What a “40 dB Choke” Number Actually Means

When somebody calls a choke “40 dB,” my first question is not whether that sounds impressive. I ask: 40 dB of what, in which circuit, at what frequency and reference plane? The answer may be a valid 50-ohm fixture reading, a calculated current ratio or an installed measurement. Without that definition, the number cannot travel from the chart to the antenna system.

This is also why I object when Jeff, K6JCA's Y21 work is reduced to “S21 and Y21 give the same answer”. Agreement in one fixture is not a declaration that the two calculations have identical error behaviour. The conditions are the point of the experiment.

Decibels Express a Ratio, Not a Device Type

For a power ratio, the decibel value is 10 log₁₀(P₂/P₁). For compatible voltage or current amplitudes under the stated impedance conditions, it is 20 log₁₀(A₂/A₁). A 40 dB amplitude ratio is 100:1; a 40 dB power ratio is 10,000:1. Those arithmetic statements still do not identify what the numerator and denominator represent.

A graph labelled “40 dB rejection” might mean |S₂₁| in a laboratory fixture, an inferred current ratio for an assumed common-mode loop, a measured before/after cable current, or an unrelated receiver CMRR quantity. The same numeric value can describe different experiments.

The sign convention also matters. A transmission plot may show −40 dB, while marketing copy calls the result “40 dB attenuation.” Both can refer to the same magnitude ratio, but the axis, direction and reference must be explicit.

Complex Zcm Is the Useful Starting Point

For a defined small-signal common-mode path, write the choke impedance as:

R_CM is the dissipative part at that operating point; X_CM is the reactive part. Their magnitude and phase change with frequency, ferrite material, geometry, turns, winding capacitance, cable construction, temperature and drive level. A VNA usually characterises the linear small-signal state, so the test conditions remain part of the result.

|Z_CM| is more transferable than a bare dB label because it can be inserted into a declared circuit model. It is still not a complete power rating or an automatic prediction of installed suppression. Differential insertion loss, voltage, current, heating and parasitic current paths require separate evidence.

A 50-Ohm S21 Trace Belongs to Its Fixture

Suppose a calibrated series fixture places an ideal two-terminal impedance Z_DUT between two equal VNA ports of impedance Z₀. In that specific circuit:

The conversion uses complex S₂₁, not only its magnitude in dB. Fixture shunt capacitance, direct coupling, cable common mode, mismatch and distributed behaviour can violate the single-series-branch model. Calibration and de-embedding must reach the declared DUT planes, and known standards should verify the useful impedance range.

The resulting S₂₁ is a legitimate measurement of that fixture. It is not the amount by which every antenna installation will reduce exterior-shield current, because the installed source, load, return path and parallel bypasses are different.

What Jeff's Y21 Comparison Actually Shows

In The Y21 Method of Measuring Common-Mode Impedance, Jeff credits Dick Benson, W1QG, with the method. His 1–30 MHz comparison found close S21/Y21 results with local shunt capacitances around 2.3 pF. He then simulated 100 pF at each shunt: the simple S21-derived impedance changed substantially, while Y21 retained the same series-branch result. With the shunts removed, the calculations agreed.

That is a useful demonstration of why I prefer the full Y21 extraction when local fixture shunts matter. It is not a dismissal of a well-controlled S21 measurement; Jeff also describes that simpler method as adequate for many applications. Nor does the demonstration say that Y21 removes every kind of stray coupling.

A full calibrated two-port S matrix can be converted to an admittance matrix. If the DUT and fixture are adequately represented by a π network, the transfer branch can be extracted from Y₂₁:

This model can separate local terminal-to-reference shunts into Y₁₁ and Y₂₂. It does not automatically remove capacitance or magnetic coupling directly between the ports, radiation, cable-exterior current or calibration residuals; those transfer paths remain in Y₂₁. Y21 is powerful when its circuit model is valid, not because the parameter name guarantees de-embedding.

A direct S21 series conversion and a Y21 extraction can agree when fixture shunts are negligible. They can diverge simply because those local shunts increase, even before the fixture becomes distributed. Full Y21 then has a specific advantage: it separates the local shunt branches instead of silently treating the network as one series impedance. If direct transfer bypasses or distributed effects dominate, changing the mathematical extraction is not enough.

Installed Current Reduction Uses the Whole Loop

For a simple Thevenin common-mode loop, let Z_S + Z_L represent the external source and return network at the choke plane. Before adding the choke:

Because the impedances are complex, phase matters. Replacing the expression with 20 log₁₀(1 + |Z_CM|/R) is only a special approximation; it is not a universal choke equation. If the installation has several return paths, one series-loop model is not enough.

The common-mode source and load impedances can change with frequency, coax length and route, antenna geometry, mast, bonding, equipment, mains wiring and nearby conductors. The same choke can therefore produce different current reductions in different installations without changing its bench impedance.

“6 dB” Is Not a Universal Correction

Factors of two appear naturally in equal-port networks, matched-source conventions and conversions between available, incident, delivered and accepted quantities. That does not create a universal 6 dB penalty to add to or subtract from every choke trace.

If a calculation seems to require a fixed correction, draw the source, both port impedances, DUT branch and measured wave quantities. Then derive the result from that circuit. The correction belongs to the stated definition, not to S21, Y21 or common-mode impedance as categories.

Resistance, Reactance and Heat Must Stay Visible

Two chokes with the same |Z_CM| can have different R/X composition. Both may reduce current in a particular loop, but the voltage distribution, resonance and dissipation can differ. Under a measured RMS common-mode current, the small-signal resistive estimate is P ≈ I_CM²R_CM. Real ferrite and winding behaviour can change with temperature and drive, so powered testing must check the finished assembly.

A high dB notch on a low-power trace does not rate the choke for transmitter power. It also says nothing by itself about wanted differential insertion loss. Publish complex Z_CM, differential S-parameters, exact construction and controlled electrical/thermal limits as separate evidence.

A Choke Report Readers Can Reuse

• Name the measurand. State whether the dB axis is S21, voltage, current, power, CMRR or an installed before/after result.
• Draw the circuit. Include source/load/reference impedances, fixture grounds, cable exteriors and every return path.
• Declare the planes. List calibration, fixture compensation, port impedance and any mathematical conversion.
• Publish complex data. Provide R_CM, X_CM and |Z_CM| across frequency, with a usable uncertainty range.
• Measure the wanted mode. Record differential insertion loss, return loss and mode conversion.
• Verify the installation. Map exterior current at several positions before and after the choke, then restore the first state in an A/B/A check.
• Qualify power separately. State current, voltage, waveform, duty cycle, load, ambient, cooling, equilibrium temperature and post-test results.

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