CE testing requirements (EMC, safety, radio)

Author Hugues Orgitello Last updated 25 May 2026

CE · Pillar

Tests are the technical materialisation of conformity. For a Wi-Fi/BLE-connected electronic product, count on three to five days of EMC lab time, as much for radio, and several weeks of in-house pre-tests to avoid surprises. This page details what is expected, in what order, with what accreditation, and how to organise it without blowing up the budget or schedule.

Overview: three test families, one common logic

For a typical electronic product, three families of tests structure the whole campaign:

Electromagnetic compatibility (EMC), Directive 2014/30/EU, verifies that the product does not emit too much, and that it withstands the normal disturbances of its environment.
Electrical safety (LVD), Directive 2014/35/EU, verifies that the product presents no electrical, mechanical, thermal or chemical hazard.
Radio (RED), Directive 2014/53/EU, if the product intentionally transmits or receives radio waves, verifies safety, radio EMC, efficient use of spectrum, and (since 2025) cybersecurity.

Each family follows the same underlying logic: select the applicable harmonised standards (see Standards), perform the tests per their procedures, document results with full metrological traceability, and retain the report in the technical file for 10 years.

Electromagnetic compatibility (EMC)

Directive 2014/30/EU imposes two families of requirements: emissions (the product must not disturb its electromagnetic environment) and immunity (the product must function correctly when exposed to normal disturbances).

Emissions

Emission tests break down into four families:

Type	Method	Frequency range	Limit (Class B / residential)
Conducted emissions	Measured at the mains via LISN	150 kHz – 30 MHz	56–60 dB(µV) average, 66–70 dB(µV) quasi-peak
Radiated emissions	Biconical + log-periodic antenna at 3 or 10 m, semi-anechoic chamber	30 MHz – 1 GHz	30 dB(µV/m) at 30 MHz decreasing to 37 dB(µV/m) above 230 MHz
Radiated emissions above 1 GHz	Horn antenna at 3 m	1 GHz – 6 GHz (Wi-Fi), up to 18 GHz depending on product	Per-band limits, typically 50–70 dB(µV/m)
Current harmonics	Precision current clamp on neutral	DC – 2 kHz (40 harmonics)	EN 61000-3-2 tables by Class A, B, C or D
Flicker	Flicker analyser, simulated network load	Voltage variations	P_st ≤ 1.0; P_lt ≤ 0.65

Class A vs Class B: Class B (residential) is 10 dB to 20 dB stricter than Class A depending on frequency. Class A (industrial) is more permissive but requires an explicit warning to the user in the manual. For an IoT product likely to be used in residential or tertiary environments, always target Class B by default. Declaring Class A on a consumer product is an almost systematic ground for rejection in market surveillance.

A few classic emission pitfalls:

The mains cable acts as antenna. At 30 MHz, a 1 m cable is a quarter wavelength, it efficiently radiates common-mode noise. An input mains filter and careful PCB layout around the switching supply are indispensable.
The microcontroller clock and its harmonics typically appear at 16, 24, 48, 96 MHz and multiples. If the crystal hits a strict limit, shifting it by a few MHz sometimes suffices to pass.
LCD or OLED displays often emit strongly between 100 and 300 MHz (differential drive signals). Local shielding or series filtering on data lines is the antidote.

Immunity

Immunity covers the disturbing phenomena the product must withstand without major malfunction:

Phenomenon	Standard	Typical level (residential)
Electrostatic discharge (ESD)	EN 61000-4-2	± 8 kV contact, ± 15 kV air
Radiated RF electromagnetic field	EN 61000-4-3	3 V/m from 80 MHz to 1 GHz; 3 V/m from 1.4 to 6 GHz
Electrical fast transients (EFT/burst)	EN 61000-4-4	± 2 kV on power supply, ± 1 kV on I/O
Surge	EN 61000-4-5	± 2 kV differential, ± 4 kV common mode
Conducted RF	EN 61000-4-6	3 V (3 mA equivalent current) from 150 kHz to 80 MHz
50 Hz magnetic field	EN 61000-4-8	3 A/m continuous, 30 A/m transient (3 s)
Voltage dips and interruptions	EN 61000-4-11	Dips at 0 % (10 ms), 40 % (200 ms), 70 % (500 ms), interruption 5 s
Pulse bursts (at 100 kHz)	EN 61000-4-18	± 1 kV (industrial only)
Pulsed and oscillatory magnetic fields	EN 61000-4-9 / -10	Industrial only

For each test, the performance criterion must be defined before the lab campaign:

Criterion A: the product operates normally, within specified tolerances, during the disturbance.
Criterion B: temporary acceptable degradation, automatic return to nominal operation after the disturbance.
Criterion C: manual intervention required for recovery (e.g. user reset).
Criterion D: unacceptable, data loss, hardware damage, potential injury.

