LLC Transformer Design: Simulation and Selection

This article summarizes a technical webinar delivered by power electronics consultant Nicola Rosano, who walked engineers through the entire LLC transformer design workflow using the Frenetic Collect simulation platform, from circuit fundamentals to validated physical schematics. LLC transformer design requires the resonant tank, magnetic construction and switching conditions to be developed together. This article explains the electrical design and simulation workflow for isolated power converters, using Nicola Rosano’s Frenetic webinar as a practical reference and extending it with design checks for conventional, matrix and planar transformers. It is intended for design engineers and component engineers specifying or sourcing high-frequency power magnetics.

Key Takeaways

The central design task is to obtain the required voltage gain without exceeding switching, magnetic, thermal or insulation limits.

The LLC tank contains series resonant inductance L_r, magnetising inductance L_m and resonant capacitance C_r. L_r may be external, integrated into transformer leakage, or shared between both.
Operating between the two characteristic resonant frequencies does not automatically ensure soft switching. Check the inductive operating region, commutation current, device capacitances and dead time.
The inductance ratio and load-dependent quality factor jointly shape the gain curve. Lower magnetising inductance can increase available boost gain but also increases circulating current.
Electrical scaling preserves an ideal normalised tank response only under defined assumptions. Core size, winding losses, capacitances and temperature require a new physical evaluation.
Matrix and planar constructions are design options, not universal upgrades. Compare the complete magnetic assembly, rectifiers, cooling, insulation and manufacturing cost.
Simulation should lead to a measurable component specification and a validated prototype, not directly to production approval.

Why LLC Transformer Design Is Non-Trivial

An LLC transformer provides isolation and voltage transformation while its magnetising inductance participates in the resonant network. When leakage supplies some or all of the series resonant inductance, winding construction also becomes a functional tank-design variable. These requirements apply to isolated switch-mode power converters, rather than signal transformers or mains-frequency transformers. [1, 2] The design must cover the required input voltage, output voltage and load range while controlling RMS current, core flux excursion, winding temperature and semiconductor switching stress. A change intended to improve one parameter can worsen another: separating windings can increase leakage but alters coupling and AC loss, while stronger interleaving can reduce leakage and winding loss but increase winding capacitance. [1, 3] The LLC network can be understood as an extension of a series LC resonant circuit with an additional magnetising branch in parallel with the transformer-referred load. This circuit model is more useful for design than treating a parallel-to-series load transformation as a physical rearrangement of the converter. The reference-kernel approach in Rosano’s webinar uses this network to select an operating region before scaling the electrical values and developing the magnetic construction. [1, 4]

Figure 1. Original LLC design illustration accompanying the webinar-based article. Source: Frenetic.

Figure 2. Original illustration for the LLC transformer design webinar presented by Nicola Rosano. Source: Frenetic.

The LLC Kernel Circuit and Switching Behaviour

A useful first design model refers the transformer and rectifier load to the primary side. It contains the series branch L_r and C_r, feeding the parallel combination of magnetising inductance L_m and equivalent AC load resistance R_ac. All inductances and impedances must use the same reference side. [1]

Characteristic Frequencies and Normalised Parameters

The series resonant frequency f_r is set by L_r and C_r. The lower characteristic frequency f_p includes L_m and describes the ideal no-load resonance; the loaded gain peak and inductive-to-capacitive boundary depend on load. Eq. (1) therefore identifies two useful reference frequencies, not two fixed boundaries enclosing an automatically safe operating region. [1]

f_{r}=\frac{1}{2\pi\sqrt{L_{r}C_{r}}},\qquad f_{p}=\frac{1}{2\pi\sqrt{(L_{r}+L_{m})C_{r}}}\tag{1}

Define the dimensionless inductance ratio L_n, normalised switching frequency f_n and effective quality factor Q_e as in Eq. (2). Here f_s is the switching frequency and R_ac is the equivalent AC load resistance referred to the primary. Some references use a different inductance-ratio convention, so compare definitions before transferring a numerical value. [1]

L_{n}=\frac{L_{m}}{L_{r}},\qquad f_{n}=\frac{f_{s}}{f_{r}},\qquad Q_{e}=\frac{\sqrt{L_{r}/C_{r}}}{R_{ac}}\tag{2}

