Skip to content

Radiated emissions EMC test: pre-scan and final scan

Guide. EMC test method

Radiated emissions are the most-cited cause of schedule slip in CE marking and FCC certification campaigns. The measurement looks straightforward, an EUT, an antenna, a receiver, a graph, but it concentrates a rare number of methodological traps: antenna choice by band, measurement distance, detector, polarisation, cable arrangement, EUT operating mode. This guide describes the two-pass strategy, fast pre-scan in peak followed by slow final-scan in quasi-peak, which remains the reference practice of accredited labs. It covers the 30 MHz - 1 GHz band and the above-1 GHz band, with CISPR 32 and FCC Part 15 limits placed side by side.

Normative frame and boundaries between standards

Section titled “Normative frame and boundaries between standards”

Radiated emissions denote the unintentional radio-frequency energy that an equipment under test (EUT) releases into its environment, as opposed to conducted emissions, which propagate through power and signal cables. Three families of standards bound the measurement.

CISPR 32 (published as EN 55032 in the EU) covers information technology, audio, video and television-reception equipment. It replaced CISPR 22 in 2017. It defines limits by class (A for commercial / industrial, B for residential) and refers to CISPR 16 for measurement methods.

CISPR 11 (published as EN 55011) covers industrial, scientific and medical equipment. It applies to RF heating, welding and plasma equipment, and to power converters, inverters and variable-speed drives. Its class structure (groups 1 and 2, classes A and B) overlaps with CISPR 32 but with group and band subtleties of its own.

Above these two generic standards, product-specific standards may take precedence. Examples: EN 55014-1 for household appliances, EN 55015 for lighting, EN 55025 for automotive (a CISPR 25 derivative), EN 50121 for railway. A product standard takes precedence over the generic one where it exists.

On the US side, FCC Part 15 Subpart B covers unintentional radiators (digital devices). The referenced method is ANSI C63.4, a functional counterpart of CISPR 16 with its own distance and detector conventions.

DomainEU genericUS generic
Multimedia / IT equipmentCISPR 32 (EN 55032)47 CFR 15 Subpart B
ISM equipmentCISPR 11 (EN 55011)47 CFR 18
Measurement methodologyCISPR 16-2-3ANSI C63.4
InstrumentationCISPR 16-1-1ANSI C63.2

The full test range typically spans 30 MHz to 6 GHz for most products, or up to 40 GHz for equipment with very fast clocks. No single antenna covers the whole range. The split into sub-bands is dictated by the physics of radiation and by the availability of calibrated antennas.

Frequency bandTypical antennaMeasurement distanceSite
30 - 200 MHzBiconical or hybrid biconilog3 m or 10 mSAC or OATS
200 MHz - 1 GHzLog-periodic or hybrid biconilog3 m or 10 mSAC or OATS
1 - 6 GHzHorn (double-ridge)3 mSAC dedicated above 1 GHz
6 - 18 GHzHigher-frequency horn1 m (typical)SAC or FAR
18 - 40 GHzSpecific horn or open waveguide1 mFAR (fully anechoic room)

Quick legend:

  • SAC = semi-anechoic chamber with a reflecting ground plane. The CISPR standard below 1 GHz.
  • OATS = open area test site, an open-air range with a ground plane. The historical reference, still in service but sensitive to weather and ambient noise.
  • FAR = fully anechoic room, with no ground plane, required by CISPR above 1 GHz to limit reflections.

Antenna height typically varies from 1 to 4 m below 1 GHz to sweep the worst-case radiation angle. Above 1 GHz, antenna height is fixed (typically 1.5 m), and the EUT is explored in azimuth and elevation instead.

CISPR 16-1-1 specifies the EMI receiver resolution bandwidth (RBW), which is not a free parameter.

BandCISPR RBWPrimary detector
30 MHz - 1 GHz120 kHzQuasi-peak (CISPR)
1 GHz - 18 GHz1 MHzPeak + average
18 - 40 GHz1 MHzPeak + average

The FCC, through ANSI C63.4, uses the same RBW values in practice. A measurement taken with a different RBW (for example 100 kHz on a generic spectrum analyser instead of 120 kHz) is non-compliant and cannot stand in a certification report.

The three CISPR detectors are defined in CISPR 16-1-1 with precise time constants.

