Insights · Technical note

How CISPR 25 learned to validate its own chamber

CISPR 25 has changed its mind three times about how to prove that an absorber-lined shielded enclosure is fit for component measurements. A short history across four editions, then the method that survived: the long wire antenna method of Edition 5, how it works, what it is sensitive to, and where it does and does not apply.

A CISPR 25 component measurement is not made the way a CISPR 32 measurement is. The antenna stands one metre from a harness that lies 50 mm above a bonded metal table, inside a chamber that is often only a few metres across. Normalised site attenuation, which validates a 3 m or 10 m site with two antennas over a floor, says little about that arrangement. The standard has had four different answers to that problem in twenty years, and for eight of those years it had none at all, so laboratories whose results disagreed below 100 MHz had no agreed way to find out which chamber was responsible.

In short
  • The second edition (2002) limited the error from reflections to 6 dB and offered an informative comparison with an open site using a noise source. The third edition (2008) deleted that procedure and asked only for 6 dB of absorber performance, while a joint task force worked on a replacement.
  • The fourth edition (2016) introduced ALSE performance validation from 150 kHz to 1 GHz with two alternative methods: a reference measurement method and a modelled long wire antenna method.
  • The fifth edition (2021) keeps only the long wire antenna method, in Annex I, and clause 4.3.4.2 now says it shall be used for a component ALSE. The criterion is a deviation within ±6 dB from the modelled reference at 90 % or more of the 481 frequencies.

Four editions, four answers

EditionWhat the standard asked of the ALSEWhat that left open
Edition 2
2002
Clause 4.4.1: the maximum error caused by energy reflected from the walls and ceiling shall be less than 6 dB from 70 MHz to 1 000 MHz. An informative Annex G described how to check it: a standard noise source in place of the EUT, on the standard 1 500 mm harness 50 mm above the ground plane, measured at 1 m on an open site and again in the enclosure. It was recommended for enclosures of at least 7.0 m by 6.5 m by 4.0 m, and the difference could not be used as a calibration factor.The noise source was not specified beyond "stable", nothing below 70 MHz was covered, and the procedure was informative.
Edition 3
2008
The reflection requirement became a material requirement: absorber performance of 6 dB or better from 70 MHz to 2 500 MHz, with IEEE Std 1128 noted as a test method. The foreword records the deletion of the annex on characterisation of the shielded enclosure, to be replaced when the joint CISPR/D and CISPR/A task force on chamber validation finished its work. Absorber on the floor was no longer allowed.Everything about the chamber as set up: the ground plane, its bonding, the room size and all frequencies below 70 MHz, which is where component chambers disagree most. This gap lasted eight years.
Edition 4
2016
The foreword lists "methods for chamber validation" as a main change. An informative Annex J, 150 kHz to 1 GHz, offered two alternatives, either of which could be used: a reference measurement method, comparing the ALSE with the same measurement on a reference site using a small monopole, biconical or shortened dipole source, and a modelled long wire antenna method. Both used ±6 dB at 90 % of data points.The clause said the procedure "may be used". The reference method also depended on access to a reference site meeting CISPR 16-1-4 and on a transmit antenna whose construction was only loosely specified, so two laboratories could build different references.
Edition 5
2021
The reference measurement method is gone. Annex I carries the long wire antenna method alone, and clause 4.3.4.2 states that it shall be used to evaluate the ALSE as configured for the component radiated emission test of clause 6.5. The absorber requirement is extended to 5 925 MHz. No absorber on the floor; flat ferrite tiles up to 25 mm are allowed for component testing only if the chamber still meets Annex I.Validation above 1 GHz. The annex of items under consideration lists it as future work, so above 1 GHz the absorber requirement is still the only one.

Read in order, the editions show what the committee learned. A comparison with an open site, first with a noise source and later with a small antenna, depends on a reference that every laboratory has to create for itself. A modelled reference is the same for everyone. Two ideas from 2002 survive in 2021 unchanged: the source sits where the harness sits, and the difference between reference and chamber is a verdict on the site, never a correction for the data.

Why a long wire

In a CISPR 25 component test the thing that radiates is mostly the harness. It runs parallel to the front edge of the reference ground plane, 100 mm in from the edge and 50 mm above the metal. The validation source copies that geometry with something that can be built to a drawing and solved in a field simulator: a straight brass rod over the ground plane, fed at one end and terminated at the other. It is a short transmission line above a plane, and its field at the antenna position can be calculated. The standard did so with a method-of-moments code and printed the result as Table I.1.

