Abstract: This paper provides a detailed analysis and comparison of the factors affecting the reliability of military PWM switching power supplies, and proposes some suggestions for improving the reliability of switching power supplies based on engineering practice. Keywords: Switching power supply, reliability 1 Introduction The design of electronic products, especially military regulated power supplies, is a systematic project. It not only requires consideration of the power supply's own parameter design, but also electrical design, electromagnetic compatibility design, thermal design, safety design, and tri-proof design. Because even the slightest oversight in any aspect can lead to the collapse of the entire power supply, we should fully recognize the importance of power supply product reliability design. 2 Electrical Reliability Design of Switching Power Supplies 2.1 Selection of Power Supply Method Centralized power supply systems suffer from reduced power quality due to deviations between outputs and voltage differences caused by different transmission distances. Moreover, the use of a single power supply can lead to system paralysis when the power supply fails. Distributed power supply systems, because the power supply units are close to the load, improve dynamic response characteristics, have good power quality, low transmission loss, high efficiency, energy saving, high reliability, and are easy to form N+1 redundant power supply systems. Power expansion is also relatively easy. Therefore, the use of distributed power supply systems can meet the requirements of high-reliability equipment. 2.2 Circuit Topology Selection Switching power supplies generally employ eight topologies: single-ended forward, single-ended flyback, two-transistor forward, dual single-ended forward, dual forward, push-pull, half-bridge, and full-bridge. In single-ended forward, single-ended flyback, dual single-ended forward, and push-pull topologies, the switching transistors must withstand voltages exceeding twice the input voltage. If used with a 60% derating, transistor selection becomes difficult. In push-pull and full-bridge topologies, unidirectional magnetic saturation may occur, damaging the switching transistors. The half-bridge circuit, due to its automatic anti-imbalance capability, avoids this problem. In two-transistor forward and half-bridge circuits, the switching transistors only withstand the maximum input voltage of the power supply; even with a 60% derating, transistor selection is relatively easy. These two circuit topologies are generally chosen for high-reliability engineering applications. 2.3 Selection of Control Strategy In small and medium power supplies, current-mode PWM control is widely used. It offers the following advantages over voltage-mode control: cycle-by-cycle current limiting, which is faster than voltage-mode control; it prevents damage to the switching transistor due to overcurrent, significantly reducing overload and short-circuit protection; excellent grid voltage regulation; rapid transient response; stable loop, easy compensation; and much smaller ripple than voltage-mode control. Production practice shows that the output ripple of a 50W current-mode switching power supply is around 25mV, far superior to that of voltage-mode control. Due to the limitation of switching losses, the switching frequency of hard switching technology is generally below 350kHz. Soft switching technology uses the principle of resonance to make the switching device turn on and off in a zero voltage or zero current state, so that the switching loss is zero and the switching frequency can be increased to the megahertz level. This converter using soft switching technology combines the advantages of both PWM converter and resonant converter, and has near-ideal characteristics, such as low switching loss, constant frequency control, suitable energy storage element size, and wide control range and load range. However, this technology is mainly used in high power power supplies, while PWM technology is still the main technology in small and medium power supplies. 2.4 Selection of components Because the components directly determine the reliability of the power supply, the selection of components is very important. The failure of components mainly focuses on the following four aspects: (1) Manufacturing quality problems Failure caused by quality problems is unrelated to working stress. Unqualified products can be eliminated through strict inspection. In engineering applications, mature products from designated manufacturers should be selected, and products that have not been certified are not allowed to be used. (2) Component Reliability Issues Component reliability issues refer to the basic failure rate, which is a random type of failure. The difference between this and quality issues is that the failure rate of components depends on the operating stress level. Under a certain stress level, the failure rate of components will decrease significantly. To eliminate components that do not meet usage requirements, including those with unqualified electrical parameters, unqualified sealing performance, unqualified appearance, poor stability, and early failure, screening tests should be conducted. This is a non-destructive test. Screening can reduce the component failure rate by 1-2 orders of magnitude. Of course, the cost of screening tests (time and expense) is significant, but it is still worthwhile for comprehensive maintenance, logistical support, and overall system testing, and the development cycle will not be extended. The general requirements for screening tests of major components in power supply equipment are: ① Resistors