EMC: What fails first in the lab

EMC issues rarely come from “mystery noise.” They usually come from predictable design choices: grounding, cabling, enclosure seams, and I/O filtering.

Failure pattern #1: Cable + return path mismatch

Long cables turn into antennas. If the return path is uncontrolled (or forced through a noisy area), emissions rise and immunity gets fragile.

• Define cable types, lengths, and routing early (and test that configuration).
• Control return paths for high-speed and switching currents.
• Avoid “incidental” grounding through random fasteners or paint.

Failure pattern #2: Shielding assumptions that don’t hold

“Metal enclosure = shielded” is not enough. Seams, vents, coatings, and gaskets decide reality.

• Check seam continuity (especially around doors/plates).
• Validate conductive coatings and bonding points.
• Plan for apertures: displays, vents, connector cutouts.

Failure pattern #3: I/O filtering placed too far from the boundary

If filters aren’t at the interface, the cable/harness can inject noise into internal circuits.

• Put protection/filtering at the boundary (connector side), not “somewhere on the board.”
• Keep filter grounds short and low impedance.
• Separate “dirty” and “clean” zones with clear layout rules.

Simple lab-ready checklist

• Worst-case operating modes are defined (max load, PWM, radios, motors, switching).
• Final-ish harness/cables are ready (not dev jumpers).
• Grounding strategy is documented (what bonds to chassis/PE and where).
• Enclosure bonding points are intentional (not accidental).