How to design a switching power supply?

The editor recently discovered that some netizens mentioned the parameter settings and working status analysis of the feedback loop of the switching power supply. Because I was relatively poor in advanced mathematics when I was in school, I almost took the make-up exam for "Principles of Automatic Control". I am still afraid of this subject. I still cannot fully write the transfer function of a closed-loop system, and I have a feeling for the concepts of zeros and poles of the system. It's very vague. Looking at the Bode plot, you can only roughly tell whether it is diverging or converging, so I don't dare to talk nonsense about feedback compensation, but I have some suggestions. If you have some mathematical skills and some study time, you can find the university textbook "Principles of Automatic Control" and digest it carefully, and analyze it according to the working status based on the actual switching power supply circuit. You will definitely gain something. There is a post on the forum "Apprenticeship to Learn Feedback Loop Design and Adjustment" where CMG gave a good answer. I think it can be used as a reference.

Let’s start with the description of the design and production process of the switching power supply. First, let’s talk about the design of the printed board. Switching power supply works in high frequency and high pulse state, which is a special type of analog circuit. High frequency circuit wiring principles must be followed when laying out the board.

1. Layout: The pulse voltage connection should be as short as possible, including the connection between the input switch tube and the transformer, and the connection between the output transformer and the rectifier tube. The pulse current loop should be as small as possible, such as the positive input filter capacitance to the negative return capacitance of the transformer to the switching tube. From the output end of the transformer to the rectifier to the output inductor to the output capacitor and back to the transformer circuit, the X capacitor in the circuit should be as close as possible to the input end of the switching power supply. The input line should avoid being parallel to other circuits and should be avoided. The Y capacitor should be placed at the chassis ground terminal or FG connection. Keep a certain distance between the common inductor and the transformer to avoid magnetic coupling. If it is difficult to handle, you can add a shield between the common inductor and the transformer. The above items have a greater impact on the EMC performance of the switching power supply.

Generally, two output capacitors can be used, one close to the rectifier and the other close to the output terminal, which can affect the power output ripple index. The effect of two small-capacity capacitors in parallel should be better than using one large-capacity capacitor. Heating devices should be kept at a certain distance from electrolytic capacitors to extend the life of the whole machine. Electrolytic capacitors are the bottleneck of the life of switching power supplies. For example, transformers, power tubes, and high-power resistors should be kept at a distance from electrolytic capacitors, and space for heat dissipation must also be left between electrolytic capacitors. , it can be placed at the air inlet if conditions permit.

Attention should be paid to the control part: high-impedance weak signal circuit connections should be as short as possible, such as sampling feedback loops. During processing, try to avoid interference. Current sampling signal circuits, especially current control circuits, may be prone to unexpected events if not handled properly. Unexpectedly, there are some tricks. Now take the 3843 circuit as an example as shown in Figure (1). The effect of Figure 1 is better than Figure 2. In Figure 2, when using an oscilloscope to observe the current waveform at full load, there are obvious spikes superimposed on it. Due to the interference, the current limit point is larger than the design value. On the low side, there is no such phenomenon in Figure 1. There is also a switch tube drive signal circuit. The switch tube drive resistor should be close to the switch tube, which can improve the working reliability of the switch tube. This is related to the high DC impedance voltage drive characteristics of the power MOSFET.

Let’s talk about some principles of printed board wiring.

Line spacing: With the continuous improvement and improvement of printed circuit board manufacturing technology, there is no problem in general processing plants producing line spacing equal to or even less than 0.1mm, which can fully meet most applications. Considering the components and production process used in the switching power supply, the minimum line spacing of double-sided panels is generally set to 0.3mm, and the minimum line spacing of single-sided panels is set to 0.5mm. Pads and pads, pads and vias, or vias and vias holes, the minimum spacing is set to 0.5mm to avoid "bridging" during the welding operation. , so that most board manufacturers can easily meet production requirements, control the yield to a very high level, and achieve reasonable wiring density and a more economical cost.

The minimum line spacing is only suitable for signal control circuits and low-voltage circuits with voltages lower than 63V. When the voltage between lines is greater than this value, the line spacing can generally be determined based on the empirical value of 500V/1mm.