should be tested 100% at room temperature according to technical conditions, and unqualified products should be eliminated. ② Ordinary capacitors should be tested 100% at room temperature according to technical conditions, and unqualified products should be eliminated. ③ Connectors should be sampled and tested for various parameters according to technical conditions. ④ Semiconductor devices are screened according to the following procedure: visual inspection → initial test → high temperature storage → high and low temperature shock → power aging → high temperature test → low temperature test → room temperature test. After screening, the rejection rate Q should be calculated. Q = (n / N) × 100% Where: N - total number of test samples; n - number of rejected samples; If Q exceeds the upper limit specified in the standard, all components in this batch are not allowed to be used in the machine and shall be handled in accordance with relevant regulations. If they meet the standard requirements, the qualified components will be marked with paint dots and then put into a special warehouse for installation. (3) Design issues First, it is important to select appropriate components: ① Use silicon semiconductor devices as much as possible and use fewer or no germanium semiconductor devices. ② Use more integrated circuits to reduce the number of discrete devices. ③ Use MOSFETs for switching to simplify the drive circuit and reduce losses. ④ Use diodes with soft recovery characteristics for output rectifiers as much as possible. ⑤ Select devices with metal packaging, ceramic packaging, or glass packaging. Do not use devices with plastic packaging. ⑥ Integrated circuits must be Class I products or military-grade products meeting or exceeding the quality level of MIL-M-38510 or MIL-S-19500 standards B-1. ⑦ Minimize the use of relays in the design; if necessary, use sealed relays with good contact. ⑧ Potentiometers should generally not be used; those that must be retained should be solidified. ⑨ The absorption capacitor should be very close to the switching transistor and output rectifier transistor. Because high-frequency current flows through them, they are prone to overheating; therefore, these capacitors must have high-frequency, low-loss, and high-temperature resistance characteristics. In humid and salt spray environments, aluminum electrolytic capacitors are susceptible to shell corrosion, capacitance drift, and increased leakage current. Therefore, aluminum electrolytic capacitors should not be used on ships or in humid environments. Because the electrolyte decomposes when bombarded by space particles, aluminum electrolytic capacitors are also unsuitable for power supplies in aerospace electronic equipment. Tantalum electrolytic capacitors have good temperature and frequency characteristics, are resistant to high and low temperatures, have long storage time, and are stable and reliable. However, tantalum electrolytic capacitors are heavier, have a low volume ratio, are not resistant to reverse voltage, have fewer high-voltage varieties (>125V), and are expensive. Regarding derating design: The basic failure rate of electronic components depends on their operating stress (including electrical, temperature, vibration, shock, frequency, speed, collision, etc.). Except for a few components that fail under low stress, most exhibit a higher failure rate with higher operating stress. To reduce the failure rate of components, derating design is necessary in circuit design. The degree of derating must consider factors such as size, weight, and cost in addition to reliability. Different components have different derating standards. Practice shows that the basic failure rate of most electronic components depends on electrical stress and temperature; therefore, derating mainly controls these two stresses. The following are the derating factors for commonly used components in switching power supplies: ① Resistors: power derating factor between 0.1 and 0.5. ② Diodes: power derating factor below 0.4, reverse withstand voltage below 0.5. ③ Light-emitting diodes: voltage derating factor below 0.6, power derating factor below 0.6. ④ Power switching transistors: voltage derating factor below 0.6, current derating factor below 0.5. ⑤ The voltage derating factor of ordinary aluminum electrolytic capacitors and non-polar capacitors is between 0.3 and 0.7. ⑥ The voltage derating factor of tantalum capacitors is below 0.3. ⑦ The current derating factor of inductors and transformers is below 0.6. (4) Loss problem The failure of components caused by loss depends on the length of working time and is not related to working stress. Aluminum electrolytic capacitors will gradually lose electrolyte when working at high frequency for a long time, and the capacity will also decrease synchronously. When the electrolyte loss is 40%, the capacity decreases by 20%; when the electrolyte loss is 0%, the capacity decreases by 40%. At this time, the capacitor core is basically dry and can no longer be used. In order to prevent failure, the replacement time of aluminum electrolytic capacitors should be marked on the drawings in general, and forced replacement should be carried out when the time expires. 