Why run pre-compliance EMC tests in-house?

A pre-compliance bench catches the bulk of design issues before booking a formal accredited lab. Industry feedback consistently puts this share around 80 percent of issues found on first formal pass that could have been seen on a basic in-house bench: harmonic spurs from a switching regulator, common-mode current on a cable harness, an antenna detuned by a nearby ground plane. A re-do of a formal radiated emissions campaign at an accredited 3 m chamber typically costs more than the starter bench itself, before counting schedule slippage on the program. Running pre-compliance early also shortens design iteration: the hardware team measures a fix the same afternoon rather than waiting weeks for the next lab slot.

What is the difference between a TEM cell and a GTEM cell?

A TEM cell (Transverse ElectroMagnetic) is a rectangular shielded enclosure with two ports, a septum running through it and tapered ends on both sides. It generates a transverse plane wave inside its useful volume and works from DC up to around 1 GHz before higher-order modes appear. A GTEM cell (Gigahertz TEM) is a single-tapered structure terminated on one end by a matched load and absorbers, opening the useful range up to 18 GHz and beyond. The trade-off is volume: a GTEM capable of holding a small system is large and heavy, while a tabletop TEM cell fits next to a workbench but limits the EUT (Equipment Under Test) to roughly one third to one quarter of the septum-to-floor distance. Both are covered by IEC 61000-4-20.

Is a spectrum analyser enough for pre-compliance, or do I need an EMI receiver?

A modern spectrum analyser with an EMI measurement option is enough for pre-compliance, on the explicit condition that it offers the CISPR-compliant detectors required by CISPR 16-1-1: quasi-peak, average, RMS-average and peak-hold. Resolution bandwidths must match the CISPR bands (200 Hz for band A, 9 kHz for band B, 120 kHz for bands C and D). A general-purpose spectrum analyser without the EMI option will measure peak detector only, which overestimates the margin to the limit line because formal CISPR 32 limits are quasi-peak and average. A dedicated EMI receiver, slower and more expensive, remains the reference for formal measurements; for pre-compliance the analyser plus EMI option is the usual compromise.

What does a LISN actually do?

A LISN (Line Impedance Stabilisation Network) sits between the mains and the EUT power input. It performs three functions at once: it presents a defined RF impedance to the EUT (typically 50 ohm in parallel with 50 microhenry above 150 kHz, per CISPR 16-1-2), it isolates the measurement from mains noise upstream, and it routes the RF noise current to a 50 ohm port for the spectrum analyser or EMI receiver. Without a LISN, conducted emissions measurements between 150 kHz and 30 MHz under CISPR 32 or FCC Part 15 Subpart B are not reproducible, since the impedance seen by the EUT depends on cable length and on the lab installation. The LISN is mandatory equipment for any serious pre-compliance bench.

How do near-field probes fit in the workflow?

Near-field probes do not measure compliance, they localise the radiator. The standard set is an H-field loop probe (loops of 5, 10, 20, 50 and 100 mm), an E-field stub probe, and a low-noise amplifier (LNA) of around 20 to 30 dB. The probe is moved over the powered board while the spectrum analyser is tuned to the suspect frequency seen on a TEM or chamber measurement. The hot spot reveals the source: a switching regulator inductor, a clock trace, a connector launch, a decoupling loop. The workflow is iterative: identify the offending frequency from the radiated measurement, locate the source with the probe, photograph the board with the probe position annotated, apply a layout or filtering change, re-measure.

Which antennas do I need to cover 30 MHz to 6 GHz?

Three antenna types cover the usual range. A biconical antenna handles 30 MHz to 200 MHz, with a typical gain that requires a calibration table over the band. A log-periodic dipole array (LPDA) covers 200 MHz to 1 GHz, and a hybrid bilog combines the two from 30 MHz to 1 GHz to spare an antenna change. A double-ridged horn covers 1 GHz to 18 GHz for radiated emissions above the 80 cm wavelength. Each antenna comes with a free-space antenna factor (dB/m) curve from its calibration certificate, which is added to the analyser reading to give the field strength at the reference distance. For pre-compliance a 3 m semi-anechoic chamber is the usual setup; in a TEM cell the antenna factor is replaced by the cell factor, traceable to IEC 61000-4-20.

How does pre-compliance fit between bench measurement and formal test?

The lab sequence has four steps. Step one is bare-board bench measurement with near-field probes, during bring-up, to flag obvious radiators before any enclosure. Step two is system measurement in a TEM or GTEM cell on the assembled product, to estimate the field strength at a reference distance and check broad spectral behaviour. Step three is conducted emissions on the LISN, plus harness scan with a current probe, to characterise the noise paths. Step four is a 3 m semi-anechoic chamber dry-run, either in-house if available or outsourced to an unaccredited but equipped lab, before booking the formal accredited campaign. Each step filters out a category of issues, so the formal test focuses on the residual margin rather than chasing surprises.

