A question we hear in different forms: "The certificate says the LISN was calibrated. Why did the assessor raise a finding?" The answer is usually in what was measured, not who measured it. Line impedance stabilisation networks, impedance stabilisation networks and coupling/decoupling networks are not ordinary RF components. The standards that define them do not describe them by S-parameters. They describe them by parameters at a defined reference plane, under defined terminations, against a tolerance mask. A calibration is complete when it reports those parameters against that mask.
- The instrument is common. The method is specific to the device.
- A certificate has to show four things: the parameter as the standard defines it, the reference plane and terminations, the result against the standard's tolerance, and an uncertainty that belongs to that parameter.
- A scope line that says "insertion loss" or "S-parameters, 50 Ω" describes a measurement. It does not by itself describe a LISN, ISN or CDN calibration.
This note is not about any laboratory. Plenty of general calibration laboratories do this well. It is about knowing what to look for, because the scope line, not the logo on the certificate, tells you whether the method is covered.
The standard defines the parameter, not the instrument
CISPR 16-1-2 specifies artificial mains networks (the AMN or LISN) and impedance stabilisation networks (ISNs) for conducted emission measurement. IEC 61000-4-6, in its Annex A, specifies coupling/decoupling networks (CDNs) for conducted immunity. None of these clauses mention a network analyser. They state the quantity, the port it is defined at, what the other ports are connected to while it is measured, and the tolerance.
| Device | What the standard defines | Where and under what conditions |
|---|---|---|
| LISN / AMN CISPR 16-1-2 |
Impedance magnitude and phase at the EUT terminals against the 50 Ω / 50 µH curve, with a tolerance of ±20 % and ±11.5°. Voltage division factor between the EUT terminals and the receiver port. Isolation between the mains terminals and the receiver port. | 9 kHz to 30 MHz for the 50 µH network. Receiver port terminated in 50 Ω. Mains terminals in the conditions the standard states. Reference plane at the EUT terminals, which means the adapter that connects them to a coaxial port is part of the measurement. |
| ISN CISPR 16-1-2, CISPR 32 |
Common-mode impedance at the EUT port, 150 Ω with a magnitude and phase tolerance. Longitudinal conversion loss (LCL) to the category value the ISN is built for. Voltage division factor to the receiver port. Decoupling between the auxiliary equipment port and the EUT port. | Over the conducted emission band, with the AE port terminated as the standard requires. LCL is a balance measurement per the ITU-T method the standard references, not a two-port transmission. |
| CDN IEC 61000-4-6 Annex A |
Common-mode impedance at the EUT port, 150 Ω with a tolerance that changes with frequency band. Voltage division factor from the RF input to the EUT port, which the level-setting procedure relies on. Decoupling towards the AE port. | 150 kHz to 80 MHz at least, 230 MHz where the test goes that far. All EUT terminals connected together to the 150 Ω to 50 Ω adapter, AE port terminated as specified. The adapter is part of the definition, and it needs its own calibration. |
Why a plain S21 is not a voltage division factor
A network analyser reports S21: the wave leaving port 2 divided by the wave incident on port 1, both referenced to 50 Ω. The voltage division factor of a LISN is something else: the voltage at the EUT terminals divided by the voltage at the receiver port. The two are equal only when the EUT port looks like 50 Ω. A 50 µH network does not. Its EUT-port impedance is about 5 Ω at 9 kHz, around 34 Ω at 150 kHz, and only approaches 50 Ω towards 30 MHz. Below that, part of the analyser's incident wave is reflected before it ever becomes voltage at the terminals, and a raw S21 reading absorbs that reflection into the "division factor".
