The Development of High-Frequency Switching Power Supply Technology in the Communications Industry

1. Development of High-Frequency Switching Power Supply Technology in the Communication Industry The development of high-frequency switching power supply technology for communication applications can be summarized in several aspects: converter topology, modeling and simulation, digital control, and magnetic integration.
1.1 Soft-switching, power factor correction (PFC), and multilevel converter topology technologies have been hot topics in converter topology in recent years. Soft-switching technology can effectively reduce switching losses and stress, contributing to improved converter efficiency; PFC technology can improve the input power factor of AC/DC converters and reduce harmonic pollution to the power grid; while multilevel technology is mainly used in three-phase input converters for communication power supplies, effectively reducing voltage stress on the switching transistors. Furthermore, due to the high input voltage, employing appropriate soft-switching techniques to reduce switching losses is an important future research direction for multilevel technology.
To reduce the size of the converter, it is necessary to increase the switching frequency to achieve high power density. This requires the use of smaller magnetic materials and passive components. However, increasing the frequency will significantly increase the switching and drive losses of the MOSFETs. The application of soft-switching technology can reduce switching losses. Currently, the most widely used technologies in communication power engineering are active clamp ZVS technology, ZVS phase-shifted full-bridge technology which emerged in the early 1990s, and synchronous rectification technology proposed in the late 1990s.
1.1.1 ZVS Active Clamping Active clamping technology has evolved through three generations, all of which have been patented. The first generation was the active clamping ZVS technology from VICOR Corporation in the United States, which increased the operating frequency of DC/DC converters to 1MHz and achieved a power density of nearly 200W/in³, but its conversion efficiency did not exceed 90%. To reduce the cost of the first-generation active clamping technology, IPD Corporation patented the second-generation active clamping technology, which used P-channel MOSFETs and employed active clamping on the secondary side of the transformer for the forward circuit topology, significantly reducing product costs. However, this method resulted in a narrower zero-voltage switching (ZVS) boundary condition for the MOSFETs, and the PMOS operating frequency was not ideal. To prevent magnetic energy from being wasted during core reset, a Chinese-American engineer patented the third-generation active clamping technology in 2001. Its key feature was that, based on the second-generation active clamping technology, the energy released during core reset was transferred to the load, thus achieving higher conversion efficiency. It has three circuit schemes: one of which can use N-channel MOSFETs, so the operating frequency can be higher. This technology can combine ZVS soft switching and synchronous rectification technology, thus achieving an efficiency of up to 92% and a power density of over 250W/in3.
1.1.2 ZVS Phase-Shifted Full-Bridge Since the mid-1990s, ZVS phase-shifted full-bridge soft-switching technology has been widely used in medium and high power power supplies. This technology has played a significant role in improving converter efficiency when the switching speed of MOSFETs is not ideal, but it also has several drawbacks. The first drawback is the addition of a resonant inductor, which leads to a certain size and losses, and the electrical parameters of the resonant inductor need to be kept consistent, which is difficult to control during manufacturing. The second drawback is the loss of an effective duty cycle. Furthermore, since synchronous rectification is more conducive to improving converter efficiency, the control effect of phase-shifted full-bridge on the secondary-side synchronous rectification is not ideal. The initial PWM ZVS phase-shifted full-bridge controllers, UC3875/9 and UCC3895, only controlled the primary side, requiring additional logic circuitry to provide accurate secondary synchronous rectification control signals. While the latest phase-shifted full-bridge PWM controllers, such as LTC1922/1 and LTC3722-1/-2, have added secondary-side synchronous rectification control signals, they still cannot effectively achieve ZVS/ZCS synchronous rectification on the secondary side. However, this is one of the most effective measures to improve converter efficiency. Another significant improvement of the LTC3722-1/-2 is its ability to reduce the inductance of the resonant inductor. This not only reduces the size and losses of the resonant inductor but also improves duty cycle loss.
1.1.3 Synchronous Rectification Synchronous rectification includes self-driven and externally driven methods. Self-driven synchronous rectification is simple and easy to implement, but the secondary voltage waveform is easily affected by factors such as transformer leakage inductance, resulting in lower reliability in mass production and thus its limited application in actual products. For output voltage conversion from 12V to around 20V, dedicated external driver ICs are often used, which can achieve better electrical performance and higher reliability.
