Among the many converter topologies available today, the flyback topology is the most commonly used. Despite its simplicity, this converter design offers significant advantages for many applications. In recent years, many newer and more complex topologies have emerged, but the flyback converter design remains popular.
This switch-mode power converter offers a highly competitive size, cost, and efficiency ratio across a low to medium power range (approximately 2W to 100W). The flyback converter operates based on a coupled inductor, which enables power conversion while isolating the converter's input and output. The coupled inductor also supports multiple outputs, making the flyback converter ideal for a variety of applications.
The first step is the design process.
I. Confirm basic technical parameters;
Friendly reminder: Develop good work habits. No matter how low the power consumption or how simple the technology, always strive to create a detailed technical specification for each product. First, clarify what kind of product you are making. This will make your design thinking clearer and guide you on how to proceed with the next steps.
Technical parameters are divided into two types: basic and detailed.
The basic technical parameters that generally need to be listed are as follows (taking a 60W product as an example):
Minimum input voltage: 85VAC
Maximum input voltage: 265VAC
Output voltage and current: 12V 5A (1% accuracy)
Minimum efficiency: 85%
Operating temperature: -25～+60℃
Detailed technical specifications are quite complex, and the number of parameters required varies depending on the specific situation. Generally, they include: input/output characteristics, protection characteristics, safety regulations, EMC, reliability, application environment, product dimensions, input/output port definitions, product labels, housing labels, product packaging, etc.
Input/output characteristics
Input voltage range, input frequency, power factor, maximum input current, inrush current, output voltage range, output current range, voltage regulation, load regulation, voltage regulation accuracy, peak-to-peak ripple, overall efficiency, standby power consumption, power-on delay time, output voltage rise time, capacitive load, power-on/off overshoot, dynamic response time, dynamic response amplitude, and minimum startup voltage.
Protective features
Input undervoltage protection point, input undervoltage recovery point, input overvoltage protection point, input overvoltage recovery point, output overvoltage protection point, output short circuit protection mode, overtemperature protection point, overtemperature recovery point
Friendly reminder: For non-standard products, if you are unsure which parameters to list, it is recommended to refer to competitor product information or the most influential suppliers in the industry. If neither of these resources is available, try to match the technical specifications of standard products as closely as possible.
II. Design Approach (Developing Design Schemes and Reference Calculations): Identify design challenges and solutions based on the product's technical specifications.
Friendly reminder: Don't be afraid of others surpassing you, and don't hold back too much technical knowledge if you want to minimize the risk of failure. The design should undergo extensive internal discussion from the initial project initiation stage. Regarding the choice of solutions (such as specific power devices, capacitors, chips, etc.), listen to the opinions of those around you; you will definitely benefit greatly over time. Since the early stages of project initiation are generally informal discussions, if you are a beginner, be sure to avoid taking up too much of others' time (avoid asking endless questions).
Which company's switching chip should I choose? How should I configure the EMI circuit? How much input capacitor should I use? What is the switching frequency? How should I choose the MOSFET? Diode? Magnetic core? Output capacitor? Many people don't know where to go from this step. The next step will focus on this analysis.
12V5A, universal input, the standard configuration is 8N60+MBR20100.
It's important to note that this parameter isn't something that can be "calculated," as calculated values often differ significantly from actual conditions, exhibiting considerable "flexibility." Regarding selection, our first consideration should be what our company has in stock and whether it's usable. When designing a product, the transformer parameters (voltage, current, stress, etc.) should be designed to meet the parameters of these components. It's not about designing the transformer first and then searching for semiconductor components; the actual development process differs from textbook descriptions. Therefore, your first consideration should be whether the company currently has suitable materials. Whether it's industrial inductors, semiconductors, or electrolytic capacitors, prioritizing the use of inventory materials will significantly shorten the development cycle and reduce various uncertainties.
The third principle, or rather, the least important principle, for material (parameter) selection is to consult the various application documents, evaluation boards, design manuals, etc., provided by the semiconductor company. TI, ON Semiconductor, Fairchild Semiconductor, PI, STMicroelectronics, and Infineon all have a wealth of technical articles, and they are much more user-friendly now than a few years ago, mostly in Chinese. It would be a shame to miss them. Relying on the internet for material selection is probably the least efficient method.
For 60W-level switching power supplies, we can use the following three input circuits. After modifying a few parameters, the first two structures should be fine for applications below 300W (in cases where surge protection needs to be considered, the input end should be reinforced). A close examination of these three structures reveals some similarities and differences, the biggest difference being in the common-mode inductor configuration.
(Note: The discharge resistor at the input terminal is omitted. The drawings were just drawn, so there may be some minor flaws.)