The target criterion depends on the product's use. An industrial temperature sensor generally accepts Criterion B on ESD; a medical device requires Criterion A on most phenomena. The test plan (TPP, Test Performance Plan) must list phenomenon by phenomenon the chosen criterion, its justification, and the operation-monitoring method during the test.

Typical test bench and equipment costs

For in-house pre-tests, the minimum equipment is:

Equipment	Function	Order of magnitude (excl. VAT, new)
Spectrum analyser 9 kHz – 6 GHz	Conducted + radiated emissions	€12,000 – €35,000
LISN 50 Ω / 50 µH single-phase 16 A	Conducted emissions	€2,500 – €5,000
ESD gun ± 30 kV	ESD immunity	€6,000 – €12,000
EFT/burst + surge generator	Conducted immunity	€18,000 – €35,000
Biconical + log-periodic antennas	RF measurement and injection	€3,000 – €8,000
Mini shielded enclosure (1 m × 1 m × 2 m)	Ambient isolation	€25,000 – €60,000
Annual metrological calibration	ISO 17025 traceability	€3,000 – €8,000/year

The good cost/effectiveness ratio for a moderate-volume manufacturer: invest in the analyser + LISN + ESD gun (≤ €50,000), and subcontract the rest. That covers 80 % of defects detectable before the external lab.

Electrical safety (LVD)

Directive 2014/35/EU applies to voltages from 50 to 1,000 V AC and 75 to 1,500 V DC. Below these thresholds the product is not covered by LVD (but safety remains assessed via the essential requirements of other directives, notably RED 3.1(a) for radio equipment).

Principal standard and the HBSE approach

EN 62368-1 has been the reference standard since it replaced EN 60065 (audio/video) and EN 60950-1 (IT) on 20 December 2020. Its distinguishing feature is the HBSE approach (Hazard-Based Safety Engineering): instead of design prescriptions, it defines energy classes and exposure classes, and imposes protection barriers based on the intersection of the two.

Energy sources are classified into three levels:

Class	Electrical energy (typical limits)	Consequence on barriers
ES1 (low)	< 30 V AC / 60 V DC continuous, < 2 W	Accessible to ordinary persons, no special protection
ES2 (moderate)	30–60 V AC / 60–120 V DC, < 100 W	Accessible to instructed persons; basic insulation required for ordinary access
ES3 (high)	> 60 V AC / 120 V DC or > 100 W	Skilled persons only; reinforced insulation

The approach extends to thermal energy (ETS1/2/3), mechanical (MS1/2/3), chemical (CS), radiation (RS) and fire (PIS1/2/3).

Key LVD tests

Dielectric strength: test voltage of 1,500 V AC for 1 minute between primary and secondary (basic insulation), 3,000 V AC for reinforced. Pass: no breakdown, leakage current < 5 mA.
Insulation resistance: measured at 500 V DC for 60 s. Pass: R > 2 MΩ basic, R > 4 MΩ reinforced.
Leakage current: measured between chassis and earth at 110 % of nominal voltage. Typical limit: 0.25 mA for Class II, 3.5 mA for Class I with earth.
Heating: test in normal conditions and single-fault conditions (component short or open); temperature measured by K-type thermocouples on all sensitive components. Limits derive from critical-component datasheets (often T_max = 100 to 130 °C for consumer semiconductors).
Overload and foreseeable anomalies: replacing a fuse with a wire, blocking a fan, shorting an output. Pass: no fire, no smoke, no ejection of incandescent particles.
Mechanical stability: drop test, impact, vibration, access to hazardous parts (IPxD probe).
Markings and instructions: verification of CE marking, manufacturer marking, warning markings, and consistency with the user manual.

Specific application cases

Product family	Main harmonised standard
Audio/video, IT, telecoms, instrumentation	EN 62368-1
Measurement and laboratory equipment	EN 61010-1 + specific parts
Household appliances	EN 60335-1 + parts 60335-2-XX
Medical electrical devices	EN 60601-1
Luminaires	EN 60598-1 + parts 60598-2-XX
Portable power tools	EN 62841-1

Radio equipment (RED)

When the product intentionally transmits or receives radio waves, the RED directive adds to the two above. Tests break down into four articles:

Article 3.1(a): Health and safety

Includes two main families:

SAR (Specific Absorption Rate) for equipment worn within 20 cm of the body. Measured in W/kg averaged over 10 g of tissue for the trunk, 1 g for the head. Limits:

Trunk / limbs: 4.0 W/kg (EU): 2.0 W/kg (medical rigour)
Head: 2.0 W/kg (EU)

Method defined by EN 50360 (phones, head), EN 50566 (worn equipment), EN 62209-1 / -2 / -3 (general and matrix methods).