Since some relevant standards have clear regulations on line spacing, the standards must be strictly followed, such as the connection from the AC inlet end to the fuse end. Some power supplies have very high volume requirements, such as module power supplies. Generally, the spacing between the input side lines of a transformer is 1mm, which has been proven feasible in practice. For power supply products with AC input and (isolated) DC output, the strict requirement is that the safety distance must be greater than or equal to 6mm. Of course, this is determined by relevant standards and implementation methods. Generally, the safe distance can be used as a reference by the distance on both sides of the feedback optocoupler, and the principle is greater than or equal to this distance. Slots can also be made on the printed board below the optocoupler to increase the creepage distance to meet the insulation requirements. Generally, the distance between the AC input side traces or components on the switching power supply and the non-insulated shell or radiator should be greater than 5mm, and the distance between the output side traces or components and the shell or radiator should be greater than 2mm, or strictly follow safety regulations.

Commonly used methods: The circuit board slotting method mentioned above is suitable for some occasions where the spacing is insufficient. By the way, this method is also commonly used as a protective discharge gap. It is commonly used in TV picture tube tail plates and power supply AC inputs. . This method has been widely used in module power supplies and can achieve good results under potting conditions.

Method 2: Use insulating paper, such as green shell paper, polyester film, polytetrafluoroethylene directional film and other insulating materials. Generally, green paper or polyester film is used for general-purpose power supplies between the circuit board and the metal casing. This material has high mechanical strength and a certain ability to resist moisture. PTFE oriented film is widely used in module power supplies due to its high temperature resistance. An insulating film can also be placed between the component and surrounding conductors to improve the electrical resistance of the insulation.

Note: The insulation coating of some devices cannot be used as an insulating medium to reduce the safe distance, such as the outer skin of an electrolytic capacitor. Under high temperature conditions, the outer skin may shrink due to heat. Space should be left at the front of the large electrolytic explosion-proof tank to ensure that the electrolytic capacitor can relieve pressure without hindrance in emergency situations.

Let’s talk about some matters about printed circuit board copper wiring:

Wiring current density: Most electronic circuits are now made of insulating boards bound with copper. The commonly used circuit board copper thickness is 35μm, and the current density value for wiring can be determined based on the empirical value of 1A/mm. For specific calculations, please refer to the textbook. In order to ensure the mechanical strength of the wiring, the line width should be greater than or equal to 0.3mm (other non-power circuit boards may have a smaller minimum line width). The copper thickness is 70μm. Circuit boards are also common in switching power supplies, so the current density can be higher.

In addition, commonly used circuit board design tool software generally has design specification items, such as line width, line spacing, dry plate via size and other parameters that can be set. When designing circuit boards, the design software can automatically execute according to specifications, which can save a lot of time, reduce part of the workload, and reduce the error rate.

Generally, double-sided panels can be used for lines with high reliability requirements or high wiring density. It is characterized by moderate cost, high reliability, and can meet most applications.

Some products in the modular power supply industry also use multi-layer boards, which is mainly to facilitate the integration of power devices such as transformers and inductors, and to optimize wiring and power tube heat dissipation. It has the advantages of beautiful and consistent process and good heat dissipation of the transformer, but its disadvantages are higher cost and poor flexibility, and it is only suitable for industrial mass production.

Single-sided circuit boards are almost all common switching power supplies circulating in the market. They have the advantage of low cost, and some measures can be taken in the design and production process to ensure their performance.

Let’s talk about some experiences in the design of single-sided printed boards. Since single-sided boards are low-cost and easy to manufacture, they are widely used in switching power supply circuits. Since they have only one side with copper, the electrical connection and mechanical fixation of the device must rely on That layer of copper must be handled with care.

In order to ensure good welding mechanical structural performance, the single-sided pad should be slightly larger to ensure good bonding force between the copper and the substrate, so as to prevent the copper from peeling off or breaking off when subjected to vibration. Generally, the width of the welding ring should be greater than 0.3mm. The diameter of the pad hole should be slightly larger than the diameter of the device pin, but not too large to ensure that the solder connection distance between the pin and the pad is the shortest. The size of the pad hole should not hinder normal inspection. The diameter of the pad hole is generally larger than the pin. Diameter 0.1-0.2mm. Multi-pin devices can also be larger to ensure smooth inspection.

