G3TXQ “Common-Mode Chokes” Revisited

The original “Common-mode chokes” page by G3TXQ is one of the most influential amateur-radio resources of the last two decades. It pushed the community toward measuring impedance versus frequency and away from pure “it worked for me” folklore.

But that same page (last updated in 2017) is now routinely treated as a measurement standard — and that’s where the trouble starts. This article keeps the good engineering instincts, corrects the measurement model, and updates the conclusions for modern stations and tools.

Executive Summary

• Still true: A choke is defined by its common-mode impedance Z_CM(f) in ohms, not by a marketing “dB” number.
• Outdated leap: “Prefer resistive impedance (R > |X|)” became a meme. High minimum |Z_CM| with damped behavior is what survives real stations.
• Main technical issue: Series-element extraction (often called the “G3TXQ method”) is fixture-sensitive because it assumes away shunt parasitics.
• Modern fix: Use Y21 π-model extraction: Z_CM = −1 / Y21. This isolates the choke even with shunt capacitance present.
• System reality: dB is not a choke property. Attenuation depends on the common-mode loop resistance R_CM.

What G3TXQ Got Right

Stop guessing — measure impedance versus frequency.
The original work clearly showed that choke behavior is frequency-dependent, material-dependent, and geometry-dependent. That insight alone raised the technical baseline of the hobby.

Reactive cancellation is real.
If the choke reactance cancels the reactance of the common-mode path, current can increase. That warning is mathematically correct — and still valid.

Load-impedance myths were rejected.
Tying required choke impedance to the differential-mode antenna impedance was correctly called out as nonsense.

Where the Page Aged Poorly

“Aim for resistive impedance” became an overcorrection.
Resistive impedance damps resonances — that part is true. But common-mode suppression depends first on the magnitude of Z_CM, not on whether it is resistive or reactive. Loss is a tool, not a goal.

The black-bar chart encoding misled readers.
“Mostly resistive” was widely interpreted as “safe and superior.” In practice, a 6–10 kΩ inductive choke often outperforms a 1–2 kΩ resistive one.

The feedline-length example was over-generalized.
Real stations are distributed systems — coax exterior, mast, ground, shack wiring. Designing around a single lumped reactance is not robust engineering.

The Measurement Problem

G3TXQ correctly noted that one-port measurements collapse under a few picofarads of stray capacitance. Unfortunately, the follow-up two-port S21 extraction still assumes a pure series element.

At kilohm impedances, tiny shunt parasitics dominate. The real fixture behaves as a π network, not as “50 Ω → Z → 50 Ω.” This is why results vary wildly between seemingly identical setups.

The Modern Replacement: Y21 Extraction

Treat the DUT and fixture as a π network with shunt admittances at both ports. In the Y-matrix, the series element appears directly:

Z_CM = −1 / Y21

This cleanly removes shunt parasitics and yields a repeatable, fixture-robust result. Jeff Anderson (K6JCA) and others explicitly recommend this method for that reason.

Why the “6 dB Loss” Argument Is Wrong

A choke does not “have dB.” dB appears only after you assume a system model. Confusing a 50 Ω insertion-loss model with a single common-mode current loop is the source of the infamous “missing 6 dB.”

Y21 does not add or subtract dB — it removes the wrong circuit model.

How Much Choking Do You Actually Need?

If you want attenuation in dB, you must define the loop resistance:

Att(dB) = 20 · log₁₀(1 + Z_CM / R_CM)

Using R_CM ≈ 100 Ω as a realistic order-of-magnitude station value:

• ~20 dB → ~900 Ω
• ~30 dB → ~3.1 kΩ
• ~40 dB → ~9.9 kΩ

RX noise improvement often occurs well below TX-side requirements. TX design must consider heating, saturation, and voltage stress.

Chokes Are Not Topologies

Calling a choke “40 dB” confuses component data with system behavior. So do bare-shield VNA tests and “hybrid baluns” wound directly on transformers. Placement, current maxima, and return paths matter more than catalog numbers.

What a Modern G3TXQ Chart Should Show

• Minimum |Z_CM| per band, not just peaks
• Fixture sensitivity and repeatability
• Clear separation of component characterization and in-station validation
• Preference for damped, low-Q behavior — not “resistive at all costs”

What About Air-Cored (Coax-Only) Chokes?

Air-cored chokes — typically formed by winding several turns of coax into a solenoid — are often treated as a “clean” or “lossless” alternative to ferrite. They are neither magic nor useless… but they are frequently misunderstood.

What an air-cored choke actually is
An air-cored choke is a purely inductive common-mode impedance element. It provides no intentional resistive damping and relies entirely on inductive reactance:

Z_CM ≈ jωL

That has two immediate consequences:

• Impedance rises linearly with frequency.
• There is no intrinsic loss mechanism to damp resonances.

Why they can work very well
When physically large enough, air-cored chokes can reach several kilohms of Z_CM on the upper HF bands (like 10M). Because there is no ferrite, there is:

• No saturation
• No μ-collapse
• No thermal compression

For high-power TX on higher HF bands, a properly sized air-cored choke can be extremely robust.

Why they often fail in real stations

• High-Q behavior: With no resistive component, air-cored chokes exhibit narrow, high-Q resonances.
• Fixture and environment sensitivity: A few pF of stray capacitance can shift resonance dramatically.
• Band-limited usefulness: At 160 m to 20 m, practical coil sizes rarely produce enough Z_CM.

This is exactly where casual S21 or one-port measurements tend to over-promise performance. A sharp peak in a lab sweep often collapses once installed on a mast or routed along a structure.

Air-cored vs ferrite: the correct comparison

• Air-cored chokes offer reactive blocking with no loss.
• Ferrite chokes offer loss-assisted damping with broader, more predictable behavior.

Neither is “better” in isolation. The correct engineering question is:

What is the minimum Z_CM across the band, and how sensitive is it to installation?

Best-practice guidance

• Air-cored chokes are well-suited for upper-HF, high-power use where physical size is not constrained.
• They should be measured with Y21 extraction to avoid resonance over-interpretation.
• For low-band HF and uncontrolled installations, ferrite-based chokes with damped behavior are usually more reliable.

Air-cored chokes are not “bad” — but they demand controlled geometry, controlled routing, and honest measurement.

Closing

G3TXQ deserves credit for dragging amateur radio toward measurement. What failed was freezing that work into dogma and copying a fragile extraction method as if it were a standard.

The modern upgrade is simple:

• Specify chokes by Z_CM(f) in ohms using Y21 extraction
• Convert to dB only with a defined loop model
• Validate at power and in the station

Interested in more technical content? Subscribe to our updates for deep-dive RF articles and lab notes: https://shop.rf.guru/pages/subscribe

Questions or experiences to share? Feel free to contact RF.Guru.