This article based on Frenetic webinar summarizes a practical design workflow using Frenetic AI, Frenetic Simulator and the new Planar Design Tool to compare a conventional PQ32/35 transformer against an EQ38/8/25 planar solution in a 700 W phase‑shift full bridge (PSFB) converter.
Planar magnetics are becoming an important option for high‑density power converters where low profile, reproducible parasitics and scalable manufacturing are critical.
Key Takeaways
- This article compares conventional vs planar transformers in a 700 W PSFB converter using Frenetic tools.
- Planar magnetics offer low profiles and better control of parasitics, making them suitable for high-density applications.
- Conventional transformers provide flexibility for prototyping, while planar solutions are cost-effective for large-scale production.
- The design workflow includes specifying parameters in Frenetic AI, simulating with Frenetic Simulator, and creating PCB layouts with the Planar Design Tool.
- Ultimately, the choice depends on performance requirements, production volume, and time-to-market needs.
Key features and benefits
The Frenetic article consider a practical example of 700 W PSFB full bridge with input 370–440 VDC, 28 V output and 100 kHz switching frequency, a typical high‑power, high‑density application where transformer volume and parasitics matter.
Conventional and planar transformers are contrasted mainly through their winding technology and geometry. A conventional transformer for this class of design uses round, litz or foil wound wire on a PQ32/35 core, resulting in relatively tall magnetics with a larger height and more variability in leakage inductance and stray capacitance from build to build. The planar option instead uses PCB windings with flat core shapes such as EQ or ER families, which inherently provide low profile, more symmetric windings and better control of parasitics over production.
From a practical standpoint:
Conventional wound magnetics offer high flexibility for prototyping. Engineers can rework layer order, add or remove turns or adjust terminations quickly without waiting for new PCB tooling.
Planar transformers, in contrast, demand PCB fabrication for any topology or turns change, adding cycle time for variants but enabling better repeatability once the design is fixed.
Economically, conventional construction tends to be more attractive for prototypes and small‑volume production runs because off‑the‑shelf cores, coil formers and wire can be sourced and assembled quickly in‑house. For large‑volume production (hundreds of thousands to millions of pieces), planar transformers become cost‑effective because PCB fabrication scales efficiently once the fixed tooling cost is amortized.
Typical applications
The example in the webinar is a 700 W PSFB full bridge stage with high‑voltage input and low‑voltage, high‑current output, representative of many front‑end and isolated DC‑DC stages. Such a converter is typical in industrial and telecom power supplies, high‑power chargers and potentially automotive or server auxiliary supplies where high power density is important.
Applications where planar designs are particularly attractive include:
- High‑density AC‑DC or DC‑DC modules requiring low profile magnetics, for example in 1U power shelves or compact EV chargers.
- Designs where tight control of leakage inductance and capacitance is essential, such as resonant topologies, PSFB converters using leakage for zero‑voltage switching, or systems with strict EMC constraints.
- High‑volume products where manufacturing repeatability of parasitics and losses is as important as electrical performance.
Conversely, conventional wound transformers remain a good fit for:
- Early prototyping or proof‑of‑concept builds where the transformer design is still changing and fast iteration is required.
- Low‑to‑medium volume projects where PCB tooling and layout effort for planar does not pay off.
- Situations where custom mechanical integration (special terminations, unusual mounting) is easier to achieve with wound parts.
Technical highlights
The workflow begins in Frenetic AI, where the designer enters converter specifications and constrains the primary to 10 turns to keep a reasonable window for adapting the design. Frenetic AI returns a magnetic design and proposes both concentric (conventional) and planar candidate solutions, including suggested cores. For the conventional option, the tool proposes a PQ32/35 core, leading to about 6.1 W total losses at the target operating conditions.
For planar, the automatic suggestion starts with an ER51/10 core which, while lower in height than PQ32/35 (about 20 mm versus 35 mm), results in a larger width and depth than desired for compact layouts. The designer then moves into Frenetic Simulator and systematically explores alternative planar shapes like ER41, ER32 and EQ38, trading off core loss, peak flux density and volume. Some candidates, such as ER32, are rejected for operating too close to loss limits in the chosen temperature range, while EQ38 emerges as a better balance between increased volume and acceptable loss.