TI proposed the UCC27221/2 chip with a predictive drive strategy, dynamically adjusting the dead time to reduce the conduction loss of the body diode. STMicroelectronics also designed a similar chip, STSR2/3, which is suitable for both flyback and forward converters, while improving the performance of continuous and discontinuous conduction modes. The Center for Power Electronics Systems (CPES) studied various resonant drive topologies to reduce drive losses and proposed a new type of synchronous rectification circuit in 1997, called quasi-square wave synchronous rectification. This circuit can significantly reduce the conduction loss and reverse recovery loss of the body diode of the synchronous rectifier and easily achieve soft switching of the primary main switch. Linear Technology's synchronous rectification control chips LTC3900 and LTC3901 can be better applied to forward, push-pull, and full-bridge topologies.
ZVS and ZCS synchronous rectification technologies have also begun to be applied, such as the synchronous rectification driver of the active clamp forward circuit (NCP1560), and the synchronous rectification driver chips LTC1681 and LTC1698 of the dual transistor forward circuit, but none of them have achieved the excellent effect of symmetrical circuit topology ZVS/ZCS synchronous rectification.
1.2 Modeling and Simulation Switching converters mainly employ two modeling methods: small-signal analysis and large-signal analysis.
Small-signal analysis methods, primarily the state-space averaging method, were proposed by R.D. Middlebrook of the California Institute of Technology in 1976. This can be considered the first truly significant breakthrough in modeling and analysis in the field of power electronics. Later methods, such as the current injection equivalent circuit method, the equivalent controlled source method (proposed by Zhang Xingzhu in 1986), and the three-terminal switching device method, all fall under the category of circuit averaging methods. The disadvantages of averaging methods are obvious: averaging the signal prevents effective ripple analysis; it cannot accurately perform stability analysis; and it may not be suitable for resonant converters. Crucially, the model obtained by averaging is independent of the switching frequency, and its applicability depends on the natural frequencies generated by inductors and capacitors in the circuit being much lower than the switching frequency for high accuracy.
Large-signal analysis methods include analytical methods, phase plane methods, large-signal equivalent circuit model methods, switching signal flow methods, nth harmonic three-port model methods, KBM methods, and general averaging methods. Another method is the equivalent small-parameter signal analysis method proposed by Professor Qiu Shuisheng of South my country University of Technology in 1994. This method is applicable not only to PWM converters but also to resonant converters and can perform output ripple analysis.
The purpose of modeling is for simulation, followed by stability analysis. In 1978, R. Keller first used R. D. Middlebrook's state-space averaging theory for SPICE simulation of switching power supplies. Over the past 30 years, many scholars have established various model theories for the average SPICE model of switching power supplies, resulting in various SPICE models. These models each have their strengths; some representative ones include: Dr. Sam Ben Yaakov's switching inductor model; Dr. Ray Ridley's model; the average PSpice model of switching power supplies based on Dr. Vatche Vorperian's Orcad 9.1; the average Isspice model of switching power supplies based on Steven Sandler's ICAP4; and the average model of switching power supplies based on Dr. Vincent G. Bello's Cadence model, etc. Based on these models, a macro-model is constructed by combining the main parameters of the converter. The DC/DC converter constructed from these models is then used to perform DC analysis, small-signal analysis, and closed-loop large-signal transient analysis on professional circuit simulation software platforms (Matlab, PSpice, etc.).
Due to the rapid pace of change and development of converter topologies, the requirements for converter modeling are becoming increasingly stringent. In short, converter modeling must keep up with the evolution of converter topologies to be accurately applied in engineering practice.
1.3 Digital Control The simple applications of digital control are mainly in protection and monitoring circuits, as well as communication with the system. It is currently widely used in communication power supply systems. It can replace many analog circuits, performing power supply startup, input and output over/under voltage protection, output overcurrent and short circuit protection, and overheat protection. Through specific interface circuits, it can also achieve communication and display with the system.
