RF PCB design: layout rules, impedance control, certification

Why so many RF projects fail, and how to avoid it

A 2.4 GHz ceramic antenna placed less than 3 mm from the ground plane sees its S11 go from -15 dB to -3 dB: half the power is reflected instead of radiated, and range collapses. In RF PCB design, placement details like this one decide the product. A Radio Frequency (RF) PCB integrates wireless communication components (transceivers, antennas, filters, amplifiers, matching networks) to transmit and receive radio signals. These boards are the hardware foundation of Long Range (LoRa), Bluetooth Low Energy (BLE), Wireless Fidelity (Wi-Fi), Narrowband Internet of Things (NB-IoT) and Sub-GHz technologies used in industrial IoT, connected medical devices and logistics.

Most RF project failures trace back to a handful of design errors: an antenna placed too close to the ground plane, a missing matching network, wrong-impedance routing, an unfiltered power supply and certification pushed to the end of the project. The sections below cover the technologies we deploy, a comparison grid and the corrective practices for each error.

In short

An electronic board with radio connectivity is not something you improvise.

Between interference, signal loss, failed certifications and badly integrated antennas, RF errors cost time, money and sometimes an entire market. The most frustrating part: many of these errors are avoidable at the design stage, provided you master the subtleties of radio propagation on PCB.

At AESTECHNO, we design RF boards for connected products in production - industrial sensors, medical devices, logistics trackers. We have learned, often the hard way, that the difference between a prototype that works on the bench and a product that passes Radio Equipment Directive (RED) certification comes down to design details most engineering offices underestimate. The critical tests to anticipate cover Effective Radiated Power (ERP), Total Radiated Power (TRP) and Voltage Standing Wave Ratio (VSWR).

RF technologies we deploy in production

Our RF expertise rests on technologies we integrate into real products, in volume, with certifications obtained per FCC Part 15 and the RED 2014/53/EU directive. We do not do theoretical R&D: every architectural choice has been validated in production and in the certification lab.

The four subsections below cover BLE 5 on Nordic nRF52 and nRF53 chipsets, LoRa and LoRaWAN on Semtech SX126x and STM32WL, Wi-Fi on Espressif ESP32 modules, and the antenna matching work that underpins RED certification. The constraints differ for each: BLE targets short-range, very-low-power links, LoRa trades throughput for range in the 868 MHz Sub-GHz band, and Wi-Fi brings high peak current and coexistence issues with other radios on the same PCB.

BLE 5 (Bluetooth Low Energy)

We design boards based on Nordic Semiconductor nRF chipsets (nRF52, nRF53) for applications requiring short-range, very-low-power communication. Per Nordic in its official datasheets, the nRF52832 delivers up to +4 dBm transmit power with 5.4 mA RX consumption, and the Bluetooth SIG Core Specification 5.x defines the Long Range and High Throughput profiles. The Generic Attribute Profile (GATT) application layer and mesh topologies, along with alternatives from Silicon Labs or Infineon, complete the industrial BLE landscape. Our experience covers ceramic and trace antenna integration on compact PCBs, matching network tuning at the VNA, and link-budget validation in real conditions (plastic enclosure, metal enclosure, proximity to the human body). We have shipped BLE 5 products in volume for monitoring applications with multi-year battery life. To go deeper into Bluetooth 5.4 and PAwR specifics, see our dedicated article on Bluetooth Low Energy. Our decibel converter (dBm, W, dBW) also returns an EIRP compliance verdict against EN 300 328 and FCC 15.247.

LoRa / LoRaWAN

We integrate LoRa modules (Semtech SX126x, STM32WL) for long-range, very-low-power sensors. Per the Semtech SX1261/2 datasheet, the SX1262 reaches about -137 dBm sensitivity at SF12/BW125 (down to -148 dBm only at the narrowest 7.8 kHz bandwidth). On the Kineis side, satellite UHF links impose even tighter link-budget constraints. The main LoRa challenge is antenna integration: the Sub-GHz band (868 MHz in Europe) requires physically larger antennas than 2.4 GHz, and the slightest mismatch translates into a critical range loss. We have hands-on experience with satellite connectivity through Kineis, which pushes RF constraints further still. For a detailed comparison of LPWAN networks, see our article on LoRaWAN, NB-IoT and Sigfox.

