Briefly describe the shortcomings and countermeasures in the development of miniaturization of DC-DC switching power supplies

summary

This article will briefly analyze the success and shortcomings of high-frequency DC-DC switching power supplies in the process of miniaturization (the second basic goal), and propose measures to improve and deal with them. This will lead to the advent of switching power supplies with high power density, large current output, and high ripple noise suppression capabilities.

In terms of the most basic indicators of power supply, a high-frequency DC-DC module or power chip with such indicators may be the final achievement of our development stage: they have a power density of 180-1000W/in3, an output current of more than 100A and a power of hundreds of watts to 1 kilowatt, and at the same time, under all loads, they have a ripple factor of less than 0.1-0.05%. Moreover, according to needs, basically no external components are needed, and power supplies of any output voltage, current and power, various specifications, uses and indicators can be formed through series and parallel connection and feedback.

A high current output inductive filter NIF with a very high inductance/volume ratio L/V is connected to the output of the 100A power chip VTM. Its inductance L is sufficient to make the ripple factor of VTM lower than 0.1-0.05% in full load, while its volume is only equivalent to the 3.5MHz power transformer in VTM. Moreover, since NIF is not an energy storage element, it is not a limiting factor in improving the response speed.

Table of contents

Preface

High-frequency DC-DC switching power supplies have made great achievements in achieving miniaturization and improving power density

However, the high power density, high frequency DC-DC low voltage and high current output capability and ripple noise suppression capability are limited

Briefly analyze the reasons for these two restrictions

Improvement and response measures

A high current output inductive filter NIF with extremely high inductance/volume ratio L/V

About response speed

Outlook on the development of switching power supplies

1. Introduction

People hope that the DC power supply is as clean as a large-capacity primary battery, with extremely low ripple and noise, and extremely small internal resistance. They also hope that it has a fast response speed. This has been the first basic goal that people have been striving to pursue for decades.

At the same time, improving the power density and efficiency of the power supply, making it small, light and efficient; improving the power output of the power supply, especially the low-voltage and high-current output, is also the second basic goal that people have been striving for for decades.

In the development stage of linear regulated power supplies and SCR phase-controlled power supplies, we have achieved great success in pursuit of the first basic goal. So far, it is not difficult to manufacture a DC power supply with a ripple factor of 0.01% and a voltage regulation rate of 0.05%. If sufficient open-loop gain and reasonable open-loop frequency characteristics are designed, a high closed-loop response speed can also be achieved. However, it is large in size, low in efficiency, and therefore the output current is not large.

The high-frequency DC-DC switching power supply technology that emerged to achieve the second basic goal is at the core of the application of power electronics technology and various power supply systems. It has achieved great success in improving power density and efficiency, making it small, lightweight, efficient and current output. However, due to the commutation problem in its inverter and the imbalance in the development of its components, it also leaves obvious shortcomings and regrets: ultra-small modules or power chips cannot achieve large current output and high ripple noise suppression capabilities. In other words, a switching power supply that requires low-voltage, high-current output and high ripple noise suppression capabilities cannot have a high power density.

This article will briefly analyze the success and shortcomings of high-frequency DC-DC switching power supplies in the process of miniaturization (the second basic goal), and propose measures to improve and deal with them. This will lead to the advent of switching power supplies with high power density, large output current, and high ripple noise suppression capabilities.

2. High-frequency DC-DC switching power supplies have made great achievements in achieving miniaturization and improving power density

More than half a century ago, people actually knew the principle of switching power supply: using the inverter method to achieve DC-AC-DC conversion. In this way, the output DC voltage can be adjusted through transformers, frequency modulation-width modulation, etc.; the volume and weight of transformers, energy storage components inductors, filters, and capacitors in the circuit can be reduced by increasing the operating frequency. Such power supply has much higher efficiency and power density than linear regulated power supply and SCR phase-controlled power supply.

With the rapid development of electronic technology and semiconductor devices, the high-frequency and soft-switching technology of power supplies have become one of the main research hotspots in the international power electronics community for decades. In many aspects, switching power supplies have gradually replaced linear regulated power supplies and SCR phase-controlled power supplies. The ideal of high-frequency switching power supplies has begun to be realized, and the operating frequency of DC/DC converter modules using PWM control technology has reached the range from 20kHz to 400kHz. At the same time, in order to solve the problems of high power consumption, reduced efficiency and increased noise caused by the commutation in the inverter, in 1997, after nearly three decades of global soft switching basic theoretical research, the "second-generation product" was based on zero current switching (ZCS) and zero voltage switching (ZVS) soft switching control technology, combined with the latest scientific and technological achievements in control integration, packaging, ferrite, noise and heat dissipation technology, so that the power density reached 120-180W/in3, the efficiency reached 90%, and the operating frequency was close to 1MHz. In the past few years, full ZVS synchronous rectification has been realized and research and development of integrated power modules has been carried out. The high-power density and high-frequency DC-DC switching power supply products have reached a state very close to the "ideal power device". Many circuit topologies have emerged. Currently, there are even 1000W/in? power chips with an operating frequency of 3MHz, which have developed rapidly.