More advanced digital applications include not only comprehensive protection and monitoring functions, but also the output of PWM waves to control power switching devices via drive circuits and achieve closed-loop control. Currently, companies such as TI, ST, and Motorola have launched dedicated DSP chips for motor and motion control. At present, the digitalization of communication power supplies mainly adopts a combination of analog and digital methods. The PWM section still uses dedicated analog chips, while the DSP chip mainly participates in duty cycle control, frequency setting, output voltage adjustment, and protection and monitoring functions.
To achieve faster dynamic response, many advanced control methods have been proposed. For example, ON Semiconductor proposed improved V2 control, Intel proposed Active-droop control, Semtech proposed charge control, Fairchild proposed Valley current control, IR proposed multiphase control, and many other universities in the United States have also proposed various other control concepts [7, 8, 9]. Digital control can improve system flexibility, provide better communication interfaces, fault diagnosis capabilities, and anti-interference capabilities. However, in precision communication power supplies, control accuracy, parameter drift, current detection and current sharing, and control delay are practical problems that urgently need to be solved.
1.4 Magnetic Integration As the switching frequency increases, the size of the switching converter decreases and the power density is greatly improved. However, the switching losses will increase, and more magnetic components will be used, thus occupying more space.
Research on magnetic component integration technology is relatively mature abroad, and some manufacturers have already applied this technology to practical communication power supplies. In fact, magnetic integration is not a new concept; the idea of ​​magnetic integration was first proposed by Cuk in the late 1970s when he introduced the Cuk converter. Since 1995, the Center for Power Electronics Systems (CPES) in the United States has conducted extensive research on magnetic device integration, using the concept of coupled inductors to conduct in-depth research on the integration of multiphase BUCK inductors, and applying it to various types of converters. In 2002, Yim-Shu Lee et al. from the University of Hong Kong also proposed a series of explorations and designs regarding magnetic integration technology.
Conventional magnetic component design methods are extremely complex and require consideration from various perspectives, such as core size selection, material and winding determination, and assessment of iron and copper losses. However, magnetic integration technology, in addition to these considerations, must also address the issue of magnetic flux imbalance, as the equivalent total magnetic flux distributed across the iron core varies, and some parts may saturate prematurely. Therefore, the analysis and research of integrated magnetic devices will be even more complex and challenging. Nevertheless, the resulting high power density advantage will undoubtedly be a major development trend in future communication power supplies.
1.5 Manufacturing Process The manufacturing process of high-frequency switching power supplies for communication applications is quite complex and directly affects the electrical functions, electromagnetic compatibility, and reliability of the power supply system. Reliability is the primary indicator for communication power supplies. Comprehensive testing methods, complete process monitoring points, and the adoption of anti-static measures during production largely ensure optimal product design performance. The widespread use of SMD (Surface Mount Device) components can significantly improve soldering reliability. Starting in 2006, Europe and the United States will require lead-free manufacturing processes for electronic products, which will place higher and stricter demands on the selection of components and the control of the manufacturing process in communication power supplies.
A more attractive technology currently is the concept of Integrated Power Electronics Modules (IPEMs), commonly known as "building blocks," proposed in recent years by the Center for Power Electronics Systems (CPEC). This technology utilizes advanced packaging techniques to reduce parasitic factors, improving voltage ringing and efficiency in circuits. It integrates drive circuits and power devices to increase drive speed and thus reduce switching losses. Power electronics integration technology can improve not only transient voltage regulation but also power density and system efficiency. However, such integrated modules currently face many challenges, primarily the integration of passive and active devices and the difficulty in achieving optimal thermal design. CPEC has conducted years of research on power electronics integration technology, proposing many useful methods, structures, and models.
2. Conclusion The main future trend for high-frequency switching power supplies in communication applications will be towards integration and miniaturization, leading to increasingly higher power densities and more demanding manufacturing processes. Without new breakthroughs in semiconductor devices and magnetic materials, significant technological advancements may be difficult to achieve, and the focus of technological innovation will be on improving efficiency and reducing weight. Therefore, manufacturing processes will play an increasingly important role in power supply manufacturing. Furthermore, the application of digital control integrated circuits is another future direction for switching power supply development, which will depend on further improvements in DSP operating speed and anti-interference technology.