Sevick’s Transmission Line Transformers, Re-Read

I returned to the fourth edition of Jerry Sevick’s Transmission Line Transformers expecting the usual experience: a respected historical text, still useful but surrounded by caveats. Instead, the strongest impression was how much of the practical method remains right. Sevick builds, measures, compares and changes the geometry. That habit has aged better than many transformer slogans written after him.

My agreement is direct: the book remains essential for understanding the transformer itself. My disagreement is with the way its examples are sometimes promoted into complete antenna-system prescriptions. A two-port result into defined terminations does not, by itself, settle current balance, outside-shield current, return paths, installation asymmetry or high-power behavior on a reactive load.

Which Edition Is on the Bench?

This rereading uses Sevick’s fourth edition, published by Noble Publishing under ISBN 1-884932-18-5. That matters because Transmission Line Transformers changed across editions as practical designs, construction details and material information were expanded. “Sevick says” is not a useful citation unless the edition, topology and operating conditions are identified.

I am not treating every winding in the book as a timeless parts recipe. I am treating the book as what it does best: a topology and measurement framework, illustrated with hardware and materials available to the author.

What Makes It a Transmission-Line Transformer?

In a conventional flux-coupled transformer, mutual magnetic flux is the main mechanism linking windings. In a transmission-line transformer, closely coupled conductors also carry a defined transmission-line mode. The distributed inductance and capacitance form the line rather than appearing only as unwanted leakage and stray capacitance.

That distinction explains the potential bandwidth, but it does not abolish ordinary transformer physics. Low-frequency performance still depends on sufficient magnetizing or longitudinal impedance. High-frequency performance still encounters electrical length, discontinuities, leakage, conductor loss, dielectric loss and parasitic coupling outside the intended line model.

The clean statement is therefore not “the core transfers no power” or “flux is irrelevant.” It is that the wanted differential transmission-line current can produce substantial magnetic-field cancellation in the core. Any net magnetizing current, imbalance or common-mode current produces core excitation that the ideal cancellation picture does not remove.

Flux Cancellation Has a Boundary

Equal and opposite currents in the intended line mode create opposing magnetomotive force. That is one reason a transmission-line transformer can carry substantial differential power without treating the core like an ordinary low-frequency power transformer.

But the cancellation belongs to a mode, not to the component label. Finite magnetizing current is required at the low end of the band. Unequal conductor currents, asymmetrical capacitance, a poorly defined return path or common-mode current can produce net flux. Core loss and temperature then depend on frequency, flux amplitude, material, geometry and duty cycle.

This is where I agree with Sevick’s attention to core and construction, and reject the later folklore that “the line carries the power, so the ferrite does not matter.” The intended line mode and the unwanted common mode do not excite the core in the same way.

Optimum Z₀ Is a Conditional Design Result

Sevick’s characteristic-impedance treatment remains one of the book’s most useful lessons. For a simple single-section transformation between real terminations, a geometric-mean line impedance can be the natural optimum. It is not a universal winding prescription.

The required effective Z₀ depends on topology, impedance ratio, port terminations, how constituent lines are series- or parallel-connected and which mode is being evaluated. Once the load is complex and frequency-dependent, “optimum” also depends on the band and the performance criterion: return loss, insertion loss, voltage stress, current stress, balance or common-mode isolation.

Winding a pair on ferrite changes its physical cross-section. Conductor diameter, insulation thickness, spacing, twist pitch, core curvature and neighboring turns alter the effective line impedance and delay. The value measured on a straight sample is evidence, not a guarantee for the wound structure.

Guanella and Ruthroff Share a Family, Not a Personality

Gustav Guanella’s series/parallel transmission-line structures and C. L. Ruthroff’s voltage-adding structures are both foundational transmission-line-transformer families. They do not impose the same port relationships.

A Guanella implementation uses one or more transmission lines whose inputs and outputs can be combined in series and parallel. With sufficiently high common-mode impedance and symmetric construction, it can enforce useful current relationships and provide balanced-to-unbalanced conversion. That behavior is finite and frequency-dependent; it is not guaranteed by drawing the word “current” on the box.

A Ruthroff implementation uses voltage addition or inversion along a line and commonly retains a conductive relationship between ports. It can be an excellent broadband impedance transformer. It does not inherently supply the same common-mode choking or galvanic isolation as a separately specified isolation function.

So I keep the familiar “current-mode” and “voltage-mode” labels as shorthand, but I do not let them replace a circuit diagram. State the exact port connections, the common reference, the intended differential relationship and the unwanted common-mode path.