2.5 Setting of protection circuit In order for the power supply to work reliably in various harsh environments, multiple protection circuits should be set up, such as surge protection, overvoltage, undervoltage, overload, short circuit, overheat protection circuits, etc. 3 Electromagnetic Compatibility (EMC) Design Switching power supplies employ pulse width modulation (PWM) technology, resulting in rectangular pulse waveforms with numerous harmonic components on both the rising and falling edges. Furthermore, the reverse recovery of the output rectifier diodes also generates electromagnetic interference (EMI), which negatively impacts reliability. Therefore, EMC becomes a critical system issue. As shown in Figure 1, three necessary conditions exist for EMI generation: an interference source, a transmission medium, and a sensitive receiving unit. EMC design addresses the disruption of one of these conditions. Figure 1 illustrates the three conditions for EMI formation. For switching power supplies, the primary focus is on suppressing the interference source, which is concentrated in the switching circuit and output rectifier circuit. Techniques employed include filtering, layout and wiring, shielding, grounding, and sealing. EMI is categorized into conducted interference and radiated interference based on its propagation path. Conducted noise has a wide frequency range, from 10kHz to 30MHz. While the causes of interference are known, controlling the rise and fall times of the pulse waveform may not be an efficient solution. One solution is to install power supply EMI filters, output filters, and absorption circuits, as shown in Figure 2. A power EMI filter is essentially a low-pass filter. It transmits 50Hz or 400Hz AC power to electronic equipment without attenuation, while significantly attenuating incoming interference signals. Simultaneously, it suppresses interference signals generated by the equipment itself, preventing them from entering the power grid and harming other equipment. Selecting an EMI filter involves choosing the filter network structure and component parameters based on the insertion loss, and selecting parameters such as rated voltage, rated current, leakage current, insulation resistance, and temperature conditions according to actual requirements. The power EMI filter is best installed near the socket where the power cord enters the chassis. Countermeasures for suppressing output noise are generally addressed in three frequency bands: 10kHz–150kHz, 150kHz–10MHz, and above 10MHz. The 10kHz–150kHz range mainly contains normal noise, which is typically addressed using a general-purpose LC filter. The 150kHz–10MHz range mainly contains common-mode noise, which is usually addressed using a common-mode rejection filter. Common-mode chokes should use ferrite magnetic materials with high permeability and good frequency characteristics, with an inductance of (1~2) mH and a capacitance between 3300pF and 4700pF. To control noise in the low-frequency range, the value of LC can be appropriately increased. For frequencies above 10MHz, the solution is to improve the filter's shape. Reverse recovery of the output rectifier diode can also cause electromagnetic interference. In this case, an RC absorption circuit can be used to suppress the rate of current rise. Typically, R is between (2~20) Ω, and C is between 1000pF and 10nF. C should be a high-frequency ceramic capacitor. Good layout and wiring techniques are also important means of controlling noise. To reduce noise and prevent malfunctions caused by noise, the following points should be noted: ① Minimize the area surrounded by high-frequency pulse current. ② Place the buffer circuit as close as possible to the switching transistor and the output rectifier diode. ③ Keep the area through which the pulse current flows away from the input and output terminals to separate the noise source from the output. ④ The control circuit and power circuit should be separated, using a single-point grounding method. Large-area grounding can easily cause antenna effects, so it is recommended to avoid large-area grounding. ⑤ If necessary, the output filter inductor can be placed on the ground loop. ⑥ Use multiple low ESR (equivalent series resistance) capacitors in parallel for filtering. ⑦ Use copper foil for low-inductance, low-impedance wiring. ⑧ There should be no excessively long parallel lines between adjacent printed lines. Avoid parallel routing as much as possible; use a perpendicular crossing method. Avoid abrupt changes in line width or sudden corners. Loop routing is prohibited. ⑨ The input and output lines of the filter must be separated. Do not bundle the input and output lines of the switching power supply together. For radiated interference, sealed shielding technology is mainly used. Electromagnetic enclosure is implemented in the structure, requiring good electromagnetic contact between all parts of the casing to ensure electromagnetic continuity. Currently, aluminum alloy casings are mostly used to reduce weight, but aluminum alloy has poor magnetic permeability. Therefore, the casing needs to be plated with nickel or sprayed with conductive paint, and the inner wall needs to be covered with a high-permeability shielding material. Permanent connections in the casing are secured with conductive adhesive or employ a continuous welded structure. For connections requiring disassembly, conductive rubber strips can be used to ensure electromagnetic continuity. Conductive materials must possess high conductivity, elasticity, and a minimal width-to-thickness ratio. 