Transformer Turns Ratio: What It Changes—and What Sets Efficiency

No ideal turns ratio carries an efficiency penalty. Ratio, absolute turns, topology, winding geometry and load are separate variables. Changing any one may alter match, magnetizing admittance, flux, RMS current, parasitics and temperature, but there is no universal “gentle ratio,” “extreme-ratio penalty” or ratio-only efficiency curve.

Ratio, Absolute Turns and Topology Are Different Specifications

For an ideal two-winding transformer, define a = Ns/Np. With consistent dot polarity:

The impedance relation applies to complex impedance as well as resistance. An ideal ratio can transform a load reactance; it does not erase it. The ideal current relation is a winding-current relation under the model assumptions—not permission to substitute rated input current, output DC current or unrelated RMS waveforms from a switching converter.

Two builds with Np:Ns = 10:1 and Np:Ns = 20:2 have the same ratio. They do not have the same magnetizing inductance, flux per applied volt-second, winding length, leakage, capacitance or AC resistance. Conversely, changing a secondary ratio while holding primary turns and its applied waveform constant need not change primary flux. “Ratio” and “number of turns” must not be used interchangeably.

A ratio can change system power without changing intrinsic efficiency

If source and load impedances are fixed, changing a changes the impedance presented to the source. That can change accepted and delivered power because the match changed—even in a lossless transformer. A higher output meter reading or lower input SWR is not proof of higher efficiency. Efficiency compares real power delivered to the intended load with real power accepted at the transformer input, at declared planes.

First Name the Magnetic Component

Analog Devices’ MAXREFDES1249 documentation explicitly describes the flyback element as a coupled inductor: energy builds in the primary during switch on-time and transfers during off-time. TI’s magnetic-field seminar makes the same distinction between a true transformer and a flyback energy-storage component. Therefore a forward-converter voltage equation, a sinusoidal RF transformer relation and a flyback RMS-current calculation are not interchangeable.

Flux Comes From Winding Volt-Seconds, Not Ratio Alone

Faraday’s law supplies the general starting point for a winding of N turns on effective core area Ae:

Use the actual voltage across that winding during each switching or RF interval. Bus voltage, output voltage and winding voltage can differ because of switches, duty, dead time, reset networks, leakage spikes and load. Any net volt-seconds produce flux walk or DC bias.

For a zero-DC sinusoid, the familiar conditional result is

For an ideal symmetric bipolar square voltage of magnitude V, 50% duty and complete symmetry, Bpk = V/(4 f N Ae). Neither coefficient may be applied to an arbitrary forward, push-pull, bridge or flyback waveform without reconstructing its volt-seconds and reset.

Saturation and core loss are separate limits

Keeping peak flux below a material’s saturation curve does not establish low loss or safe temperature. Core loss depends on exact material and core shape, frequency, flux excursion, waveform, DC bias and temperature. TDK’s current magnetic design tool explicitly calculates core loss as a function of signal form and temperature; IEC 62044-3:2023 defines current high-excitation methods for core power loss and amplitude permeability.

A generic Steinmetz expression can interpolate within a characterized region, but its coefficients, waveform correction and valid range belong to the manufacturer data and test method. No universal exponent or verbal description of rapidly rising loss is a design limit. Use the relevant measured curve or model and include tolerance and temperature.

Magnetizing Current Is a Complex, Operating-Point Quantity

At low excitation, a manufacturer may give AL so that L ≈ AL N². That relation inherits the specified frequency, test level, temperature, core assembly and tolerance. Under RF or power excitation, permeability is complex, frequency-dependent and nonlinear. A useful shunt branch has both stored and dissipative parts:

For a linear sinusoidal inductance, the reactive RMS component is V/(ωLm). For a constant applied voltage interval, its current ramp is approximately ΔI = V Δt/Lm. Real current also includes core loss and any DC-bias or nonlinear effects. More primary turns often increase small-signal Lm approximately with N², but gap, permeability, excitation and winding distribution remain part of the result.

Magnetizing current is not automatically wasted power: its reactive component returns energy. It still raises winding RMS current and device stress; the in-phase core-loss component consumes power. Measure or model the phase rather than multiplying RMS voltage by RMS current and calling the result watts.