  • Peak (Pk): instantaneous maximum in the resolution bandwidth. The fastest, the most pessimistic. Used in pre-scan and above 1 GHz.
  • Quasi-peak (QP): time-weighted with the charge (1 ms) and discharge (550 ms) constants defined between 30 MHz and 1 GHz. The CISPR reference detector below 1 GHz. Slow.
  • Average (AV): integration over time. Used above 1 GHz and for some conducted emissions.

Fundamental property: for any given signal, Peak >= QP >= AV. This inequality is what makes the pre-scan / final-scan strategy described below possible.

CISPR 32 Class A and Class B limits are defined in dB(uV/m) at the reference measurement distance. Class A applies to a commercial or industrial environment, Class B to a residential environment.

BandClass A (10 m)Class B (10 m)Detector
30 - 230 MHz40 dB(uV/m)30 dB(uV/m)Quasi-peak
230 - 1000 MHz47 dB(uV/m)37 dB(uV/m)Quasi-peak

Class B limits are 10 dB tighter than Class A across both sub-bands. The discontinuity at 230 MHz reflects a physical reality: above 230 MHz, the internal structures of an EUT (cables, traces, apertures) become more efficient antennas, and the standard accepts a slightly higher emission level.

BandClass A (3 m)Class B (3 m)Detector
1 - 3 GHz56 dB(uV/m) peak / 76 dB(uV/m) peak50 dB(uV/m) peak / 70 dB(uV/m) peakPeak / Average
3 - 6 GHz60 dB(uV/m) peak / 80 dB(uV/m) peak54 dB(uV/m) peak / 74 dB(uV/m) peakPeak / Average

The CISPR 32 reference distance above 1 GHz is 3 m, against 10 m below. The change in distance and detector requires distinct instrumentation: broadband horn, tracking generator if analysis is performed in a fully anechoic chamber.

On the US side, limits are set in 47 CFR 15.109 (Class A and Class B unintentional radiators). The reference distance is 3 m for Class B and 10 m for Class A, which complicates a direct comparison with CISPR.

BandFCC Class A at 10 mFCC Class B at 3 mDetector
30 - 88 MHz39 dB(uV/m)40 dB(uV/m)Quasi-peak
88 - 216 MHz43.5 dB(uV/m)43.5 dB(uV/m)Quasi-peak
216 - 960 MHz46.4 dB(uV/m)46 dB(uV/m)Quasi-peak
Above 960 MHz49.5 dB(uV/m)54 dB(uV/m)Quasi-peak

Useful conversion between 3 m and 10 m: in the far field, 20 log(10/3) = 10.5 dB. An FCC Class B limit at 3 m of 40 dB(uV/m) maps, scaled to 10 m, to about 29.5 dB(uV/m). That is close to the CISPR Class B limit at 10 m (30 dB(uV/m)) at the low end, but the gap widens above 230 MHz where CISPR becomes the tighter regime.

Above 1 GHz, the FCC historically referenced limits in 47 CFR 15.109(g), with peak and average detectors at 3 m. The values are not strictly identical to CISPR 32 but are close, to within around 6 dB on average. A product sized for CISPR 32 generally clears the FCC limits above 1 GHz without modification.

BandTighter sideApproximate gap
30 - 88 MHzFCC Class B (scaled to 10 m)+0.5 dB FCC vs CISPR Class B
88 - 230 MHzFCC Class B (scaled to 10 m)+3 dB FCC vs CISPR Class B
230 - 960 MHzCISPR Class B-1.5 dB CISPR vs FCC
Above 960 MHzCISPR Class B-5 dB CISPR vs FCC

Design implication for dual-market: targeting CISPR Class B above 230 MHz covers FCC by margin. Between 30 and 230 MHz, FCC sets the bar. See CE vs FCC, EMC comparison for the detailed reasoning.

This is where lab time is won or lost. A full quasi-peak sweep over 30 MHz - 1 GHz, with a 50 kHz step and a dwell time sufficient for the QP discharge constant, takes hours per polarisation per antenna height. Multiplied by EUT operating modes, that is days of chamber time. The two-pass strategy brings that down to hours.

The objective is to identify candidate frequencies for overshoot, not to produce a compliance measurement. The sweep is run with:

  • Peak detector (the fastest),
  • CISPR-compliant RBW (120 kHz below 1 GHz),
  • Short dwell time per point (typically 1 to 10 ms),
  • Full coverage of both polarisations and the antenna height range,
  • EUT configured in the most-emissive mode identified by prior analysis.

