EMC Testing for Electronics: Emissions and Immunity

Illustration

EMC testing checks two things. Emissions tests measure the radio noise a device gives off, through the air and along its power cord. Immunity tests check that it keeps working when exposed to disturbances such as static discharge or radio fields. In the US, FCC Part 15 limits emissions from digital devices. The EU EMC rules cover both.

A failed lab run means a board change and a second booking, so EMC work starts at layout. This guide covers what EMC testing measures, the US and EU rules, the limits a lab scans against, the board features that drive radiated noise, and how to find problems on your own bench first.

An engineering summary, not legal advice. Figures and rules are as of November 2026, from the linked primary sources. We quote no limits from paywalled standards.

What is EMC testing, and what does it cover?

EMC stands for electromagnetic compatibility. A compatible device keeps its own radio noise low enough to leave other equipment alone, and keeps working when the noise around it is high. EMC testing checks both sides.

The European Commission puts it the same way. Its page on the EMC Directive says the directive limits electromagnetic emissions from equipment, and that it also governs the immunity of such equipment to interference (as of November 2026).

The test groups a small connected device is likely to face:

Sources: 47 CFR 15.107, 15.109, European Commission EMC Directive page and Radio Equipment Directive page, all as of November 2026.

EMI vs EMC: what is the difference?

EMI is electromagnetic interference: the unwanted noise itself. TI defines it as the disruption of an electronic device's operation caused by a field generated by a different device. EMC is the goal: a product that neither causes nor suffers from EMI. EMI testing usually means the emissions half of EMC testing.

The FCC words the US side in terms of radiators. In 47 CFR 15.3, an unintentional radiator generates radio frequency energy for use inside the device but is not intended to emit it. A digital device is an unintentional radiator that uses timing signals faster than 9,000 pulses per second and digital techniques (as of November 2026). Almost every AI wearable, tracker or voice device fits.

Section 15.3 also sets two classes. Class A devices are marketed for commercial, industrial or business use. Class B devices are marketed for residential use, such as personal computers and similar devices sold to the public. A consumer wearable is Class B.

What does EMC compliance require in the US and the EU?

United States

Under 47 CFR 15.101, other Class B digital devices and peripherals are authorized through a Supplier's Declaration of Conformity (SDoC) or certification. Under section 2.906, SDoC testing cannot be done at a lab owned, controlled or directed by a prohibited entity (both as of November 2026). For the US responsible party, prohibited entities and the rest of the process, read what FCC certification is and how it works.

The 15.103 exemptions are narrow, such as devices using no more than 6 nW. A connected AI device will rarely fit.

If the device also has a radio, the radio needs its own authorization under subpart C. How a module grant splits the work with subpart B is covered in FCC Part 15B vs 15C for a BLE wearable.

European Union

The EMC Directive 2014/30/EU was published on 29 March 2014 and repealed Directive 2004/108/EC from 20 April 2016. The Commission says harmonized standards published in the Official Journal provide a presumption of conformity. For radio products, the Radio Equipment Directive sets essential requirements that include electromagnetic compatibility (both as of November 2026). For how the FCC and EU processes differ, read FCC and CE certification for radio devices.

Which standard does the lab apply? The IEC's CISPR guide (January 2024) lists CISPR 32 for emission requirements of multimedia equipment and CISPR 35 for its immunity requirements. Your lab picks the exact harmonized standards and editions. Those standards are sold by the standards bodies, so this guide quotes no EU limit values.

How is EMI testing done, and what limits apply?

Section 15.31(a)(4) names ANSI C63.4-2014, minus a few listed clauses, as the measurement procedure for unintentional radiators. When the device provides for external accessories, 15.31(i) requires testing with them attached, configured to maximize emissions within normal operating conditions (as of November 2026). So the lab tests your device with its real cable, in its noisiest normal mode.

Radiated limits

Section 15.109(a) sets these limits at 3 meters for unintentional radiators other than Class A digital devices. The dBµV/m column is our conversion of the rule's µV/m values.

Source: 47 CFR 15.109(a) and (b), as of November 2026. The tighter limit applies at band edges, per 15.109(c).

How high does the scan go? Section 15.33(b)(1) ties it to the highest frequency generated or used in the device, or on which it operates or tunes. If that is between 108 and 500 MHz, the scan runs to 2000 MHz. If it is above 1000 MHz, the scan runs to the 5th harmonic or 40 GHz, whichever is lower (as of November 2026).