The size of the difference is not academic. Using the nominal impedance curve from CISPR 16-1-2:
| Frequency | Nominal EUT-port impedance | Raw S21 versus true division factor |
|---|---|---|
| 9 kHz | about 5 Ω, 27° | differs by roughly 14 dB |
| 150 kHz | about 34 Ω, 47° | differs by about 1 dB |
| 30 MHz | about 50 Ω, near 0° | under 0.1 dB |
One decibel at 150 kHz is the difference between a pass and a fail for a product sitting near the CISPR 32 or CISPR 14-1 limit line. Fourteen decibels at 9 kHz would make a CISPR 15 or CISPR 11 group 2 result meaningless. There are two correct ways to get the division factor from an analyser: measure S11 and S21 together and convert to the terminal voltage ratio, or measure through a defined adapter at the EUT terminals whose own behaviour has been characterised. Both are legitimate. Both require the impedance and the adapter to be part of the method and of the uncertainty budget. That is the difference between owning a network analyser and having a LISN method.
Impedance is a quantity at a reference plane, not a return loss
The same logic applies to impedance. S11 or return loss is referenced to 50 Ω at the analyser's port. The standard's tolerance is in ohms and degrees at the EUT terminals, which sit behind an adapter: banana sockets, a mains socket or a terminal block converted to a coaxial connector. The adapter has to be characterised or de-embedded, the receiver port has to be terminated in 50 Ω, and the other terminals have to be in the state the standard describes. The phase tolerance of ±11.5° is the one that catches people out. A few centimetres of unaccounted adapter shift the phase by degrees at the top of the band, and a phase failure is as much a non-conformance as a magnitude failure.
ISNs: LCL is not an attenuation
An impedance stabilisation network for a telecom or data port has one parameter that no two-port transmission measurement can produce: longitudinal conversion loss. LCL describes how well the network keeps a common-mode signal from converting into a differential signal on the pair. It is measured by the balance method CISPR 16-1-2 references from the ITU-T, with a fixture whose own balance is well beyond the value being verified. An ISN built for category 6 cabling is specified at 75 dB, and the fixture has to be better than that with margin. A calibration that reports insertion loss and calls it an ISN calibration has measured the division factor at most. The common-mode impedance, the LCL and the decoupling are separate parameters, and each has its own line on a proper scope.
CDNs: the adapter is part of the definition
For a coupling/decoupling network the standard is explicit about the arrangement. The common-mode impedance is measured with all the EUT terminals connected together and taken to the analyser through the 150 Ω to 50 Ω adapter, with the AE port terminated as specified. The tolerance is 150 Ω with a band that is tighter below 26 MHz and wider above it. The division factor is measured through the same adapter, because that is how the level-setting procedure in the test itself uses it. Measure the CDN as a bare 50 Ω two-port and the numbers on the certificate are not the ones the test method needs. The adapter, in turn, needs its own calibration for insertion loss and return loss, and it should appear on the scope in its own right.
How to read a scope in two minutes
Any accredited laboratory's scope is public. Before you accept a quotation for a LISN, ISN or CDN, or before your next assessment, open the scope of the laboratory whose certificate you hold and check five things.
What an assessor can reasonably ask
ISO/IEC 17025 asks a laboratory to calibrate equipment where the accuracy or measurement uncertainty affects the validity of its results, and to establish traceability for the quantities it actually uses. In a conducted emission test the quantity you use is the division factor you add to the receiver reading and the impedance that defines the measurement. In a conducted immunity test it is the CDN impedance and the division factor behind the level setting. A certificate that reports a 50 Ω transmission measurement of the same box does not establish traceability for those quantities, however good the analyser was. That is the finding, and it is a finding against the method, not against the laboratory that holds the network.
The accredited scope we work under lists LISN, AMN and ISN calibration and CDN calibration by parameter: impedance magnitude and phase, voltage division factor, isolation and LCL for artificial networks, and impedance, division factor and decoupling for CDNs, with the 150 Ω adapters as their own line. The LISN and ISN and CDN guides show the parameters and ranges. If you hold a certificate from elsewhere and are not sure what it covers, send it to us. We will tell you what it does and does not establish, before your assessor does.