1. Introduction
The design of high-frequency magnetic components in switching power supplies is crucial for the normal operation of the circuit and the achievement of various performance indicators. Furthermore, the design of high-frequency magnetic components involves many detailed aspects, which are difficult to fully list in one or a few so-called "design encyclopedias" [1-3]. To optimize the design of high-frequency magnetic components, multiple design variables must be comprehensively considered according to the application scenario, and repeated calculations and adjustments must be made. For this reason, the design of high-frequency magnetic components has always been a headache for designers new to the power supply field, and even a problem that perplexes power supply engineers with many years of experience.
Many documents and related technical materials often overlook the influence of certain design variables in their magnetic component design methods or formulas, deriving a set of formulas after making simplistic assumptions; or they fail to clearly explain the application conditions of the formulas, and some documents even convey incorrect information. Many power supply designers are unaware of this, directly applying formulas from design manuals, or taking certain statements from design manuals out of context and treating them as "design guidelines," without thorough analysis, reflection, or experimental verification. The result is often that the designed high-frequency magnetic components fail to meet the requirements of the application, affecting the progress of research and development and the timely completion of projects.
To help power supply designers avoid making the same mistakes during the design process, we have summarized some conceptual issues encountered in our learning and research and development, hoping to provide a reference for everyone.
2. Clarification of some misconceptions
This section presents eight common misconceptions in the design of high-frequency magnetic components for switching power supplies, with detailed explanations provided.
1) Fill the core window – an optimized design
Many power supply designers believe that filling the core window in high-frequency magnetic component design yields optimal results, but this is not the case. In numerous high-frequency transformer and inductor designs, we've found that adding one or more winding layers or using larger diameter enameled wire not only fails to optimize performance but also increases overall winding losses due to proximity effects. Therefore, in high-frequency magnetic component design, it's acceptable if the winding doesn't completely fill the core window, even if it only covers 25% of the area. It's not necessary to strive to fill the entire window area.
This misconception is primarily influenced by the design of power frequency magnetic components. In power frequency transformer design, the integrity of the core and windings is emphasized, thus avoiding gaps between them. Generally, the windings are designed to fill the entire opening to ensure mechanical stability. However, high-frequency magnetic component design does not have this requirement.
2) "Iron loss = Copper loss" - Optimized transformer design
Many power supply designers, and even in many magnetic component design reference books, list "iron loss = copper loss" as one of the standards for optimizing high-frequency transformer design. However, this is not the case. In the design of high-frequency transformers, iron loss and copper loss can differ significantly, sometimes by an order of magnitude. This does not mean that the high-frequency transformer design is poor.
This misconception is also influenced by the design of power frequency transformers. Power frequency transformers often have a large number of winding turns and occupy a large area, leading to the empirical design rule of "iron loss = copper loss" from the perspective of thermal stability and thermal uniformity. However, for high-frequency transformers, which use very fine enameled wire as windings, this empirical rule does not hold true. In the design of high-frequency transformers for switching power supplies, many factors determine the optimal design, and "iron loss = copper loss" is actually one of the least considered aspects.
3) Magnetized inductance with leakage inductance of 1%
Many power supply designers, after designing the magnetic components, often specify the required leakage inductance when submitting the relevant technical requirements to transformer manufacturers. Many technical sheets state requirements such as "leakage inductance = 1% of magnetizing inductance" or "leakage inductance < 2% of magnetizing inductance." This kind of wording or design standard is unprofessional. Power supply designers should set a numerical limit on the acceptable leakage inductance value based on the normal operating requirements of the circuit. During transformer manufacturing, the leakage inductance value should be reduced as much as possible without deteriorating other transformer parameters (such as inter-turn capacitance), rather than specifying the ratio of leakage inductance to magnetizing inductance as a technical requirement. This is because the relationship between leakage inductance and magnetizing inductance varies greatly depending on whether the transformer has an air gap. Without an air gap, the leakage inductance may be less than 0.1% of the magnetizing inductance, while with an air gap, even if the transformer windings are tightly coupled, the ratio of leakage inductance to magnetizing inductance can reach 10%.