Under the conventional first harmonic approximation, or FHA, the resonant-tank gain magnitude M is given by Eq. (3), where j is the imaginary unit. M describes the tank’s fundamental-frequency voltage transfer; it is not the complete DC conversion ratio, which also depends on bridge configuration, transformer turns ratio and rectification. The approximation is most useful near series resonance. [1, 2]

M=\left|\frac{L_{n}f_{n}^{2}}{(L_{n}+1)f_{n}^{2}-1+jQ_{e}L_{n}f_{n}(f_{n}^{2}-1)}\right|\tag{3}

At f_n = 1, Eq. (3) gives unity tank gain in the ideal model. The reactances of the series inductor and capacitor cancel, so the series LC branch has minimum impedance, not maximum impedance. The complete input impedance still includes the magnetising branch and reflected load. [1, 2]

Primary Zero-Voltage Switching

For primary zero-voltage switching, or ZVS, the tank must provide current with the appropriate direction to charge and discharge the switching-node capacitances before the next device turns on. An inductive input impedance is a necessary design condition in the conventional operating analysis, but it does not by itself prove that the transition will complete. Available current, nonlinear device output charge, transformer capacitance and dead time must also be checked. [1, 3]

Primary ZVS can occur both above series resonance and below it within the inductive part of the operating region. Between f_p and f_r, the impedance character depends on load. Moving too far below resonance can enter capacitive operation and cause hard commutation. Verify the boundary at the relevant load and component-tolerance corners rather than relying only on the two frequencies in Eq. (1). [1, 2]

Secondary Zero-Current Switching

Secondary zero-current switching, or ZCS, occurs when rectifier current returns to zero before commutation. Conventional LLC operation at or below series resonance can provide this behaviour, whereas above-resonance operation can retain primary ZVS while losing secondary ZCS. Synchronous-rectifier timing and device capacitances must be included when assessing the real secondary stage. [1, 3]

Soft switching reduces particular switching-loss mechanisms; it does not eliminate conduction loss, gate-drive loss, magnetic loss or every turn-off loss. The optimum operating point is therefore determined by the converter’s total loss and regulation requirements, rather than by a ZVS label alone. [2, 3]

Operating Points A and B, Circulating Current and Core Loss

In the webinar’s comparison, point A near the higher resonance is preferred to a lower-frequency point B because it reduces circulating-current stress within that particular design. This is not a general statement that input impedance reaches a universal maximum at resonance, or that below-resonance operation should always be avoided. Below-resonance boost operation is often needed to maintain regulation at a lower input voltage. [2, 4]

Core flux excursion follows the voltage-time integral across the magnetising branch. Core loss must consequently be evaluated using the actual excitation waveform, frequency, material and temperature. A lower switching frequency may increase flux excursion, but the resulting loss cannot be inferred from input impedance or total primary current alone. The webinar itself discusses the competing effects of flux excursion and frequency rather than establishing a universal core-loss ranking. [1, 2, 4]

Scaling Laws: Rapid Design Adaptation Across Power and Frequency

A validated electrical kernel can provide a useful starting point for related converters. Scaling preserves its ideal normalised response only when voltage levels, equivalent load, frequency and tank impedance are changed consistently. It does not automatically preserve physical losses or soft-switching margin in the finished converter. [1, 4]

Let k_f be the target-to-reference frequency ratio and k_Z the target-to-reference impedance ratio. Primed quantities denote the target design. Applying the same scaling to both inductances, the capacitor and the equivalent load gives Eq. (4), derived from the reactance and normalisation relationships used in Eqs. (1) and (2). [1]

L_{r}’=\frac{k_{Z}}{k_{f}}L_{r},\qquad L_{m}’=\frac{k_{Z}}{k_{f}}L_{m},\qquad C_{r}’=\frac{C_{r}}{k_{Z}k_{f}},\qquad R_{ac}’=k_{Z}R_{ac}\tag{4}

For unchanged voltage levels, transformer ratio and topology, the equivalent load resistance varies inversely with output power. Thus, doubling power at unchanged frequency halves both inductances and doubles capacitance in the ideal scaled tank. Doubling frequency at unchanged impedance halves both inductances and capacitance. These are electrical value changes, not instructions to halve the physical component size.

For the original illustrative product family of 500 W, 1 kW and 2 kW, unchanged voltage and frequency would give inductance multipliers of 1, 0.5 and 0.25, and capacitance multipliers of 1, 2 and 4. Each variant still requires its own winding-current, core-loss, thermal and commutation checks. If voltage levels also change, first recalculate the impedance scale from the equivalent load and establish the required turns ratio separately.

