The digital power design approach is the general trend

Preface: More than a decade ago, companies such as Texas Instruments and Microchip have already mixed digital components into traditional analog power supply designs. Today's information products require smaller size, lower cost, and higher reliability and controllability. The traditional analog power supply architecture is obviously insufficient for this application.

Power supply technology development trend

Caption: Switching power conversion system.

The information industry is moving towards smaller processes, hoping to bring a positive solution to the power consumption problem. However, the chips are integrating more and more functions and are faster and faster. The new processes often result in higher power consumption and heat generation. However, the new generation of information products is either in appearance or in size. There are several benefits to reducing the size of information products. The first is in large-scale applications. For example, when the size of the server can be effectively reduced inside the computer room of an enterprise, more equipment can be installed within the same unit area, and more services can be provided. In other words, the smaller the equipment size, the smaller the computer room area can be used to achieve the same menu display. It is just the miniaturization of the equipment size. Relatively speaking, the power supply system must also be miniaturized, and the power load that can be supplied must be able to maintain or even exceed the past level. This is a very severe challenge for power supply design manufacturers.

■ Digital power design helps reduce equipment size and enhance management capabilities

Caption: Comparison of digital and analog circuits.

Although the size of the machine has been reduced, as the performance and functions increase, the power consumption of these relatively small devices will not be reduced at all. In order to meet the power supply needs of these devices and to be able to fit the power supply modules into these thin and light cases, in addition to the use of insulated gate bipolar transistors (IGBTs), power field effect transistors (MOSFETs), intelligent IGBT power modules (IPMs), MOS gate-controlled thyristors (MCTs), electrostatic induction transistors (SITs), super recovery diodes, non-inductive capacitors, non-inductive resistors, new magnetic materials and transformers, EMI filters and other high-performance components, digital PWM and fully digital control can also help reduce the volume occupied by the power supply module.

Digital power is a power converter that uses a digital interface and is programmable. Digital interface and programmability areimportant features of digital power and are also important niches for simplifying product applications. The controller can be implemented in analog circuit mode or digital circuit mode. Digital power refers to the use of digital control to implement the control loop and interface of the switching power supply. In comparison, traditional switching power supplies mainly use analog control to implement their control loop and interface.

Generally speaking, the use of full digital control technology can effectively reduce the size of the power supply, reduce costs, and improve the reliability of the equipment and its adaptability to users. The entire power supply signal sampling, processing, control (including voltage and current, etc.), communication, etc. all use DSP technology to obtain consistent and stable control parameters.

Digital power control can be more flexible. For example, the power supply can dynamically adjust and optimize the output of the power supply at various voltages and temperatures, such as derating protection and PFC digital control harmonics. DSP technology can achieve simpler and more stable communication and current sharing, and obtain good EMC control. Digital components can provide a higher degree of intelligence, so flexible LED warning combinations, self-monitoring capabilities, and remote communication mechanisms can be easily achieved. Digital design can also effectively reduce the number of components used and increase the degree of modularity, as well as increase power density. Eliminate the discreteness and temperature drift of components in analog control technology, ensure that each module reaches the optimal indicator, and improve power reliability. The modules are more intelligent and easy to use and maintain.

■Power management of handheld devices tends to be highly integrated and digitally designed

In handheld devices, the power management mechanism is constantly improving. The convenience of handheld devices has gradually been replaced by multi-function and high performance. Before the battery technology has achieved a breakthrough, the only way to seek more efficient power management is to use them.

Figure caption: Comparison of power consumption distribution of 2G and 3G mobile phones.

Currently, mainstream mobile devices usually integrate multiple functions such as video, audio, camera/video recording, file storage/editing, etc. For example, PMP can complete tasks such as audio and video playback, video recording and file storage, and mobile phones can take pictures, listen to music and even use wireless modules to surf the Internet. These different functions are usually achieved through corresponding components, but these components require different voltage supplies to operate normally, and the voltage quality must be stable, reliable, clean and high-performance, which is also a challenge faced by power management design. In these increasingly complex power management systems, a product may require multiple voltage supplies such as 5V, 3.3V, 2.5V, 1.8V, 1.5V, 1.2V, 0.9V and 0.7V at the same time. How to effectively manage multiple voltages and prevent them from interfering with each other is a difficult problem facing power design.

