When should a chip antenna be preferred over a PCB trace antenna?

A chip antenna (LTCC ceramic, families from Johanson, Yageo, Murata, Pulse, Kyocera) is preferred when board area is constrained and when tuning sensitivity to the enclosure must be minimised. It comes with a vendor matching reference and a defined ground-clearance footprint, which shortens the design loop. A PCB trace antenna (IFA, monopole, meander) costs nothing in bill of materials and reaches comparable efficiency, but requires more clearance, a more rigorous Smith chart tuning campaign and freezes a non-trivial portion of the board edge. For 2.4 GHz BLE under tight area constraints, a chip antenna is typically the first choice; for sub-GHz LPWA, a PCB IFA or external whip remains common because chip-antenna efficiency drops at lower frequencies.

What is the role of the pi-network matching topology?

The pi-network is a three-element matching topology (shunt, series, shunt) placed between the radio output and the antenna feed. It is preferred over the simpler two-element L-network because it provides the degrees of freedom needed to move the impedance anywhere on the Smith chart and to absorb the detuning introduced by the enclosure, the battery and nearby metal. In practice, two of the three positions are populated and the third is reserved for fine tuning measured on the assembled product. The components used are high-Q 0402 or 0201 capacitors and inductors qualified at the working frequency.

Why measure return loss on the assembled product, not on the bare PCB?

The antenna sees its near-field environment: enclosure plastic, battery, display, USB cable, hand of the user. Each of these elements contributes a capacitive or inductive perturbation that shifts the resonance and degrades return loss. A matching network tuned on the bare PCB looks excellent on the bench and fails immediately once the product is closed. Best practice is to measure S11 at the antenna feed in the final mechanical configuration, including battery, display and cables, then tune. CTIA OTA chambers go further and include hand and head phantoms to capture realistic loading.

What ground-clearance area does a chip antenna require?

The vendor datasheet always specifies the recommended keep-out and ground-clearance area. The figure depends on frequency and antenna family, but a typical 2.4 GHz chip antenna asks for a rectangular clearance around 5 by 10 mm to 7 by 15 mm, placed along the long edge of the board, with no copper pour, no via stitching and no signal trace underneath. Violating this clearance is the most common cause of catastrophic efficiency loss seen on first prototypes: the antenna stops radiating and behaves as a high-Q resonator decoupled from free space.

How is GNSS antenna selection different from BLE or Wi-Fi?

GNSS L1 at 1575 MHz and L5 at 1176 MHz uses right-hand circular polarisation and weak signals (around minus 130 dBm at the receiver input). The dominant choice is a ceramic patch antenna on top of a ground plane, sized 15 by 15 mm or 25 by 25 mm depending on sensitivity targets. A low-noise amplifier sits at the antenna feed, preceded or followed by a SAW filter depending on the link budget and the rejection needed against nearby cellular emitters. Active patches integrate the LNA. Passive patches need an external LNA in the front-end. The ground plane below the patch must be at least as wide as the patch, ideally larger, to avoid pattern distortion.

What does CTIA OTA testing measure?

CTIA Test Plan for Wireless Device Over-the-Air Performance measures Total Radiated Power (TRP) and Total Isotropic Sensitivity (TIS) in an anechoic chamber, with the device active and communicating with a base station emulator. TRP integrates radiated power over the complete sphere, capturing antenna efficiency and chain losses. TIS integrates receiver sensitivity over the sphere, capturing self-noise and desensitisation by digital activity. CTIA is the reference for US carrier acceptance (AT and T, Verizon, T-Mobile) and is mirrored by 3GPP TS 38.521 for cellular RF performance and by carrier-specific programmes worldwide. PTCRB testing draws on the same chambers.

Why does a SAW filter sit between the LNA and the receiver?

In a multi-radio product (GNSS plus cellular plus Wi-Fi plus BLE on the same board), the GNSS LNA receives not only the wanted signal but also high-power emissions from adjacent radios that leak through the antenna and through the housing. Without filtering, the LNA saturates and the GNSS receiver loses sensitivity by tens of dB. A SAW or BAW filter rejects out-of-band signals before they reach the receiver. Placement order (filter before LNA, or LNA before filter) depends on the link budget: filter-first preserves linearity at the cost of noise figure, LNA-first preserves sensitivity at the cost of linearity. Both topologies are valid and are chosen per product.

What MIMO and diversity rules apply to compact IoT enclosures?