The electrical connection should be as wide as possible. In principle, the width should be greater than the diameter of the pad. In special cases, the line must be widened when the connection intersects with the pad (commonly known as teardrops) to avoid breakage of the line and the pad under certain conditions. In principle, the minimum line width should be greater than 0.5mm.

单面板上元器件应紧贴线路板。需要架空散热的器件，要在器件与线路板之间的管脚上加套管，可起到支撑器件和增加绝缘的双重作用，要最大限度减少或避免外力冲击对焊盘与管脚连接处造成的影响，增强焊接的牢固性。线路板上重量较大的部件可增加支撑连接点，可加强与线路板间连接强度，如变压器，功率器件散热器。

单面板焊接面引脚在不影响与外壳间距的前题条件下，可留得长一些，其优点是可增 加焊接部位的强度，加大焊接面积、有虚焊现象可即时发现。引脚长剪腿时，焊接部位受力较小。在台湾、日本常采用把器件引脚在焊接面弯成与线路板成45度 角，然后再焊接的工艺，的其道理同上。今天谈一谈双面板设计中的一些事项，在一 些要求比较高，或走线密度比较大的应用环境中采用双面印制板，其性能及各方面指标要比单面板好很多。

双面板焊盘由于孔已作金属化处理强度较高，焊环可比单面板小一些，焊盘孔孔径可 比管脚直径略微大一些，因为在焊接过程中有利于焊锡溶液通过焊孔渗透到顶层焊盘，以增加焊接可靠性。但是有一个弊端，如果孔过大，波峰焊时在射流锡冲击下部分器件可能上浮，产生一些缺陷。

大电流走线的处理，线宽可按照前帖处理，如宽度不够，一般可采用在走线上镀锡增加厚度进行解决，其方法有好多种

1， 将走线设置成焊盘属性，这样在线路板制造时该走线不会被阻焊剂覆盖，热风整平时会被镀上锡。

2， 在布线处放置焊盘，将该焊盘设置成需要走线的形状，要注意把焊盘孔设置为零。

3， 在阻焊层放置线，此方法最灵活，但不是所有线路板生产商都会明白你的意图，需用文字说明。在阻焊层放置线的部位会不涂阻焊剂。

线路镀锡的几种方法如上，要注意的是，如果很宽的的走线全部镀上锡，在焊接以后，会粘接大量焊锡，并且分布很不均匀，影响美观。一般可采用细长条镀锡宽度在1~1.5mm，长度可根据线路来确定，镀锡部分间隔0.5~1mm双面线路板为布局、走线提供了很大的选择性，可使布线更趋于合理。关于接地，功率地与信号地一定要分开，两个地可在滤波电容处汇合，以避免大脉冲电流通过信号地连线而导致出现不稳定的意外因素，信号控制回路尽量采用一点接地法，有一个技巧，尽量把非接地的走线放置在同一布线层，最后在另外一层铺地线。输出 线一般先经过滤波电容处，再到负载，输入线也必须先通过电容，再到变压器，理论依据是让纹波电流都通过旅滤波电容。

电压反馈取样，为避免大电流通过走线的影响，反馈电压的取样点一定要放在电源输出最末梢，以提高整机负载效应指标。

走线从一个布线层变到另外一个布线层一般用过孔连通，不宜通过器件管脚焊盘实现，因为在插装器件时有可能破坏这种连接关系，还有在每1A电流通过时，至少应有2个过孔，过孔孔径原则要大于0.5mm，一般0.8mm可确保加工可靠性。

器件散热，在一些小功率电源中，线路板走线也可兼散热功能，其特点是走线尽量宽大，以增加散热面积，并不涂阻焊剂，有条件可均匀放置过孔，增强导热性能。

谈谈铝基板在开关电源中的应用和多层印制板在开关电源电路中的应用。

铝基板由其本身构造，具有以下特点：导热性能非常优良、单面缚铜、器件只能放置在缚铜面、不能开电器连线孔所以不能按照单面板那样放置跳线。

铝基板上一般都放置贴片器件，开关管，输出整流管通过基板把热量传导出去，热阻很低，可取得较高可靠性。变压器采用平面贴片结构，也可通过基板散热，其温升比常规要低，同样规格变压器采用铝基板结构可得到较大的输出功率。铝基板跳线可以采用搭桥的方式处理。铝基板电源一般由由两块印制板组成，另外一块板放 置控制电路，两块板之间通过物理连接合成一体。

由于铝基板优良的导热性，在小量手工焊接时比较困难，焊料冷却过快，容易出现问题现有一个简单实用的方法，将一个烫衣服的普通电熨斗(最好有调温功能)，翻过来，熨烫面向上，固定好，温度调到150℃左右，把铝基板放在熨斗上面，加温一段时间，然后按照常规方法将元件贴上并焊接，熨斗温度以器件易于焊接为宜，太高有可能时器件损坏，甚至铝基板铜皮剥离，温度太低焊接效果不好，要灵活掌握.

