Common-Mode Choke Measurements With a VNA: What the Trace Really Means

The question I ask before believing any choke plot is simple: where does the test current actually flow? “S21,” “differential fixture,” “balun jig” and “current injection” are labels. The wiring, reference conductor and calibration planes decide which mode the instrument excites and which quantity the trace represents.

Start With the Mode, Not the Instrument Menu

In a coaxial feed line, the wanted differential mode carries equal and opposite current on the centre conductor and the inside surface of the shield. The unwanted exterior mode involves net current on the outside of the shield and whatever return path the installation supplies through the antenna, mast, station wiring, bonding and surrounding capacitance.

A common-mode choke is meant to insert impedance in that exterior path while disturbing the wanted differential transmission as little as practical. Those are two different measurands:

• Common-mode impedance, Z_CM(f) = R_CM + jX_CM: the complex small-signal series impedance presented to the defined common-mode current path.
• Differential transmission: insertion loss, return loss and any mode conversion seen by the wanted signal through the feed line.

A two-port coaxial VNA can support both experiments, but not in the same connection. A series fixture can place the defined common-mode path between the two VNA ports. A normal through-line connection can characterize the wanted coaxial transmission. A balanced multiport measurement can separate differential, common and conversion terms mathematically when every conductor is measured and the mixed-mode reference impedances are declared.

Rohde & Schwarz’s balanced-component guidance makes the same distinction: single-ended measurements can be transformed into mixed-mode parameters, but the port definitions and reference impedances must be part of the measurement.

Fixture Names Do Not Decide What Was Measured

Amateur discussions often dismiss or endorse a setup by name: a DG8SAQ/VNWA differential-port fixture, a VE2AZX-style balun jig, or a simple S21 through fixture with 50 Ω ports. None of those names is a complete circuit diagram.

A differential fixture measures differential behaviour when its two conductors are driven in opposition. It can also participate in a common-mode measurement when its conductors and references are connected to force in-phase current through the intended common-mode branch. A centre-tapped transformer can create a useful modal stimulus, but it is not the only valid way to make a two-terminal series-impedance measurement.

Conversely, a fixture described as “common mode” can still be wrong for the intended DUT if VNA cable exteriors, exposed leads, enclosure capacitance or the bench provide a lower-impedance bypass. The schematic must include:

• both VNA ports and their common chassis reference;
• every conductor entering and leaving the choke;
• connector shells, shields, fixture ground and enclosure;
• the calibration reference planes;
• lead length, conductor spacing and DUT orientation; and
• any path through cable exteriors, nearby metal or displacement capacitance.

If current can reach port 2 without passing through the intended common-mode branch, that leakage belongs to S21. A smooth trace does not reveal the bypass by itself.

Series-Through S21 Can Yield Absolute Impedance

For an ideal series impedance Z_DUT placed between two equal, matched VNA ports with reference impedance Z₀, the complex transmission is:

This is not an improvised ham-radio shortcut. Keysight documents the port-1-to-port-2 series method for middle-to-high impedances and gives the same conversion for 50 Ω ports. Keysight also specifies calibration or fixture compensation at the DUT terminals and publishes a bounded accuracy range; the equation is not permission to ignore the fixture.

The formula uses the complex S21 value, not only its magnitude in decibels. If the fixture has significant shunt capacitance, direct transfer, mismatch or distributed behaviour, the simple one-branch model no longer contains the whole circuit.

Y21 Is a Model-Based Branch Extraction

A full two-port S matrix can be converted to an admittance matrix. If the calibrated DUT and fixture are adequately represented by a π network with one transfer branch between the ports and separate local shunts to the reference conductor, then:

This is why Y21 is useful for high choke impedance: separate capacitance from each terminal to the local fixture reference appears mainly in Y11 or Y22 rather than in the transfer branch. The result remains conditional on the π model. Capacitance directly between the ports, magnetic coupling between fixture halves, radiation and cable common mode still appear in Y21 because they transfer current from one port to the other.

Y21 and series-through S21 are not competing physical truths. They are two ways of interpreting calibrated network data under different equivalent-circuit assumptions. Keysight’s full two-port conversions use S11, S21, S12 and S22 to form admittance and other network parameters. That makes the assumed data set worth checking: a file containing placeholders or copied reverse terms is not a measured full matrix.

Complex Zcm Says More Than a dB Number

A choke curve should publish R_CM, X_CM and |Z_CM| against frequency. The resistive and reactive parts matter for different reasons:

• R_CM represents the dissipative part of the small-signal common-mode impedance. Under operating current it contributes to heating, although material nonlinearity and temperature can change the value.
• X_CM represents stored magnetic or electric energy. A large reactance can reduce current in a defined circuit without the same dissipation, but its sign and magnitude can change around resonance.
• |Z_CM| is convenient for current-divider estimates only when the rest of the installed common-mode loop is known.

A number such as “40 dB choke” is incomplete unless it states the circuit, reference impedance, frequency and definition. S21 insertion loss between two 50 Ω ports is not an intrinsic attenuation rating for every antenna installation. In the station, current reduction depends on the choke impedance in series with the complete external source and return network, plus any parallel bypass paths.