The practical sequence is to characterise a reference kernel, choose the new frequency and impedance scales, derive the target tank values, and then check the switching waveforms. Core geometry, integer turn counts, insulation, AC resistance and winding capacitance are physical design variables that must be reassessed rather than directly obtained from two scaling factors. [1, 3, 4]

Frenetic Collect Platform: Workflow and Key Features

Rosano’s webinar demonstrates a browser-based magnetic-design workflow using the Frenetic Collect platform. Its useful engineering sequence is to establish and validate the converter operating point, enter the resulting magnetic requirements, compare cores and windings, and prepare a construction handover. The demonstration is a worked workflow, not a guarantee of development time or final manufacturing performance. [4]

Input Parameter Entry

Begin with a converter specification that distinguishes electrical requirements from the assumptions used in the magnetic model. The demonstrated interface includes operating-point, winding, temperature and insulation inputs. [4]

Input voltage range, required output voltage and output-power range.
Switching-frequency operating points and the series resonant frequency.
Magnetising inductance, target resonant inductance and whether the resonant inductor is integrated or external.
Primary-to-secondary turns ratio, with the secondary winding definition stated explicitly for centre-tapped rectification.
Ambient temperature, cooling conditions, permissible temperature rise and insulation requirements.
Additional load points and any auxiliary windings required by the converter.

The secondary power entry demonstrated in the webinar is useful for inspecting another load operating point without rebuilding the magnetic design. Changing a power entry and regenerating waveforms should not be confused with validating a closed-loop load-step transient, which also depends on the controller and output network.

Graphical Design Approach

The webinar first uses gain and input-impedance information to choose the electrical operating point, then transfers the resulting requirements into the magnetic-design environment. Gain alone is insufficient: an apparently suitable voltage-transfer point can lie in the wrong impedance region or lack commutation margin. The inductance ratio and quality factor must therefore be considered together. [1, 4]

Reducing L_m relative to a fixed L_r generally increases available boost gain and can obtain the required gain change over a smaller frequency excursion, at the expense of higher magnetising current. Increasing L_m reduces that current but can require a wider frequency span or restrict available boost gain. These trends are not independent of load and the chosen tank impedance. [1, 2]

Simulation and Validation

The webinar uses TINA circuit simulations to validate the electrical design and compare waveforms with the magnetic platform. This is an external cross-check, not evidence that every platform calculation is a built-in SPICE simulation. FHA, switching-circuit simulation and magnetic simulation answer different questions and should agree within their respective assumptions. [4]

Map the input inverter correctly. For the same DC bus and conventional symmetrical excitation, a full bridge applies twice the AC tank excitation amplitude of a half bridge. A half-bridge switching node moves between the bus rails, but its AC excitation after removal of the DC component has half the full-bridge amplitude. Mixing these conventions produces a factor-of-two error. [1, 2]

Specify whether the secondary uses a full-bridge rectifier or a centre-tapped full-wave rectifier. Both provide full-wave rectification, but they differ in winding utilisation, conducting-device count and device voltage and current stress. For a centre-tapped secondary, state whether the turns ratio refers to one half winding or the complete end-to-end winding. [1, 4]

Inspect primary switch voltage and current during dead time and at turn-on.
Check rectifier-current extinction and synchronous-rectifier timing across the operating range.
Compare the actual magnetising-branch voltage, tank current and secondary current between models.
Evaluate startup, no-load, overload and load-step conditions separately from nominal steady-state operation.
Use the resulting voltage and current waveforms for magnetic-loss and temperature calculations. [1, 3, 4]

Core Optimiser

The demonstrated tools allow core candidates and winding alternatives to be explored, including loss-related comparisons against turns and core geometry. Use these results to identify feasible constructions rather than assuming that the lowest predicted loss is automatically the best component. Winding-window utilisation, insulation space, manufacturability and available core and bobbin combinations remain constraints. [4]

For a transformer, core selection is not solely an energy-storage calculation based on inductance and peak current. Applied volt-seconds, flux excursion, core-loss characteristics, winding RMS current and cooling also matter. Where resonant inductance is integrated, the magnetic assembly must additionally accommodate the required leakage field and its associated losses. [1, 3]

Mechanical Design Integration

The webinar’s mechanical handout brings together winding arrangement, core and coil-former information, waveforms and predicted performance. It is a useful starting point for supplier discussion. A production drawing must additionally define tolerances, approved insulation materials, winding connections, terminal identification, test conditions and acceptance limits. [4]

Pay particular attention to winding placement when leakage is intentional. A drawing that states only turns, wire size and core type does not fully define an integrated LLC transformer: chamber spacing, interleaving and insulation build can materially alter the realised resonant inductance.

