Galvanic Isolation at HF: RF Transfer, Common Mode and Safety

Transformers are routine HF components, but “isolated” is not one complete specification. A design must state which differential energy should cross the barrier, how much common-mode RF may cross it, the electrical and thermal operating envelope, and whether the barrier has a certified protective-safety function.

Start by Naming the Three Different Requirements

In radio, HF conventionally means 3–30 MHz. Power-electronics engineers also use “high frequency” informally for switching rates from kilohertz upward. The physics overlaps, but the circuit objectives and applicable safety standards may not.

A transformer may satisfy the first requirement, deliberately pass differential HF energy, and still couple common-mode HF through capacitance. It may also use only functional insulation—insulation needed for operation—rather than insulation credited as a shock-protection safeguard. Those are not contradictions; they are different specifications.

Wanted Differential Transfer and Unwanted Common-Mode Transfer

A conventional isolated RF transformer transfers the wanted signal by mutual magnetic flux. With appropriate winding polarity, impedance and termination, the differential signal crosses the winding barrier by design. That is why “galvanically isolated” must never be read as “blocks all AC.”

The same physical assembly also contains primary-to-secondary capacitance, winding-to-core and winding-to-shield capacitance, package capacitance and capacitance through the PCB and enclosure. These provide electric-field paths for common-mode current. For a fast common-mode voltage transition, the first-order relation is:

For an illustrative 10 pF primary-to-secondary branch, the sinusoidal reactance is:

This arithmetic is not an isolation specification. Actual current also needs a complete return path, and actual transfer depends on source impedance, load impedance, winding voltage distribution, other stray capacitances, shields, cable common mode and the measurement reference.

Texas Instruments’ measured converter work shows why this matters with fast SiC or GaN edges: reducing transformer interwinding capacitance reduced measured common-mode current, while converter topology and winding construction changed the leakage-capacitance tradeoff. The same displacement-current mechanism applies in HF equipment even when the waveform and voltage are different.

Electrostatic shields redirect current; they do not erase it

A correctly designed screen can intercept electric-field current and return it to a chosen quiet reference. It also creates capacitance to that screen, can alter leakage and resonance, and may worsen performance if its connection becomes an RF antenna or crosses the intended safety barrier incorrectly. A screen or Y-capacitor is part of the complete EMC and safety design—not a universal retrofit.

The Real HF Equivalent Circuit

The useful model contains an ideal transformer plus finite magnetising inductance, core-loss resistance, winding resistance, leakage inductances and distributed capacitances. Package, PCB, connector, cable and enclosure parasitics complete the network.

• At the low-frequency edge, finite magnetising reactance draws current and can reduce input impedance, return loss and transferred voltage.
• Through the passband, turns ratio, coupling, source/load impedance, winding structure and losses set insertion loss, balance and phase.
• At the high-frequency edge, leakage inductance, capacitance, conductor loss and core behavior can produce roll-off, peaking and resonance.

For example, |XL| = 2πfL gives 18.85 Ω for 100 nH at 30 MHz. That number alone is not mismatch or insertion loss. Leakage is distributed between ports, interacts with capacitance and termination, and must be referred through the turns ratio consistently.

Mini-Circuits’ RF-transformer measurement guidance makes the boundary explicit: insertion loss is measured in a properly impedance-matched system relative to an ideal transformer of the same ratio, and return loss requires the opposite winding to see its intended termination. A headline frequency range without test impedance, loss criterion and termination is incomplete.

Topology changes both the conductive path and the model

• A two-winding flux transformer can provide a DC break between windings while magnetically transferring the wanted signal.
• A transmission-line transformer uses coupled line modes; some configurations provide a DC break and some do not. Its high-frequency behavior is not captured by a single lumped turns ratio.
• An autotransformer or many unun connections share a conductor and therefore do not provide galvanic separation between ports.
• A flyback magnetic component stores and releases energy and is not adequately analysed as an ideal instantaneous power transformer.

Choose topology from the required differential transfer, common-mode boundary, DC path, bandwidth and safety function—not from the word “transformer.”

Flux, Voltage and Current Set Different Limits

Faraday’s law ties core flux change to the winding volt-seconds:

The first expression is the safer design starting point because waveform, duty cycle, reset interval and DC imbalance matter. A tiny average-voltage imbalance in a driven power transformer can walk the flux toward saturation. At RF, low-frequency band-edge voltage is often the severe flux case, but core loss can become the governing thermal limit as frequency rises. Material, effective core area, turns, temperature and waveform must all be known.