To enforce a “true planar” low‑profile transformer, the EQ38 core is combined with a PLT style to reduce total height to approximately 10.7 mm by adjusting E‑core dimensions and stacking. Core losses are re‑evaluated at higher temperatures, leveraging the typical reduction of ferrite loss between about 60 °C and 100 °C to confirm acceptable thermal behavior.
Planar design parameterization
Once the core is chosen, the planar design tool requires a set of core window parameters: window height, window width and PCB window utilization. The tool allows defining the PCB window as, for example, 80–90% of the core width, shifting the PCB within the window, and leaving mechanical margins to accommodate core dimensional tolerances and PCB manufacturing tolerances. Ensuring sufficient margins is emphasized as crucial to avoid assembly issues in long‑term production.
The winding definition is highly parametric:
- Number of turns per winding (e.g., 20 turns on the primary, 2 turns on the secondary in the example).
- Copper thickness and trace geometry to manage current density and achievable clearances.
- Distance between turns and distance from copper to PCB edges, constrained by manufacturer rules and insulation requirements.
The design tool presents current density as an early sanity check prior to full finite‑element simulation. For instance, a 5 A current in a 0.1 mm² cross‑section would produce about 48 A/mm², clearly too high for acceptable losses, while guideline target values are typically on the order of 5–7 A/mm² for many designs, depending on cooling and manufacturer recommendations.
In the PSFB example, the primary RMS current is around 2.7 A, and the secondary carries about 25.2 A, which is entered into the tool to evaluate current densities for each layer. Copper thickness and turns‑per‑layer are adjusted to keep primary current density around 11 A/mm², which is acceptable for PCB windings in this use case, and the secondary is implemented with parallel conductors where needed to reduce current density.
Interleaving and layer ordering
Interleaving is a primary lever for controlling leakage inductance and proximity losses. The design tool supports arbitrary layer order customization and recommends maintaining an even number of layers per winding to simplify PCB layout: entering on one layer, moving through a via to another layer, and exiting on an outer bar becomes more straightforward with even layer counts. While not strictly mandatory, this guideline significantly eases route planning in CAD and reduces the need for additional connection layers.
The layer order directly influences leakage inductance and AC losses. Designers can first define a basic stack‑up, then use the tool’s leakage inductance estimate to iterate interleaving patterns before investing in full 2D finite‑element simulations. For example, in PSFB applications, a certain amount of leakage can be beneficial to achieve zero‑voltage switching on the primary, while in flyback designs, leakage is often minimized.
Waveform‑aware loss modeling
The tool imports current waveforms from Frenetic Simulator (CSV) or other circuit simulators like LTspice, supporting up to five harmonics in the finite‑element analysis. A practical recommendation is to simulate up to the third harmonic for a good balance between accuracy and run time for typical converter waveforms. For waveforms with substantial higher‑frequency components, the designer can include more harmonics or even the full spectrum; for sinusoidal currents, a single harmonic is usually sufficient.
The simulation engine uses 2D finite‑element analysis powered by COMSOL Multiphysics, then translates 2D loss results into 3D predictions using an internal scaling model. The output separates DC and AC losses, where pure transformers with ideal AC excitation show zero DC loss and all conduction, proximity and skin effect contributions appear under AC losses. Inductors can be treated similarly by configuring a single winding and inspecting the resulting DC and AC components.
Leakage inductance and parasitic capacitances can be computed independently of the full loss analysis. Leakage is derived via analytical models consistent with those used in Frenetic Simulator, yielding fast results suitable for rapid iteration on layer ordering. The open‑circuit primary capacitance, combined with the magnetizing inductance calculated from turns count and the core’s AL value, enables estimation of the transformer self‑resonant frequency before building any hardware.