For Wi-Fi MIMO and cellular diversity, the two antennas must be decorrelated. The two main levers are spatial separation, ideally at least a quarter wavelength (around 30 mm at 2.4 GHz, 12 mm at 5 GHz), and polarisation diversity (one antenna horizontally polarised, one vertically). The envelope correlation coefficient (ECC) is the figure of merit, with a target below 0.5 to deliver MIMO gain. Inside a compact enclosure, achieving this is hard and typically requires careful placement at opposite corners of the board, with ground-plane shaping to break common-mode currents. Vendor reference designs and anechoic-chamber measurement of ECC are the practical path.

Antenna design and impedance matching for IoT

Author Hugues Orgitello Last updated 27 May 2026 ~14 min read

Guide, antennas and matching

Antenna integration is the layer at which a connected product wins or loses its radio link budget. A poorly chosen antenna, a missing ground clearance, a matching network tuned on the bare PCB instead of on the assembled enclosure, or a misplaced SAW filter all translate immediately into a measurable loss of radiated power, a degradation of receive sensitivity and a higher probability of failing certification campaigns under ETSI EN 300 328, FCC Part 15, 3GPP TS 38.521 or carrier OTA test plans. This page covers antenna family selection (chip, PIFA, IFA, monopole, patch, helix), band-specific considerations from sub-GHz LPWA up to UWB, the matching network workflow on a vector network analyser, ground-clearance requirements, detuning factors brought by enclosure and user, the CTIA OTA test plan, GNSS-specific front-end choices, MIMO and diversity for compact IoT enclosures, and the recurring pitfalls observed at the antenna stage.

Antenna families and when to choose which

The starting point of any radio integration is selecting an antenna family that matches the frequency band, the available volume, the bill of materials target and the radiation pattern needed. The five families below cover the majority of connected product designs.

Family	Typical bands	Strengths	Weaknesses
Chip antenna (LTCC ceramic)	2.4 GHz, 5 GHz, sub-GHz, GNSS	Small footprint, vendor reference design, defined ground clearance, low BOM cost	Lower efficiency than a well-designed PCB antenna at the same frequency, sensitive to placement
PIFA (Planar Inverted-F Antenna)	Cellular 700, 2700 MHz, multi-band	Wide bandwidth, integrates above ground plane, moderate volume	Requires 3D space above the PCB, mechanical complexity
IFA (Inverted-F)	2.4 GHz, 5 GHz, sub-GHz	Free in BOM (PCB trace), efficient when clearance is respected	Footprint freezes a portion of the board edge, sensitive to tuning
Monopole (PCB trace or wire)	Wide range, sub-GHz to 6 GHz	Simple, omnidirectional pattern, easy to scale	Requires a ground plane as counterpoise, polarisation-sensitive
Patch (ceramic, microstrip)	GNSS L1 and L5, UWB 6, 8 GHz, mmWave	Directional gain (3, 6 dBi), circular polarisation possible	Requires a ground plane, larger surface, narrow bandwidth
Helix (external)	UHF cellular, GNSS, professional radio	Robust, high gain, broadband variants	External, mechanical exposure, cost

The first triage is by frequency and volume. A wearable at 2.4 GHz with no external antenna almost always uses a chip antenna or a PCB IFA. A cellular IoT tracker covering 700 to 2700 MHz typically uses a PIFA or a multi-band chip antenna with a careful matching network. A GNSS module embeds a ceramic patch. UWB at 6 to 8 GHz uses either a chip antenna or a custom patch. UHF asset trackers commonly use external helices.

The vendor families to evaluate first are Johanson Technology (extensive 2.4 GHz and GNSS catalogue), Yageo (broad coverage), Murata (LTCC ceramics and integrated antenna-in-package), Pulse Electronics (cellular PIFA), Taoglas (multi-band and external antennas) and Kyocera AVX (broad chip antenna and ceramic patch lines). Each family ships with reference designs, recommended matching networks, ground-clearance footprints and OTA performance plots that anchor the early-design decisions.

Bands and link-budget targets

The frequency band drives every downstream decision: antenna size, matching network topology, regulatory framework and test plan.

Technology	Band	Wavelength (free space)	Typical antenna size
BLE, Wi-Fi 2.4 GHz	2400, 2483.5 MHz	125 mm	Chip antenna 7, 15 mm or PCB IFA 30, 40 mm
Wi-Fi 5 GHz	5150, 5875 MHz	55 mm	Chip antenna 4, 8 mm or PCB IFA 15, 25 mm
Wi-Fi 6 GHz	5925, 7125 MHz	47 mm	Chip antenna or PCB IFA, dual-band reuse common
LoRa, Sigfox sub-GHz	868 MHz (EU), 915 MHz (US, AU)	345, 327 mm	PCB IFA, monopole, external whip, chip antenna lower efficiency
NB-IoT, Cat-M cellular	700, 2700 MHz	428, 111 mm	Multi-band PIFA or chip antenna with band-specific matching
GNSS L1	1575 MHz	190 mm	Ceramic patch 15, 25 mm
GNSS L5	1176 MHz	255 mm	Ceramic patch, larger than L1, often dual-band L1, L5
UWB	6, 8.5 GHz	35, 47 mm	Chip antenna or patch