3. However, the high power density, high frequency DC-DC low voltage and high current output capability and ripple noise suppression capability are limited

However, we should also see that in the process of continuously improving the operating frequency, power density and efficiency, high power density and high frequency DC-DC switching power supply modules or power chips will be subject to two limitations: its low voltage and high current output capability is limited. When outputting low voltage and high current, its ripple noise suppression capability is also limited. However, as pointed out earlier, in any low frequency switching power supply that does not pursue high power density, there are no insurmountable limitations in these aspects.

At present, whether it is a high power density DC/DC module or a power chip in a factorized power architecture, the power is mostly within hundreds of watts and the maximum output current is below 100A. Instead of pursuing high power density switching power supplies, for example, frame-type complete switching power supply products with an operating frequency below 100-200KHz and a power density of about 6-10W/in?. Its output power can reach more than several kilowatts, and the output current can easily reach hundreds of amperes, or even greater. If you want to manufacture a DC power supply with an output current of hundreds of amperes or thousands of amperes and a ripple factor of 0.05%-0.01%, there will be no difficulties in practice or theory. What does this show? This shows that in the process of continuously improving power density, there has always been a limiting factor of outputting low voltage and high current capabilities. As for increasing power and current by connecting modules in series and parallel, that is another issue. In fact, it is impossible to solve it completely through series and parallel. In an n+1 redundant system, a power supply with higher current and higher power is formed by connecting modules in series and parallel, but its overall power density is much lower than that of each module. The greater the power of a single power module, the more the power density of the entire machine decreases.

Similarly, as mentioned above, the ripple noise of linear regulated power supplies and SCR phase-controlled power supplies, or frame-type complete switching power supplies that do not pursue high power density and output hundreds of amperes, can easily reach below 0.2%-0.05%. However, whether it is a high-power density high-frequency DC/DC module or a power chip in a factorized power architecture, its ripple noise can even reach more than 5-10% when its maximum output current is 80-100A. What does this show? This shows that in the process of continuously improving the output high current capability, high-power density high-frequency DC/DC modules and chips have always had a limiting factor in ripple and noise levels. As a product, it is not so easy to solve the problem with external capacitors, and the larger the output current, the more so.

Therefore, although the power density and efficiency of high power density high frequency DC/DC or power chip in factorized power architecture are so high, it is at the expense of certain performance. Therefore, it cannot completely replace those larger and less efficient linear regulated power supplies or lower operating frequency switching power supplies.

4. Brief analysis of the reasons for these two restrictions

In this article, I will first briefly analyze the causes of the above-mentioned deficiencies or defects that have occurred in the development process of high power density and high frequency DC/DC power supplies, and put forward constructive suggestions on some of the problems. If necessary, I will discuss it in detail in subsequent articles.

The main reasons why its low-voltage and high-current output capability is limited are:

First of all, the switching power supply that adopts the inversion method to realize the DC-AC-DC conversion has the problem of reverse direction conversion as its principle defect, which is the same as the commutation of DC motor in theory. If one day, we invent a new DC/DC conversion method that does not have the problem of reverse direction conversion, then this kind of principle defect will be gone. At present, although we can overcome it to a certain extent, it is impossible to completely solve the restrictive factors brought about by the principle defect. This defect will inevitably limit the further improvement of the output current capacity and power density of the DC/DC converter. As far as the current situation is concerned, high power density and high frequency DC/DC module products, or power chips in the factorized power architecture, may not have much valuable development possibilities at such a high power density. Of course, we can further study how much this possibility is. In future articles, I am also going to discuss the estimation of this issue in more detail.

Secondly, for the above reasons, if the high current output capability is to be further improved, the electromagnetic environment will be further deteriorated, and the ripple and noise will increase. In such a small space, there is no room to install a suitable filter. In order to maintain the most basic DC output quality, manufacturers must either reduce the power density or reduce the current output. This issue will be discussed later.

Third, in the "microelectronics technology" era we are in, the problem of room-temperature superconducting materials may not be solved for quite a long time.

The main reasons why its ripple and noise suppression capabilities are limited at low voltage and high current output are:

In the DC/DC conversion method, reverse commutation takes time, and the greater the output current, the more time it takes. This is the root cause of the large ripple of high-power density and high-frequency DC/DC modules, or power chips in the factorized power architecture, when low-voltage and high-current output. In other words, if commutation does not take time, then the output voltage waveform of the DC/DC converter will theoretically be a smooth straight line without ripple and noise. We know that in any DC/DC converter, there are actually ripples and noise caused by commutation with a base frequency of twice the operating frequency. Especially for large current commutation, whether it is zero voltage, zero current switching, or any other circuit topology, this kind of principled double frequency ripple and noise cannot be eliminated. The larger the output current, the greater the ripple noise, and the only hope is filtering.

