Antenna Design for IoT

Antenna Design for IoT Applications

Antenna performance in compact IoT devices can change dramatically between an open-board prototype and the final product. Enclosures, batteries, nearby plastics or metal, and even a user’s hand can detune the antenna and reduce efficiency, which shows up as reduced range and higher power consumption for a given link margin.

IoT antenna design is a trade-off between size, bandwidth, efficiency, and real-world environment. Many products also need to support multiple bands (for example 868/915 MHz and 2.4 GHz) and maintain acceptable performance across installation and handling conditions.

Many antenna problems stem from treating the antenna as an afterthought. This guide explains the key constraints, common antenna options, matching and ground-plane effects, and practical measurement approaches so you can converge on a design that performs reliably in the finished product.

The Physics You Can't Cheat

The Chu-Harrington limit gives a useful reality check for electrically small, lossless, linearly polarised antennas: $Q \geq \frac{1}{k^3a^3} + \frac{1}{ka}$, where $k = 2\pi/\lambda$ and $a$ is the radius of the smallest sphere enclosing the antenna and its current distribution. Smaller antennas drive Q up, which means bandwidth gets tighter for a given match. Circular polarisation has a slightly looser bound and so allows a smaller antenna for the same Q. Efficiency isn't set by the Chu limit directly, but because radiation resistance gets small, even modest loss resistance in copper, dielectric, or the matching network can dominate. There are clever tricks to trade size, bandwidth, and efficiency, but you can't have all three at once.

At 2.4 GHz (λ ≈ 125mm), a 15mm antenna is already electrically small. At 868 MHz (λ ≈ 345mm), that same antenna is tiny. The radiation resistance drops, losses dominate, and your battery dies faster because the radio has to work harder to compensate.

Antenna Types: Know Your Options

Quarter-wave monopole: Best performance, but at 868 MHz that's 86mm of wire sticking out. Not happening in most products.

Inverted-F (IFA/PIFA): Folds the monopole to save space. Popular for a reason: good balance of size and performance. My go-to for most designs.

Chip antennas: Tiny and convenient, but efficiency suffers. Fine for short-range Bluetooth; questionable for LoRa where every dB matters.

PCB trace antennas: Free in terms of BOM cost, but you're paying in board space and engineering time to get them working.

Your Ground Plane IS Part of the Antenna

This is where most designs go wrong. That "antenna" in the corner of your PCB? It's actually the antenna plus the entire ground plane acting together. Change your board size, and your antenna performance changes. I've seen 6dB swings from a 10mm change in PCB dimensions.

For monopole-type antennas, current flows along the ground plane edges. If your ground is too small, performance tanks. If the ground plane dimensions happen to resonate at your operating frequency, you get weird pattern distortions. Test your actual PCB, not an eval board with different dimensions.

Matching: Where Theory Meets Reality

Your electrically small antenna probably has an impedance nowhere near 50Ω. Maybe 15Ω resistive with -j200Ω reactive. You need a matching network to transform that to 50Ω, and every component in that network has losses.

I've seen matching networks that looked perfect in simulation burn 2dB of signal in real life because someone used cheap inductors. Use high-Q components. An extra five cents per inductor is worth it when you're fighting for range.

The Hand Effect and Other Surprises

Proximity effects can detune antennas more than you'd expect. Published measurements on 2.45 GHz wearable and on-body antennas show resonance shifts of about 2 to 7 percent for typical body positions, with worst-case grip or skin-contact geometries pushing toward ~10 percent. Implants and embedded antennas inside high-permittivity tissue can shift much more. The human body (εr ≈ 40 to 50, lossy) is a strong near-field perturber, so validate tuning and efficiency with representative use conditions.

Your plastic enclosure? Also shifts the frequency down. Metal mounting bracket? Creates reflections that either help or hurt depending on spacing. Test in the actual operating environment, not just on the bench with the case off.

Multi-Band: Pick Your Battles

Covering 868 MHz, 915 MHz, and 2.4 GHz with one antenna? Possible, but you'll compromise everywhere. A dual-band design with separate resonant elements usually outperforms a wideband compromise.