Case Study 1: Standard LLC Transformer Design

The first webinar example develops a conventional wound transformer after selecting and scaling an electrical kernel. It demonstrates two important checks: the difference between an FHA operating estimate and a switching simulation, and the difference between requested inductances and a physically modelled winding arrangement. [4]

Electrical Operating-Point Validation

At the minimum input-bus condition, the webinar reports an FHA frequency estimate of approximately 53 kHz, while the TINA model regulates at approximately 60 kHz. The difference is 7 kHz, or 13.2% relative to the estimate. It illustrates why the initial gain calculation should be followed by time-domain simulation rather than treated as an exact controller frequency. [4]

The magnetic-design inputs discussed in the demonstration include magnetising inductance of approximately 52 µH and requested leakage inductance of approximately 5.7 µH. Their ratio is approximately 9.12. This is consistent with a nominal ratio near nine, rather than the ratio of six associated with the second kernel. [4]

Realised Inductance and Winding Adjustment

The example uses a separated winding arrangement to obtain intentional leakage. Adjusting the distance between winding chambers changes the modelled leakage, with approximately 5.2 µH discussed against a requested 5.7 µH. That is approximately 8.8% below the target; it is a design difference to assess at converter level, not an automatic acceptance. [4]

The example also compares a calculated magnetising inductance with the requested value and explores winding type, parallel conductors, layers and auxiliary windings. Each revision should feed back into the tank model because the realised inductances, rather than the initial requested values, determine the actual gain and switching behaviour.

Keep a one-page electrical summary alongside the magnetic construction record. It should identify the input and output operating points, bridge and rectifier configuration, reference side of each inductance, requested and realised values, frequency range and waveform files used for loss calculation. This preserves the connection between circuit requirements and the component handed to the supplier.

Case Study 2: Matrix Transformer Design

The second example compares a single-transformer solution with a matrix arrangement using several transformer elements. In the demonstrated series-primary arrangement, the elements share the primary excitation and divide the secondary current-handling task. The electrical connection and rectifier arrangement, not simply the number of cores, define the equivalent transformer. [4]

Figure 3. Original single-core and matrix-transformer comparison accompanying the webinar example. Source: Frenetic.

Reference-Kernel Adaptation

The webinar changes its illustrative reference kernel from an inductance pair of 100 and 900 to 143 and 857, with both values expressed on the same scale. The first pair gives L_m/L_r = 9; the second gives 857/143 = 5.993, approximately six. An exact ratio of six with L_r = 143 would require L_m = 858 on that same scale. [4]

This is a deliberate change of kernel and inductance ratio, not a demonstration that increasing output power necessarily increases the physical leakage inductance from 100 µH to 143 µH. The dimensional tank values must be obtained from the complete scaling calculation. The lower ratio is then checked for gain coverage and switching behaviour in the new example.

The second example also shows a frequency correction from approximately 85 kHz in the initial estimate to approximately 95 kHz in the switching simulation. The 10 kHz difference is approximately 11.8% of the estimate, reinforcing the need to validate the operating point before completing the magnetics. [4]

Mapping One Equivalent Transformer into Three Elements

For the specific three-element mapping demonstrated, the presenter divides the equivalent resonant inductance, magnetising inductance and primary-to-secondary turns ratio by three when specifying each transformer element. This mapping relies on the demonstrated winding connections and secondary rectification. It must not be applied unchanged to a different series, parallel or integrated-core arrangement. [4]

Confirm the mapping by comparing the individual transformer voltages and currents with the complete circuit simulation. Check that the assembled network reproduces the intended equivalent inductances and conversion ratio, and that each secondary carries its intended share of current.

