Insights · Technical note

An EMI receiver is not a spectrum analyser with a CISPR badge

Why a sine-wave calibration, however good, cannot show that a receiver complies with CISPR 16-1-1, what Annex K actually asks for, and what to look for on the certificate.

The certificate looks complete. Frequency accuracy, amplitude accuracy at a dozen frequencies, attenuator steps, resolution bandwidths, displayed noise. Every number is in tolerance and the header says "calibrated in accordance with CISPR 16-1-1". Then the assessor asks for the quasi-peak pulse response and there is nothing to show. Nothing was hidden. The receiver was calibrated as a spectrum analyser, and a spectrum analyser calibration has no reason to apply a pulse to the input.

In short
  • A spectrum analyser calibration verifies the instrument with sine waves. CISPR 16-1-1 defines an EMI receiver by its response to pulses: the detector weighting, the impulse bandwidth and the meter time constants.
  • Since Edition 4.0 (2015) the normative Annex K sets a minimum verification set, Table K.1: VSWR, sine-wave voltage tolerance, response to pulses and selectivity. Three of the four cannot be done with a sine wave alone.
  • A certificate that reports only the sine-wave parameters can be correct and traceable and still say nothing about whether the receiver measures a CISPR emission correctly.

This note is not about any laboratory. Spectrum analyser calibration is a mature, well-run service almost everywhere. The point is that an EMI receiver is a different instrument in the one respect that matters for a compliance measurement, and the standard that defines it says so in its own terms.

What makes a receiver a receiver

A spectrum analyser is specified as a linear measuring instrument for sine waves: how accurately it reads the level of a CW signal, how accurately it tunes, how its filters are shaped, how much it compresses. An EMI receiver is specified by CISPR 16-1-1 for something else: how it indicates a disturbance, which is usually a train of pulses, through a defined bandwidth and a defined detector. The quasi-peak detector exists to weight pulses by their repetition rate, the way a listener would be annoyed by them. The average and RMS-average detectors have specified time constants so that intermittent and drifting signals read the same on every receiver. The measurement bandwidths are defined as 6 dB bandwidths with a shape mask, and above 1 GHz as an impulse bandwidth. None of this is exercised by a sine wave, because a sine wave has no repetition rate and no pulse spectrum. Every detector reads the same level for a CW signal. That is exactly why a CW calibration cannot tell the detectors apart.

+20 dB0 dB−20 dB−40 dB 1 Hz10 Hz100 Hz1 kHz10 kHz Pulse repetition frequency (log scale) Indication relative to the reading at 100 Hz, CISPR band B Peak Quasi-peak Average A sine-wave calibration is not on this chart. Every detector reads the same for CW.
The quasi-peak curve is drawn from the band B values in CISPR 16-1-1: +4.5 dB at 1 kHz, 0 dB at 100 Hz, −6.5 dB at 20 Hz, −10 dB at 10 Hz, −22.5 dB at 1 Hz. The average detector follows the repetition frequency directly. The peak detector does not weight at all. These curves are what a receiver calibration verifies, and what a CW calibration never touches.

What Annex K asks for

Until 2015 the calibration of a receiver was left to the manufacturer's performance test and the laboratory's judgement. Edition 4.0 of CISPR 16-1-1 added a normative Annex K, and its foreword names it as the main technical change. Annex K says the calibration shall determine, traceably, whether the receiver complies with its published specifications, and it sets a minimum set of parameters that shall be in the verification, whichever process is used. That set is Table K.1.

Table K.1 parameterWhat the standard requiresCan a sine-wave calibration cover it?
VSWR of the inputNot to exceed 2.0:1 at 0 dB attenuation and 1.2:1 at 10 dB or more, 9 kHz to 1 GHz, with wider limits above 1 GHz. Both attenuator settings at each suggested frequency.Yes, if both attenuator settings are measured. The 10 dB condition is the one most often missing.
Sine-wave voltage tolerance±2 dB to 1 GHz and ±2.5 dB above, from a 50 Ω source, at the start, centre and stop of each CISPR band, for every detector the receiver implements.Yes, if it is done at the CISPR band points and per detector, not at a generic set of spectrum analyser frequencies.
Response to pulsesAbsolute: a pulse of specified impulse area at each band's repetition rate shall indicate within ±1.5 dB of a 60 dB(µV) sine wave, for quasi-peak, peak, average and RMS-average. Relative: the indication shall follow the repetition-frequency curves. Intermittent: the meter time constants shall give the specified drop for a 1.6 s gated signal.No. This needs a calibrated pulse generator and a pulse-modulated signal generator whose pulsed amplitude has been verified (Annex L in Edition 5.0). A CW calibration has no stimulus for it.
SelectivityThe overall pass-band shape at the centre of each band against the CISPR mask, at the 200 Hz, 9 kHz and 120 kHz reference bandwidths and the 1 MHz impulse bandwidth. The 3 dB width from this curve sets the ceiling for the average repetition test, so K.3 makes the sequence mandatory.Partly at best. A spectrum analyser calibration checks the 3 dB width of its own filters. It does not trace the CISPR pass-band mask, and it does not feed the result into the pulse test that depends on it.