4. Thermal Design for Power Supply Reliability Besides electrical stress, temperature is the most critical factor affecting equipment reliability. Temperature rise within the power supply will lead to component failure. When the temperature exceeds a certain value, the failure rate increases exponentially, and exceeding the limit will result in component failure. Foreign statistics show that for every 2°C increase in electronic component temperature, reliability decreases by 10%; the lifespan at a 50°C temperature rise is only 1/6 of that at a 25°C temperature rise. Technical measures are needed to limit the temperature rise of the chassis and components; this is thermal design. The principles of thermal design are: firstly, to reduce heat generation, i.e., to select superior control methods and technologies, such as phase-shift control technology and synchronous rectification technology; secondly, to select low-power devices, reduce the number of heat-generating components, increase and thicken the width of printed lines, and improve power supply efficiency. Secondly, heat dissipation should be enhanced by utilizing conduction, radiation, and convection technologies to transfer heat. This includes methods such as radiators, air cooling (natural convection and forced air cooling), liquid cooling (water, oil), thermoelectric cooling, and heat pipes. Forced air cooling can dissipate more than ten times the heat of natural cooling, but it requires additional fans, fan power supplies, and interlocking devices. This not only increases the cost and complexity of the equipment but also reduces the reliability of the system and increases noise and vibration. Therefore, natural cooling should be used as much as possible under normal circumstances, rather than air cooling or liquid cooling. When arranging components, heat-generating devices should be placed downwind or on top of the printed circuit board. Heat sinks should be treated with an oxidation blackening process to improve emissivity; black paint coating is not allowed. Applying conformal coating will affect heat dissipation, so an appropriate margin should be added. The surface of the components mounted on the heat sink should be smooth and flat. Generally, silicone grease should be applied to the contact surface to improve thermal conductivity. Transformers and inductors should use thicker wires to suppress temperature rise. 5. Safety Design For power supplies, safety has always been considered one of the most important performance characteristics. Unsafe products not only fail to perform their intended functions but may also cause serious accidents, resulting in significant losses such as damage to equipment and loss of life. To ensure a high level of product safety, safety design is essential. The main focus of power supply product safety design is preventing electric shock and burns. For the commercial equipment market, representative safety standards include UL, CSA, and VDE, with content varying depending on the application. Permissible leakage current is between 0.5mA and 5mA. China's military standard GJB1412 specifies a leakage current of less than 5mA. The magnitude of the power supply equipment's leakage current to ground depends on the capacitance of the EMI filter capacitor Cy, as shown in Figure 2. From an EMI filter perspective, a larger capacitance of capacitor Cy is better; however, from a safety perspective, a smaller capacitance of capacitor Cy is better. The capacitance of capacitor Cy is determined according to safety standards. If the safety performance of capacitor Cx is inadequate, it may break down when transient peaks occur in the power grid. While its breakdown may not endanger personal safety, it will cause the filter to lose its filtering function. To prevent accidental electric shock, the plug/socket should, in principle, have pins on the product end (non-power end) and holes on the mains end (power end); the input end of the power supply device should be a pin, and the output end a hole. To prevent burns, for exposed parts that may come into contact with the human body (heat sink, casing, etc.), the maximum temperature should not exceed 60℃ when the ambient temperature is 25℃, and the maximum temperature of the panel and manually adjustable parts should not exceed 50℃. 6. Three-proof design Three-proof design refers to moisture-proof design, salt spray-proof design, and mildew-proof design. During the design phase, sealing measures should be implemented for components requiring airtightness; epoxy resin potting can be used for irreparable modular devices; the moisture absorption of components and raw materials used should be low, and products containing easily mold-prone materials such as cotton, hemp, and silk should not be used; protective nets should be installed on sealed enclosures and cabinets to prevent insects and rodents from entering; the outer top of devices directly exposed to the atmosphere should not have a recessed structure to avoid water accumulation and corrosion; corrosion-resistant materials can be selected, and a layer of metallic or non-metallic protective film can be applied to the surface of electronic equipment and its components through plating, coating, or chemical treatment to isolate the surrounding medium; sealed or semi-sealed structures should be used to isolate adverse external environments; applying a special three-proof varnish to the surface of printed circuit boards and components can effectively prevent corona and breakdown between wires and improve the reliability of the power supply; inductors and transformers should be impregnated with varnish and end-sealed to prevent moisture from entering and causing short circuits. 7 Conclusion The above recommendations only apply to military power supplies; different choices may be made for commercial and industrial products in certain aspects. In summary, the reliability of power supply equipment depends not only on electrical design but also on components, structure, assembly, manufacturing processes, and processing quality. Reliability is based on design; in practical engineering applications, feedback data obtained through various tests should be used to refine the design and further improve the reliability of the power supply.