Generic electromagnetic compatibility of health aspects (radiative fields close to user, prolonged exposure): covered by ICNIRP standards transposed at European level.

Article 3.1(b): Radio electromagnetic compatibility

Covered by the EN 301 489 series, comprising a generic part -1 and specific parts -X by radio type. The most common:

Part	Coverage
EN 301 489-1	Common requirements for all radio equipment
EN 301 489-3	SRD ≤ 1 GHz (LoRa, Sigfox 868 MHz, remote controls...)
EN 301 489-17	Wi-Fi 2.4/5/6 GHz and Bluetooth
EN 301 489-19	GNSS receivers (GPS, Galileo)
EN 301 489-50	Cellular base stations
EN 301 489-52	Cellular user equipment
EN 301 489-53	UWB

The methodology differs slightly from generic EMC: radio immunity tests are performed in both transmit AND receive modes, with a performance criterion defined per band.

Article 3.2: Efficient use of spectrum

Standard specific to the radio type. For the most common cases:

Band	Standard	Main limits	Specifics
2.4 GHz Wi-Fi/BLE	EN 300 328	EIRP ≤ 20 dBm, occupancy < 10 % per 50 ms	Adaptive frequency hopping or channel detection for BLE
5 GHz Wi-Fi (5150–5350)	EN 301 893	EIRP 23 dBm	TPC (Transmit Power Control) required
5 GHz Wi-Fi (5470–5725)	EN 301 893	EIRP 30 dBm	DFS (Dynamic Frequency Selection) required, avoid weather radars
6 GHz Wi-Fi (5945–6425)	EN 303 687	EIRP 23 dBm (LPI), 36 dBm (VLP outdoor)	New EU band since 2021
Sub-GHz SRD 868 MHz	EN 300 220	25 mW EIRP, duty cycle 0.1 %, 1 % or 10 % per sub-band	863–870 MHz sub-bands precisely divided
Sub-GHz SRD 433 MHz	EN 300 220	10 mW EIRP, duty cycle 10 %	433.05–434.79 MHz sub-bands
Cellular 2G/3G/4G/5G	EN 301 511 / EN 301 908	3GPP profiles	Tests against simulated base stations
LoRa / Sigfox 868 MHz	EN 300 220	Channel and duty-cycle limits	G3 sub-band typically for LoRa
UWB 6–8 GHz	EN 302 065	−41.3 dBm/MHz EIRP	Very short range, high precision
Z-Wave 868 MHz	EN 300 220 (G3 sub-band)	25 mW EIRP	Duty cycle 1 %

For each band, the radio test report lists: tested frequencies, output power EIRP and conducted, spectral occupancy, out-of-band emission mask, spurious emissions, frequency tolerance, occupancy duration, interference behaviour (LBT/AFA/AFH depending on standard).

Article 3.3: Specific requirements

Article 3.3 activates particular requirements per Delegated Regulation (EU) 2022/30, applicable since 1 August 2025. Three sub-articles concern the cybersecurity of nearly all connected radio products:

3.3(d), protection of the network functionalities the equipment is connected to (do not degrade the network, do not act as an attack relay).
3.3(e), protection of the user's personal data and privacy (encryption of sensitive communications, access control).
3.3(f), protection against fraud (payment integrity, authentication).

Standards EN 18031-1 (3.3(d)), EN 18031-2 (3.3(e)) and EN 18031-3 (3.3(f)) harmonised in 2024 cover these requirements. The evaluation methodology includes:

Analysis of the product's security architecture
Verification of control implementation (authentication, signed updates, hardening, secret management)
Review of user documentation (security notice, support duration)
Targeted penetration testing (per the chosen assurance level)

This is the most expensive test to add to an existing product, and the least mastered by labs in 2026.

Accreditation: what can be done in-house, what must go out

The classic manufacturer mistake is to believe everything must be done in an external lab. The actual rule is more nuanced, the CE directive nowhere requires a third party, except for assessment modules with a Notified Body (see Procedure).