Final comparison: PQ32/35 vs EQ38/8/25 planar
In the final summary, the conventional PQ32/35 solution is compared with the optimized EQ38/8/25 planar design. The PQ32/35 implementation is larger in overall volume and weight, has higher total losses and higher leakage inductance than the planar solution. The planar EQ38/8/25 transformer achieves:
- Lower profile: roughly a third of the height (around 10.7 mm) compared to 35 mm for PQ32/35.
- Reduced overall weight, approximately three times lighter.
- Lower total losses and reduced leakage inductance, improving efficiency and potentially easing snubber or clamp design.
The trade‑off is that the planar core has larger width and depth than PQ32/35, so board footprint grows even while height and weight are reduced. The choice between these solutions therefore depends not only on electrical performance but also on mechanical envelope, cost targets and schedule.
Summary table of main design trade‑offs
| Aspect | Conventional PQ32/35 | Planar EQ38/8/25 |
|---|---|---|
| Winding technology | Round, litz or foil wire | PCB windings |
| Profile (height) | Approx. 35 mm | Approx. 10.7 mm |
| Width/depth footprint | Smaller than planar | Larger width and depth |
| Weight | Higher | About three times lower |
| Total losses | Around 6.1 W (example design) | Lower than PQ32/35 in optimized design |
| Leakage inductance | Higher and more variable | Lower, better controlled by interleaving |
| Prototype flexibility | Very high (easy to rework) | Low (PCB redesign needed for changes) |
| High‑volume cost | Higher per unit | Lower per unit once PCB is tooled |
Values are as shown in the webinar example; exact figures should be confirmed against the manufacturer datasheet and current versions of the Frenetic tools.
Design‑in notes for engineers
A recurring theme in the webinar is “let the requirements decide”. Rather than starting from the hype around planar magnetics or sticking rigidly to conventional windings, the suggested approach is to let electrical specifications, mechanical envelope, manufacturing volume and time‑to‑market drive the selection.
Some practical design‑in considerations:
- Prototype schedule: If management expects working hardware within a week, a PQ32/35‑class wound transformer is often the pragmatic choice. Cores and coil formers can be ordered from broadline distributors, and windings can be implemented quickly in the lab using available wire.
- Production volume: If the plan is to build, for example, 1 million units after validation, investing in a planar EQ38/8/25 solution is likely worthwhile, even if the first prototype is more expensive and slower to produce.
- EMC and parasitics: For applications where repeatable leakage inductance and interwinding capacitance are crucial (EMI compliance, resonant behavior, controlled self‑resonant frequency), planar designs with CAD‑controlled interleaving offer clear advantages.
- Thermal management: Evaluate core losses over the expected temperature range using the simulator, noting that ferrite core losses typically decrease between about 60 °C and 100 °C in the example presented. Combine this with copper loss estimates from current density and FEA outputs to ensure junction and winding temperatures remain within limits.
The workflow also highlights the benefit of a unified toolchain: Frenetic AI proposes initial designs, Frenetic Simulator refines core selection and operating points, and the Planar Design Tool translates those parameters into a detailed PCB winding structure with accurate loss and parasitic predictions. For purchasing engineers, this pipeline provides early visibility into core shapes, PCB complexity and likely manufacturing constraints, helping align cost and supply chain decisions with the design team.
Conclusion
This webinar illustrates a complete path from converter specifications to a detailed comparison of conventional and planar transformers using a concrete 700 W PSFB example. Planar magnetics on an EQ38/8/25 core can deliver significantly reduced profile, lower weight and improved control over losses and leakage compared to a traditional PQ32/35 design, at the expense of a larger footprint and more complex upfront layout work.
By combining Frenetic AI, Frenetic Simulator and the Planar Design Tool, engineers can move from converter specifications to a detailed, data‑driven comparison of conventional and planar transformers for a 700 W PSFB stage. This structured workflow helps de‑risk magnetics design, shorten iteration cycles and base the final choice on predicted losses, parasitics and manufacturing implications instead of intuition alone.