As far as current developments are concerned, the integration of power-system-on-a-chip is a long-term trend in the entire mobile application and even the industry. Naturally, the power supply design of handheld mobile devices also follows this trend and develops towards a higher level of chip integration.

In the past, the power supply system of mobile devices was composed of many scattered analog parts, but such outdated designs are no longer seen in recent products. The original design trend of mobile device power supply modules in terms of functions was mainly to integrate and minimize the size of these discrete analog power supply parts, and then integrate these parts into one or a few power management components. These power supply parts began to add more and more functions and intelligent management capabilities, and gradually spread to other fields in terms of application scope, and entered the design of various types of mobile devices (such as GPS, handheld game consoles, PMP, mobile TV and other products).

The reasons for chip integration are the same for the entire electronics industry: by integrating more functions into fewer parts, product costs can be greatly reduced because the number of components and circuit board area required to manufacture these products will be reduced, and the assembly and manufacturing procedures of the products will be simpler, thereby achieving higher system reliability and significantly shortening testing time. In addition, higher chip integration can also improve the efficiency of the R&D process, thereby shortening the time for manufacturers to bring new products to market, which is of great help in improving product competitiveness.

Design Method of Digital Switching Power Supply

Generally, there are two main methods to achieve digital control of switching power supplies:

The first type: The single-chip controller samples through an external A/D conversion chip, calculates and adjusts the data after sampling, and then transmits the result to the PWM chip after digital/analog (D/A) conversion, so as to achieve indirect control of the switching power supply by the single-chip controller. The technology of this method is relatively mature, the design method is easy to master, and the requirements for the single-chip controller are not high, and the cost is relatively low. However, the control circuit is more complicated because it uses multiple chips; and after the steps of A/D and D/A conversion, it will cause a relatively large signal delay, which will inevitably affect the dynamic performance and voltage regulation accuracy of the power supply. Some single-chip controllers integrate PWM output, but the switching power supply is developing towards high frequency. The frequency of the general single-chip controller is limited, and the generated PWM output frequency is inversely proportional to the accuracy, so it is impossible to generate a PWM output signal with sufficient frequency and accuracy.

Caption: Structural block diagram of PWM.

The second method is to directly control the power supply through high-performance digital chips such as DSP . The digital chip completes signal sampling AD conversion and PWM output. Since the output digital PWM signal power is not enough to drive the switch tube, a driver chip is required to drive the switch tube. This can simplify the design of the control circuit. Since these chips have relatively high sampling speed (the 10-bit AD converter inside TMS320LF2407 only takes 500ns to complete an AD conversion, compared to the fastest 8-bit single-chip controller that takes several microseconds) and computing speed, they can quickly and effectively implement various complex control algorithms and achieve effective control of the power supply system. Such a design has high dynamic performance and voltage regulation accuracy. However, the DSP chip has a complex structure and a relatively high cost; and the DSP control technology is relatively difficult to master, and the requirements for designers are relatively high, so it is difficult to be widely used in the mainstream switching power supply field. Although DSP technology has begun to be used in switching power supplies, it is currently still mainly limited to application fields that require high power performance and are relatively expensive.

Problems to be faced after digitalization of power supply control

Digitally controlled switching power supplies inevitably have the following problems: The speed and accuracy of the A/D (analog/digital) converter are inversely proportional. In order to ensure that the switching power supply has a higher voltage regulation accuracy, the A/D converter must have a relatively high-precision sampling, but the high-precision sampling frequency requires a longer A/D conversion time. As part of the feedback loop, a long A/D conversion time will inevitably cause additional phase delay time. In addition to the phase delay existing in analog control, the delay time of the conversion process will inevitably cause additional waiting cycles, causing the real-time response capability of the loop to deteriorate.

Just like the method of using RC (resistance and capacitance) compensation for PI regulation in analog chips, the method of introducing PI regulation in the control loop is used to improve the real-time response capability of the control loop. This approach requires a large amount of system resources for digital chips, because digital control is different from analog control. Signal sampling is not continuous, but regular and discrete. There will be a period of time between two samples, and the value during this period cannot be obtained. In order to achieve precise control, the time interval between each sampling cannot be too long, that is, the sampling frequency cannot be too low. As a digital chip, after each AD conversion, the result will be sent to the system's central processor, and then the processor will calculate and PI regulate the sampled value.