Therefore, do not provide the ratio of leakage inductance to magnetizing inductance as a transformer design specification to magnetic component manufacturers. Doing so would demonstrate a lack of understanding of leakage inductance or a lack of genuine concern about actual leakage inductance values. The correct approach is to clearly define an acceptable absolute value for leakage inductance, with a certain percentage added or subtracted, typically around 20%.
4) Leakage inductance is related to the permeability of the magnetic core.
Some power supply designers believe that adding a magnetic core to the windings will make the windings more tightly coupled, which can reduce the leakage inductance between the windings; while others believe that after adding a magnetic core to the windings, the magnetic core will couple with the field between the windings, which can increase the leakage inductance.
In fact, in switching power supply design, the leakage inductance of a two-coaxial-winding transformer is unrelated to the presence or absence of a magnetic core. This result may be difficult to understand because a material with a relative permeability of several thousand has minimal impact on leakage inductance when placed near the coil. Actual measurements of hundreds of transformers show that the presence or absence of a magnetic core generally does not change the leakage inductance by more than 10%, with many changes only around 2%.
5) The optimized value for transformer winding current density is 2A/mm²～3.1A/mm²
Many power supply designers often consider the current density in the windings as the standard for optimizing high-frequency magnetic components. However, optimal design is not actually related to the current density in the windings. What truly matters is the amount of loss in the windings and whether the heat dissipation measures are sufficient to ensure that the temperature rise remains within acceptable limits.
We can consider two extreme cases of heat dissipation measures in switching power supplies. When liquid immersion and vacuum are used for heat dissipation, the corresponding current densities in the windings will differ significantly.
In the actual development of switching power supplies, we are not concerned with the current density, but only with how hot the coil gets and whether the temperature rise is acceptable.
This misconception stems from a limitation imposed by designers to avoid tedious trial and error, simplifying the number of variables and thus the calculation process. However, this simplification does not specify the conditions for application.
6) Primary winding loss = Secondary winding loss — Optimized transformer design
Many power supply designers believe that an optimized transformer design corresponds to equal primary winding losses and secondary winding losses. This is even used as a standard for optimization in many magnetic component design manuals. However, this is not a true standard for optimization. In some cases, the iron and copper losses of a transformer may be similar. But a significant difference between the primary and secondary winding losses is not a major concern. It must be emphasized again that for high-frequency magnetic component design, our concern is how hot the windings get under the chosen cooling method. The rule that primary winding losses equal secondary winding losses is merely an empirical rule for power frequency transformer design.
7) When the winding diameter is smaller than the penetration depth, high-frequency losses will be very small.
Having a winding diameter smaller than the penetration depth does not necessarily mean there is no significant high-frequency loss. If a transformer winding has many layers, even if the winding uses enameled wire with a diameter much thinner than the penetration depth, there may still be significant high-frequency losses due to the strong proximity effect. Therefore, when considering winding losses, one cannot judge the magnitude of the loss solely based on the thickness of the enameled wire; the entire winding structure must be considered, including the winding method, the number of winding layers, and the wire thickness.
8) In a forward converter circuit, the open-circuit resonant frequency of the transformer must be much higher than the switching frequency.
Many power supply designers believe that the open-circuit resonant frequency of a transformer must be much higher than the switching frequency of the converter when designing and testing transformers. This is not the case. The open-circuit resonant frequency of a transformer is unrelated to the switching frequency. We can consider an extreme case: for an ideal magnetic core, its inductance is infinite, but it will still have a relatively small inter-turn capacitance, and its resonant frequency will be approximately zero, much lower than the switching frequency.
What truly relates to the circuit is the transformer's short-circuit resonant frequency. Generally, the transformer's short-circuit resonant frequency should be at least two orders of magnitude higher than the switching frequency.
3. Conclusion
To help power supply designers avoid making the same mistakes during the power supply design process, we have summarized some conceptual issues related to the design of high-frequency magnetic components that we encountered in the research and development of switching power supplies. We hope this will serve as a starting point for further discussion.
Disclaimer: This article is a reprint. If there are any copyright issues, please contact us promptly for deletion (QQ: 2737591964). We apologize for any inconvenience.