Isolation Is Three Different Questions

Transformer discussions often use “isolation” without saying which isolation:

• Galvanic isolation: no DC conductive path between specified ports;
• Common-mode impedance: opposition to a current that moves conductors together relative to the environment; and
• Differential port decoupling: the intended topology keeps input and output behavior acceptably independent within its design model.

These are not interchangeable. High longitudinal reactance in a transformer model may support the intended voltage relationship while a parasitic capacitance still provides a common-mode bypass. A component can transform impedance beautifully and provide little useful choking in the installed current path.

This is the most important station-level extension to the book: identify the complete common-mode loop, then measure the impedance and mode conversion that the actual installation presents.

Twisting Is Geometry, Not Virtue

Twisting can improve coupling, repeatability and symmetry. It also changes conductor spacing, Z₀, capacitance, delay, voltage stress and proximity loss. Tighter is therefore not automatically better.

The correct pitch is the one that gives the required effective line impedance and balance without exceeding dielectric, conductor or manufacturing limits. Parallel, twisted, coaxial and other line forms can all work when their geometry matches the topology. Sevick’s experimental comparisons remain valuable precisely because they refuse to turn one construction style into a religious rule.

Ferrite Is a Complex Material and a Thermal Component

Initial permeability is not enough to select a core. Ferrite permeability is complex and frequency-dependent. Its real component describes inductive energy storage in a chosen equivalent representation; its imaginary component represents magnetic loss. Both vary with material, frequency, temperature and excitation.

Small-signal impedance curves do not establish high-power loss. At power, flux density, waveform, frequency and temperature matter, as do core volume and heat removal. Manufacturer data are measured on stated core shapes and conditions; transferring them to another geometry or excitation requires care.

More turns can improve low-frequency magnetizing impedance, yet also increase length, capacitance, leakage paths, copper loss and voltage between turns. More permeability can reduce turns, yet the chosen material may have unsuitable loss or temperature behavior elsewhere in the band. There is no single “HF mix” answer without topology, frequency range, load and thermal envelope.

There Is No One-Number Power Rating

The transmission line sets conductor current, dielectric voltage and part of the heating limit. The core sets magnetizing/common-mode flux and magnetic loss limits. Connections set contact loss and local field concentration. The enclosure and ambient conditions set cooling.

Mismatch and reactive loads can raise peak voltage or current far above the matched dummy-load case. Duty cycle determines whether loss becomes a short pulse or sustained temperature rise. Common-mode current can add another core-excitation mechanism without appearing in a differential insertion-loss test.

A defensible power statement therefore includes frequency, source and load impedances, SWR or complex-load envelope, waveform/duty cycle, ambient temperature, permitted temperature rise, conductor and dielectric ratings, and measured harmonic or saturation behavior where relevant.

Sevick’s Measurement Habit Survives; The Fixture Must Evolve

Sevick’s strongest modern lesson is still “build it and measure it.” A VNA makes that easier, not automatic.

• Differential transfer: measure insertion loss and return loss at declared reference impedances and planes.
• Balance: measure output amplitude and phase balance under the intended termination.
• Mode behavior: use mixed-mode S-parameters or an equivalent characterized setup to separate differential, common and conversion terms.
• Common-mode impedance: choose a valid series, shunt or transfer-admittance method for the topology, then de-embed fixture delay and capacitance.
• Load envelope: repeat with representative complex loads, not only a matched resistor.
• Power and temperature: verify loss, waveform, voltage, current and thermal rise at the intended duty cycle.

A back-to-back pair is useful for comparative loss and bandwidth tests. Dividing the result by two estimates one device only when the two devices are sufficiently identical, independently loaded and operating in the same modes. The pair can hide imbalance, interaction, common-mode bypass and unequal voltage stress.

Y-parameters can be an excellent way to extract a particular impedance from a two-port fixture. They are not the only valid representation, and “Y₂₁” is not a substitute for defining port polarity, reference impedance, fixture, mode and de-embedding. The trustworthy method is the one whose network model matches the quantity being claimed.

Where I Agree, and Where I Draw the Line

I agree with Sevick on the essentials:

• transmission-line geometry and effective Z₀ are design variables;
• electrical length limits the upper end of a topology’s useful band;
• magnetizing or longitudinal impedance matters at the lower end;
• core, winding form and construction change measured performance; and
• a design should be built, measured and compared rather than copied by appearance.

I draw a harder boundary around the conclusion:

• a two-port match does not prove balanced antenna current;
• a transformer label does not prove galvanic isolation or common-mode choking;
• a dummy-load sweep does not establish behavior on every reactive load;
• a small-signal ferrite sweep does not establish power or temperature margin; and
• one measurement representation does not own the truth.

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