最近几年，随着多层线路板在开关电源电路中应用，使得印制线路变压器成为可能，由于多层板，层间距较小，也可以充分利用变压器窗口截面，可在主线路板上再加一到两片由多层板组成的印制线圈达到利用窗口，降低线路电流密度的目的，由于采用印制线圈，减少了人工干预，变压器一致性好，平面结构，漏感低，偶合 好。开启式磁芯，良好的散热条件。由于其具有诸多的优势，有利于大批量生产，所以得到广泛的应用。但研制开发初期投入较大，不适合小规模生。

开关电源分为，隔离与非隔离两种形式，在这里主要谈一谈隔离式开关电源的拓扑形式，在下文中，非特别说明，均指隔离电源。隔离电源按照结构形式不同，可分为两大类：正激式和反激式。反激式指在变压器原边导通时副边截止，变压器储能。原边截止时，副边导通，能量释放到负载的工作状态，一般常规反激式电源单管 多，双管的不常见。正激式指在变压器原边导通同时副边感应出对应电压输出到负载，能量通过变压器直接传递。按规格又可分为常规正激，包括单管正激，双管正激。半桥、桥式电路都属于正激电路。

正激和反激电路各有其特点，在设计电路的过程中为达到最优性价比，可以灵活运用。一般在小功率场合可选用反激式。稍微大一些可采用单管正激电路，中等功率可采用双管正激电路或半桥电路，低电压时采用推挽电路，与半桥工作状态相同。大功率输出，一般采用桥式电路，低压也可采用推挽电路。

反激式电源因其结构简单，省掉了一个和变压器体积大小差不多的电感，而在中小功率电源中得到广泛的应用。在有些介绍中讲到反激式电源功率只能做到几十瓦，输出功率超过100瓦就没有优势，实现起来有难度。本人认为一般情况下是这样的，但也不能一概而论，PI公司的TOP芯片就可做到300瓦，有文章介绍反激电源可做到上千瓦，但没见过实物。输出功率大小与输出电压高低有关。

反激电源变压器漏感是一个非常关键的参数，由于反激电源需要变压器储存能量，要 使变压器铁芯得到充分利用，一般都要在磁路中开气隙，其目的是改变铁芯磁滞回线的斜率，使变压器能够承受大的脉冲电流冲击，而不至于铁芯进入饱和非线形状态，磁路中气隙处于高磁阻状态，在磁路中产生漏磁远大于完全闭合磁路。

变压器初次极间的偶合，也是确定漏感的关键因素，要尽量使初次极线圈靠近，可采用三明治绕法，但这样会使变压器分布电容增大。选用铁芯尽量用窗口比较长的磁芯，可减小漏感，如用EE、EF、EER、PQ型磁芯效果要比EI型的好。

Regarding the duty cycle of the flyback power supply, in principle, the maximum duty cycle of the flyback power supply should be less than 0.5, otherwise the loop will not be easily compensated and may be unstable. However, there are some exceptions. For example, the TOP series chips launched by the American PI Company are Can work under the condition that the duty cycle is greater than 0.5. The duty cycle is determined by the primary and secondary turns ratio of the transformer. My opinion on doing flyback is to first determine the reflected voltage (the voltage value at which the output voltage is reflected to the primary through transformer coupling). When the reflected voltage increases within a certain voltage range, The working duty cycle increases and the switching tube loss decreases. When the reflected voltage decreases, the operating duty cycle decreases and the switching tube loss increases. Of course, this also has prerequisites. When the duty cycle increases, it means that the output diode conduction time is shortened. In order to maintain output stability, it will be guaranteed by the output capacitor discharge current more often, and the output capacitor will withstand greater high frequency. Ripple current washes away and heats up, which is not allowed under many conditions. Increasing the duty cycle and changing the turns ratio of the transformer will increase the leakage inductance of the transformer and change its overall performance. When the leakage inductance energy reaches a certain level, it can fully offset the low loss caused by the large duty cycle of the switching tube. At this time, there is no point in increasing the duty cycle, and the switching tube may even be broken down because the leakage inductance reverse peak voltage is too high. Due to the large leakage inductance, the output ripple and other electromagnetic indicators may deteriorate. When the duty cycle is small, the effective value of the switching tube current is high, and the effective value of the transformer primary current is large, which reduces the efficiency of the converter, but can improve the working conditions of the output capacitor and reduce heat generation.