Long wire radiator, side view reference ground plane brass rod, 4 mm diameter 10 dB feed cable laid on the plane and routed to the back 50 Ω termination attenuator 500 ±5 mm between the sheet angles 50 ±2 mm sheet angles bonded to the plane, ≤ 2.5 mΩ Heights are exaggerated. The rod stands in for the harness: same height, same distance from the front edge.
After Figure I.4 of CISPR 25. The construction is fixed closely because the printed reference data is only valid for it.
  • The radiator. A brass rod 4 mm in diameter and 500 mm between two metallic sheet angles, 50 mm above the reference ground plane and 100 mm in from its front edge, centred where the harness centre sits during testing. The angles are bonded to the plane, with no more than 2.5 mΩ of DC resistance.
  • Feed and load. One end is fed through a 10 dB, 50 Ω attenuator, the other is terminated in 50 Ω, both through type N connectors in the angles. The attenuator is part of the reference data.
  • The receive side. The same antennas, at the same 1 m distance and positions, as clause 6.5: the rod antenna below 30 MHz, the biconical to 200 MHz and the log-periodic to 1 GHz. Above 30 MHz both polarisations are measured and the larger is taken.
  • The quantity. An equivalent field strength for 1 V into the attenuator: 120 dB(µV) plus the measured transmission, plus the receive antenna factor. With a network analyser the transmission is simply S21 after a two-port calibration at the cable ends.
  • The frequencies. 481 points: 150 from 150 kHz to 29.95 MHz in 200 kHz steps, 170 from 30 MHz to 199 MHz in 1 MHz steps, and 161 from 200 MHz to 1 GHz in 5 MHz steps.
  • The chamber. Configured exactly as for a product test: the same ground plane size, bonding, absorbers and antenna counterpoise connection. The battery and artificial networks are left off the table.
Validation set-up, plan view absorber-lined walls, chamber as configured for testing reference ground plane size and bonding as used for testing bonding to the enclosure long wire radiator, 500 mm, at the centre of the harness position rod 100 mm in from the front edge receive antenna of clause 6.5 rod, biconical or log-periodic 1 000 ±10 mm feed cable with ferrites every 20 cm, on the plane, out at the rear battery and artificial networks removed
After Figure I.7 of CISPR 25. The modelled reference assumes a 2.5 m by 1 m plane with a single rear strap; the chamber is measured with whatever plane and bonding it really uses, and the comparison shows what that difference costs.

The criterion in I.2.4 is statistical. The deviation between the measured and the reference equivalent field strength is calculated at every frequency, and the ALSE complies if 90 % or more of the points are within ±6 dB. Where a chamber is used for only part of the range, only those ranges of Table I.1 are counted. The standard adds two restrictions that are easy to overlook: the difference shall not be used as a correction factor for product measurements, and it shall not be used to derive an antenna factor.

What a validated chamber looks like

The plots below are from a long wire validation we performed in a semi-anechoic chamber set up for CISPR 25 component measurements: reference ground plane, bonding and antennas exactly as used for testing, battery and artificial networks removed. The upper plot sets the measured equivalent field strength against the modelled reference. The lower plot is the difference between the two, which is the quantity the criterion judges.

Equivalent field strength Eeq,max, measured in the ALSE against the modelled reference of Table I.1 150 kHz 1 MHz 10 MHz 30 MHz 200 MHz 1 GHz 50 60 70 80 90 100 110 dB(µV/m) rod antenna biconical log-periodic measured in the chamber reference, Table I.1 Deviation Δ from the reference, against the ±6 dB criterion 150 kHz 1 MHz 10 MHz 30 MHz 200 MHz 1 GHz -8 -6 -4 -2 0 +2 +4 +6 +8 +6 dB −6 dB largest: +5.5 dB at 25.15 MHz 481 of 481 points within ±6 dB: 100 %, against a requirement of 90 %
Long wire validation of a chamber configured for CISPR 25 component testing, at the 481 frequencies of Table I.1. Below 30 MHz the rod antenna is used; from 30 MHz the larger of the horizontal and vertical results is plotted, as the annex requires.