Compare the Complete Magnetic System

In the demonstrated comparison, the single-core approach needs a separate resonant inductor, while the matrix solution can integrate the required resonant inductance into its transformer elements. A fair comparison therefore includes the external inductor, its losses, volume and cooling, rather than comparing transformer cores alone. [4]

The webinar also demonstrates importing voltage and current waveforms from a circuit simulator through CSV files to design the separate resonant inductor. This is useful where the inductor’s actual resonant waveform differs from the assumptions of a standard hard-switched converter template.

The matrix solution is preferred within that example because of its combined volume, current distribution and cooling result. This is not a universal claim that multiple cores always reduce losses or parasitics. The presenter also identifies cost estimation and finite-element checking as important parts of a more complete design assessment. [4]

Benefits and Limits of the Matrix Approach

Symmetry can improve current sharing, but requires matched magnetic elements, controlled winding connections and balanced PCB paths.
Distributed cores and secondary windings can spread heat, but the actual hot spots depend on airflow, mounting and local copper loss.
Shorter secondary current paths can be useful, but total leakage and capacitance depend on the complete arrangement rather than core count alone.
Adding elements changes the equivalent electrical network, rectifier layout and assembly cost; it requires remapping and validation rather than simple mechanical replication. [4]

CLLC Extension: Bidirectional Converter Support

For bidirectional power conversion, topology selection must precede detailed transformer optimisation. The webinar introduces CLLC support as an available premium feature at the time of recording, rather than as a future addition. [4]

A CLLC arrangement adds secondary-side resonant elements to support resonant conversion in both power-flow directions. Potential application classes include bidirectional battery chargers, vehicle-to-grid interfaces, battery energy storage and microgrid DC–DC stages. Soft switching still depends on operating point and control; adding the secondary tank does not guarantee it throughout both operating envelopes.

Define voltage range, power direction, turns ratio and resonant-element reference sides for both modes. Check the semiconductor commutation, transformer currents and thermal limits separately in each direction. A transformer optimised only for one-way LLC operation is not automatically an acceptable bidirectional design.

Planar Magnetics Integration and High-Frequency Design

Planar magnetics are no longer only an upcoming feature in the Frenetic product description. Frenetic now presents a dedicated Planar Tool for entering core and winding-layer geometry and calculating losses and parasitic behaviour using a FEM-based engine. This provides a current extension to the conventional-winding workflow demonstrated in the webinar. [5]

PCB Windings Are a Different Optimisation Problem

PCB windings offer low-profile construction and repeatable layer geometry, but they must be assessed for AC resistance, proximity effect, gap-fringing fields, interwinding capacitance and heat extraction. Copper thickness, layer overlap, spacing and interleaving are coupled variables; improving leakage or resistance can worsen another parameter. [3, 5]

Do not transfer a wound-transformer current-density rule directly to a planar stack without evaluating its loss and cooling paths. Internal copper, vias, core contact and the surrounding assembly influence temperature. A smaller magnetic volume is useful only if the resulting hot spots and insulation temperatures remain acceptable.

Capacitance and Dead Time Must Be Co-Designed

Higher-frequency LLC designs make transformer capacitance part of the semiconductor commutation problem. Infineon’s high-frequency soft-switching application note identifies primary and secondary device capacitances, transformer winding capacitance and layout capacitance as contributors to the charge that must be moved during dead time. Selecting L_m without these contributions can overestimate ZVS margin. [3]

Lower-output-charge switches may permit less magnetising current or shorter dead time, but they do not remove the need to optimise the transformer. The same application note discusses the interaction between planar winding construction, capacitance and device loss. The practical update is to co-design semiconductor selection, winding geometry and commutation timing rather than treating them as independent tasks. [3]

At very light load, parasitic capacitances can introduce an additional high-frequency resonance and alter the expected gain-versus-frequency behaviour. Include the transformer and rectifier capacitances when checking maximum-frequency operation and no-load regulation. Also review the common-mode current path across the isolation barrier as part of the converter’s EMC assessment. [3]

Design-In Notes for Engineers

A useful LLC transformer specification describes how the component behaves in its converter, not just its nominal inductance and power rating. The following checklist connects electrical modelling, mechanical construction and supplier acceptance.