How to Determine Transformer Reflected Voltage (i.e. Duty Cycle)

Next, let’s talk about the duty cycle of the flyback power supply (I focus on the reflected voltage, which is consistent with the duty cycle). The duty cycle is also related to the withstand voltage of the selector switch. Some early flyback power supplies use relatively low withstand voltage switch tubes. For example, 600V or 650V is used as a switching tube for AC 220V input power supply. This may be related to the production process at that time. High-voltage tubes are difficult to manufacture, or low-voltage tubes have more reasonable conduction losses and switching characteristics, such as line reflections. The voltage cannot be too high, otherwise in order to make the switching tube work within a safe range, the power absorbed by the circuit loss is also considerable. Practice has proved that the reflected voltage of a 600V tube should not be greater than 100V, and the reflected voltage of a 650V tube should not be greater than 120V. When the leakage inductance peak voltage value is clamped to 50V, the tube still has a working margin of 50V. Nowadays, due to the improvement of MOS tube manufacturing technology, generally flyback power supplies use 700V or 750V or even 800-900V switch tubes. For a circuit like this, the switching transformer's reflection voltage can be made higher due to its stronger ability to withstand overvoltage. The maximum reflection voltage is 150V, which is more appropriate and can achieve better overall performance. PI's TOP chip recommends using transient voltage suppression diode clamps for 135V. However, the reflected voltage of his evaluation board is generally lower than this value, around 110V. Both types have advantages and disadvantages:

Category 1: Disadvantages: weak overvoltage resistance, small duty cycle, and large primary pulse current of the transformer. Advantages: The transformer has small leakage inductance, low electromagnetic radiation, high ripple index, small switching tube loss, and the conversion efficiency is not necessarily lower than the second type.

Category 2: Disadvantages: The switching tube loss is larger, the transformer leakage inductance is larger, and the ripple is worse. Advantages: stronger overvoltage resistance, large duty cycle, lower transformer loss, and higher efficiency.

There is another determining factor for the reflected voltage of the flyback power supply.

The reflected voltage of the flyback power supply is also related to one parameter, that is, the output voltage. The lower the output voltage, the greater the turns ratio of the transformer, the greater the leakage inductance of the transformer, and the higher the voltage the switch tube can withstand, which may cause breakdown of the switch tube and absorb the voltage. The greater the power consumed by the circuit, the more likely it is that the circuit power absorption device will permanently fail (especially circuits using transient voltage suppression diodes). The optimization process of designing a low-voltage output low-power flyback power supply must be handled carefully. There are several methods for this:

1. Use a magnetic core with a larger power level to reduce leakage inductance, which can improve the conversion efficiency of low-voltage flyback power supplies, reduce losses, reduce output ripple, and improve the cross-regulation rate of multi-channel output power supplies. It is generally common in switches for home appliances. Power supply, such as CD player, DVB set-top box, etc.

2. If conditions do not allow increasing the magnetic core, the reflected voltage can only be reduced and the duty cycle reduced. Reducing the reflected voltage can reduce the leakage inductance but may reduce the power conversion efficiency. The two are contradictory. There must be a replacement process to find a suitable point. During the transformer replacement experiment, the primary side of the transformer can be detected. Reverse peak voltage, try to reduce the width and amplitude of the reverse peak voltage pulse, which can increase the operating safety margin of the converter. Generally, the reflected voltage is more suitable when it is 110V.