Copper Loss Uses RMS Waveforms and AC Resistance

For each winding:

Rac includes conductor resistivity and length, skin effect, proximity effect, layer fields, gap-fringing fields, terminations and temperature. It may be much greater than DC resistance. TDK’s current design material explicitly adjusts transferable power for skin and proximity effects and warns that actual thermal resistance and AC/DC resistance ratio must be determined for the design.

Current follows transferred power, load and waveform. At the same delivered power, a lower winding voltage implies more current, but a ratio alone states neither power nor load. A one-turn secondary is not inherently inefficient, and a many-turn secondary is not inherently lossy. Conductor cross-section, foil/litz construction, parallel strands, window allocation, termination and field geometry decide.

Nor does a thicker conductor automatically worsen high-frequency loss. Diameter, strand size, layer count, field orientation and current sharing determine the AC result. Calculate the actual RMS waveform for the named topology. In a flyback converter, primary and secondary currents occur in different intervals; the ideal simultaneous winding-current relation does not directly supply their RMS values.

Leakage and Capacitance Trade Against Each Other

Leakage inductance comes from flux that does not link the intended windings. Capacitance comes from electric fields within and between windings, core, shield and enclosure. Both depend on absolute turns and physical arrangement—not ratio in isolation.

• Interleaving often lowers leakage but can increase interwinding capacitance and common-mode current.
• More separation can improve isolation capacitance while increasing leakage and voltage overshoot.
• More turns increase length, yet sectioning and distribution can move several resonances in non-monotonic ways.
• Foil may improve current distribution in one design and worsen proximity loss or capacitance in another.

In switching converters, leakage energy can be dissipated in clamps/snubber networks or raise device switching loss. In RF transformers, leakage, magnetizing admittance, winding capacitance and transmission-line behavior shape insertion loss, return loss, balance and common mode. “Efficiency” without bandwidth and load is incomplete.

HF UNUN and Balun Ratios Transform Complex Networks

For an ideal 1:a² impedance transformer, a complex load RL + jXL refers to the primary as (RL + jXL)/a². An EFHW or OCF feed impedance varies with frequency, wire geometry, height, ground, surroundings and return path. A nominal 49:1, 64:1 or other label is not an efficiency value and does not prove a 50 Ω resistive input.

An unun may also include the feed line, enclosure or counterpoise in its current path. A balun is a multi-conductor, multi-mode device whose differential transmission, amplitude/phase balance and common-mode impedance all matter. A two-port 50 Ω sweep can miss those modes.

Switching-Converter Ratio Advice Is Topology-Specific

The ideal forward-converter relation Vout ≈ D Vin Ns/Np omits rectifier/switch drops, reset limits, ripple and conduction mode, but it can be a useful first-order equation for that named topology. It must not be generalized to flyback, half-bridge, full-bridge, push-pull or resonant converters.

Changing ratio can trade duty cycle against switch voltage, rectifier voltage, peak/RMS currents, conduction loss, switching loss and control range. The direction and optimum depend on input range, output, topology, semiconductor properties, dead time, leakage/clamp, magnetizing design and regulation mode. There is no universal claim that too-small ratio raises magnetizing current or that too-large ratio necessarily produces high peak current.

Define Efficiency at Accepted and Delivered Power Planes

For a transformer under a stated source, load, frequency, waveform and thermal condition:

Use average real power, not apparent volt-amperes. Declare whether input power is available from the source, incident on the transformer, or accepted after reflection. Declare whether output power is incident on the load or absorbed by it. At RF, mismatch, internal dissipation and power leaving through an unintended mode or port are different. Desired-path efficiency can count the last item as unavailable to the intended load, but it must not mislabel it as transformer heat. In a converter, gate-drive/control and semiconductor loss belong to system efficiency but not transformer efficiency.

A useful loss budget includes:

• core loss under the actual flux waveform and temperature;
• AC winding, lead, joint and connector loss under actual RMS waveforms;
• dielectric, shield and contact loss;
• leakage/clamp or snubber loss assigned to the agreed component boundary;
• power delivered into unintended common-mode or fixture paths;
• measurement uncertainty and thermal drift.

What Back-to-Back Testing Can Actually Prove

Connecting a step-up transformer to a step-down transformer can restore a 50 Ω instrument interface, but the measured pair is one cascaded network. Dividing its dB loss by two is valid only if the units are effectively identical at the operating point, individually matched as intended, linear, stable and connected through a negligible or characterized intermediate fixture. Even then it estimates one unit in that resistive back-to-back configuration—not efficiency into a different complex antenna or converter load.