Electrical and Datasheet Checklist

State bridge configuration, rectifier configuration, input and output ranges, continuous and peak loads, switching-frequency limits and cooling conditions.
Specify turns ratio, polarity and pin assignment; define each half winding explicitly for centre-tapped secondaries.
Specify L_m, effective L_r and their tolerance ranges, including the reference side and whether an external resonant inductor is included.
Compare winding DC resistance, predicted AC loss under the specified waveforms, winding capacitance, core material, core-loss estimate and hot-spot temperature.
Check C_r tolerance, RMS current, peak voltage and loss because resonant-capacitor variation also changes the operating range.
Evaluate gain and switching behaviour at combined component-tolerance, voltage, load and temperature corners, not only at nominal values. [1, 3]

Measurement and Acceptance Conditions

Agree the inductance measurement method with the supplier. Record instrument frequency, test amplitude, fixture compensation, temperature and the state of every other winding. An open-circuit primary-inductance measurement is not automatically identical to the magnetising-branch value of every equivalent-circuit model; likewise, a short-circuit leakage reading must be interpreted using the agreed model and winding configuration.

Measure the prototype’s inductances, resistance and relevant capacitances, then correlate the component model with converter waveforms. Verify ZVS at the switch terminals, rectifier commutation, winding and core temperatures, and output regulation under the intended operating conditions. Use finite-element analysis when geometry-dependent leakage, fringing or local AC loss makes a simple analytical model inadequate. [1, 3, 4]

Catalog Versus Custom Parts

A catalog transformer is suitable only if its specified ratio, inductances, current handling, frequency range and insulation construction match the intended converter. A nominal power label does not establish compatibility with an LLC tank. An external resonant inductor can provide an additional tuning variable when the transformer’s leakage is not the required L_r.

A custom transformer provides more control over leakage, winding arrangement, auxiliaries and mechanical interfaces, but requires a controlled drawing and acceptance specification. For purchasing, compare the complete solution: transformer elements, external inductors, rectifier interconnects, cooling hardware, tooling, prototype cost and production test. Require approval before changes to core material, winding wire, insulation build or assembly geometry.

Insulation, Qualification and Safety

Determine the applicable end-product safety requirements before fixing the winding stack or bobbin. Specify the required insulation function, working voltage, transient environment, creepage, clearance, solid insulation and dielectric tests for that application. A dielectric-withstand voltage by itself is not a complete insulation specification.

Confirm the temperature ratings and compatibility of winding insulation, bobbin, tape, PCB materials, adhesives and potting compounds. Define environmental and mechanical qualification from the product’s mission profile. For a matrix design, include element-to-element tolerances and current-sharing checks; for a planar design, include the approved PCB stack-up and manufacturing tolerances. Do not infer end-product compliance from the simulation tool’s insulation menu.

Conclusion

LLC transformer design is a coordinated electrical, magnetic and thermal task. The gain curve establishes a feasible operating range, but the realised inductances, winding capacitances, semiconductor commutation and cooling determine whether that range is practical.

The webinar’s workflow remains useful: establish a reference kernel, scale it consistently, validate the switching circuit, and then develop the core and winding construction. Its two examples also show why estimated frequency and requested inductance should not be mistaken for final hardware values.

Matrix and planar implementations extend the available design choices, while bidirectional applications require their own topology and operating-envelope analysis. The final deliverable is a controlled component specification supported by prototype measurements and converter-level validation, rather than a simulation result alone.

FAQ

These answers summarise the main decisions involved in designing and specifying an LLC transformer.

What makes an LLC transformer different from a standard power transformer?

An LLC transformer has a deliberately specified magnetising inductance that participates in the resonant network. Its leakage inductance may also supply some or all of the series resonant inductance. The required turns ratio, inductances, winding losses, capacitances and insulation must therefore be developed together rather than selecting the component from voltage ratio and nominal power alone.

How does the inductance ratio affect LLC converter design?

The inductance ratio is the magnetising inductance divided by the total series resonant inductance in the chosen equivalent circuit. Reducing magnetising inductance can increase boost gain and reduce the frequency excursion needed for a given gain change, but raises magnetising current. Select the ratio together with the load-dependent quality factor, frequency limits and acceptable conduction losses.

Does operating between the two LLC resonant frequencies guarantee ZVS and ZCS?

No. Between the two characteristic resonant frequencies, the input impedance can be inductive or capacitive depending on load. Primary zero-voltage switching also requires sufficient commutation current and suitable dead time to move the relevant capacitive charge. Secondary zero-current switching depends on the rectifier current waveform and timing. Check both conditions in switching simulation and prototype measurements.

What does Frenetic contribute to the LLC transformer design workflow?