When the sampling frequency is relatively high, this approach consumes a lot of system computing resources, so the performance requirements for digital chips are also relatively high. There are not many digital chips specifically used for power supply control. Although DSP chips are generally used in occasions with relatively high requirements, they have fast computing and sampling speeds and powerful functions, but they are relatively expensive. Moreover, general DSP chips are not specifically used as power supply control chips. The utilization rate of their chip resources in general power supply applications is not high. Under certain conditions, using DSP chips as the core of power supply digital control is a waste.

The competition between DSP and MCU architecture in power supply design

Currently, the two semiconductor companies that hold the leading position in the field of digital power supply are Texas Instruments and Microchip. However, pure MCU or pure DSP architecture has its shortcomings in application. Therefore, the two semiconductor companies have coincidentally turned to combining MCU and DSP architecture to design digital power supply . DSP has powerful digital computing and processing capabilities, while MCU has powerful control capabilities over the periphery. For the design of a comprehensive digital control power supply, both are indispensable.

Photo caption: DSC products from Microchip.

Nevertheless, both companies still believe that they have a better advantage in their respective areas of expertise. Texas Instruments naturally takes DSP as the main focus, and strongly emphasizes the real-time response capability and control accuracy that can be brought about by powerful computing power. The programmability of DSP has an excellent performance on the system architecture, portability and maintainability. Microchip emphasizes that ordinary customers do not need too powerful DSP computing power, and complex programmable design will only prolong the product development schedule. The DSC (Digital Signal Controller) architecture provided by the company successfully integrates MCU and DSP, not only successfully simplifies the instruction flow, but also effectively shortens the product design schedule through the standard C language compiler.

Can the analog components of a power supply be completely replaced? Not necessarily!

Many radical manufacturers claim that the use of digital components and circuits can completely replace the analog components in the switching regulator, which can greatly simplify the design of the switching regulator and help stabilize the entire power supply system. However, the power supply itself belongs to the analog category in terms of physical laws. Even if the ADC (analog/digital converter) or DSP is used to replace the error amplifier and pulse width modulation of the digital switching regulator, it still requires a voltage reference, current detection circuit/switch and FET driver. These components only exist in analog form and are widely used in various types of switching regulators and cannot be replaced. Even the ADC component itself is the same. The ADC is basically more analog.

Analog design has always been compared to art. Often, the adjustment of analog components and the overall architecture design always rely on the designer's experience and skills to adjust the perfect proportions, just like a chef who must rely on long-term experience to control the heat when cooking to cook a dish with good color, aroma and taste. Although the analog circuit architecture is simple, the layout is often a domino effect. Since the power supply cannot abandon the burden of analog components, further research and development is needed in analog technology. After all, most semiconductor companies only dominate the digital field, but they often have only a superficial understanding of analog architecture. Take Taiwan as an example. Although Taiwan is a major IC design company, it rarely mentions analog processes. Although there are a lot of solutions for digital ICs on the market, for analog solutions, it can only seek from foreign manufacturers.

Pursuing pure digital power supply is still out of reach. Reasonable combination of digital and analog design is the right way.

Digital power has been widely discussed in recent years, but the industry generally has different views on this industry. Although the stringent requirements of mobile devices for power management have allowed digital power to show its strengths, traditional analog power solutions still dominate in most application fields after decades of development. Even though analog solutions are relatively weak in some aspects, such as the number of control loop components, system stability, flexible configurability and communication capabilities, power supply manufacturers are gradually moving towards different design thinking and starting to add digital components or design methods in order to break through the traditional analog power design barriers.

Traditional analog power supplies are simple and easy to use. Although there are not many changeable parameters, simplicity is its biggest advantage. In higher-end applications, system administrators may need additional control functions to monitor the status of the power supply, which may include temperature, input and output current, input and output voltage, etc., and report to the central control system regularly according to the system administrator's settings. In addition, some ID tags, fault status messages, time tags, etc. can be stored in the flash memory or other non-volatile storage architecture on the microcontroller, and these messages are reported at a specified time. These designs require a large number of integrated digital circuits. These digital power supplies are usually more commonly seen in high-end servers. In general affordable consumer products, these additional control functions are not needed.