Switchable antennas using PIN diodes or RF switches let you optimize for each band separately. More complex, but if you need both sub-GHz and 2.4 GHz performance, often the better approach.

Keep Your Noise Away from the Antenna

The antenna's near-field extends about λ/2π from the element. At 2.4 GHz, that's roughly 20mm. Any noisy digital circuit within that zone couples directly into your receiver, killing sensitivity.

I've debugged designs where a switching regulator 15mm from the antenna added 10dB to the noise floor. Move the antenna to a corner, keep the ground plane clear underneath, and route high-speed signals away from it.

PCB Antennas: Free Isn't Free

No BOM cost for a trace antenna, but you pay in other ways. FR-4 has losses that eat into efficiency, especially above 1 GHz. The effective dielectric constant ($\varepsilon_{eff}$ ≈ 3.3 for typical microstrip) shrinks your antenna, which sounds good until you realize it also shrinks your bandwidth.

If you need every dB of performance, consider a ceramic-loaded antenna or a dedicated RF substrate in the antenna area. For cost-sensitive high-volume products where "good enough" range suffices, PCB traces work fine.

Diversity: When One Antenna Isn't Enough

In multipath environments, two antennas with λ/4 spacing rarely fade simultaneously. Selection diversity (switching to whichever antenna has better signal) can recover 10dB in deep fades. WiFi and Bluetooth chips often support this natively.

The challenge in small devices: getting enough isolation between antennas. They need to be far enough apart to see independent fading, but close enough to fit on your PCB. Aim for at least 15dB isolation; use decoupling slots or neutralization lines if needed.

Simulation tools help, but they don't replace prototyping. Full-wave EM simulators (FEM, MoM) can predict antenna behavior including ground plane and component interactions, but your simulation is only as good as your model. Oversimplify and you miss real effects. Over-detail and it takes forever to run. FR-4 properties vary with frequency and between manufacturers, so don't trust simulation alone for final performance numbers. Use parametric sweeps to find which dimensions are sensitive and need tight tolerances, then prototype and measure.

Manufacturing Tolerance Reality

Your simulated antenna assumes perfect dimensions. Reality is ±10% on trace width, ±5% on dielectric constant, and substrate thickness variations. Each of these shifts your resonant frequency. A design that works perfectly in simulation may ship with 10% yield loss if you don't account for this.

Design for the middle of the tolerance window, not the edges. Add tunability (component values in the matching network) where you can't control variations. For high-volume production, characterize your manufacturing variation statistically and design with that margin built in.

Measuring IoT antennas is harder than measuring standalone antennas because the antenna is part of the product. Anechoic chamber measurements give accurate far-field patterns but may not reflect real-world performance with the enclosure, battery, and user proximity. Over-The-Air (OTA) testing measures complete radio performance including the antenna, which is more representative but requires a more sophisticated test setup. For production, conducted measurements through temporary connectors are faster and repeatable, but you need to establish how they correlate with radiated performance. The hard part is finding tests that are fast enough for production but still catch real problems.

For battery-powered devices, antenna efficiency isn't the only factor in power consumption. Total energy per transmitted bit depends on transmit power, data rate, protocol overhead, and sleep current between transmissions. A more efficient antenna reduces required transmit power for a given range, but the PA efficiency curve makes this non-linear. Below a certain transmit power, you might actually increase total energy consumption if it causes retransmissions or longer active periods. Think about the radiation pattern too. An IoT sensor mounted on a ceiling benefits from a downward-directed pattern, not omnidirectional. Optimize the antenna, radio, and protocol together rather than independently.

Techniques like metamaterial-inspired structures and optimization algorithms may push the size-performance trade-off further in the future, but for now the fundamentals above are what make or break most designs.

I design antennas for IoT products, from initial concept through production validation. Whether you need a custom PCB antenna, help with multi-band integration, or solutions for challenging environmental conditions, get in touch to discuss your antenna design challenges.

Antenna integration is one of the most common areas where wireless projects run into trouble. These RF design principles are applied across all my RF design services, from initial schematic through production validation.