3. Enhance coupling, reduce losses, and adopt new technologies and winding processes. In order to meet safety regulations, the transformer will take insulation measures between the primary side and the secondary side, such as padding with insulating tape and adding insulating terminal tape. These will affect the leakage inductance performance of the transformer. In actual production, the primary winding can be used to wrap the secondary winding. Or the secondary is wound with triple insulated wire, eliminating the insulation between the primary and secondary, which can enhance the coupling, or even be wound with a wide copper sheet.

In this article, low-voltage output refers to an output less than or equal to 5V. For this type of low-power power supply, my experience is that if the power output is greater than 20W, the forward mode can be used to obtain the best cost performance. Of course, this is not absolutely correct. It depends on personal habits and application environment.

The magnetic core of the flyback power transformer works in a unidirectional magnetization state, so the magnetic circuit needs an air gap, similar to a pulsating DC inductor. Part of the magnetic circuit is coupled through the air gap. The principle of why the air gap is opened is as follows: Since power ferrite also has an approximately rectangular operating characteristic curve (hysteresis loop), the Y-axis on the operating characteristic curve represents the magnetic induction intensity (B). The current production process is generally The saturation point is above 400mT. Generally, this value should be 200-300mT in design. The X-axis represents the magnetic field intensity (H). This value is proportional to the magnetizing current intensity. Opening an air gap in the magnetic circuit is equivalent to tilting the magnet's hysteresis loop toward the X-axis. Under the same magnetic induction intensity, it can withstand a larger magnetizing current, which is equivalent to the magnetic core storing more energy. This energy is cut off at the switching tube. When it is discharged to the load circuit through the secondary side of the transformer, the air gap in the flyback power core has two functions. One is to transfer more energy, and the other is to prevent the core from entering saturation.

The transformer of the flyback power supply works in a unidirectional magnetization state. It not only transfers energy through magnetic coupling, but also plays the multiple roles of voltage conversion input and output isolation. Therefore, the handling of the air gap needs to be very careful. If the air gap is too large, the leakage inductance will increase, the hysteresis loss will increase, the iron loss and copper loss will increase, which will affect the overall performance of the power supply. An air gap that is too small may saturate the transformer core, causing damage to the power supply.

The so-called continuous and intermittent modes of flyback power supply refer to the working state of the transformer. In the full load state, the transformer works in the working mode of complete energy transfer or incomplete transfer. Generally, the design should be based on the working environment. Conventional flyback power supplies should work in continuous mode, so that the losses of switching tubes and circuits are relatively small, and the working stress of the input and output capacitors can be reduced. However, there are some exceptions. It needs to be pointed out here: due to the characteristics of the flyback power supply, it is more suitable to be designed as a high-voltage power supply, and high-voltage power transformers generally work in intermittent mode. I understand that the high-voltage power supply output requires the use of high-voltage rectifier diodes. Due to the characteristics of the manufacturing process, high reverse voltage diodes have long reverse recovery time and low speed. In the continuous current state, the diode recovers when there is forward bias. The energy loss during reverse recovery is very large, which is not conducive to the performance of the converter. The improvement will at least reduce the conversion efficiency, cause the rectifier tube to become seriously heated, and at worst, even burn the rectifier tube. Since in discontinuous mode, the diode is reverse biased at zero bias, the loss can be reduced to a relatively low level. Therefore, the high-voltage power supply operates in intermittent mode, and the operating frequency cannot be too high. There is also a type of flyback power supply that works in a critical state. Generally, this type of power supply works in frequency modulation mode, or frequency modulation and width modulation dual mode. Some low-cost self-excited power supplies (RCC) often use this form. In order to ensure output stability, the transformer The operating frequency changes with the output current or input voltage. When it is close to full load, the transformer always remains between continuous and intermittent. This kind of power supply is only suitable for low-power output, otherwise the electromagnetic compatibility characteristics will be a headache.

The flyback switching power supply transformer should work in continuous mode, which requires a relatively large winding inductance. Of course, there is also a certain degree of continuity. It is unrealistic to pursue absolute continuity excessively. It may require a large magnetic core, and a lot of The number of coil turns, along with large leakage inductance and distributed capacitance, may not be worth the trade-off. So how to determine this parameter? After many practices and analysis of peer designs, I believe that when the nominal voltage is input, it is more appropriate for the output to reach 50%~60% and the transformer transitions from intermittent to continuous state. Or in the highest input voltage state, at full load output, the transformer can transition to the continuous state.