The output is a peak vs frequency plot. Any frequency whose peak level is below the standard's QP limit minus 6 dB is considered passed, since QP <= Peak by construction. Only frequencies above that threshold are retained as candidates.

For each candidate frequency identified in pre-scan, the receiver switches to quasi-peak mode and the measurement is repeated point by point. Dwell time here is 1 to 2 seconds per frequency (550 ms discharge constant plus stabilisation). The number of frequencies to measure is typically 5 to 50 across the whole band, against thousands in a continuous sweep.

For each QP frequency, the operator sweeps:

  • Both antenna polarisations (vertical, horizontal),
  • Antenna height range (1 to 4 m below 1 GHz),
  • Turntable azimuth in steps (typically 22.5 or 45 degrees),
  • EUT operating modes relevant to the frequency (a peak at 480 MHz appears with USB 2.0 active, not necessarily at idle).

The value retained per frequency is the maximum across all those parameters. That value is compared against the limit.

StepDetectorTypical timeOutput
Fast pre-scanPeak5 - 30 min per EUT modeCandidate-frequency list
Threshold filter-ImmediateFrequencies above QP_limit - 6 dB
Final-scan QPQuasi-peak1 - 2 s per candidate pointCompliance measurement
MaximisationQPIncludedWorst case across polarisation / height / azimuth
Limit comparison-ImmediatePass / fail per frequency

The physical configuration of the EUT during the test is not a detail. CISPR 16-2-3 and ANSI C63.4 set a rule: the EUT must be tested in the most-emissive configuration reasonably expected in normal use. That covers:

  • Cable arrangement (excess length folded in a 30 to 40 cm serpentine per the standard),
  • Representative connected peripherals (keyboard, screen, power supply, sensors),
  • Software operating mode (data transfer, screen active, processor load),
  • Enclosure orientation on the turntable.

A small change in cable routing can shift a peak by 6 to 10 dB. That is the single biggest cause of divergence between in-house pre-compliance and lab compliance.

Pre-compliance is run in-house or in a non-accredited chamber, with simplified instrumentation (spectrum analyser plus biconilog antenna, without a calibrated ground plane). Its purpose is to flag gross overshoots before paying for lab time. Its typical uncertainty is 6 to 10 dB. Pre-compliance carries no presumption of conformity.

Full-compliance is carried out in an ISO/IEC 17025-accredited lab recognised by the competent authority (a notified body for CE, a TCB for FCC). The report it produces is legally binding.

Engineering practice is to target 6 dB below the standard limit in pre-compliance, which absorbs:

  • Lab measurement uncertainty (CISPR 16-4-2 quotes up to 5.2 dB for radiated emissions 30 MHz - 1 GHz),
  • Unit-to-unit dispersion in manufacturing,
  • Lifetime drift (component ageing, shielding-gasket contamination),
  • Risk of a peripheral change (mains cable, third-party power supply).

A margin below 3 dB is treated as a significant non-conformity risk at serial production.

Common pitfalls in radiated-emissions measurement

Section titled “Common pitfalls in radiated-emissions measurement”

Six errors recur in failed campaigns.

  1. Cable routing modified between pre-scan and final-scan. An operator re-arranges cables to ease access to the EUT, and the peak found in pre-scan no longer reproduces in QP. The frequency is not measured in final, and the product fails on a later serial unit or under market surveillance. Photographing the setup before each scan is the minimum prevention.
  2. Missed polarisation. A quick pre-scan covers only one polarisation. If final-scan does not include cross polarisation, a major peak can be ignored. CISPR 16-2-3 mandates both polarisations per retained frequency.
  3. Mis-calibrated antenna. The antenna factor (in dB/m) is specific to each antenna and to each polarisation, and must be traceable to a recent calibration. An antenna whose certificate has expired, or whose factor is applied with the wrong sign, introduces a 10 to 20 dB error without any warning.
  4. EUT operating mode not exhaustive. The EUT is tested idle, while the real application involves data transfer, active radio communication, or sensor scanning. The emission frequencies differ. The rule: identify all possible modes before entering the chamber, and measure in each.
  5. Confusion between 120 kHz and 100 kHz RBW. A generic spectrum analyser defaults to 100 kHz. An EMI receiver runs at 120 kHz per CISPR. Conversion between the two is not linear and depends on signal nature (CW, modulated, pulsed). Measuring at 100 kHz and applying a correction factor is not accepted in compliance.
  6. Insufficient margin in pre-compliance. A product clearing pre-compliance at -1 dB from the limit is a product that will not pass. Cumulative uncertainty exceeds that gap. The pragmatic rule: if the margin is below 6 dB, treat the peak as a fail and apply mitigation before paying the lab.