Mini-Circuits AN20-001 gives a stronger three-unit method. Measure pairs AB, AC and BC with their high-impedance windings properly joined and low-impedance ports matched. If cascade loss is additive under those conditions:

This avoids assuming A, B and C have equal loss. It still inherits connection repeatability, matching, fixture, calibration and linearity uncertainty. Pair measurements can conceal a resonant intermediate node, transformer-to-transformer tolerance interaction or loss that changes under the intended load.

Use the right multiport description

Mini-Circuits also requires the designed terminations when measuring a non-1:1 transformer. Raw S21 between two 50 Ω VNA ports is not automatically insertion loss for a 50-to-3,200 Ω device. Use physical terminations/matching networks with known loss, or an appropriate renormalized network calculation, and report the reference impedances.

For a centre-tapped or balanced transformer, measure the relevant three-port or mixed-mode S-parameters, amplitude/phase balance, output interaction and common-mode conversion. Calibrate or de-embed to the transformer terminals; include adapters, high-impedance fixture capacitance and connection repeats; state the uncertainty and instrument dynamic range.

A Complete Validation Sequence

1. Freeze the definition. Record topology, dot convention, ports/modes, source/load network, ratio, absolute turns, core/gap, winding and insulation drawings.
2. Verify low excitation. Measure complex impedance, magnetizing branch, leakage, winding resistance, turns ratio, return loss, transmission and resonances at declared planes and temperatures.
3. Measure the intended modes. Use full multiport/mixed-mode data for balanced structures; measure common-mode behavior separately.
4. Use load matrices. Test the intended complex loads and credible mismatch boundaries, not one nominal resistor alone.
5. Step large signal safely. Use calibrated voltage, current and power sensors; record waveform, frequency, PEP or average, duty, duration, ambient, enclosure and cooling.
6. Close the power balance. Compare accepted input, delivered output, reflected/unintended paths and loss with uncertainty. Measure temperature after thermal equilibrium and during credible transients.
7. Check stress. Verify flux including reset/DC bias, winding RMS/peak current, terminal and interwinding voltage, dielectric clearance, local hot spots, saturation margin and post-test drift.
8. Repeat tolerance corners. Include material/core/winding lots, temperature, assembly stress, ageing and connection repeatability.

A small-signal VNA cannot assign a power rating. A thermal camera alone cannot measure total loss without emissivity, hidden hot-spot and heat-flow controls. Back-to-back insertion loss alone cannot separate core, winding, fixture, mismatch and common-mode loss. The result becomes defensible when all three views—network, power balance and thermal/stress—agree within uncertainty.

Primary Sources and Measurement References

• Mini-Circuits AN20-001, How RF Transformers Work and How They Are Measured: manufacturer turns/impedance relationships, equivalent-circuit loss terms, designed terminations, three-unit back-to-back method and balanced-output measurement.
• TDK Ferrite Magnetic Design Tool: current manufacturer calculations for complex permeability, waveform-dependent core loss, skin/proximity-adjusted power, distortion and temperature.
• TDK/EPCOS Ferrites and Accessories application notes: manufacturer thermal-resistance, AC winding-loss and core power-capacity boundaries.
• TDK/EPCOS Ferrites and Accessories data book: complex/amplitude permeability, flux/core-loss, temperature, winding, skin/proximity and insulation definitions.
• IEC 62044-2:2005 with 2021 corrigendum: current low-excitation measuring methods for magnetic-core properties.
• IEC 62044-3:2023: current high-excitation methods for core power loss and amplitude permeability.
• Analog Devices MAXREFDES1249: primary reference-design documentation distinguishing flyback coupled-inductor energy storage, magnetizing inductance and leakage from direct transformer action.
• Texas Instruments, Magnetic Field Evaluation in Transformers and Inductors: manufacturer seminar distinction between true-transformer and flyback magnetic-field/current behavior.
• Keysight Impedance Measurement Handbook: complex impedance, method, calibration, cabling, fixture and transformer-measurement guidance.
• Keysight, De-Embedding and Embedding S-Parameter Networks: measurement-plane versus device-plane separation and fixture removal.
• IEEE 370-2020: active standard for high-frequency fixture design, measured-data quality, accuracy and consistency.

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