bits in dB). If sidelobe requirements are −25 dB or tighter, amplitude tapering (Taylor or Chebyshev) must supplement the phase taper, and each channel needs a variable-gain amplifier or attenuator.
Feed Network and Beamformer PCB Layout
Corporate vs. Series Feed
| Attribute | Corporate (parallel) feed | Series feed |
|---|---|---|
| Bandwidth | Wide (equal path lengths) | Narrow (frequency-dependent beam squint) |
| Routing complexity | High—binary tree with many junctions | Low—single meandered line |
| Amplitude taper | Easy via unequal power dividers | Achieved by varying coupling coefficients |
| Typical use | Active arrays with per-element MMICs | Passive arrays, fixed-beam applications |
Active phased arrays almost always use corporate feed because each element already has its own transmit/receive module (TRM). The layout challenge is routing the RF, LO, IF, DC bias, and digital control lines to every TRM without violating antenna keepout rules around each radiating element. A common stackup for a 28 GHz active array is 8–10 layers: antenna patches on the top surface, ground plane immediately below, two RF stripline layers for the corporate feed, a power-distribution plane, and digital/control layers at the bottom.
Impedance-Controlled Routing
Every transmission-line segment in the feed network must be impedance-controlled—typically 50 Ω for single-ended paths. At 28 GHz on RO4350B (Dk 3.48, 5 mil core), a 50 Ω microstrip is approximately 0.30 mm (12 mil) wide. Trace-width tolerance of ±0.5 mil shifts impedance by roughly ±1.5 Ω, which is acceptable for most designs but must be verified against the S11 and VSWR measurement budget after fabrication.
Amplitude and Phase Matching
For sidelobe control to hold, the feed network must deliver amplitude balance within ±0.5 dB and phase balance within ±5° across all paths at the design frequency. Length-matching all corporate-feed arms to within ±50 µm is routine on modern high-frequency laminates, but via transitions between layers can introduce 0.5–2° of untracked phase per transition at Ka-band. Minimize layer changes in the RF path, and when a via is unavoidable, use back-drilled or laser-drilled microvias with controlled stub lengths.
Material Selection for Phased Array PCBs

Laminate choice affects loss, dimensional stability, and cost—three vectors that pull in different directions.
| Laminate | Dk | Df | Typical Band | Relative Cost |
|---|---|---|---|---|
| FR-4 (high-speed grade) | 4.2–4.5 | 0.018–0.020 | Below 3 GHz | 1× |
| Rogers RO4350B | 3.48 | 0.0037 | 3–30 GHz | 3–5× |
| Rogers RO3003 | 3.00 | 0.0010 | 24–81 GHz | 6–8× |
| Isola Astra MT77 | 3.00 | 0.0017 | 24–77 GHz | 5–7× |
At 77 GHz, every 0.001 increase in Df adds roughly 0.15 dB/cm of additional insertion loss in microstrip—significant when the corporate feed tree may span 30 mm or more. Rogers RO3003 laminate data shows a Dk tolerance of ±0.04, tight enough to keep patch resonant frequency within 1 % across a production panel.
System-Level Design Checklist
Before committing a phased array antenna design to layout, verify every item on this checklist. Missing a single entry—especially grating-lobe clearance or thermal dissipation—can force a full respin. Enter your array parameters below to generate a tailored checklist:
A typical output for a 256-element, 28 GHz active array on RO4350B might flag: element spacing 5.35 mm (safe for ±60° scan), 8-layer stackup with two RF stripline layers, thermal via array under each TRM pad rated for 0.25 W dissipation, and a minimum of 6-bit phase resolution to meet −20 dB sidelobe specification. Walk through every flagged item with your PCB fabricator before releasing Gerbers; the prototype-to-production transition guide covers the tolerance and yield conversations you need to have.
Evaluating Design Changes Before Respin
Phased array boards are expensive to fabricate and even more expensive to respin. A single layer-count change on a 10-layer mmWave stackup can add weeks of lead time and significant NRE. Before approving any engineering change order, quantify the return: does switching from RO4350B to RO3003 on the antenna layers save enough loss to justify the laminate cost delta and the hybrid bonding process it requires?
If the calculator shows that moving to a lower-loss laminate saves 1.2 dB of array gain (equivalent to roughly 30 % more elements to recover the same EIRP), the cost of the material upgrade is almost certainly justified. Conversely, if the gain delta is under 0.3 dB, the change adds risk without meaningful system benefit. Use results like these to anchor design-review discussions in numbers, not opinions.
Measurement and Validation
A phased array cannot be validated with a single return-loss sweep. You need:
- Embedded-element pattern: Measure each element while all others are terminated in 50 Ω. Compare with simulation to catch fabrication-induced coupling anomalies.
- Active S-parameters: Sweep active VSWR at multiple scan angles. The worst-case mismatch usually appears near the scan limit.
- Far-field beam pattern: Verify main-beam pointing accuracy, sidelobe level, and null depth at broadside and at the maximum scan angle. Antenna pattern measurement techniques using planar near-field scanning are the most practical for large flat arrays.
- EVM / ACLR (communication arrays): Beamformer nonlinearity and quantization noise affect modulation quality; measure at the system level with a modulated signal, not just CW.
Frequently Asked Questions
What is the maximum scan angle of a phased array?
Theoretically ±90° from broadside, but practical limits are ±60° to ±70°. Beyond that, projected aperture shrinks rapidly (gain drops as cos θ), active impedance mismatch increases, and element patterns roll off. Most specifications define the usable scan volume as the region where gain drops no more than 3–4 dB from broadside.
Can FR-4 be used for a phased array?
Yes, at frequencies below about 3 GHz where its loss tangent (Df ≈ 0.020) is tolerable and its Dk variation across a panel (±5–10 %) does not push element resonances out of band. Above 6 GHz, a controlled-Dk laminate like RO4350B is effectively mandatory. Refer to IEEE Antennas and Propagation Society publications for peer-reviewed comparisons of substrate impact on array performance.
How many elements does a typical 5G mmWave phased array have?
Base-station panels commonly use 256 to 1024 dual-polarized elements; user-equipment modules in handsets use 4 to 16 elements per sub-array, with two to three sub-arrays covering different angular sectors.
What causes scan blindness in a phased array?
Scan blindness occurs when a surface wave or substrate mode is phase-matched to the Floquet mode of the array at a particular scan angle, causing near-total reflection. It is most common on thick, high-Dk substrates. Mitigation strategies include substrate thinning, adding metallic via fences between elements, or switching to cavity-backed elements. The IEEE Transactions on Antennas and Propagation archive contains extensive studies on scan-blindness prediction and suppression.
Ready for a quotation?
Send your requirements for a material, stackup and manufacturing review. Price and schedule are confirmed for your project.