The demonstrated workflow uses electrical requirements and validated operating waveforms to explore core and winding constructions, predicted losses and temperature. It also supports preparation of a mechanical handover. Circuit simulation remains a separate validation task in the webinar, and the final component still needs an agreed manufacturing specification, measured parameters and converter-level prototype testing before approval.

Can an existing LLC transformer design be scaled to a higher power?

An existing electrical tank can provide a starting point when frequency, impedance and load are scaled consistently. At unchanged voltage levels and frequency, doubling power halves the ideal scaled inductances and doubles resonant capacitance. This does not establish the new core size or temperature. Recheck winding current, losses, capacitance, insulation and switching margin for the physical design.

When should a matrix transformer be considered for an LLC converter?

Consider a matrix arrangement when secondary current distribution, magnetic volume, layout or cooling makes a single transformer difficult to optimise. There is no universal power threshold. Compare the complete assembly, including any external resonant inductor, rectifiers and interconnects. The winding connections determine how each element must be specified, and current sharing and manufacturing cost require explicit evaluation.

What should a buyer request from an LLC transformer supplier?

Request a controlled drawing stating turns ratio, polarity, inductance tolerances, winding construction, resistance limits, insulation materials and test conditions. Include the operating waveforms, frequency range, cooling conditions and acceptance temperatures. Agree how leakage and magnetising inductance are measured, and require approval for material or construction changes that could alter the resonant behaviour or insulation system.

How to Design and Validate an LLC Transformer

Use this procedure to connect the converter specification to a physically verified transformer. It follows the webinar’s electrical-to-mechanical sequence while adding prototype and supplier acceptance checks.

Develop an LLC transformer from electrical requirements to a measured and controlled component specification.

Define the converter specification
Record input and output voltage ranges, load range, switching-frequency limits, bridge configuration, rectifier arrangement, cooling conditions and insulation requirements. Define the turns-ratio convention and the reference side for every inductance.
Select the electrical tank
Choose magnetising inductance, series resonant inductance and resonant capacitance together with the turns ratio. Evaluate gain, quality factor and the inductive operating region across the required load and voltage range rather than selecting the inductance ratio alone.
Scale a reference design when appropriate
If a validated electrical kernel is available, establish consistent frequency and impedance scaling factors. Recalculate the inductances, capacitance and equivalent load. Treat the result as an electrical starting point and reassess the physical core and winding design.
Validate the switching circuit
Run time-domain simulation with the correct bridge excitation and rectifier configuration. Check regulation, commutation current, dead time and rectifier timing. Include relevant capacitances and evaluate light load, startup, overload and load-step behaviour separately from nominal steady state.
Develop the magnetic construction
Compare core materials, core sizes, gaps and winding arrangements using the validated voltage and current waveforms. Evaluate realised inductances, AC winding loss, core loss, capacitance and hot-spot temperature. Include any external resonant inductor when comparing alternative constructions.
Prepare the supplier handover
Create a controlled drawing with winding connections, polarity, construction details, approved materials, tolerances and insulation requirements. Agree inductance and resistance measurement conditions, dielectric tests and acceptance limits. For matrix or planar designs, define element matching or the approved PCB stack-up.
Measure and approve the prototype
Measure the component parameters and correlate them with the circuit and magnetic models. Verify converter switching waveforms, regulation and temperature across the operating envelope. Update the design and acceptance specification before approving production, including any required environmental and safety qualification.

References

References [1]–[3] support the circuit and switching discussion; [4] is the original worked demonstration; [5] documents the current planar-tool description.

Hong Huang, Designing an LLC Resonant Half-Bridge Power Converter, Texas Instruments Power Supply Design Seminar, SLUP263, 2010–2011.
Mladen Ivankovic, LLC Design Guide: 3300 W Converter — Simplified Algorithm Based on FHA and Vector Method, Infineon Technologies, application note AN_2302_PL52_2302_230440, 2023.
Infineon Technologies, Optimizing Soft-Switching Operation of GaN at High Frequency, application note AN152548, version 1.0, 2025.
Nicola Rosano and Frenetic, Design an LLC Transformer Easily in Frenetic Simulator with Nicola Rosano, Frenetic webinar, YouTube, 2024.
Frenetic, Planar Tool: Design and Simulate Planars, Frenetic product documentation, undated; accessed 2026.

Source: Frenetic

Trending Tags