Wi-Fi

Wi-Fi integration on a custom board carries specific constraints: high peak current (up to 300 mA), sensitivity to power-supply noise and tricky coexistence with other radios (BLE, LoRa) on the same PCB. We use Espressif modules (ESP32) and dedicated chipsets depending on throughput and security requirements, with particular care for power filtering and ground-plane isolation.

Antenna matching and RED certification

Every RF design we ship includes an antenna matching network sized and validated at the VNA (vector network analyser). We check the reflection coefficient (S11), the -10 dB bandwidth and the VSWR before freezing the design for production, using a calibrated two-port SOLT sweep. This rigour lets us reach the performance levels demanded by the RED 2014/53/EU directive, published in the Official Journal of the European Union, on the first lab pass. The applicable tests are notably defined in ETSI EN 300 328 for the 2.4 GHz band and ETSI EN 300 220 for Sub-GHz. To learn more about the certification process, see our guide on CE and RED certification for IoT products.

Radio-technology comparison for IoT

The radio-technology comparison is a mapping table between typical range, throughput, power consumption, cost and use case for each family (BLE, LoRa, Wi-Fi, NB-IoT, Sigfox). It is the first structuring decision of any IoT project, because it conditions the product business model.

The grid below compares BLE 5.x, LoRa/LoRaWAN, Wi-Fi, NB-IoT and Sigfox on typical range, throughput, power, module cost and typical use. Range spans from 50-200 m for BLE to 5-15 km for LoRa, while throughput runs from 100 bps for Sigfox to over 100 Mbps for Wi-Fi. Power is very low for BLE, LoRa and Sigfox, moderate for NB-IoT and high for Wi-Fi. Here is our decision grid, drawn from our field projects.

Our advice: For long-range industrial projects with battery life, prefer LoRa or NB-IoT. For wearables and connected medical devices, BLE 5 remains the standard. For gateways and applications needing throughput, Wi-Fi is the right call. Contact us for a personalised analysis of your use case.

Classic RF errors and how to avoid them

Classic RF errors are recurring design defects that degrade radio range, fail RED certification or eat into the link budget, observed systematically during prototype reviews. We see them on the prototypes clients bring us after a first unsuccessful trial. Here are the five most common, with concrete explanations to understand and avoid them.

The five errors are a poorly integrated antenna, sloppy RF routing, a neglected power supply, reliance on a certified module without thinking about EMC, and RF/EMC certification pushed to the end of the project. Each one costs link-budget margin or a lab pass: an antenna too close to the ground plane can shift S11 from -15 dB to -3 dB, and a missing matching network costs 2 to 3 dB on a tight LoRa budget.

1. Poorly integrated antenna - the classic that kills your range

The antenna is not "an extra component". It is the conversion point between electrical signal and radio wave, and it is the most sensitive element of the entire RF chain.

Concrete error: antenna too close to the ground plane. A 2.4 GHz ceramic antenna placed less than 3 mm from the ground plane sees its impedance drop and its resonant frequency shift. As a result, the reflection coefficient (S11) goes from -15 dB to -3 dB, meaning half the power is reflected instead of radiated. Range collapses. Contrary to the idea that "any 2.4 GHz antenna will work if the PCB is solid", the ground-plane keepout specified in the supplier datasheet (typically 5 to 10 mm) remains the dominant constraint. The antenna length calculator gives the quarter-wave length for a given frequency.

Concrete error: missing matching network. Without a matching network between the transceiver and the antenna, reflected power increases. On a tight link budget (typical for long-range LoRa), the 2-3 dB loss from poor VSWR is the difference between a sensor that reports its data and a sensor that goes silent.