For design engineers, the key is to treat planar as one more tool in the magnetics toolbox rather than a universal solution: it is especially compelling for high‑density, high‑volume designs with tight parasitic requirements, while conventional wound parts remain very efficient for rapid prototyping and low‑volume production. Tools such as Frenetic AI, Frenetic Simulator and the Planar Design Tool make it feasible to iterate both options quickly, quantify trade‑offs and base decisions on predicted losses, parasitics and manufacturing implications rather than intuition alone. Engineers considering similar designs should review the latest manufacturer datasheets and run parametric sweeps in their preferred simulator to fine‑tune current densities, interleaving and core choices for their own converters.
FAQ: Conventional vs planar transformers
Conventional transformer as shown in the application example (PQ32/35) is taller, heavier and has higher, less predictable leakage inductance and total losses, while the planar (EQ38/8/25) solution achieves a much lower profile, reduced weight, lower losses and better‑controlled parasitics at the cost of a larger board footprint.
Conventional wound transformers such as PQ32/35 make sense during early prototyping and low‑to‑medium volume production when design changes are frequent and time‑to‑first‑hardware is critical. Engineers can quickly rework turns, layer order and terminations without waiting for new PCB tooling, and off‑the‑shelf cores, coil formers and wire are easy to source from distributors.
A planar transformer such as EQ38/8/25 is particularly attractive for high‑density, high‑volume designs that demand a low profile, tightly controlled leakage inductance and capacitance, and highly repeatable parasitics. Once the PCB is tooled, planar magnetics scale well in manufacturing, offer lower weight and losses than the conventional solution, and help with EMI and resonant behavior control.
The workflow starts in Frenetic AI, which proposes initial magnetic designs and core options for the PSFB converter based on electrical specifications and constraints like primary turns. Frenetic Simulator then refines the choice of core shape and operating point, while the Planar Design Tool parameterizes PCB window usage, winding geometry, current density, interleaving and parasitics, and uses 2D FEA (COMSOL‑based) to predict AC and DC losses before hardware is built.
The conventional transformer in the example has higher profile (around 35 mm), larger weight, about 6.1 W total losses in the example design, and higher, more variable leakage inductance.[ The optimized EQ38/8/25 planar transformer reduces height to about 10.7 mm, is roughly three times lighter, lowers total losses and leakage inductance, but occupies a larger width‑and‑depth footprint on the PCB, so the best choice depends on mechanical envelope, cost targets and schedule.
Planar transformers are very suitable for high‑density AC‑DC and DC‑DC modules that require low profile magnetics, such as 1U power shelves, compact EV chargers, industrial and telecom front‑end supplies and server auxiliary converters. They are also useful where repeatable leakage inductance and capacitance are critical for EMI compliance, resonant topologies or controlled self‑resonant frequency in production.
The Frenetic Planar Design Tool allows the designer to set copper thickness, turns per layer and spacing, then evaluates current density using RMS currents such as about 2.7 A on the primary and 25.2 A on the secondary in the PSFB example. Copper cross‑section and parallel conductors are adjusted to keep current density around 11 A/mm² on the primary and acceptable levels on the secondary, while core losses are re‑checked at elevated temperatures, using the typical decrease of ferrite losses between about 60 °C and 100 °C to confirm safe thermal margins.
The tool supports flexible layer ordering and suggests using an even number of layers per winding to simplify routing while enabling interleaving patterns that tune leakage inductance and proximity losses. Designers can iterate stack‑ups with fast analytical leakage estimates and waveform‑aware FEA that imports harmonics from circuit simulators, so they can deliberately use leakage for zero‑voltage switching in PSFB or minimize it in flyback topologies.
How to decide between conventional and planar transformers in a 700 W PSFB design
- Step 1: Define converter specifications and constraints in Frenetic AI
First, enter the main converter specifications into Frenetic AI, including topology (PSFB), power level (700 W), input range (370–440 VDC), output voltage (28 V), switching frequency (around 100 kHz) and any additional constraints such as limiting the primary to about 10 turns. Frenetic AI uses these inputs to synthesize initial magnetic designs and propose both conventional and planar candidate solutions with suitable core families.