What are the most common pitfalls on an in-house pre-compliance bench?

Recurring traps: an EUT too large for the TEM cell useful volume, which puts it inside the non-uniform field region and breaks the cell factor; a LISN with a floating ground plane, which leaks common-mode current through the measurement path; a near-field probe held at random orientation rather than orthogonal to the suspect trace, which under-reads the actual coupling; a spectrum analyser used in peak detector only, which overestimates the safety margin compared to the quasi-peak limit; a measurement chain without a low-noise preamplifier on the near-field path, which buries the low-amplitude radiators under the analyser noise floor; cable common-mode current ignored, so a harness change at formal test undoes the careful EUT optimisation; and an analyser configured with the wrong resolution bandwidth (RBW) and video bandwidth (VBW), which does not match the CISPR band rules and produces incomparable numbers.

Pre-compliance EMC: TEM cell, near-field probes, LISN

Author Hugues Orgitello Last updated 27 May 2026 ~14 min read

Guide, pre-compliance EMC bench

An in-house pre-compliance EMC bench is the single most effective de-risking investment for an electronics product development team. A typical breakdown puts roughly 80 percent of issues found on a first formal accredited campaign within reach of a basic bench: a switching regulator harmonic above the limit line, a common-mode current on a cable harness, an antenna detuned by a nearby ground plane, a clock spur radiating from a connector launch. This guide walks through the equipment families that make up a credible pre-compliance bench, from TEM and GTEM cells under IEC 61000-4-20, through LISN and current probes under CISPR 16-1-2, to spectrum analysers, EMI receivers and near-field probes. It also positions the bench within the product development workflow and lists the recurring pitfalls observed in the field.

Why a pre-compliance bench

The economics of pre-compliance are straightforward. A formal accredited radiated emissions campaign in a 3 m semi-anechoic chamber, with a full test plan covering CISPR 32 emissions and the immunity suite of IEC 61000-4-x, runs into several day-rates and is booked weeks ahead. A re-do after a fail typically costs the same again, plus the engineering time to investigate and the slippage on the program timeline. A pre-compliance bench moves most of the failure modes upstream, where they are cheap to fix: a bare board on the bench is a minute away from a probe scan, an assembled product fits in a tabletop cell, a layout iteration is a same-day re-measurement.

The other motivation is iteration speed. During board bring-up the hardware team needs to know, within an afternoon, whether a layout change reduced the emissions of a switching regulator. Booking the accredited lab for that loop is operationally impossible. The pre-compliance bench is the only place that loop closes in real time.

A consistent observation across electronics teams: products that arrive at formal test without a pre-compliance history routinely fail on first pass, while products tracked from bring-up on a bench rarely surprise. The bench does not replace the formal lab, it filters what reaches it.

Equipment families

A pre-compliance bench is a stack of measurement chains, each tailored to a specific failure mode. The minimum credible set looks like this.

Equipment	Role	Reference standard
TEM or GTEM cell	Radiated emissions and immunity on small EUTs, in a controlled volume	IEC 61000-4-20
LISN (Line Impedance Stabilisation Network)	Conducted emissions on the mains port, 150 kHz to 30 MHz	CISPR 16-1-2
Current probe (clamp-on)	Common-mode current on cables and harnesses	CISPR 16-1-2
Near-field probes (H-field loops, E-field stub)	Localising radiators on PCB, ICs, harness segments	Engineering use
Spectrum analyser with EMI option, or EMI receiver	The measurement instrument behind every probe and antenna	CISPR 16-1-1
Low-noise amplifier (LNA)	Boosts low-level signals on the near-field path	Engineering use
Biconical, log-periodic, horn antennas	Radiated measurements in a chamber, 30 MHz to 18 GHz	Free-space antenna factor
Semi-anechoic chamber (SAC) or Open Area Test Site (OATS)	Final pre-compliance dry-run before formal	See sibling guide

The stack is modular: a team can start with a TEM cell, a LISN, a portable spectrum analyser and a near-field probe set, and add layers (GTEM, EMI receiver, 3 m SAC) as the product mix grows. For a comparative view of the chamber types themselves, see EMC chamber types: SAC, FAR, OATS, reverberation.