Note: Regardless of whether it is mandatory or not, and regardless of whether your PCB board's input and output lines are terminal connections or wire connections, please make clear and correct silkscreen markings for L, N, PE, and other ports.
Let's start with the input circuit ① and work our way up from beginning to end.
I rarely perform precise calculations when designing; in fact, input filter circuits are also difficult to calculate precisely. Some design approaches that may seem unscientific (or unpopular) are often very useful, which is one of the reasons I'm posting this. Of course, please point out any obvious errors.
F1: The lifespan of a fuse is affected by both input surge voltage and surge current. Slow-recovery fuses should be used whenever possible, generally selected at two to three times the maximum input current. The impact of surge voltage may be more severe with AC input. With battery input (low voltage), if input suppression is insufficient, the impact of surge current on the fuse may be more severe. In industrial applications with AC input, surge voltage is far more severe than in civilian applications. In these cases, the design of surge protection devices (parameters and structural configuration) has a particularly significant impact on the fuse, and double (triple) fuses may be necessary. For relevant design processes, refer to design literature specifically for surge protection circuits and surge current suppression circuits. A single fuse should be connected to the L line, and the glass tube lead package should ideally have an additional heat-shrink tubing layer, with the capacity marked on the PCB board.
RT1: The main function of the thermistor is to suppress input inrush current. An excessively large RT1 will cause severe overheating. An excessively small RT1 may affect the lifespan of the fuse and input electrolytic capacitor. Input inrush current is generally a hard requirement; when selecting RT1, the maximum inrush current limit must be carefully verified. If this requirement is not specified, refer to other products of the same power rating. Under fully sealed conditions, RT may overheat significantly. Additionally, if the product requires low-temperature start-up testing, the RT resistance will become considerably large, potentially preventing the product from starting normally.
X capacitors: For a 60W product, using a 0.47uF X capacitor is relatively safe. In other words, a 30W product should use a 0.22uF FX capacitor, and a 120W product should use a 1uF X capacitor. Although this method has no scientific basis, it has been tried and tested many times. If you enjoy more challenging work, then that's another story. Unlike Y capacitors, a larger X capacitor capacitance will not make other parts worse. When cost is not a major factor, it's good to be lenient and leave yourself an extra safety net. Additionally, in Figure ①, most people do not agree on the role of C4, and there is considerable controversy here.
Y capacitors: Y capacitors come in various configurations, some with two, some with four; some are 102, some are 222, some are 472; some are connected in series with ferrite beads, some with series with resistors. As long as EMI is passed and leakage current is within limits, they're considered excellent! In short, they come in all sorts of configurations, reflecting a deep-seated fear of Y capacitors. Actually, Y capacitors themselves are not inherently bad; their performance is quite good. The culprits lie in the magnetic materials (common-mode inductors, transformers) and the grounding method, which will be analyzed later.
MOV1: There is no unified standard for calculating the voltage of varistors. Incorrect estimation of actual conditions (low breakdown voltage) can seriously damage the product. In civilian products with low surge protection requirements, the 14K471 is commonly used, while industrial applications typically require voltages above 500V, such as the 14K511 and 14K561. If you are unsure about the actual power environment of your product (non-residential), avoid using varistors below 500V. Different industries employ different surge protection measures, which are rarely discussed in forums; careful research is essential, especially regarding the configuration with multiple fuses. Furthermore, after configuring surge protectors, false tripping often occurs during withstand voltage testing, which is a major headache. MOV1 requires the addition of heat-shrink tubing.
DB1: Low-power products, selection is relatively simple. From a heat dissipation perspective, for a wide range of 60W products, the minimum rectifier specification should not be lower than 2A. Under conditions where cost is not a major constraint, 4A is generally sufficient.
In certain special applications, such as those with transient high surge voltages, the rectifier specifications should be further increased. There is a rare (but not uncommon) situation where some engineers choose excessively large capacitance input electrolytic capacitors (perhaps for small quantities or for their own use), while using excessively small surge suppression (thermostat) resistors. In this case, the powerful inrush current poses a fatal threat to the fuse and rectifier. Professional power supply manufacturers would not engage in this practice, while non-professional manufacturers, due to insufficient development experience and a lack of rigorous testing standards, often overlook these potential hazards when developing system-related products.