- Step 2: Compare initial conventional and planar core proposals
Review the AI‑generated options, typically a PQ32/35 wound design for the conventional transformer and an ER‑class core such as ER51/10 for planar, along with estimated losses and dimensions. Identify early whether the planar option meets height and footprint goals; in the webinar example, ER51/10 is low in height but too large in width and depth, prompting exploration of alternative shapes like ER41, ER32 or EQ38.
- Step 3: Use Frenetic Simulator to refine planar core selection
In Frenetic Simulator, sweep candidate planar cores (for example ER41, ER32 and EQ38) and check core volume, peak flux density and total losses across the expected operating and temperature range. Reject options that run too close to loss limits, as with ER32 in the example, and converge on a core such as EQ38 that offers an acceptable compromise between volume, loss and manufacturability.
- Step 4: Enforce low profile and finalize the planar core stack‑up
To achieve a “true planar” low profile, adapt the chosen EQ38 core into an EQ38/8/25 or similar PLT style stack‑up that reduces total height to around 10.7 mm while keeping sufficient cross‑section. Recalculate core losses at higher temperatures, leveraging the typical ferrite loss reduction between roughly 60 °C and 100 °C, to confirm that thermal performance remains within design margins.
- Step 5: Parameterize PCB window usage and mechanical clearances
Open the Planar Design Tool and define window height, window width and PCB window utilization, typically using about 80–90% of the core width while leaving margins for tolerances. Adjust PCB offset within the window to reserve mechanical clearance for core dimensional variation and PCB manufacturing tolerances, avoiding assembly issues in long‑term production.
- Step 6: Define winding turns, copper geometry and clearances
Specify turns per winding (for example, 20 turns on the primary and 2 on the secondary), copper thickness, trace width and spacing between turns and edges based on insulation rules and manufacturer capabilities. Enter the expected RMS currents (around 2.7 A primary and 25.2 A secondary in the example) so the tool can compute current densities and help you keep values near guideline targets such as 5–11 A/mm² for PCB traces, depending on cooling and design limits.
- Step 7: Optimize interleaving for leakage inductance and AC losses
Use the tool’s flexible layer ordering to create interleaved structures with an even number of layers per winding, which simplifies routing and via placement. Iterate different stack‑ups using the built‑in leakage inductance calculator and proximity loss estimates, tuning leakage to support requirements such as zero‑voltage switching in PSFB or minimizing it for other topologies.
- Step 8: Import current waveforms and run waveform‑aware FEA
Export current waveforms from Frenetic Simulator or circuit tools like LTspice as CSV files and import them into the Planar Design Tool, including up to several harmonics for realistic excitation. Run the 2D COMSOL‑based finite‑element analysis to separate DC and AC losses, then extrapolate to 3D using the tool’s model to confirm that temperature rise, skin effect and proximity losses are within acceptable limits.
- Step 9: Evaluate parasitic capacitance and self‑resonant behavior
Calculate open‑circuit primary capacitance and combine it with the magnetizing inductance (from turns and core AL value) to estimate the transformer’s self‑resonant frequency before building hardware. Use this information to check compatibility with EMI, resonant operation and snubber or clamp networks, refining layer spacing and interleaving as needed to hit target parasitic values.
- Step 10: Compare conventional and planar solutions
Summarize the conventional design and the planar covering winding technology, profile, footprint, weight, total losses, leakage inductance, prototype flexibility and high‑volume cost. In the example, the planar transformer delivers about one‑third the height, roughly three times lower weight, reduced losses and tighter leakage control, while the PQ32/35 remains superior for fast prototyping and small to medium volumes where PCB tooling cost and effort are hard to justify.
- Step 11: Make the final choice based on requirements
Use the quantified trade‑offs to decide whether the project prioritizes rapid prototyping and flexibility, favoring a conventional wound transformer, or high‑density, high‑volume manufacturing with strict parasitic control, favoring a planar design. Align the decision with mechanical envelope, cost targets and schedule, treating planar as an additional tool in the magnetics toolbox rather than a universal replacement for conventional transformers.
Source
This article is based on technical information and design demonstrations presented by Frenetic in the webinar “Frenetic Planar Tool: From Conventional to Planar Transformer”, supplemented by the associated tool outputs and recommendations from their engineering team.