Ferrite behaviour is also frequency- and material-dependent. Fair-Rite’s technical catalogue separates the real and imaginary parts of complex permeability and shows that their relative contributions change with frequency. Core material, geometry, turns, winding capacitance and the actual cable construction therefore belong with the curve.

Resonance Can Belong to the DUT or the Fixture

An impedance maximum can be the useful parallel or self-resonant region of the finished choke. It can also be moved, sharpened or even created by fixture capacitance and a leakage path. The answer comes from controlled changes, not from the peak alone.

Repeat the sweep after reversing the DUT, changing fixture spacing, shortening exposed leads and moving nearby metal. Measure an empty fixture, suitable open/short/through checks and known R, R-L or R-C standards near the intended impedance range. A real DUT feature should remain explainable as the fixture is varied; a moving peak is evidence that the measurement boundary needs work.

As Z_CM rises in a series-through test, S21 falls toward the receiver noise and leakage floor. Averaging and narrower IF bandwidth can improve repeatability, but they cannot remove a deterministic bypass. State the instrument settings, residual directivity and transmission floor, and stop assigning precise impedance after the uncertainty becomes comparable to the result.

Calibration and De-Embedding Must Reach the Choke Terminals

A calibration removes systematic errors only to its declared planes. Adapters, launch boards, clips, exposed conductors and the common-mode fixture remain part of the result when they sit beyond those planes. Keysight’s impedance-calibration procedure uses open, short and load standards in the impedance domain for this reason.

De-embedding is valuable only when the fixture model has been independently characterized over the required range. Port extension can correct delay; it cannot subtract arbitrary shunt capacitance or direct electromagnetic coupling. A responsible report keeps the raw and corrected data, names the standards and planes, and shows that a known DUT is recovered after the same processing.

Differential Loss Needs Its Own Test

A feed-line choke must impede exterior current without needlessly degrading the wanted differential signal. Common-mode Z alone does not establish differential insertion loss, return loss or power handling.

Measure the completed choke as the transmission line it will carry, with the intended connectors, cable and reference impedance. Record differential S21 and S11—or the corresponding mixed-mode Sdd21 and Sdd11 when using a balanced multiport setup—across the operating range. Also inspect differential-to-common conversion where an asymmetric winding or fixture could convert the wanted mode.

A low differential insertion loss is good, but it does not prove high common-mode impedance. A high common-mode impedance is good, but it does not prove low differential loss. Publish both.

Current Injection Answers a Different Question

A calibrated injection transformer or clamp and a current probe can test how a defined cable network responds to an imposed disturbance. That is useful evidence, especially when the source impedance, load, cable layout, probe transfer impedance and measurement positions are declared. It is not automatically a direct measurement of the choke’s standalone Z_CM.

The broad statement that one current-injection arrangement is “the EMC gold standard” for choke impedance goes too far. IEC 61000-4-6:2023, for example, defines repeatable conducted-RF immunity testing of electrical and electronic equipment from 150 kHz to 80 MHz. Its injection devices and level-setting procedures assess equipment behaviour in a specified cable system; they do not standardize intrinsic amateur-radio choke impedance.

CISPR, IEC, ISO and military standards each define particular equipment, cable, emission, immunity or susceptibility tests. Cite the exact document and edition only when following that test. A family name does not validate an unrelated fixture.

Small-Signal Data Is Not a Power Rating

A VNA normally characterizes the linear small-signal state. Operating power adds common-mode current, differential current, RF voltage, core flux, conductor loss, dielectric stress, temperature rise, duty cycle and cooling. Those conditions can move impedance and resonance or expose a failure mode that a milliwatt sweep cannot show.

Power qualification requires a separately rated fixture, calibrated power planes, the intended waveform and duty cycle, representative mismatch or common-mode excitation, remote temperature measurement and stated limits. Current and thermal tests must cover the completed assembly, not just an unloaded core on the bench.

The construction should be judged by those measurements. Semi-rigid coax, flexible coax, tubing, winding form and core arrangement are neither credible nor incredible because they look unusual. Novelty is not evidence; a complete set of modal, loss and stress data is.

Installed Proof Closes the Loop

Even a well-measured choke is only one impedance inside a larger common-mode circuit. To test the installation, clamp a calibrated current probe around the complete feed line and map net exterior current at several positions before and after the choke is installed. Keep frequency, power, antenna configuration, tuner state, feed-line route and station bonding unchanged.

One probe position can be misleading because adding impedance redistributes current and can move a maximum or minimum. A multi-position map shows whether current fell along the relevant feed-line section. Receiver noise, shack RFI or pattern claims need their own controlled A/B measurements; they are not guaranteed consequences of one component trace.

A Measurement Report Worth Trusting

For every choke curve, publish enough information for another operator to reproduce the boundary:

• a circuit and physical photograph of the common-mode fixture;
• instrument, calibration kit, reference planes and sweep settings;
• the complete extraction equation and equivalent-circuit assumption;
• R_CM, X_CM and |Z_CM|, not only S21 magnitude;
• fixture-leakage, reversal, spacing and known-standard checks;
• separate differential insertion and return-loss data;
• powered loss, temperature and stress boundaries where a power claim is made; and
• multi-position installed current measurements for an installation claim.

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