The Wi-Fi 6 GHz band is regulated under FCC U-NII-5 to U-NII-8 in the United States and under harmonised EU rules through the RED and EN 303 687. For 2.4 GHz, EN 300 328 applies in the EU and FCC Part 15.247 in the United States; the cellular bands follow 3GPP TS 38.521 for NR and 3GPP TS 36.521 for LTE. Compliance with these radio standards depends materially on antenna performance: spurious emissions are dominated by harmonic radiation through the antenna, and adjacent-channel leakage is driven by the matching network.

Key antenna metrics

A consistent set of figures of merit is needed to specify, qualify and trade off antennas.

Return loss (S11): the fraction of power reflected back from the antenna feed. A target of S11 below minus 10 dB across the operating band is standard, equivalent to a voltage standing wave ratio (VSWR) below 2:1. A measurement of minus 6 dB still works but accepts a 25 percent loss to reflection.
Efficiency: the ratio of radiated power to power accepted at the feed, expressed in percent or in dB. Anechoic-chamber measurements (Satimo, NSI-MI MARS or equivalent) are the reference. A 60 to 70 percent efficiency is realistic for a well-designed chip antenna at 2.4 GHz in a compact enclosure; below 30 percent, the radio link budget collapses.
Gain: the directional efficiency, expressed in dBi (decibels over isotropic). A 2.4 GHz chip antenna is typically 0 to 2 dBi peak; a GNSS patch is 3 to 5 dBi; a directional UWB or mmWave patch can reach 6 to 8 dBi.
Bandwidth: the frequency range over which S11 stays below minus 10 dB. A 2.4 GHz chip antenna typically covers 100 MHz; a multi-band PIFA covers 700, 2700 MHz across several resonances tuned by the matching network.
Polarisation: linear (vertical or horizontal) for most BLE, Wi-Fi and cellular antennas; right-hand circular polarisation for GNSS.
Radiation pattern: omnidirectional (monopole, IFA, chip) or directional (patch, helix, mmWave array). A wearable benefits from omnidirectional patterns; a GNSS receiver benefits from a hemispherical pattern oriented toward the sky.

Matching network workflow

The matching network adapts the antenna impedance to the 50 ohm reference of the radio output. Without matching, even a good antenna reflects a large fraction of the power back to the transmitter, wastes battery and saturates the front-end.

Topologies

Topology	Components	Use case
L-network	2 components (one series, one shunt)	Simple antennas already close to 50 ohm, narrow band
Pi-network	3 components (shunt, series, shunt)	Compact antennas with notable detuning, wide tuning range, most flexible
T-network	3 components (series, shunt, series)	Less common, mainly for high-impedance loads
Shunt-series-shunt	Equivalent to pi-network	Default starting topology for chip antennas

The default starting topology is the pi-network. Two of the three positions are populated with capacitors and inductors; the third is reserved as a fine-tune slot, populated only after measurement on the assembled product. Components are high-Q 0402 or 0201 capacitors and inductors (Murata GJM, GRM, LQW; Coilcraft 0402DC, 0402CT) qualified at the working frequency, because at 2.4 GHz a generic 100 nH inductor self-resonates well below the band and behaves as a capacitor.

Smith chart workflow

Solder a calibrated U.FL connector at the antenna feed of the prototype, replacing the antenna trace bend if needed.
Calibrate the vector network analyser (VNA) at the connector end using an open-short-load (OSL) calibration kit.
Connect the antenna in its final mechanical environment (PCB assembled, enclosure closed, battery present, cables routed).
Measure S11 across the band of interest. Read the complex impedance at band centre on the Smith chart.
Identify the impedance offset from the 50 ohm centre. The chart maps directly to the matching elements to use.
Populate the matching network with calculated values, then iterate. A typical campaign converges in two to four iterations.
Validate over temperature if needed, and over a population of units if the matching tolerance is tight.

The same workflow is documented in vendor application notes (Johanson, Murata, Pulse). Free Smith chart software (smith-chart.org, AWR Microwave Office, Keysight ADS) speeds up the calculation.

Ground plane and clearance requirements

The ground plane is the radiating counterpoise of every printed and chip antenna. Without a sufficient ground plane, the antenna does not radiate at its designed frequency: it sees a high-impedance counterpoise and either fails to resonate or radiates into a distorted pattern.

