Common-Mode Choke Test Jigs: Prove You Measured the DUT

The problem is not using a VNA or a 50 Ω reference. The problem is failing to state which network the VNA measured and whether that network represents the intended common-mode DUT.

Raw S21 Is a Valid Measurand—with a Narrow Name

S21 is the complex forward travelling-wave ratio through the entire network between two calibrated reference planes. S21 LOGMAG is its magnitude in dB. Both are valid measurements.

What is invalid is silently renaming fixture transmission as:

• intrinsic common-mode rejection;
• installed cable-current reduction;
• CMRR of connected equipment;
• power-handling capability; or
• a universal “25 dB” or “40 dB” choke property.

Report S21 as fixture transmission unless a validated network model converts it to another quantity.

Using a 50 Ω VNA Is Not the Error

S-parameters need reference impedances, and most RF VNAs use 50 Ω. That does not restrict the instrument to devices that operate in 50 Ω systems. A calibrated 50 Ω VNA can characterise antennas, high impedances, low impedances and arbitrary networks when the correct topology and conversion are used.

The wanted coaxial line mode may also have a 50 Ω characteristic impedance, but that is a separate fact. The external mode on a cable has an installation-dependent return through conductors, structures and fields. Do not equate the VNA's wave reference with that installed external-mode path.

First Draw Every Terminal

A fixture advertised as a two-port jig may be a three-terminal or distributed network once VNA chassis, connector shells, shields, unused sockets, bench capacitance and nearby metal are included.

“The bench is part of the circuit” becomes actionable only after these paths and the calibration boundary are named.

Terminal-to-Terminal Capacitance Is Not Automatically Fixture Error

A choke's winding and cable have distributed capacitance. If that capacitance connects the two DUT terminals, it belongs to the finished two-terminal impedance. It can create self-resonance or an upper-frequency bypass, but the effect is real component behaviour.

Capacitance from either terminal to an unintended third reference is different. It can bypass the series model through the VNA chassis, bench or ground plane. The endpoints, not the word “stray,” decide whether capacitance belongs to the DUT or fixture.

The Valid Series-Through Extraction

For a two-terminal series impedance between two equal real port impedances Z0:

This is a legitimate impedance method, not merely a rough comparison. Keysight's current series-through measurement guidance gives the same conversion and requires calibration or fixture compensation at the DUT terminals.

Use complex linear S21, not only LOGMAG. Validate the range with known impedances and establish where residual port coupling or the receiver noise floor dominates.

How a Jig Starts Measuring Itself

• Reference planes are too far away. Leads, switches and adapters become part of the extracted impedance.
• Port nodes have independent shunts. Capacitance to chassis or bench creates a third-terminal network.
• Ports couple directly. Electromagnetic leakage bypasses the intended DUT path.
• Unused connectors remain active. Their stubs and shells alter current and voltage distribution.
• Common-mode cable paths are uncontrolled. VNA leads and fixture shields form the return circuit.
• The DUT becomes distributed. One lumped series impedance no longer represents the structure.
• S21 reaches the fixture floor. A stable leakage trace is mistaken for high choke impedance.

Frequency Does Not Set a Universal Failure Point

For 2 pF, capacitive reactance magnitude is approximately 2.65 kΩ at 30 MHz, 796 Ω at 100 MHz and 265 Ω at 300 MHz. Those values matter only in relation to the rest of the network and the capacitance endpoints.

A carefully designed fixture can remain useful at a higher frequency than a large improvised one. A poor fixture can fail on lower HF when it attempts to resolve very high impedance. Replace “higher frequency is less believable” with a validated range based on standards, perturbation and uncertainty.

Five Tests That Expose Fixture Dominance

Repeatability under one untouched setup proves precision, not necessarily validity.

What Y21 Adds

For equal real port references, a full complex S matrix can be converted to an admittance matrix:

Y = (I − S) × inverse(I + S) / Z0

In a lumped reciprocal π network with independent shunt admittances at the ports and one branch between them:

Zseries = −1 / Y21

Equivalently, for equal Z0:

Zseries = Z0 × ((1 + S11 + S22 + S11S22 − S12S21) / (2S21))

Y21 can therefore remove the effect of independent port-to-reference shunts within that validated model.

What Y21 Does Not Add

Y21 cannot tell wanted choke transfer admittance from unintended direct port-to-port coupling; both contribute to the off-diagonal term. It cannot make a distributed multi-conductor fixture into a unique lumped series branch, restore missing dynamic range or correct bad calibration.

The method also requires the full two-port data set. A forward-only T/R instrument does not directly supply S12 and S22. Any reversed measurement must preserve calibration, state and reference planes.

Use Y21 as a model-based extraction and validate the model. Do not replace an S21 slogan with a Y21 slogan.

Coax Common Mode Is Easier to Excite—Not Trivial

For coax, the complete cable can pass through a current probe or ferrite aperture. The intended internal equal-and-opposite mode largely cancels, so net longitudinal current indicates an external mode.

But the return path through VNA bodies, shields, fixtures and surrounding capacitance still determines the current. A jig must define that path or explicitly report its dependence.

A low-SWR differential coax connection does not make the external-mode fixture 50 Ω.

USB, Ethernet, Audio and DC Need Modal Definitions

“Outside-shield current” is not a general definition for every cable. An unshielded pair has no shield exterior; a USB or Ethernet cable can support differential and common combinations on signal pairs, shields and ground conductors.

A serious fixture must define:

• which conductors are tied or driven together;
• which conductor or structure provides the return;
• port impedances for the chosen mode;
• whether the cable is terminated in its operational differential impedances;
• shield-to-chassis connections at both ends; and
• which mixed-mode quantities are being measured.

Placing several connector types in parallel and plotting one single-ended S21 trace does not create equivalent common-mode tests.

Current Injection Answers a System Question

A defined injection fixture plus a calibrated current probe can measure how external cable current changes when a choke is inserted. That is a current-reduction result for the specified cable, return path, position and termination—not an intrinsic DUT dB rating.

The current IEC 61000-4-6:2023 standard concerns conducted RF immunity of equipment. It illustrates the care needed in injection devices, cable conditions and repeatability. A ham jig inspired by the topology is not automatically an IEC test or a direct impedance measurement.

For absolute current, apply the probe transfer impedance and path-loss corrections. For relative before/after results, keep probe plane, geometry, source state and receiver settings unchanged, then scan additional positions to see how the distributed current moved.

What to Report

A Fixture-Audit Workflow

1. Draw the full network. Include VNA grounds, shells, unused connectors and environmental capacitance.
2. Choose calibration planes. Move them to the DUT terminals or apply validated compensation.
3. Measure an open and known standards. Establish leakage and accuracy limits.
4. Acquire complex data. Use the full S matrix when applying Y21.
5. Perturb suspected fixture paths. A DUT-only result should not follow an unintended bench change.
6. Compare models. Series S21 and Y21 should agree where their assumptions overlap.
7. Define the cable mode. Especially for multi-conductor digital and power interfaces.
8. Verify current separately. Use a controlled fixture and then the installed system.

The Practical Verdict

A common-mode choke jig can provide accurate, useful data. Raw S21 is a valid description of that fixture's transmission, and it can yield impedance when the series model is valid.

The proof comes from boundary definition, standards, leakage tests, model validation and current verification. Without them, a beautiful graph may indeed be the jig measuring itself.

Questions, antenna-factor records or height trials to share? Contact RF.Guru.