Detectors matter too. Section 15.35 bases the limits up to 1000 MHz on a CISPR quasi-peak detector. The responsible party may instead show compliance with a peak detector, using the same measurement bandwidths as the quasi-peak method. Above 1000 MHz the limits use an average detector, and peak emissions may not exceed the average limit by more than 20 dB (as of November 2026).

Conducted limits

For equipment designed to connect to the AC power line, other than Class A digital devices, section 15.107(a) limits the voltage put back on the line, measured through a 50 µH/50 ohm line impedance stabilization network. From 0.5 to 5 MHz the limits are 56 dBµV quasi-peak and 46 dBµV average. From 5 to 30 MHz they are 60 and 50. From 0.15 to 0.5 MHz they fall with the logarithm of frequency, from 66 to 56 quasi-peak and 56 to 46 average (as of November 2026). Whether a battery device needs this test depends on how it charges (see the FAQ).

What usually fails EMI testing in small battery devices?

TI's application note AN-2155 (SNVA638A, revised April 2013) gives a simple model for radiated noise from a small current loop. The field rises with frequency squared, with loop area and with current, and falls with distance (as of November 2026). The sources in the table all fit that model. TI's measurements came from eight layout variants of one board built around its LM3102, a 1 MHz, 2.5 A step-down regulator, so treat its numbers as direction for your design.

TI's measured results, as of November 2026: a ground plane added 4 dBµV/m of margin against the CISPR 22 limit. An input filter on a buck powered by short wires from a battery cut radiated EMI by up to 20 dB. A shielded inductor helped near and far field. Slower switching edges also lower EMI, at the cost of switching loss.

Fix loop area and return paths first. In TI's tests, rounding trace corners moved the margin by less than 1 dBµV/m.

How do you run pre-compliance scans before the lab?

Pre-compliance means scanning your own boards with simpler equipment. A bench setup will not match an accredited chamber, so treat its numbers as a trend.

• Scan the first powered boards with a near-field probe and a spectrum analyzer. Find which part of the board each peak comes from.
• Run each mode: idle, recording, streaming, charging. Give the lab the noisiest one.
• Attach every cable and accessory that ships in the box. The FCC rule tests with them attached, so your scan should too.
• Scan with a peak detector at the quasi-peak bandwidths. Section 15.35 lets the responsible party show compliance with one up to 1000 MHz.
• Agree in advance on the margin below the limit you want to see before you book lab time.
• Repeat the scan after any board or firmware change that touches clocks, regulators or radio duty cycle.

For the EU, plan immunity tests on the same samples. For static discharge, the next read is ESD testing to IEC 61000-4-2.

At Duobay, FCC, lab and module rules are planned from Gate 0 Feasibility. See what Gate 0 Feasibility covers.

What should you ask your hardware team before EMC testing?

Bring this list to the layout review and again before you book a lab.

• Which FCC device class do we market under, and does the device also contain a radio that needs its own grant?
• What is the highest frequency used on the board, and how high will the lab scan?
• Can the device run while charging? If yes, who books the conducted emissions test?
• Which cables ship in the box, and will the lab test with each attached?
• What margin below the limit will we accept in pre-compliance before we book the lab?
• If we use SDoC, who is our US responsible party, and has the lab confirmed it is not owned or controlled by a prohibited entity?
• If we will sell in the EU, which harmonized standards does the lab plan to apply for emissions and immunity?

If you are planning a device and its FCC testing, tell us about your device.

Sources. All pages accessed November 6, 2026. Rules and standards are revised, so confirm current text before you rely on it.

• 47 CFR 15.3, definitions (eCFR)
• 47 CFR 15.31, measurement standards (eCFR)
• 47 CFR 15.33, frequency range of radiated measurements (eCFR)
• 47 CFR 15.35, measurement detector functions (eCFR)
• 47 CFR 15.101, equipment authorization of unintentional radiators (eCFR)
• 47 CFR 15.103, exempted devices (eCFR)
• 47 CFR 15.107, conducted limits (eCFR)
• 47 CFR 15.109, radiated emission limits (eCFR)
• 47 CFR 2.902, 2.906 and 2.1077, prohibited entities, SDoC and compliance information (2.902, 2.906, 2.1077)
• European Commission, Electromagnetic Compatibility (EMC) Directive (link)
• European Commission, Radio Equipment Directive (RED) (link)
• IEC CISPR, Guidance for users of the CISPR Standards (January 2024) (link)
• Texas Instruments, AN-2155 Layout Tips for EMI Reduction in DC/DC Converters, SNVA638A (April 2012, revised April 2013) (link)