Rules for chip antennas

Place the antenna along the long edge of the board, ideally at a corner, with the ground-clearance footprint protruding from the ground plane.
Respect the vendor keep-out area, typically 5 by 10 mm to 7 by 15 mm at 2.4 GHz, with no copper pour, no via stitching, no signal trace underneath.
Provide a ground-plane area of at least one quarter wavelength on the side adjacent to the antenna feed. At 2.4 GHz this is 30 mm; at sub-GHz 868 MHz, this is 86 mm, which is the main reason chip antennas at sub-GHz are inefficient on small boards.
Avoid placing the chip antenna at the centre of the board, surrounded by copper pour. This is a common error on first prototypes and produces a non-radiating high-Q resonator.

Rules for PCB IFA and monopole

The total length of the printed trace defines the resonant frequency, with a quarter wavelength as a starting point.
The ground-plane extension along the trace direction acts as the counterpoise.
A matching network at the feed compensates for tolerance on PCB dielectric constant and trace dimensions.

Rules for GNSS patches

The ceramic patch sits on the top side of the PCB.
The ground plane below the patch must be at least as large as the patch itself, preferably 1.5 to 2 times larger.
No copper pour or active component on the top side within the patch footprint.
The LNA sits as close as possible to the patch feed to preserve noise figure.

Detuning factors and the assembled-product reality

The bench measurement on a bare PCB is only an early indication. The real radio performance is determined by the assembled product.

Source of detuning	Mechanism	Mitigation
Plastic enclosure	Dielectric loading, shifts resonance downward by 50, 200 MHz at 2.4 GHz	Tune on closed enclosure
Metal parts (battery, screen, USB shield)	Mirror currents, pattern distortion, efficiency loss	Reserve clearance, validate on assembled product
Hand effect (capacitive loading)	User grip drops efficiency by 3, 10 dB	Measure with hand phantom (CTIA), design for worst-case grip
USB cable in operation	Cable acts as parasitic radiator, distorts pattern	Test with and without cable, ferrite if needed
Nearby crystal oscillator or DC-DC converter	Conducted noise into the antenna feed	Filter on antenna line, decoupling on RF supply, layout separation
ESD strike on antenna feed	Damages front-end LNA or matching components	TVS diode or spark gap on antenna line

The operational rule is consistent: the antenna campaign ends when the product is measured assembled, with battery, with cables, in the worst-case user posture, and the OTA performance meets the link-budget target.

OTA measurement: TRP, TIS and the CTIA test plan

Antenna performance on a connected product is qualified end-to-end through Over-The-Air (OTA) measurements in an anechoic chamber.

Total Radiated Power (TRP)

TRP is the integral of radiated power over the complete sphere around the device, measured with the device transmitting at full power and the antenna being rotated through azimuth and elevation. It captures both the antenna efficiency and the chain losses from the radio output to the antenna feed (matching, cable, connector). For Wi-Fi at 2.4 GHz, a realistic target in a compact IoT enclosure is 13, 18 dBm TRP at the band centre, against a radio output of 18, 20 dBm; the delta is the chain loss budget.

Total Isotropic Sensitivity (TIS)

TIS is the integral of receiver sensitivity over the sphere, measured by reducing the base-station signal until the device starts to lose packets at a defined error rate. It captures antenna efficiency on receive but also self-noise generated by the digital activity of the product: a switching DC-DC converter close to the receiver front-end can degrade TIS by 5, 10 dB even when the antenna is good.

CTIA Test Plan for Wireless Device Over-the-Air Performance

The CTIA Test Plan defines the methodology for TRP and TIS measurement, with anechoic chamber requirements, calibration procedures and head and hand phantom protocols. Version 3.x is the current reference as of 2024. It is used by US carrier acceptance programmes (AT and T, Verizon, T-Mobile) and broadly aligned with carrier OTA programmes worldwide. PTCRB OTA testing for cellular devices builds on the same chambers and procedures. For carrier-specific programmes see AT and T NAF cellular IoT, Verizon OPC and T-Mobile IoT lab.

3GPP TS 38.521 for cellular RF

For cellular NR (5G) devices, 3GPP TS 38.521 specifies the radio transmission and reception conformance tests, including OTA performance for FR1 and FR2 (mmWave). For LTE devices, 3GPP TS 36.521 is the equivalent. These tests are conducted by PTCRB-recognised laboratories and feed PTCRB module certification for North American operator acceptance.

GNSS front-end design

GNSS antennas have specific requirements that differ from the rest of the radio chain.

