Key Points on Conducted Emission Spectrum Analysis

Conducted emission testing is a fundamental pillar of Electromagnetic Compatibility (EMC) evaluation. It measures the electromagnetic energy unintentionally propagated by a device through its power cords or signal lines. However, merely acquiring test data is only the preliminary step; the real challenge—and the key to successful mitigation—lies in performing an in-depth spectral analysis of the emissions to pinpoint root causes.

A rigorous analysis begins with a thorough understanding of the regulatory framework and test limits. Conducted emissions are typically evaluated across a defined frequency spectrum:

150 kHz to 30 MHz: Mandated by major commercial standards such as CISPR 32 and CISPR 22, this frequency range is where most conducted disturbances manifest as either differential-mode or common-mode noise.
Quasi-Peak (QP) and Average (AV) Limits: Standards enforce dual-limit criteria. The QP detector simulates the human ear's response to pulse-like interference, whereas the AV detector evaluates narrowband continuous noise. Passing compliance requires test waveforms to stay comfortably beneath both threshold curves.

The primary objective during initial spectrum review is to map out failing frequencies against these limit curves, determining whether violations stem from broadband or narrowband sources. This distinction fundamentally dictates the subsequent engineering strategy.
A core competency in spectral interpretation is separating narrowband emissions from broadband disturbances, as their generation mechanisms and suppression techniques are entirely distinct.

Narrowband Interference: Appears on the spectrum as distinct, isolated spikes at specific frequencies. These are typically generated by single-frequency oscillators, such as a microcontroller's crystal frequency or the fundamental switching frequency of a power supply and its associated harmonics.
Broadband Interference: Characterized on the display as an overall elevated noise floor or a dense cluster of high-frequency grass across a broad span. This behavior is commonly driven by random noise, reverse-recovery transients of rectifier diodes, or simultaneous switching noise (SSN) in dense digital logic.

Accurately identifying the noise typology guides component selection—such as deploying a tuned LC trap for a stubborn narrowband spike versus applying a high-frequency ferrite core for broadband dampening.

Isolating Common-Mode and Differential-Mode Noise

Within the 150 kHz to 30 MHz window, conducted noise propagates via two distinct paths. While telling them apart purely from a raw spectrum graph can be challenging, engineers can deduce their origins using contextual clues:

Differential-Mode (DM) Noise: Flows between the line (L) and neutral (N) conductors. It is primarily driven by current ripple from power supply capacitor charging/discharging and voltage drops across equivalent series resistance (ESR). DM noise predominantly dominates the lower end of the spectrum (150 kHz to a few MHz).
Common-Mode (CM) Noise: Flows between the current-carrying conductors and the protective earth (PE) ground. It is typically excited by parasitic capacitive coupling between fast-switching power devices (like MOSFET heat sinks) and chassis ground. CM noise usually dominates the upper end of the spectrum (above a few MHz).

Diagnostic Rule of Thumb: Low-frequency failures generally point to shortcomings in X-capacitors and differential choke designs. Conversely, high-frequency failures demand a rigorous review of Y-capacitors, common-mode chokes, and transformer shielding.

Spectral Signature Analysis

Specific patterns on an EMC spectrum often act as a direct fingerprint pointing toward vulnerable circuit nodes.

Harmonic Sequences of Switching Frequencies: Seeing regular spikes—such as a fundamental at 65 kHz accompanied by harmonics at 130 kHz, 195 kHz, and so on—indicates the primary PWM controller. If these harmonics decay too slowly across the spectrum, the switching edges are unnecessarily aggressive, signaling a need to adjust gate-drive resistors.
Clock Multipliers: Spikes appearing precisely at standard digital clock intervals (e.g., 12 MHz, 24 MHz) reveal impedance discontinuities in power or ground planes, allowing internal digital noise to couple out through external cabling.
Low-Frequency Floor Elevation: A generalized rise in the lower frequency spectrum frequently points to severe ringing caused by diode reverse recovery, often linked to substandard bridge rectifiers or bulk electrolytic capacitors.

A Practical Diagnostic Case Study

Consider a scenario where a power adapter fails its conducted emission scan, exhibiting a Quasi-Peak violation exceeding the limit by 6 dB at the 500 kHz mark on the Line (L) conductor.

Spectral Signature: The 500 kHz point appears as a sharp, isolated peak, with a much weaker second harmonic visible at 1 MHz. This points directly to narrowband interference.
Frequency Correlation: The adapter's internal PWM switching frequency is 100 kHz, making 500 kHz its exact fifth harmonic.
Mode Deduction: Because 500 kHz resides in the lower frequency band, differential-mode noise is suspected as the primary culprit.
Mitigation Strategy: The engineering team places a larger X2 safety capacitor in parallel directly after the bridge rectifier and increases the inductance of the differential choke to lower low-frequency impedance.
Verification: Re-running the compliance scan shows the QP level at 500 kHz dropping by 8 dB, comfortably clearing the regulatory limit.

Conclusion

Conducted emission spectrum analysis bridges rigorous electromagnetic theory with hands-on hardware engineering. By mastering the ability to distinguish between narrowband and broadband phenomena, separating common-mode and differential-mode pathways, and decoding spectral signatures, engineers can bypass guesswork-driven troubleshooting. Integrating these analytical steps with tools like near-field probes and LISN-based mode separation techniques will consistently yield faster, more reliable EMC compliance.