Electronics Guide

Field Solvers and Their Domains

EMC-relevant solvers fall into a few families, each suited to a different question:

• Full-wave 3D solvers: Method-of-moments, finite-element, and finite-difference time-domain (FDTD) tools model radiation from board geometry, cables, and enclosures. They are accurate but computationally expensive, so they are reserved for specific structures rather than whole systems.
• 2D and 2.5D field solvers: Fast extraction of trace impedance, crosstalk, and plane behavior, used to control the high-frequency content that ultimately drives emissions.
• Circuit and SPICE-class simulation: Time-domain analysis of switching edges, ringing, and filter response, which sets the source spectrum that every downstream emission inherits.
• Cable and harness models: Common-mode current estimation on attached cables, frequently the dominant radiator in a real product.

What Simulation Predicts Well, and What It Does Not

Simulation is strongest at relative comparison and mechanism identification, and weakest at absolute prediction. It reliably shows whether a guard trace, a stitching-via pattern, or a slower edge rate reduces a problem, and it exposes resonances and current paths that are hard to see on the bench. It is far less reliable at predicting an exact field strength in dBµV/m, because real emissions depend on cable dress, enclosure seams, and ground paths that are difficult to model completely. The practical workflow treats simulation as a way to rank design options and to form hypotheses that bench measurement later confirms.

Resolution Bandwidth and the CISPR Bands

EMC limits are defined for specific measurement bandwidths, specified at the −6 dB points rather than the −3 dB points common in ordinary spectrum analysis. Using the wrong bandwidth shifts broadband readings and makes results incomparable to the limit line. CISPR 16-1-1 divides the spectrum into four bands:

• Band A (9 kHz to 150 kHz): 200 Hz measurement bandwidth.
• Band B (150 kHz to 30 MHz): 9 kHz measurement bandwidth — the conducted-emissions range.
• Band C (30 MHz to 300 MHz): 120 kHz measurement bandwidth.
• Band D (300 MHz to 1 GHz): 120 kHz measurement bandwidth — bands C and D together cover most radiated-emissions testing.

A common pre-compliance error is leaving the analyzer in auto-bandwidth mode; setting the bandwidth manually to the value for the band under test is the first step toward correlated results.

Detectors and What They Reveal

The choice of detector changes the reading even when the signal is identical, because each detector weights the signal differently in time:

• Peak: Fastest and most conservative. It never reads low, so it is ideal for survey sweeps; anything that passes against the limit on peak needs no further measurement at that frequency.
• Quasi-peak: A weighted detector intended to approximate the annoyance of impulsive interference to analog receivers, and the basis of most commercial emissions limits. The CISPR quasi-peak detector uses a 1 ms charge (attack) time with a 160 ms discharge time in the conducted range (0.15–30 MHz) and a 550 ms discharge time in the radiated range (30–1000 MHz). Because it dwells, quasi-peak measurement is slow and is reserved for the handful of frequencies that fail or approach the limit on peak.
• Average: Reads lower for impulsive signals and is used together with quasi-peak for many limits; the gap between peak and average is itself diagnostic of how impulsive a source is.
• RMS-average: A detector that better characterizes broadband and noise-like emissions, used in newer limit definitions.

The efficient workflow exploits these differences directly: sweep wide with peak (and average) detectors, identify the worst-case frequencies, then re-measure only those points with the slow quasi-peak detector. Running quasi-peak across the whole band wastes hours for no added information.

Sweep, Overload, and Noise-Floor Management

Three practical errors dominate pre-compliance spectrum measurements. Sweeping too fast relative to the resolution bandwidth causes the analyzer to under-read narrowband signals, so sweep time must respect the dwell each bin needs. Overload from a strong out-of-band signal generates internal intermodulation products that masquerade as real emissions; switching in input attenuation and confirming that suspect peaks scale correctly distinguishes a true emission from an artifact. Finally, the instrument noise floor must sit comfortably below the limit, which a low-noise preamplifier provides — at the cost of reduced headroom, so its gain must be accounted for in every reading.

Reading H-Field and E-Field Results

Magnetic-field (H-field) loop probes respond to current and are strongest over current loops, switching nodes, and return-path discontinuities; electric-field (E-field) probes respond to voltage and are strongest over high-impedance, high-dV/dt nodes such as switch-node copper and unterminated lines. Reading both, and noting which one lights up at a given frequency, already narrows the mechanism: a hot H-field reading points to a loop-area or return-current problem, while a hot E-field reading points to a voltage-driven or capacitive-coupling problem.

Scanning Technique and Frequency Correlation

A repeatable scan is what turns probing from guesswork into evidence. The core disciplines are consistent probe-to-surface distance, systematic coverage rather than spot-checking, and rotating the probe to find the orientation of maximum coupling, which reveals how the source is oriented. The decisive step is correlation: the analyzer is tuned to the exact frequency that exceeds the limit in the far field, and the probe is then used to find the hottest location at that same frequency. Only an emission that correlates in frequency with the far-field problem is worth chasing; many strong near-field readings never radiate efficiently and are distractions.

From Manual Probing to Scan Data

Automated scanners that raster a probe over a board and build a spatial heat map at each frequency turn the same measurement into analyzable data. The interpretation extends naturally: a heat map shows hot spots and their spatial extent, an overlay onto the board layout ties each hot spot to a net or component, and a before-and-after comparison quantifies whether a fix actually reduced the source rather than merely moving it. The output is a ranked list of contributors, which is far more actionable than a single far-field spectrum.