Passive versus active patches

A passive patch is a ceramic resonator with a feed pin and no integrated electronics. It is followed in the PCB front-end by an external LNA and, depending on the link budget, by a SAW filter.

An active patch integrates the LNA directly under the ceramic element. It powers up via a bias-tee on the RF coaxial cable. Active patches simplify the PCB layout and reduce noise figure but cost more and require careful DC supply filtering to avoid injecting noise.

LNA and SAW filter ordering

The two valid topologies, both used in commercial designs, differ on the trade-off between noise figure and linearity.

Topology	Order	Trade-off
LNA first	Patch, LNA, SAW, receiver	Best noise figure, lowest sensitivity loss, but LNA may saturate on nearby cellular emitters
SAW first	Patch, SAW, LNA, receiver	Best linearity and rejection, but higher noise figure due to filter insertion loss

A multi-radio product (GNSS plus LTE plus Wi-Fi plus BLE on one board) typically uses SAW-first to protect the LNA from the strong cellular bands at 700, 900, 1800 and 2100 MHz. A clean GNSS-only product can use LNA-first for the noise figure gain. The decision is documented in the link budget.

Ground plane and placement

The patch and its ground plane sit on the top side of the PCB, with the patch pointing toward the sky in the final mechanical position. Active components on the top side of the PCB within the patch footprint distort the pattern. The LNA is placed within a few millimetres of the patch feed.

MIMO and diversity in compact enclosures

Wi-Fi MIMO (2x2 or higher) and cellular diversity require two or more antennas that are decorrelated.

Decorrelation levers

Spatial separation: at least a quarter wavelength between antenna phase centres. At 2.4 GHz this is 30 mm, at 5 GHz 12 mm, at sub-GHz unrealistic on a small product.
Polarisation diversity: one antenna vertically polarised, one horizontally. A common layout is to use two orthogonal PCB IFAs at opposite corners.
Pattern diversity: each antenna oriented to cover a different part of the sphere.
Ground-plane shaping: slots and notches to break the common-mode currents that otherwise couple the two antennas through the ground plane.

Envelope Correlation Coefficient (ECC)

ECC is the figure of merit for MIMO diversity, with a target below 0.5 to deliver MIMO gain. It is derived from the 3D radiation patterns of both antennas measured in an anechoic chamber and is reported in the OTA measurement campaign. Above ECC 0.7, the second antenna brings no MIMO benefit and the product effectively operates in SISO with extra cost.

Filters: diplexers, duplexers, SAW and BAW

Multi-radio products use filters to share antennas, reject out-of-band emissions and protect receivers.

Filter type	Function
Diplexer	Combines two non-overlapping bands on a single antenna (e.g. 2.4 GHz BLE plus 5 GHz Wi-Fi)
Duplexer	Separates transmit and receive on a single antenna for FDD cellular, with high isolation
SAW filter (Surface Acoustic Wave)	Narrowband rejection, low insertion loss, used at frequencies up to 3 GHz
BAW filter (Bulk Acoustic Wave)	High-frequency rejection, lower temperature drift, used above 3 GHz and for high-power cellular front-ends

Out-of-band emission requirements under EN 300 328, FCC Part 15 and 3GPP TS 38.521 are typically the driver behind filter selection. Without proper filtering, harmonic emissions from a 2.4 GHz transmitter contaminate the 5 GHz and 6 GHz bands and the product fails radio compliance.

Recurring pitfalls

Pitfall	Consequence
Matching tuned on bare PCB without enclosure	Real efficiency 5, 15 dB below bench measurement, OTA failure
Ground-clearance violated by copper pour or vias	Catastrophic efficiency loss, antenna behaves as non-radiating resonator
Chip antenna placed at centre of board	Surrounded by ground plane, no radiation, link budget collapse
GNSS LNA before SAW saturating on cellular	Sensitivity degraded by 10, 20 dB, fix-time unacceptable in field
Generic 0603 inductors in the matching network	Self-resonance below the band, network behaves unpredictably
Hand effect ignored	Field efficiency 6, 10 dB below chamber result, drops calls
Missing ESD protection on antenna feed	LNA destroyed on first electrostatic strike, returns rate spikes
No regulatory-domain entry for target market	Antenna may radiate out of band, type approval refused

Going further

Bluetooth SIG qualification: RF + protocol qualification of the BLE radio behind the antenna
UWB FiRa consortium certification: UWB radio integration and ranging accuracy at 6, 8 GHz
Wi-Fi Alliance certification: Wi-Fi interoperability across 2.4, 5, 6 GHz
PTCRB module certification: cellular RF and OTA programme for North American operators
Glossary: definitions of S11, VSWR, TRP, TIS, ECC, SAW, BAW, PIFA, IFA