Stacked Hybrid 4:1 Baluns: Test the Assembly, Not the Label

The useful engineering claim is conditional: separating impedance transformation from common-mode suppression can be sound, and a compact stacked construction can work. Physical contact alone does not prove excessive magnetic coupling, excessive temperature or a capacitive bypass. Equally, acceptable SWR or a brief low-power transmission does not prove a QRO rating.

Start with the Exact Circuit, Not “Hybrid”

In amateur usage, “hybrid 4:1 balun” often means a 4:1 voltage-transforming network cascaded with a 1:1 common-mode choke. That description is incomplete until the schematic identifies:

• whether the transformer is Ruthroff, flux-linked, another autotransformer, or a transmission-line arrangement;
• whether the 1:1 choke is on the nominal 50 Ω side or the transformed side;
• which port is intended to be balanced and what reference defines that balance;
• the winding conductors, characteristic impedance, connections and core part numbers; and
• the intended load-impedance region, frequency range and common-mode environment.

“Followed by” changes with signal direction and does not specify the stress on either component. A 1:1 choke on a 50 Ω coax port sees different differential voltage and current from a winding on the nominal 200 Ω port. A 4:1 Guanella current balun is also not equivalent to a Ruthroff voltage transformer plus a separate choke.

Ruthroff's original broadband-transformer paper shows why topology matters: at high frequency, winding length and transmission-line impedance bound performance, while low-frequency response depends on magnetizing inductance. Close conductor coupling inside the intended transmission line can be beneficial. “Capacitance” and “coupling” therefore cannot be labelled good or bad without identifying their nodes and modes.

The Transformer and Choke Have Different Specifications

4:1 impedance transformation

An ideal 4:1 impedance ratio corresponds to a 2:1 voltage ratio and a 1:2 current ratio. The real transformer must also meet insertion-loss, return-loss, balance, isolation and complex-load requirements over its band. A 200 Ω resistor proves only one operating point.

1:1 common-mode choking

On coax, wanted differential current flows on the centre conductor and the inner surface of the shield. Current on the shield exterior is a common-mode current relative to the environment. An ideal coax choke leaves the differential mode nearly unchanged while adding common-mode impedance:

The resistive part can damp a common-mode resonance, but it is also where common-mode power becomes heat. The differential cable conductors, dielectric and connectors still carry full line current and voltage. A choke does not by itself guarantee equal amplitude and opposite phase at an antenna's two terminals; balance and common-mode suppression require separate measurements.

“100 W” and “QRO” Are Not Test Conditions

Power, waveform and duration must be stated together. PEP does not determine average heat without the modulation envelope, processing, key-down pattern and transmit/rest cycle. A digital-mode label also does not define duty cycle; use measured average power and an explicit schedule.

For an ideal lossless 50:200 Ω transformation into a purely resistive load:

The arithmetic is exact only at the named reference planes for sinusoidal, resistive, lossless operation. In a perfectly balanced 200 Ω port, each terminal may have half the conductor-to-conductor voltage relative to the symmetry plane; common-mode displacement can make terminal-to-enclosure stress unequal. Reactive loads, feed-line transformation and internal standing waves can produce different winding and terminal maxima.

Moving from 100 W to 1.5 kW multiplies ideal voltage and current by √15 ≈ 3.87. Copper loss at fixed resistance then scales by 15, but completed ferrite loss need not scale linearly: permeability, core loss, common-mode current and temperature can change with field and load. It is therefore unsupported to declare a construction safe at 100 W SSB or unsafe at 1.5 kW solely from those two numbers.

Transformer Flux Needs the Correct Winding Voltage

For a sinusoidal flux-linked winding in an appropriate low-frequency model:

Vw is the voltage actually applied across the excited turns, Nexc is that section's turn count, and Ae belongs to the exact core geometry. In an autotransformer or transmission-line transformer, blindly inserting the total port voltage and total turns can be wrong. At the upper end of HF, propagation, characteristic impedance, leakage inductance and winding capacitance become part of the model.

Fair-Rite's broadband-transformer guidance explicitly concerns low-power designs and describes low-, mid- and high-frequency limits; it is not a transmitter watt table. Current Fair-Rite high-frequency power-material data show power-loss density changing with frequency, flux density and temperature. Curie temperature, initial permeability or a low-frequency B-H point is not a safe RF operating limit.

Physical Stacking Is Not a Verdict

First define what is stacked

Several identical toroids carrying one common winding may be deliberately stacked to increase effective magnetic cross-section and core volume. That is different from placing two independently wound devices—one transformer and one choke—face to face. The phrase “stacked cores” must not confuse those two constructions.