TEM and GTEM cells

The TEM (Transverse ElectroMagnetic) cell is the workhorse of in-house radiated emissions pre-compliance. Its principle is simple: a rectangular shielded enclosure with a flat conductor (septum) running through it, terminated on both ends by tapered transition sections and 50 ohm loads. A signal injected on one port propagates as a transverse plane wave between the septum and the floor, with a field strength predictable from the port voltage and the septum-to-floor distance.

TEM cell, two-port

A classical TEM cell has two ports, one at each tapered end, and works from DC up to roughly 1 GHz. Above that frequency, higher-order modes appear and the field is no longer purely transverse, which breaks the cell factor calibration. The useful EUT volume is limited to roughly one third to one quarter of the septum-to-floor distance, so that the EUT does not perturb the propagating mode. A tabletop TEM cell with a septum height of 15 cm typically accepts an EUT footprint of around 5 cm by 5 cm by 5 cm, enough for a bare board or a small subassembly but not for a full system.

GTEM cell, single-port

The GTEM (Gigahertz TEM) cell removes the high-frequency mode limit by replacing the second tapered transition with a matched termination, made of resistive loads and RF absorbers. The structure becomes a single-tapered horn-like geometry, with a useful range that opens up to 18 GHz and beyond, depending on the cell design. The price is volume: a GTEM capable of accepting a small system measures several metres in length and weighs hundreds of kilograms, with a footprint that competes with a chamber room.

The GTEM is the natural step up from the TEM cell when the product spectrum extends above 1 GHz, which is the case for any product with a Wi-Fi, Bluetooth or cellular radio. Below 1 GHz the TEM cell is enough.

Cell factor and EUT volume

Both cell types are covered by IEC 61000-4-20 (2010 plus amendment 2022), which defines the field uniformity requirements, the EUT volume limits and the procedure to convert a cell port reading into an equivalent radiated field at a reference distance (typically 3 m or 10 m equivalent). The cell manufacturer provides a calibration table or formula. The reading is only valid inside the declared useful volume; an EUT too large for the cell sits in the non-uniform region and produces an unreliable number. This is the single most frequent TEM mistake.

Reverberation chamber

For products with high field strength immunity requirements (typically immunity testing under IEC 61000-4-3), a reverberation chamber under IEC 61000-4-21 (2011) is an alternative to the semi-anechoic chamber. The principle is the opposite of the TEM cell: a metallic room with a mechanical mode stirrer creates a statistically uniform field, where the EUT sees an averaged field independent of its orientation. The chamber is more efficient than an anechoic room in terms of field strength per watt of input power, which makes high-field immunity feasible without a multi-kilowatt amplifier. For pre-compliance, however, a reverberation chamber is a heavier investment than a TEM cell and is justified only when the product mix calls for it routinely. See IEC 61000-4-3 radiated RF immunity for the radiated immunity context.

LISN, conducted emissions on the mains port

The LISN (Line Impedance Stabilisation Network) is the equipment that turns a noisy mains environment into a reproducible measurement. It is mandatory for conducted emissions testing under CISPR 32, CISPR 22 (now superseded but still cited in older dossiers) and FCC Part 15 Subpart B in the 150 kHz to 30 MHz range.

The standard topology is a 50 microhenry inductor in series with the mains line, a 50 ohm impedance referenced to the LISN chassis through a coupling capacitor, and a measurement port that taps the RF current toward the spectrum analyser. Above 150 kHz the network presents a 50 ohm impedance to the EUT, regardless of mains harmonic content upstream. The LISN chassis is bonded to a copper ground plane (the reference ground), which is the single most important detail of the installation: a LISN with a floating chassis or a poor ground bond leaks common-mode current into the measurement path and adds tens of decibels of measurement uncertainty.

LISN coupling networks are specified in CISPR 16-1-2 (2017), which defines several variants depending on the supply type: single-phase, three-phase, DC, and telecom port AAN (Asymmetric Artificial Network) for twisted pair lines. Choosing the wrong LISN topology produces a valid measurement on a non-representative network and is a recurring source of mismatch between in-house and formal lab results.

Current probes and clamps

Beyond the LISN, conducted emissions also flow as common-mode current on cables and harnesses. A clamp-on current probe under CISPR 16-1-2 measures the RF current on a bundle without breaking the wiring, with a transfer impedance (ohm) that converts the analyser reading into a current. The current probe is associated with an ICN (Impedance Calibration Network) for cable common-mode current characterisation, particularly for telecom and signal cables where a LISN is not applicable.

The current probe is the natural complement to the LISN: the LISN sees the mains port, the current probe sees the harness. A product with low LISN readings but high field emissions at formal test is almost always a harness common-mode problem that the current probe would have flagged early.