Setup Factors That Change the Result

Conducted measurements are sensitive to arrangement, and inconsistent setup is the leading cause of pre-compliance results that do not repeat. A ground reference plane, controlled distances between the equipment, the plane, and the LISN, and fixed cable routing are prerequisites for comparable measurements. The ambient conducted noise should sit several decibels below the applicable limit, and the equipment should be operated in the mode that produces its worst-case emissions; a result taken in an idle state can mask a failure that appears only under load.

Diagnosing Common-Mode Versus Differential-Mode

The most useful interpretive distinction is the mode of the emission, because the two modes call for opposite fixes. Differential-mode noise flows out on one supply conductor and back on the other and responds to X-capacitors and differential filtering; common-mode noise flows in the same direction on both conductors, returns through ground, and responds to Y-capacitors and common-mode chokes. A mode-separation measurement, or simply comparing line-to-line and line-to-ground behavior, tells the engineer which filter element to change. Discrete spikes at a switching frequency and its harmonics point to the converter; a broadband hump points to switch-node ringing or layout; narrow lines at clock frequencies point to digital coupling onto the supply.

Converting a Reading to Field Strength

A radiated measurement is only meaningful after the raw analyzer reading is corrected. The antenna factor, in decibels, is added to convert terminal voltage in dBµV to field strength in dBµV/m, and it varies with frequency, so the correct value must be applied at each point. Cable loss between the antenna and the analyzer is added as well, while any preamplifier gain is subtracted. Many standards require measurement in both horizontal and vertical polarization, and the reported value is the worst case found while scanning antenna height and rotating the equipment to expose the maximum-emission direction. Performing these corrections in software, rather than by hand, removes a common source of error.

Working Around an Imperfect Environment

In a non-anechoic space, reflections and ambient signals corrupt absolute readings. The realistic strategy is to use radiated pre-compliance for relative comparison — verifying that a change reduced a peak — and to capture the ambient spectrum with the equipment powered off so that broadcast and Wi-Fi signals are not mistaken for emissions. Absolute pass-or-fail judgments are then made conservatively, against a margin that absorbs the environment's uncertainty, and confirmed at an accredited site before certification.

Instrument Control and Correction

EMC measurement software drives the analyzer or receiver, configures bandwidths and detectors per band, and applies the full correction chain — antenna factors, cable loss, preamplifier gain, and LISN coupling — automatically to every point. It overlays the applicable limit line so that a corrected trace can be read directly against the standard, and it automates the peak-search-then-quasi-peak sequence that would otherwise be done by hand. By keeping correction tables in one place, the software ensures that the same factors are applied consistently across sessions and operators.

Analysis, Reporting, and Traceability

Beyond capture, the analytical value is in comparison and record-keeping. Useful capabilities include overlaying spectra from successive design revisions to confirm that a fix held, marking and annotating the frequencies that drive the result, and computing margin to the limit at each peak so attention goes to the smallest margins first. Automated report generation assembles corrected plots, limit lines, and configuration details into a consistent document, and a measurement database preserves the configuration, correction set, and result for each test so that a later result can be reproduced and defended.

Why Pre-Compliance and Compliance Differ

The two measurements diverge for understood reasons: the pre-compliance environment has different absorber performance, shielding, and ambient noise; a spectrum analyzer's detectors and bandwidths may differ subtly from a calibrated receiver's; setup details such as cable dress and ground-plane quality vary; and every correction factor carries its own uncertainty. These contributions accumulate into a total measurement uncertainty that any honest pass-or-fail judgment must respect.

Setting and Validating a Margin

The standard response is to require pre-compliance results to clear the regulatory limit by a margin, commonly on the order of 6 dB or more, with a larger margin at frequencies where correlation is known to be weaker and for newer or less characterized setups. The margin is not a guess: it is calibrated by comparative testing, in which the same unit is measured both in-house and at an accredited laboratory so that frequency-dependent correction factors can be derived. That correlation is revalidated periodically as instruments age, and the methodology is documented so the in-house result can be trusted by a quality system.

From Spectrum to Root Cause

The loop is consistent: identify the failing frequencies in the far field, classify the emission by its spectral signature, localize the source with near-field probing at those exact frequencies, hypothesize a mechanism, apply a reversible mitigation, and re-measure to confirm. Characteristic signatures speed the classification:

• Clock and digital harmonics: Discrete lines at integer multiples of a clock frequency, often stronger on odd harmonics, with amplitude that grows as edge rates increase. Mitigations include slower edges where timing allows, series termination, controlled-impedance routing, spread-spectrum clocking, and shielding.
• Switching-converter emissions: Lines at the switching frequency and its harmonics plus broadband content from switch-node ringing, coupled to chassis and cables through parasitic capacitance. Mitigations include input filtering, snubbing, tighter loop layout, and shielded magnetics.
• Cable radiation: Peaks aligned with cable resonances, driven by common-mode current; cables are frequently the dominant radiator. Mitigations include common-mode chokes and ferrites, improved shield termination, and filtered connectors.
• Enclosure leakage: Emissions escaping through seams, ventilation apertures, and unfiltered penetrations. Mitigations include conductive gaskets, waveguide-below-cutoff venting, filtered connectors, and proper bonding.

Reversible Experiments and Confirmation

The discipline that keeps debugging honest is the reversible experiment combined with re-measurement. Temporary copper tape, clip-on ferrites, and hand-held absorber let an engineer test a hypothesis in seconds, but the change must be confirmed against the same calibrated measurement that revealed the problem, with the same bandwidth and detector. A fix that lowers one peak while raising another, or that merely shifts a resonance, is exposed only by re-measuring the full band rather than the single frequency under investigation.