From the start, we fold antenna constraints into the PCB geometry - ceramic antennas, trace, patch or U.FL/SMA connectors - with a clear ground plane that complies with the antenna manufacturer's specifications.

2. Sloppy RF routing = unpredictable performance

In RF, signals are extremely sensitive to impedance discontinuities. A misplaced via, the wrong trace width, or a floating ground, and your whole bandwidth collapses.

Concrete error: RF trace with wrong characteristic impedance. A trace meant to be 50 ohms but routed at 75 ohms (wrong width for the chosen stack-up) creates a reflection at every impedance transition. Each 50-to-75 ohm transition reflects about 4 percent of the power (reflection coefficient about 0.2, roughly 0.18 dB of mismatch loss); the real damage comes when the wrong impedance detunes the matching network and the antenna, which an S11 check reveals.

At AESTECHNO:

• We physically isolate analogue, digital and RF domains on the PCB.
• We route in controlled impedance (50 ohms typical), and we work upstream with the PCB fabricator to validate the stack-up and etch parameters.
• We use EM simulators (ANSYS HFSS) to validate critical transitions before fabrication.

These skills overlap directly with those of high-speed design, where signal integrity is just as critical.

3. Neglected power supply = guaranteed radio noise

Power supply in RF is a subsystem in its own right that conditions phase noise and receiver sensitivity. Per Texas Instruments in its application notes on ultra-low-noise LDOs, the integrated output noise must stay below 10 uVrms in the 10 Hz to 100 kHz band for a high-performance 2.4 GHz receiver. A noisy switching regulator can inject harmonics directly into your transceiver's frequency band. The result: a high noise floor that degrades receiver sensitivity.

We treat the power supply as a full-fledged RF subsystem, with:

• Targeted decoupling (low-ESR MLCC ceramics as close as possible to the supply pins),
• LC filters tuned to the regulator's measured harmonics,
• Layout optimised to minimise current loops and parasitic coupling.

4. Relying on a certified module without thinking about EMC

"The module is certified, so no problem." Wrong.

Just because a module carries a CE certification does not mean it will not disturb your board, or be disturbed by it. The module certification covers the module in isolation, in the manufacturer's test conditions. Your finished product, with its motors, power lines, cables and enclosure, is a different electromagnetic environment.

Concrete error: missing RF shielding. A BLE module integrated without a shielding can next to a 1 MHz DC-DC converter sees its noise floor rise. BLE throughput drops, disconnections multiply, and the product fails immunity testing in the electromagnetic compatibility (EMC) lab.

We analyse every integration in its real environment - motors, switching supplies, long cables, proximity to other radios - to identify and treat parasitic couplings before the lab pass.

5. Pushing RF/EMC certification to the end of the project

An untested board going into certification is a one-way ticket to delays and cost overruns. We have seen projects lose several months and require two or three PCB iterations because certification was treated as a project-end formality.

At AESTECHNO, we anticipate EMC and RF testing as soon as routing begins:

• Critical zones isolated with guard ring and integrated shielding from the layout stage,
• RF test pads accessible for VNA measurement without re-spin,
• In-house pre-EMC scans to flag issues before the accredited lab,
• Compliance with the RED directive anticipated from component selection onwards.

Our approach: locking in radio performance from the design stage

Our approach to RF board design is a measurement-first methodology that integrates antenna characterisation, controlled-impedance routing, supply-noise budgeting and pre-compliance scanning from the schematic stage. We treat radio connectivity as a systems-engineering problem requiring high-frequency routing, thermal management, supply integrity and certification constraints to be solved simultaneously, not sequentially.

The two subsections below detail the pillars of this approach: electromagnetic simulation with ANSYS HFSS, including AI-assisted antenna optimisation, before fabrication, and the selection of the RF PCB material, from Rogers RO4350B to FR-4 High-Tg, according to Dk, Df, Tg and cost. Both are complemented by VNA validation of the matching network and in-house pre-EMC scans, so that RED, CE and FCC requirements are prepared from the design stage rather than discovered in the accredited lab.