Antennas for radiated measurement

When the pre-compliance bench grows beyond a TEM or GTEM cell, the next layer is a 3 m semi-anechoic chamber or, more rarely, an Open Area Test Site (OATS) for products that cannot be enclosed. The chamber needs an antenna set covering the regulated range.

Antenna	Useful range	Use
Biconical	30 MHz to 200 MHz	Low-frequency radiated emissions
Log-periodic dipole array (LPDA)	200 MHz to 1 GHz	Mid-band radiated emissions
Hybrid bilog	30 MHz to 1 GHz	Single antenna combining biconical and LPDA
Double-ridged horn	1 GHz to 18 GHz	High-frequency radiated emissions

Each antenna comes with a calibration certificate giving its free-space antenna factor (in dB/m) as a function of frequency, which is added to the analyser reading (in dBmicroV) to give the field strength at the reference distance (in dBmicroV/m). The calibration is by substitution against a reference antenna, traceable to a national metrology institute. For a deeper view of calibration and measurement uncertainty, see calibration and measurement uncertainty (GUM).

Near-field probes, localising the radiator

A near-field probe does not measure compliance, it localises the source. The standard set is a series of H-field loop probes of diameters 5, 10, 20, 50 and 100 mm, plus an E-field stub probe. The small loops localise to a few millimetres on a PCB, the large loops integrate over a wider area (a connector, an inductor, a cable segment). The E-field stub picks up high-impedance radiators (high dV/dt traces, switching node).

Probe chain

A near-field probe outputs a low-amplitude signal, typically tens of microvolts on a quiet board to a few millivolts on a hot regulator. The probe chain therefore requires a low-noise preamplifier (LNA) of 20 to 30 dB between the probe and the spectrum analyser, otherwise the analyser noise floor buries the radiator. The chain is then: probe, LNA, coaxial cable, spectrum analyser.

Workflow

The near-field probe workflow is iterative.

Identify the suspect frequency on a TEM or chamber measurement: a peak above the limit line or a noticeable harmonic.
Tune the spectrum analyser to that frequency with a narrow span and a peak hold.
Move the H-field loop over the powered board, keeping the loop plane orthogonal to the suspect trace direction.
Find the hot spot: the position where the analyser reads maximum.
Photograph the board with the probe position annotated, repeat for the suspect harmonics.
Apply a layout, filter or shielding change.
Re-measure on the cell or in the chamber, then re-probe.

The orientation discipline is the single most common error: an H-field loop held at random angles under-reads the actual coupling by up to 20 dB. The plane of the loop must be orthogonal to the magnetic field, which is itself orthogonal to the current flow on the suspect trace.

Spectrum analyser versus EMI receiver

The measurement instrument behind every probe and antenna is either a spectrum analyser or an EMI receiver. The two converge on the same display but differ on detector type, dynamic range and CISPR conformity.

Spectrum analyser

A general-purpose spectrum analyser is fast and affordable. It sweeps a span in milliseconds, displays the peak amplitude per frequency bin and is the natural instrument for a bench. The pre-compliance use requires the EMI measurement option, which adds the CISPR-compliant detectors: quasi-peak, average, RMS-average and peak-hold, per CISPR 16-1-1 (2019). Without the EMI option the analyser only measures peak, which overestimates the margin to the limit line because CISPR 32 limits are quasi-peak and average.

Modern mid-range analysers (around 7 GHz or 13 GHz upper limit, 100 dB dynamic range) with the EMI option cover most pre-compliance needs. A higher-range model (above 26 GHz) becomes useful for products with mmWave radios or harmonics extending above the radio fundamental.

EMI receiver

The EMI receiver is the formal reference. It implements the CISPR-compliant detector chain in hardware, with the bandwidths and dwell times specified in CISPR 16-1-1: 200 Hz RBW for band A (9 to 150 kHz), 9 kHz for band B (150 kHz to 30 MHz), 120 kHz for bands C and D (30 MHz to 1 GHz). It is slower than a spectrum analyser in scan mode, since the quasi-peak dwell time per frequency is on the order of a second. It is also more expensive and more specialised.

For pre-compliance the spectrum analyser plus EMI option is the usual compromise. The accredited lab uses the EMI receiver for the formal test, and the small delta in detector behaviour is one of the contributions to measurement uncertainty (see calibration and measurement uncertainty (GUM)).

RBW and VBW settings

A recurring pitfall: an analyser configured with the wrong resolution bandwidth produces numbers that do not match the CISPR scale. The rule is simple: the RBW must match the CISPR band (200 Hz, 9 kHz or 120 kHz), and the video bandwidth (VBW) must be at least three times the RBW for peak measurements, or set to the dedicated CISPR detector for quasi-peak and average. An analyser with the EMI option handles this automatically when the CISPR mode is selected.