The measured curve follows the reference across all three antenna ranges, and every one of the 481 points lies within ±6 dB, where the annex asks for 90 %. The shape of the deviation is as informative as the verdict. It is largest between 10 MHz and 50 MHz, reaching +5.5 dB near 25 MHz, which is the region the annex attributes to the ground plane, its bonding and the absorbers, and it stays within 4 dB everywhere else and within 3 dB above 300 MHz. A chamber that fails almost always fails in that same low-frequency region, and the plot shows where to look.

What the method is sensitive to

Annex I names three things that decide the result: the size of the reference ground plane, its bonding straps, meaning their number, their size and whether they run horizontally or vertically, and the absorber performance. Their influence is concentrated between 10 MHz and 100 MHz. Below 30 MHz the ground plane and the rod antenna's counterpoise form a resonant structure with the chamber, and the annex allows two remedies: a block of absorber between the bottom of the receive antenna and the floor, or additional bonding straps from the counterpoise to the floor, with a length to width ratio of no more than 7 to 1. Whatever is done to pass the validation then has to stay in place for every product measurement.

That is also why the validation stops at 1 GHz. The studies behind the annex found that the absorber and the ground plane bonding create their largest deviations below 200 MHz. Above 1 GHz the question is a different one, and the standard has not answered it yet.

Where it applies, and where it does not

QuestionAnswer in Edition 5
Which chamberThe component ALSE used for the radiated emission measurement of clause 6.5. The vehicle ALSE of clause 5 has the absorber requirement only.
Which frequencies150 kHz to 1 GHz, and only the ranges the chamber is used for. Above 1 GHz there is no validation method; the 6 dB absorber requirement to 5 925 MHz stands alone.
Which configurationThe one used for testing. The result belongs to the chamber together with its table, its bonding and its antenna positions. Change the ground plane, the straps, the absorber layout or the floor treatment and the validation has to be repeated.
Which editionA laboratory working to Edition 4 could choose either method. A laboratory claiming Edition 5 has the long wire method only, and a validation done by the reference measurement method does not carry over.
What it is notNot a correction for product data and not an antenna calibration. It is a pass or fail statement about the site, which the standard says may be included in the test report.
UncertaintyEdition 5's new annex on measurement instrumentation uncertainty leaves site imperfection out of the budget because site validation is still under study. The validation and the uncertainty budget are separate things for now.

Vehicle manufacturers' own specifications usually call up the CISPR 25 set-up for component emissions, so a chamber validated to Annex I is validated for the arrangement those specifications inherit. It does not replace NSA or site VSWR for a chamber that is also used for CISPR 32 or CISPR 11 work at 3 m, which is a different geometry with its own validation.

Before you book a validation

Your part

Fix the configurationTable size and position, bonding straps, absorber layout, floor treatment and antenna positions as used for testing. If the chamber is run in more than one configuration, each one is a separate validation.
Decide the frequency rangesOnly the ranges of Table I.1 that the chamber is used for are counted, so a laboratory that tests from 150 kHz to 1 GHz is judged on all 481 points.
Clear the tableThe battery, the artificial networks and any load simulator come off the reference ground plane for the validation.

The calibration laboratory's part

The radiatorBuilt to the annex's dimensions and tolerances, with its VSWR checked against the typical curves the annex prints, because the reference data is only valid for that construction.
The receive antennas and their factorsThe antenna factor enters the equivalent field strength directly, for the rod, the biconical and the log-periodic antenna.
Cables, ferrites and noise floorFerrites every 20 cm on both cables inside the chamber, the feed cable routed to the back of the table on the plane, and a noise floor at least 10 dB below the measured signal.
The statistics and the statementDeviation at every frequency, the percentage within ±6 dB for each range and in total, and a conformity statement against the 90 % criterion.
Where we stand

ALSE performance validation to CISPR 25 Annex I, from 150 kHz to 1 GHz, is on the accredited scope we validate under, alongside NSA, NSIL and site VSWR, and it is performed on site with a long wire radiator built to the annex. The site validation guide lists the parameters. If your component chamber was validated by the reference measurement method under Edition 4, or has never been validated at all, send us the chamber layout and the configuration you test in.

Tell us what you need to test

We reply within one working day with a quote or a plan.

Contact us