Two of the four rows can be satisfied by a laboratory that already calibrates spectrum analysers, provided it moves its test points to the CISPR bands and repeats the level test for each detector. The other two cannot. They need a different stimulus, a different reference and a different sequence, and they are the rows that make the instrument a receiver.

Where the sine-wave certificate runs out

Set the two certificates side by side and the gap is a list, not an opinion.

Checked on a typical spectrum analyser calibrationRequired by CISPR 16-1-1 and absent
Frequency readout and span accuracyQuasi-peak absolute pulse response, 60 dB(µV) ±1.5 dB, each band
Amplitude accuracy with CW at manufacturer frequenciesPeak absolute pulse response at the impulse area 1.4 / Bimp mVs
Reference level and attenuator stepsAverage and RMS-average absolute response to a pulse-modulated carrier
Resolution bandwidth 3 dB accuracyQuasi-peak, average and RMS-average variation with repetition frequency
Displayed average noise levelIntermittent narrowband response, the meter time constants
Log fidelity, compressionSelectivity against the CISPR pass-band mask at the reference bandwidths
Input VSWR, sometimes at one settingResponse above 1 GHz to a pulse-modulated carrier, Annex E.6

The left column is worth having. It is the part of the receiver that is also a spectrum analyser, and an assessor will want it. But a certificate that stops there and carries the words "CISPR 16-1-1" in its header has verified the receiver against the clauses the receiver shares with any analyser, and skipped the clauses that exist only for it. Annex K allows a partial calibration. It does not allow a partial calibration to be presented as compliance.

The badge on the front does not decide

The reverse is also true, and it catches laboratories that use a spectrum analyser with EMI options for compliance measurements. Annex K applies to EMI receivers and to spectrum analysers used for CISPR measurements. If the analyser is the instrument whose quasi-peak reading goes in the test report, then its quasi-peak detector, its CISPR bandwidths and its pulse response have to be verified in the same way. The manufacturer's performance test for the base analyser does not do this, because the base analyser is not sold as a receiver. The question to ask is not "what is the instrument called" but "which reading from it appears in a CISPR test report", and to calibrate that reading.

The edition matters, and so does the sequence

Two details separate a certificate that would survive a technical review from one that only looks complete.

The edition the certificate claimsAnnex K arrived in Edition 4.0 (2015). A product standard with a dated reference to the 2010 vintage, such as EN 55032:2015, cannot be held to Annex K on that basis alone. CISPR 25:2021 references Edition 5.0, where Annex K applies in full and Annex L adds the verification of the pulsed stimulus. A certificate should say which edition it is claiming against, and the laboratory reading it should know which edition its product standard invokes.
Selectivity before the average repetition testThe average detector's repetition-frequency test runs up to half the 3 dB bandwidth, and that bandwidth comes from the selectivity measurement of the same band. K.3 makes the sequence mandatory. A certificate that reports the repetition test without the selectivity result it depends on has skipped a prerequisite, whatever the numbers say.
Conformity with the uncertainty appliedK.6 requires the expanded uncertainty to be applied to the result and defines four outcomes, two of which are inconclusive. If a laboratory resolves the inconclusive cases with its own decision rule, that rule belongs on the certificate, stated as the laboratory's, not as a requirement of the standard.
The stimulus verified before useSignal generators do not guarantee the relationship between CW and pulsed amplitude. In Edition 5.0 the average and RMS-average amplitude tests invoke Annex L, so the pulse amplitude verification should be visible on a certificate that claims them.

What to look for on a receiver certificate

  • All four Table K.1 parameters present, over the ranges and detectors the instrument implements.
  • VSWR reported at both 0 dB and at 10 dB or more of input attenuation.
  • Absolute pulse tests performed at the specified level, with the input level shown, not only the deviation.
  • The average repetition response tested up to half the 3 dB bandwidth, with the selectivity result that established it.
  • The intermittent narrowband test present for both the average and the RMS-average detector.
  • Response above 1 GHz present where the receiver is specified above 1 GHz.
  • Any deviation from the suggested frequencies or from the manufacturer's process declared and justified.
  • A conformity statement that gives the expanded uncertainty and the decision rule applied.

If the certificate in your file is a sine-wave calibration with a CISPR header, the practical step is not to discard it. Keep it for what it covers, and add the pulse and selectivity work for the bands and detectors you report. Annex K's partial-calibration clause allows exactly that, provided each certificate says what it is.

Where we stand

The accredited scope we calibrate under lists both halves separately: absolute amplitude accuracy, reference level, attenuator, log fidelity and frequency readout to 40 GHz, and CISPR detector response, pulse response and bandwidths from 9 kHz to 1 GHz. The EMI receiver calibration guide shows the parameters and ranges. This note draws on a technical brief on CISPR 16-1-1 Annex K prepared by Mr. Takashi Maruyama, President of LAB Support Ltd., verified against Edition 4.0 with its 2018 interpretation sheet and Edition 5.0. If you hold a receiver certificate and are not sure which half it covers, send it to us.

Tell us what you need to test

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

Contact us