Calibration and measurement uncertainty

Every chain in the bench (cell factor, antenna factor, LISN transfer impedance, current probe transfer impedance, near-field probe factor, LNA gain, cable loss) is a contributor to the overall measurement uncertainty. A calibration trace running back to a national metrology institute is mandatory for any value used to declare a margin to the limit line. For pre-compliance the calibration cadence is typically every two years on cells and antennas, every year on the analyser and EMI receiver, and on initial qualification for cables and adapters. The full uncertainty budget under the GUM framework is covered in calibration and measurement uncertainty (GUM).

Build-or-buy decision

A typical staged investment for an in-house pre-compliance bench.

Entry level: tabletop TEM cell, single-phase LISN, portable spectrum analyser with EMI option, near-field probe set with LNA, biconical and LPDA antennas. Footprint of a normal lab bench. Enough to catch the bulk of issues on board-level products up to 1 GHz.
Mid level: GTEM cell up to 18 GHz, three-phase LISN, current probes, calibration of cell and antenna factors, dedicated EMI receiver. Footprint of a small room. Covers products with Wi-Fi, Bluetooth and sub-6 GHz radios.
Advanced: 3 m semi-anechoic chamber, full antenna set including double-ridged horn, reverberation chamber for high-field immunity, RF amplifiers for IEC 61000-4-3 immunity. Footprint of a dedicated lab building. Covers the full formal pre-compliance scope, short of the accredited certification step.

Stages two and three are typically reserved for organisations whose product mix justifies the standing capital. For occasional needs, the alternative is to subcontract a 3 m chamber to an unaccredited but equipped lab for a final pre-test before the formal accredited campaign. The cost split is usually favourable: a day in a 3 m chamber dry-run, then a clean formal pass, beats two formal campaigns with a fail in between.

Workflow during product development

A coherent pre-compliance workflow tracks the product through bring-up and integration.

Layout review before fabrication: identify the EMC-critical zones (switching nodes, clock, connector launches, ground stitching), check decoupling and return paths. See PCB design for EMC.
Bare board measurement during bring-up: power up the board, scan with near-field probes, characterise the dominant radiators. Catch obvious issues (missing decoupling, unconnected ground via, oversized loop) before the enclosure is even ordered.
Assembled product in the TEM or GTEM cell: place the product in the useful volume, run radiated emissions sweeps, compare to a limit line scaled to the cell factor. Iterate on shielding, gasketing, filtering.
Harness scan with near-field probes and current probes: identify common-mode current on cables, validate cable shielding and ferrite placement.
Conducted emissions on the LISN: run the mains port emissions sweep, compare to CISPR 32 or FCC Part 15 Subpart B limits. Iterate on input filter.
Final pre-test in a 3 m SAC, in-house or outsourced, before booking the formal accredited campaign. See radiated emissions EMC test for the formal context.

Each step filters out a class of failure, so the formal test focuses on residual margin rather than chasing surprises.

Pitfalls

Pitfall	Consequence
EUT larger than the TEM cell useful volume	EUT in the non-uniform field region, cell factor invalid, unreliable measurement
LISN chassis not bonded to a copper ground plane	Common-mode current leaks into the measurement path, tens of decibels of uncertainty
Near-field probe held at random orientation	Coupling under-read by up to 20 dB, hot spot mis-localised
Spectrum analyser used in peak detector only, without EMI option	Margin overestimated by 10 to 20 dB versus quasi-peak limit
No low-noise amplifier on the near-field path	Low-amplitude radiators buried under the analyser noise floor
Cable harness common-mode current not checked	EUT optimised, harness change at formal test breaks the result
Wrong RBW or VBW for the CISPR band	Numbers not comparable to CISPR limits, false confidence
Cell factor or antenna factor not applied	Reading in dBmicroV used directly as if dBmicroV/m, off by tens of decibels
Calibration overdue on cells or antennas	Measurement no longer traceable, dossier rejected at formal lab review
Skipping the bare-board near-field scan	Obvious radiators discovered only in the chamber, expensive iteration

Going further

EMC chamber types: SAC, FAR, OATS, reverberation: comparative view of the chamber types referenced above
Calibration and measurement uncertainty (GUM): uncertainty budgets and calibration cadence for the bench
PCB design for EMC: upstream layout discipline that pre-compliance verifies
IEC 61000-4-3 radiated RF immunity: the immunity context for radiated test chambers
Radiated emissions EMC test: the formal accredited test that pre-compliance prepares
Glossary: definitions of TEM, GTEM, LISN, ICN, quasi-peak, antenna factor, cell factor