Current creates another limit. Copper loss follows winding RMS current and frequency-dependent AC resistance; skin and proximity effects make AC resistance exceed DC resistance. Load current, magnetising current and any unbalanced DC component must be separated. Connector, PCB, termination and winding hot spots can fail before the core does.

This is why “100 W transformer” is not a transferable rating. Required evidence includes:

• frequency band and waveform, including modulation envelope and duty cycle;
• source/load impedance and permitted mismatch or open/short transients;
• primary and secondary RMS/peak voltage and current;
• DC bias or volt-second imbalance;
• core material, geometry, turns, winding arrangement and enclosure;
• ambient, airflow, temperature-rise limit and insulation-system temperature class.

Validate insertion loss and distortion at power, then measure core, winding, terminal and nearby-insulation temperatures after thermal equilibrium. A small-signal VNA trace cannot establish the QRO envelope.

Safety Isolation Is a System Claim

Current IEC terminology distinguishes levels and functions of insulation. Reinforced insulation provides protection equivalent to double insulation; basic and supplementary safeguards have different roles. IEC 62368-1:2023 applies a hazard-based safeguard framework to audio/video, information and communication technology equipment. IEC 61558-1 and IEC 61558-2-16 address safety requirements and tests for relevant transformers and switch-mode power-supply transformers. The applicable standard still depends on the end product, energy source, market and jurisdiction.

IEC 60664-1 is an insulation-coordination foundation for specified low-voltage supply systems and covers creepage, clearance and solid-insulation criteria within its scope. Its published scope is limited to frequencies up to 30 kHz. IEC 60664-4 addresses periodic high-frequency voltage stress above 30 kHz through 10 MHz and is used together with Part 1 or Part 5. Even that does not span the whole 3–30 MHz radio HF band, which is a warning against copying one spacing table blindly into an RF construction. Product-standard rules and application-specific frequency effects govern.

The complete barrier includes winding wire, bobbin, tape or margin, core and clamp, potting, lead exits, PCB land pattern, slots, enclosure, contamination control and protective components. Working voltage, repetitive and non-repetitive peaks, overvoltage category, pollution degree, altitude, temperature and manufacturing tolerances all matter. UL’s electrical-insulation-system guidance also stresses that the material combination is evaluated as a system; individual material temperature ratings are not a substitute.

Certification or recognition of a transformer can support an end-product assessment only within its documented conditions of acceptability. It does not automatically certify a different PCB layout, shield connection, insulation system or completed radio.

A Defensible Selection and Test Workflow

1. Define the barrier. Identify which conductors/domains must have no DC path and whether the need is functional, EMC-related or protective.
2. Define differential transfer. State band, impedance ratio, source and load, insertion loss, return loss, balance, phase/group delay and distortion limits.
3. Define common-mode performance. State source/return structure, common-mode voltage spectrum or edge rate, permitted transfer current or voltage, and installed cable/enclosure configuration.
4. Define stress. Record peak and RMS voltage/current, waveform, duty, mismatch, DC bias, transients, startup/fault cases, ambient and cooling.
5. Define safety evidence. Identify the applicable product standard, insulation function, working/impulse voltage, pollution degree, overvoltage category, altitude, material/temperature system and required component approval or certification.
6. Select the topology and part. Reject any candidate whose datasheet omits a parameter the design relies upon; request controlled manufacturer data where needed.
7. Measure the small-signal path. With suitable fixtures, calibration/reference planes and intended terminations, measure S11/S21 across and beyond the band. Record both magnitude and phase where relevant.
8. Measure the unwanted path separately. Characterise primary-secondary capacitance and common-mode transfer with a defined fixture. Differential insertion loss does not reveal common-mode isolation.
9. Validate at operating stress. Test the exact waveform, load range, mismatch/fault cases and thermal environment with rated instruments and conservative staged power.
10. Assess the complete equipment. Keep hazardous-energy dielectric, fault and compliance tests with qualified personnel and an appropriate laboratory procedure.

When Is a Transformer the Right HF Choice?

Use one when its measured transfer function, common-mode behavior, stress margins and—if required—documented safety construction meet the actual system requirement. That may describe an RF input transformer, balun, pulse transformer, converter transformer, Ethernet magnetic component or certified isolated interface. The parts are not interchangeable merely because each has two windings.

Avoid a candidate when the only supporting fact is “the windings are separate,” when the datasheet’s band and impedances do not match the circuit, when common-mode transfer is uncharacterised, when flux or temperature margins are unknown, or when a protective-isolation claim lacks the applicable evidence.

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