Magnetic coupling is geometry- and mode-dependent

A toroid provides a largely closed magnetic path, so external field can be much smaller than the field inside the core. Separate coaxial toroids do not automatically have a large coupling coefficient merely because their faces touch. The familiar relation

defines mutual inductance once k is known; it does not calculate k from the words “same plane.” Winding loops, lead dress, asymmetry, gaps, nearby conductors and which differential or common mode is excited can dominate the external coupling.

Separation or rotation may improve one construction, but it is not a universal cure. It also changes lead length, loop area, electric-field geometry and enclosure capacitance. The honest comparison holds every practical variable as constant as possible, measures the relevant multiport response, and then changes spacing or orientation.

Thermal coupling can help or hurt

Two lossy components in contact exchange heat. Contact can reduce exposed convective area and let one component preheat the other; it can also spread heat into additional thermal mass or a deliberate heat path. Consequently, stacking does not universally increase thermal resistance.

The real temperature field depends on each loss source, contact pressure and interface, ferrite thermal conductivity and heat capacity, winding insulation, potting, enclosure, airflow, ambient and duration. Fair-Rite publishes typical ferrite thermal constants and part-specific material data, not a completed-assembly thermal rating. Measure both core and winding hot spots through thermal equilibrium; a surface reading on the accessible core can miss the buried interface.

Parasitic Capacitance Can Matter—Only Across the Right Nodes

If a stray capacitance truly bridges the input and output common-mode nodes of the choke, it forms a parallel path:

Those values verify the source arithmetic. They do not prove that every picofarad in a compact assembly bypasses the choke: capacitance from a node to a floating core, to an enclosure, between adjacent turns or directly across the choke has different effects. Measure common-mode R + jX after final wiring and enclosure assembly. Comparing the transformer present, absent, stacked, separated and rotated can identify a layout-dependent change.

Insulation Is a System, Not a Core-Coating Number

Fair-Rite's current 61 mm coated-toroid page, for example, specifies at least 1,000 V RMS uniformly across the core's coated height for that part. That is not a rating for winding-to-winding, terminal-to-enclosure, connector, cable or contaminated outdoor surface stress. It also does not establish behaviour under the actual HF field distribution.

Clearance is through air; creepage follows an insulating surface. Both can be affected by peak voltage, field concentration, material, contamination, condensation, altitude and aging. IEC 60664-1 provides low-voltage insulation-coordination principles but its public scope ends at 30 kHz, so its tables must not be copied mechanically into a 1.8–30 MHz transmitter design. Use the product's applicable safety requirements and validate RF withstand under the intended environment.

A Defensible Qualification Plan

1. Freeze the specimen. Record the schematic, port references, core manufacturer/part/lot, coatings, windings, coax, lead dress, spacing, orientation, enclosure and ambient.
2. Characterize all modes at small signal. Measure transformation ratio, return and insertion loss, output amplitude/phase balance, common-mode impedance or rejection, and port isolation over frequency. De-embed the fixture and preserve the intended port impedances.
3. Test the final assembly. Repeat with both magnetic devices mounted and the enclosure, wiring and hardware installed. Compare stacked, separated and rotated versions without silently changing lead length or routing.
4. Apply differential power. Use the stated waveform, average power, key-down schedule, frequency and representative resistive and complex loads. Measure calibrated input and delivered power, match, waveform and temperature until equilibrium or the declared time limit.
5. Apply or measure common-mode stress. A balanced dummy load may leave the choke nearly idle. Characterize common-mode current and voltage expected from the antenna, deliberately exercise that mode safely, and verify Icm²Rcm heating.
6. Verify voltage margin. Inspect conductor-to-conductor, conductor-to-core and conductor-to-enclosure peaks. Test insulation in the intended humidity, contamination, altitude and temperature range with defined pass/fail limits.
7. Derate from measured limits. Define maximum temperature, insertion loss, common-mode impedance, balance, dielectric margin and permanent-change criteria. A power label without these conditions is not reproducible.

ARRL QEX's three-port balun-measurement work is a useful reminder that CMRR, transformation accuracy and insertion loss are different quantities and that fixtures must preserve the intended impedances. A VNA is valuable for the linear network, but it cannot by itself certify thermal equilibrium, nonlinear ferrite behaviour or RF insulation margin.

Primary Sources Checked

• C. L. Ruthroff — Some Broad-Band Transformers, Proceedings of the IRE, 1959
• Fair-Rite 17th-edition catalogue — broadband transformers, complex permeability and ferrite properties
• Fair-Rite High Frequency Power Materials, revision 05 — loss versus frequency, flux and temperature
• Fair-Rite 5952003801 current toroid data — geometry, material properties, coating and test boundaries
• ARRL QEX — Balun Performance Measurements
• G3TXQ — measured common-mode choke resistance and reactance
• IEC 60664-1:2020 — insulation-coordination scope and environmental boundaries

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