RF simulation and AI-assisted antenna optimisation

At AESTECHNO we routinely use ANSYS HFSS to simulate antennas, RF lines and radiating structures before fabrication, using the FEM (Finite Element Method) approach validated by industry for 30 years. In our lab, we now use AI-assisted antenna optimisation built into the ANSYS tools: the algorithms explore the parameter space (shape, length, matching network) and converge faster towards a design that passes the RED directive on the first try. Each simulation is measured with virtual probes placed per the reference test procedure of ETSI EN 300 328. We can tell, before fabrication, whether the board will work, with good accuracy - a rare advantage on the French market, since ANSYS licences represent a significant investment that few engineering offices commit to.

Selecting the RF PCB material

The RF PCB material is the dielectric laminate that conditions Dk (permittivity), Df (loss tangent), Tg (glass transition temperature) and dimensional stability at RF and microwave frequencies. Per Rogers Corporation in the RO4350B datasheet, the material reaches Dk = 3.48 at 10 GHz and Df = 0.0037, suited to patch antennas and microstrip lines up to 20 GHz. Laminate choice conditions HF performance. We are experts in selecting the material adapted to each project: Rogers RO4350B for dimensional stability and low Df in the 2.4 GHz ISM band and beyond, Isola IS410/370HR for mixed designs, FR-4 High-Tg for cost/performance trade-offs. We arbitrate Dk, Df, Tg, CTE, thermal stability, Pb-free compatibility, fabricator availability and cost for each set of requirements. Our portfolio covers designs up to 10 GHz, with laser uVias, buried vias and integrated PCB antennas.

What we do at AESTECHNO:

• Boards optimised for LoRa, BLE, Wi-Fi, Sub-GHz, NB-IoT
• Controlled-impedance HF routing with fabricator stack-up validation
• Antenna integration matched to your mechanical and environmental constraints
• RF simulation (ANSYS HFSS) and VNA validation
• CE / RED / FCC certification prepared from the design stage
• Multi-radio coexistence (BLE + LoRa, Wi-Fi + BLE) on a single PCB

For projects combining advanced signal processing and RF connectivity, such as SDR (Software Defined Radio) applications, we can also integrate FPGAs for baseband digital processing.

We work as an electronic engineering office embedded in your R&D team, from specification through to industrialisation.

RF design: a strategic stake for connected products

Radio connectivity has become a major differentiator for industrial, medical and logistics IoT products, and it is also the single most-cited cause of late-stage product slips. We have observed that rigorous RF design from the start avoids the certification failures and time-to-market delays that can derail a commercial launch under the RED 2014/53/EU directive.

The radio performance of your product depends on interdependent technical choices: antenna integration, controlled-impedance routing, domain isolation, and electromagnetic compatibility management. These constraints overlap with those of high-speed design and call for expertise in end-to-end electronic design.

In practice, this means treating the antenna, the routing, the power supply and certification as one problem from the first schematic, and validating each choice by simulation and VNA measurement before freezing the design for production.

Bottom line

Bottom line is a five-bullet summary of the load-bearing decisions for any RF PCB project: antenna integration, controlled-impedance routing, supply-noise budgeting, multi-radio coexistence and certification anticipation.

• Antenna placement first, schematic second. A 2.4 GHz ceramic antenna closer than 3 mm to the ground plane shifts S11 from -15 dB to -3 dB, and range collapses.
• 50 ohm is a target, not a guarantee. On a thin 0.1 mm prepreg, a 25 micrometre dielectric tolerance shifts impedance from 50 to about 57 ohm (on a 0.5 mm core the same tolerance moves it by only 1 to 2 ohm), about 0.02 dB of mismatch loss on its own (reflection coefficient about 0.065), but enough to detune a tightly matched network; controlled-impedance routing with fabricator stack-up validation is mandatory above 1 GHz.
• Supply noise is RF noise. Per Texas Instruments LDO application notes, output noise must stay below 10 uVrms in the 10 Hz to 100 kHz band for a high-performance 2.4 GHz receiver; tuned LC filters and low-ESR MLCCs are not optional.
• Module certification does not mean board certification. A pre-certified Nordic Semiconductor nRF54 or NXP module integrated next to a switching regulator can still fail FCC Part 15 limits; in-system pre-EMC scanning is the only safe net.
• Anticipate the RED directive at routing time. ETSI EN 300 328, ETSI EN 301 893 and, for Sub-GHz, ETSI EN 300 220 emission limits drive layout choices, not the other way round.