Test	In-house feasible?	Technical conditions
Radiated EMC emissions	Not practically	Calibrated semi-anechoic chamber, turntable antenna, distance ≥ 3 m
Conducted emissions	Yes	LISN + spectrum analyser + shielded chamber ≥ 1 m³
ESD immunity	Yes	ESD gun calibrated annually, reference ground plane
Radiated RF immunity	Not practically	Anechoic chamber + 100 W power amplifier
Surge / EFT	Yes	Surge generator, CDN couplings
Safety (dielectric strength, insulation)	Yes	5 kV dielectric tester, megohmmeter
Heating	Yes	Climatic chamber, K-type thermocouples
Stability, impact, vibration	Equipment-dependent	Calibrated drop test, vibration table
Radio (ETSI tests)	Not practically	Accredited lab, RF anechoic chamber, specialised analysers
SAR	No	Robotic SAR system with dosimetric probe, human models

In-house tests must be documented with the same rigour as external ones:

Written procedures: detailed protocol, equipment used, test conditions, success criteria.
Calibrated equipment: annual or biannual calibration by a COFRAC or equivalent ILAC accredited lab, certificates archived.
Metrological traceability: each measurement traceable to a national standard.
Signed reports: operator, date, results, photos of the setup, raw data in annex.

Common practice is to do in-house pre-tests to clear gross issues, then move to an external lab for the final tests that go into the official file.

Criteria for selecting an external lab

A credible lab ticks at least four boxes:

ISO/IEC 17025 accreditation for the relevant tests, issued by an ILAC member body (COFRAC in France, DAkkS in Germany, UKAS in the UK, NABL in India...). The list of accredited tests is annexed to the accreditation certificate and must be explicitly checked: a lab may be accredited for EN 55032 but not for EN 61000-3-2.
Frequency and phenomena range covering your need without extrapolation. Check case by case, especially above 6 GHz for Wi-Fi 6E/Wi-Fi 7 radios.
Actual availability, labs are often saturated; 4 to 8 weeks of lead time is normal in western Europe in 2026.
MRA designation for radio tests if you target multiple markets. A lab recognised MRA can produce reports usable for US FCC and Canadian ISED without retesting. A precious time-saver for international IoT products.

Test campaign timeline for an IoT product

Here is a realistic chronology for a Wi-Fi/BLE-connected mains-powered sensor:

Week -8 to -6 : In-house EMC pre-tests (conducted emissions, basic ESD)
Week -5 to -4 : Design fixes (filtering, shielding, layout)
Week -3       : Book external lab, ship samples (× 3)
Week -2 to -1 : Draft test plan (TPP), list performance criteria
Week 0        : Lab intake, visual inspection, chamber placement
Week +1       : Emissions tests (conducted + radiated + harmonics)
Week +2       : Immunity tests (ESD, EFT, surge, radiated RF, conducted RF)
Week +3       : Radio Article 3.2 tests (EIRP, occupancy, mask)
Week +4       : LVD safety tests (dielectric, insulation, heating)
Week +5       : Receive preliminary report, internal review
Week +6       : Fixes if needed + partial retests
Week +7       : Signed final report, file archival

Plan a further 20 to 40 % margin for retests after modifications. An EMC campaign clearing in two iterations is typical; three or four iterations is the bad case.

Indicative costs

These orders of magnitude should be validated with each lab; they can vary by a factor of 2 depending on product complexity, country, and lab load at booking time.

Campaign	Range (excl. VAT)
Full residential EMC (Class B)	€4,000 – €12,000
Industrial EMC (longer durations, reinforced immunity)	€8,000 – €20,000
EN 62368-1 LVD safety	€5,000 – €15,000
EN 60335-1 LVD safety (household)	€8,000 – €25,000
Radio 2.4 GHz EN 300 328	€6,000 – €15,000
Radio 5 GHz EN 301 893	€8,000 – €18,000
Cellular LTE / 5G radio	€15,000 – €60,000
Cybersecurity EN 18031-1/2/3	€8,000 – €25,000
SAR (worn equipment)	€4,000 – €10,000
Partial retest after modification	€1,000 – €5,000

Also factor in associated costs:

Sample shipping and return (often €200–500 in international express)
On-site personnel to attend tests (recommended for the first campaign, indispensable for radio tests)
Sample reproduction in case of failure during testing (labs typically request 3 units)

Common test-campaign mistakes

Three failure patterns recur in the campaigns we see:

1. Under-investing in pre-tests. Arriving at an external lab with a product that has never seen a spectrum analyser is a guarantee of at least two iterations. In-house pre-tests, even rough, detect 80 % of defects at marginal cost.

2. Defining performance criteria badly. The test plan (TPP) must be written before the lab, not during it. Once the chamber is closed and the clock is billed by the hour, modifying criteria is expensive.

3. Confusing test report with accredited report. A lab can produce an "ISO 17025-style" report without being accredited for the specific test concerned. Verify the explicit accreditation scope before citing the report in the file, otherwise, in audit, the test will be considered non-accredited and technical equivalence will need to be justified.