See also CE pitfalls and FCC pitfalls for per-regime detail.

When a peak exceeds the limit in final-scan, three families of levers are available, from fastest to most structural.

  • Cable re-arrangement (separating power and signal, shorter length, twisted pairs),
  • Clip-on common-mode ferrites on external cables (selected for the peak frequency),
  • Additional conductive gaskets on enclosure apertures,
  • Local capacitive decoupling added in series on suspect lines.

These actions do not require a PCB revision and can be validated in a few hours of chamber time. They typically yield 3 to 8 dB of additional margin.

  • Local Faraday cage on the emissive module (oscillator, DC/DC converter),
  • Pi filters (capacitor - inductor - capacitor) in series on inputs / outputs,
  • Cable-shielding upgrade (from single braid to double shield),
  • Reduction of enclosure apertures below lambda/20 at the peak frequency.

These changes affect the mechanical enclosure and the BOM, but not the board. Typical lead time: 2 to 4 weeks.

  • Re-routing of clocks and high-speed traces to reduce current loops,
  • Continuous ground plane with no slot under critical signals,
  • Pairwise local decoupling (100 nF + 10 nF + 1 nF) at each power pin,
  • Series termination on outgoing clocks (22 to 33 ohm series resistor),
  • Spread-spectrum-clocking oscillator on main clocks.

This is the most effective mitigation but also the heaviest, with a 4 to 8 week PCB cycle. It becomes unavoidable when the required margin exceeds 10 dB.

Radiated emissions are only one test in a full EMC campaign. The table below places the test within the typical sequence.

FamilyTestGeneric standardReference
EmissionsRadiated emissionsCISPR 32 / FCC 15This guide
EmissionsConducted emissions on LISNCISPR 32 / FCC 15Guide pending
ImmunityElectrostatic discharge (ESD)EN 61000-4-2See CE tests
ImmunityRadiated RF fieldEN 61000-4-3See CE tests
ImmunityFast transients (EFT)EN 61000-4-4See CE tests
ImmunitySurgeEN 61000-4-5See CE tests
ImmunityConducted RFEN 61000-4-6See CE tests

See RED tests for additional tests on radio products, and FCC tests for the US-side sequence. The glossary covers the acronyms (RBW, QP, SAC, NSA, EUT) with their CISPR definitions.

  • The 30 MHz - 1 GHz vs above-1 GHz split dictates instrumentation (antennas, chamber, detector). No single antenna covers the range.
  • RBW is fixed by CISPR 16-1-1, not by the operator: 120 kHz below 1 GHz, 1 MHz above.
  • CISPR Class B is tighter than FCC Class B above 230 MHz, and the reverse below. Size to the tighter envelope per band.
  • Pre-scan in peak then final-scan in QP divides chamber time by ten without loss of rigour, provided the cable setup stays identical.
  • The 6 dB margin absorbs measurement uncertainty (up to 5.2 dB per CISPR 16-4-2), serial dispersion and ageing.
  • Cable routing changes between passes are the number-one cause of pre-compliance / compliance divergence.

For implementation on the EU side, see CE tests. On the US side, see FCC tests. On radio products, see RED tests.

Sources & references

  1. CISPR 32:2015+A1:2019, Multimedia equipment, emission requirements , IEC webstore.iec.ch/publication/26241
  2. CISPR 16-1-1:2019, Specifications of radio disturbance and immunity measuring apparatus , IEC webstore.iec.ch/publication/63465
  3. CISPR 16-2-3:2016, Methods of measurement, radiated disturbance measurements , IEC webstore.iec.ch/publication/26326
  4. CISPR 11:2024, Industrial, scientific and medical equipment, radio-frequency disturbance characteristics , IEC webstore.iec.ch/publication/68645
  5. 47 CFR Part 15, Radio frequency devices , FCC www.ecfr.gov/current/title-47/chapter-I/subchapter-A/part-15
  6. ANSI C63.4-2014, American National Standard for methods of measurement of radio-noise emissions , IEEE / ANSI standards.ieee.org/ieee/C63.4/5536/