FAQ: RF board design

Which radio technology should I choose for my IoT product?

The choice depends on four criteria: required range, data throughput, battery life and unit cost. For long-range industrial sensors at low data rates, LoRa or NB-IoT are the reference technologies. For medical devices or wearables requiring short-range, very-low-power communication, BLE 5 is the standard. For applications needing throughput (video, gateways), Wi-Fi is the right call. We analyse your use case to recommend the optimal technology.

How do I choose between BLE, LoRa and Wi-Fi?

BLE 5 excels in ultra-low power and short range (wearables, beacons, monitoring). LoRa offers a range of several kilometres with minimal power consumption, ideal for distant sensors transmitting little data. Wi-Fi provides high throughput but consumes far more, which restricts it to mains-powered or large-battery applications. On the same product, we know how to integrate several radios (BLE + LoRa, BLE + Wi-Fi) with clean coexistence on the same PCB.

What certifications are required for a radio product in Europe?

Any radio product placed on the European market must comply with the RED 2014/53/EU directive (Radio Equipment Directive), which covers safety, electromagnetic compatibility and efficient use of the radio spectrum. The applicable harmonised standards depend on the technology (ETSI EN 300 328 for 2.4 GHz, ETSI EN 300 220 for Sub-GHz). CE marking is mandatory. For export to the United States, FCC Part 15 certification is required. We prepare compliance from the design stage to avoid lab failures.

Why does my RF design fail certification?

The most common causes of failure are: out-of-band spurious emissions (poorly filtered transceiver or regulator harmonics), insufficient radiated power (poorly matched antenna or missing matching network) and susceptibility to conducted disturbances (insufficient decoupling). We also see failures linked to the enclosure: a metal enclosure not factored into the antenna design radically alters the resonant frequency and the radiation pattern.

Can AESTECHNO design a complete RF board?

Yes. We design the complete board: from electronic schematic (transceiver, power supply, interfaces) to high-frequency PCB routing (controlled impedance, domain isolation, antenna integration), through EM simulation and VNA validation. We also prepare the RED/CE certification package and support the project up to the accredited-lab pass. Our scope covers BLE, LoRa, Wi-Fi, NB-IoT and Sub-GHz.

How do I integrate several radio technologies on the same PCB?

Multi-radio coexistence (BLE + LoRa, Wi-Fi + BLE) on the same PCB is a challenge of isolation and interference management. The 2.4 GHz radios (BLE, Wi-Fi) can interfere with each other if antennas sit too close. The keys are physical separation of antennas, ground-plane isolation, time-sequencing transmissions when possible, and appropriate filtering. We have shipped multi-radio boards in production with performance validated in real environments.

Related articles

To go further with your wireless connectivity projects:

• High-performance electronics project: a decision-maker's guide.
• Bluetooth 5.4 and PAwR, BLE technologies for massive IoT networks
• LoRaWAN vs NB-IoT vs Sigfox, LPWAN long-range network comparison
• Electromagnetic Compatibility (EMC), EMC standards and certification
• High-speed design, Routing and signal integrity
• CE and RED certification for IoT, Radio compliance guide
• Electronic engineering office, Our design services
• FPGA board design, FPGA and digital processing for RF applications
• AESTECHNO blog, full archive of electronic-design articles
• How a PCB works, illustrated primer on substrate, copper layers and fabrication