What Is EMC? Emission, Immunity and Coupling Paths Explained

Electromagnetic compatibility (EMC) is the ability of equipment to function satisfactorily in its electromagnetic environment without introducing intolerable disturbances to anything in that environment. That definition, from the International Electrotechnical Vocabulary (IEV 161-01-07), has two halves, and every EMC requirement you will meet maps onto one of them. This article explains the two halves, the source–path–victim model used to analyse any interference problem, and the steps a product goes through before it can be sold in markets such as the EU, the United States, Korea and Japan.

Emission and immunity: the two halves of EMC

Emission requirements limit what a product puts out: noise conducted onto its cables and fields radiated from its enclosure and cables. Immunity requirements (often written EMS, for electromagnetic susceptibility, in Asian markets) check that the product keeps working when exposed to disturbances it can reasonably expect: electrostatic discharge from a user, RF fields from a nearby radio, fast transients from switched loads and surges from lightning on the mains. A product can pass one and fail the other, so the two are designed for and tested separately.

Immunity results are not numbers but judgements. Criterion A means normal operation within specification during and after the test. Criterion B allows temporary degradation during the test as long as the EUT recovers by itself afterwards. Criterion C allows a temporary loss of function that needs operator action or a restart, provided the function can be restored. The product standard says which criterion applies to each test, but the manufacturer must define what normal operation means for its product, and that definition belongs in the test plan before testing starts.

The source–path–victim model

Every interference problem has three elements: a source that generates the disturbance, a coupling path that carries it, and a victim whose function is affected. Remove or weaken any one of them and the problem disappears. The model works in both directions. In an emission test your product is the source and the receiving antenna or LISN stands in for the victim. In an immunity test the generator is the source and your product is the victim.

Four coupling paths

These effects are easy to underestimate. A stray capacitance of only 1 pF next to a node switching at 10 V/ns injects 10 mA while the edge lasts. A mutual inductance of 5 nH with a current edge of 0.1 A/ns induces 0.5 V. The near-field boundary λ/2π is about 0.48 m at 100 MHz, so inside a typical enclosure most coupling is near field, either capacitive or inductive, while the cables leaving the box are usually what radiates to the test antenna.

Where to break the chain

Fixing the source is usually the cheapest per unit and the most effective: slower edges where timing allows, spread-spectrum clocking, smaller current loops. Fixing the path comes next: filters at the boundary where cables leave, shielding, layout separation. Hardening the victim with input filtering, software plausibility checks and watchdogs is essential for immunity but does nothing for an emission failure. A ferrite clamp added to a cable at the end of a project treats the path; it often works, but it adds cost to every unit built.

From design to market: the EMC steps

A typical route for an electronic product looks like this:

• Define the markets and the product category. The category decides the standards: multimedia and IT equipment uses CISPR 32 for emission and CISPR 35 for immunity, industrial, scientific and medical (ISM) equipment uses CISPR 11, household appliances and tools use CISPR 14-1 and CISPR 14-2, and lighting uses CISPR 15. In the EU these appear as EN versions, such as EN 55032 and EN 55035.
• Design for EMC from the first schematic: ground planes and stack-up, filter footprints at every connector, enclosure seams and cable entry. Changes after tooling or certification cost far more.
• Run pre-compliance checks: a spectrum analyzer with near-field probes and a current probe on the cables, a LISN for conducted emission, and an ESD generator if one is available. Results are relative rather than absolute, so aim for margin; many teams target 3 to 6 dB below the limit.
• Write a test plan: operating modes that exercise the noisiest and most sensitive functions, cable lengths and layout, support equipment, and the pass/fail definition for each immunity test.
• Test at an accredited laboratory (ISO/IEC 17025) and keep the report together with the exact hardware and firmware versions tested.
• Complete the conformity route of each market and keep it valid when the design changes.

Radiated emission limits are tied to a specific measurement distance, usually 3 m or 10 m. Pre-compliance setups often measure closer, and converting with the 20·log(d) rule is only an approximation; our article on 3 m to 10 m limit conversion explains when it fails. The Limit Margin & Uncertainty Check calculator helps decide whether a small margin is really a pass.

Common misconceptions

• "Passing emission means the product is EMC-compliant." Immunity is a separate set of tests, and ESD and fast transients often reveal firmware lock-ups that emission testing never shows.
• "EMC is something you test at the end." By then the PCB and enclosure are frozen, and fixes become ferrites, gaskets and shielding cans that add cost to every unit.
• "Ground is ground." At high frequency a ground wire is an inductor. With the common rule of thumb of about 1 nH per mm, a 10 cm wire has roughly 100 nH, which is about 63 Ω at 100 MHz.
• "A metal enclosure solves everything." An unfiltered cable passing through the wall carries noise straight out, and a long seam or slot can leak as much as an open hole.
• "Our pre-scan passed by 1 dB, so we are fine." Radiated emission results typically differ by several dB between setups and laboratories, and production units vary. A small margin is a risk, not a pass.
• "One good sample proves compliance." Conformity applies to every unit placed on the market, so component substitutions, PCB revisions and even firmware changes that alter clock or switching behaviour need a review.