Common-mode inductors: Three configurations are given above. Option ① is the most common. We often see a small Φ8-Φ16 magnetic ring (30-1000uH) in the preamp stage and a large Φ20-Φ25 magnetic ring (15-30mH) in the power amp stage. The preamp operates at high frequencies, and the power amp at low frequencies, a perfect balance. Option ② is also quite common, with two identical common-mode coils, which is very aesthetically pleasing. With this configuration, to ensure good filtering (reducing distributed capacitance), the inductance (number of turns) of each stage cannot be too high. This not only reduces the distributed capacitance of the common-mode inductor but also simplifies the winding process and improves aesthetics, although it increases the cost. Option ③ is generally used in products with lower EMI requirements, with low-cost EE-type common-mode inductors being the most common. In some products with stringent cost requirements, many also use a single Φ18-Φ25 magnetic ring in the design, which requires developers to have sufficient experience and skills. The material, shape, and winding process of the common-mode inductor have a significant impact on the filtering effect, and the configuration of EMI filtering components is also closely related to the overall system structure. Many people don't know how to calculate the common-mode inductance value; below is a reference method (applicable to small and medium power applications).
100kHz ------ 30MHz
1.0MHz ------ 3.0MHz
10MHz ------- 300uH
100MHz ------ 30uH
5.0MHz ------ 600uH
30MHz-------100uH
During conduction testing, problems are likely to occur at four points: 3*F, 1MHz, 5MHz, and 20-30MHz.
Note: 1. This method only has regularity, but no scientific basis;
2. The material, shape, and winding process of a common-mode inductor have a significant impact on its filtering effect;
3. Common-mode inductors will not saturate (due to symmetrical winding), but they will generate higher surge voltages;
4. For common-mode magnetic rings, it is best to only wind two layers. More effort should be put into the manufacturing process of the magnetic rings.
5. The design principle of common-mode filtering is to make it more effective.
(This part is prone to ambiguity and will be supplemented later when time permits. If more practical materials are uploaded gradually, it should be easier to understand.)
The secret between Cin, Vacmin, and Vdcmin: 85-265VAC input, 12V 5A output.
①Real-world situation: Choosing a 100uF/400V electrolytic capacitor is unlikely to cause much controversy.
②3uF/W rule: 3uF*60W = 180uF. Considering efficiency factors, choose 220uF.
Assuming an ambient temperature of 25℃, a 60W output, and 85% efficiency, the calculated Vdcmin value is as follows:
(Vdcmin is affected by many factors. The data below is calculated using PI's spreadsheet and is for reference only.)
Input voltage capacity minimum DC voltage ripple voltage percentage
85VAC 100uF 68VDC 52V 43.3%.
85VAC 150uF 89VDC 31V 25.8%.
85VAC 220uF 100VDC 20V 16.6%.
Input voltage capacity minimum DC voltage ripple voltage percentage
90VAC 100uF 79VDC 48V 37.8%.
90VAC 150uF 98VDC 29V 22.8%.
90VAC 220uF 108VDC 19V 15.0%.
Input voltage capacity minimum DC voltage ripple voltage percentage
100VAC 100uF 101VDC 40V 28.4%.
100VAC 150uF 116VDC 25V 17.7%.
100VAC 220uF 125VDC 16V 11.3%.
Input voltage capacity minimum DC voltage ripple voltage percentage
175VAC 68uF 216VDC 31V 12.5%.
175VAC 100uF 227VDC 20V 8%.
Classic and authoritative textbooks invariably state that Vdcmin = Vacmin * 1.414, but this is not actually the case. So where does the problem lie?
It is safe to say that these textbooks are less likely to contain errors in the Vdcmin calculation problem.
Many people, when designing products, use Vac*1.414 without thinking, never considering the size of the Cin capacity.
Vdcmin = Vacmin * 1.414
The prerequisite for its validity is that a reasonable ripple voltage percentage must be defined.
(Ripple voltage percentage = Vdcmax - Vdcmin / Vdcmax; Vdcmax = Vacmin * 1.414)
In other words, Cin must satisfy Vdcmin, otherwise the formula is invalid. This is also the reason why Cin is selected as 3uF/W for wide-range input and 1uF/W for narrow-range input;
On a side note, many 12V 5A adapters use 100uF electrolytic capacitors, but their input voltage range is 100-265VAC. Is this the reason?
Cin selection rules:
1. Wide-range input: 3uF/W; Narrow-range input: 1uF/W;
2. Wide input range to ensure ripple voltage does not exceed 15% (i.e., ensure Vdcmin≈100V);
Narrow input range ensures ripple voltage does not exceed 20% (i.e., guarantees Vdcmin≈200V);
3. If Vdcmin is insufficient, increase the capacitance of Cin until the ripple voltage meets the requirements;
4. If lifespan is taken into account, Cin needs to be increased further.
5. The capacity of Cin is significantly affected by low temperature; therefore, Cin needs to be further increased in this case.