However, when the output is low voltage and high current, the load resistance is extremely low, close to short circuit. Only inductive filters can effectively remove ripples. The effect of parallel capacitor filtering is not obvious. The inductive filter is an energy storage element. Its volume is proportional to the square of the current (output current) passing through it. Therefore, when manufacturing a high current output switching power supply with extremely low ripple coefficient, the filter with large inductance required will have a large volume, which will be much larger than the power transformer. It cannot be installed in a high-frequency DC/DC module with extremely high power density or a power chip. This is the main reason why its ripple and noise suppression ability is limited when the output is low voltage and high current. In other words, if you want to manufacture a switching power supply with extremely large output current and extremely low ripple coefficient, you cannot have an extremely high power density. On the contrary, if you want to manufacture a high-frequency DC/DC module or power chip with extremely high power density, you cannot have an extremely low ripple coefficient. In the power chip of the factorized power architecture and the high power density high-frequency DC/DC module, due to volume limitations, it is impossible to install enough inductive filters, or inductive filtering is not considered, so it cannot achieve a very low ripple coefficient.

Based on the above discussion, I believe that in the development of miniaturization of DC-DC switching power supplies, the problem of reverse conversion is a fundamental defect. It is the main reason why the output current capacity of high-power density and high-frequency DC/DC switching power supplies or power chips in factorized power architecture is limited, and it is unlikely to have a step-by-step development. The slow development of miniaturization of inductive filters is the main reason why its ripple noise suppression ability is limited. In other words, the development of various components of the power supply is unbalanced. For example, power switches, power diodes, transformers, control integrated circuits, etc. are developing very fast, but energy storage components such as capacitors and inductors are developing slowly. Therefore, it can also be said that this unbalanced development is the ultimate reason for the unbalanced development of power density and ripple noise suppression ability in the development of miniaturization of DC-DC switching power supplies. It is impossible to solve all problems by simply increasing the frequency.

If we can come up with satisfactory solutions to these problems of high power density and high frequency DC-DC switching power supply: it not only has extremely high power density and efficiency, but also has strong high current output capability. Moreover, under any load, the quality of output voltage is also extremely high, and the ripple noise is extremely low. Then, the two basic goals that people hope to achieve with DC power supply can be fully realized. Only in this way can high frequency power supply modules or power chips completely replace those DC-DC switching power supplies with larger volume and lower efficiency.

V. Improvement and Response Measures

Due to the principle limitation of switching power supply such as "commutation problem" and the limitation of our era of "microelectronic technology", when the power density and frequency reach a certain level, the loss will be intolerable. Therefore, the maximum output current of 80-100A of soft-switching high-frequency DC/DC modules with power density of 120-180W/in3 and power chips with power density of 1000W/in3 and switching frequency of 3.5 megacycles is unlikely to be greatly improved. In other words, BCM and VTM may be close to the final results of our development stage.

However, under current technical conditions, we still have the potential to significantly improve the ripple noise suppression capabilities of the above modules and chips. Can we think that in terms of the most basic indicators of power supply, a high-frequency DC-DC module or power chip with such indicators may be the final achievement of our development stage: they have a power density of 180-1000W/in3, an output current of more than 100A and a power of hundreds of watts to 1 kilowatt. At the same time, under all loads, they have a ripple factor of less than 0.1-0.05%. Moreover, according to needs, basically no external components are needed, and power supplies of any output voltage, current and power, various specifications, uses and indicators can be formed through series and parallel connection and feedback.

The final achievement of this development stage depends on the balanced development of all components of the power supply and the possibility of miniaturization of high-current inductor filters. For example, a high-current output inductive filter NIF with an extremely high inductance/volume ratio L/V is connected to the output end of a 100A power chip VTM. Its inductance L is sufficient to make the ripple factor of the VTM lower than 0.1-0.05% at full load, while its volume is only equivalent to the 3.5MHz power transformer in the VTM. Then, the power chip equipped with this NIF not only has an extremely high power density, but also has an extremely low ripple factor.

However, we also know from the history of power supply development that the research on output filters, especially the miniaturization of high-current inductive filters, is very insufficient: traditional inductive filters cannot meet the requirements, they are large in size, and the inductance/volume ratio L/V is extremely low. The mathematical and physical model of power supply technology theory established decades ago tells us that in the case of high current output, the volume of the output inductor filter occupies the largest part of the power supply. The larger the output current, the larger the volume ratio, and the lower the L/V. If we can make breakthrough progress in this regard, have new research results in basic theory, and apply them to high-power density power products. Then, we may be able to obtain satisfactory indicators in terms of power density, high current output, and very low ripple coefficient.

In view of the above reasons, I am very interested in the research of miniaturization of inductive filters with large current output, and have achieved practical results. This article will announce the advent of a large current output inductive filter NIF with a very high inductance/volume ratio L/V, which will satisfactorily solve the above problems. I will discuss the more detailed situation of NIF in future articles.

6. A high current output inductive filter NIF with extremely high inductance/volume ratio L/V

We know that the traditional inductive filter is an energy storage element, and its volume is proportional to the square of its output current, that is, V=kI2, that is, its volume is proportional to the energy it stores. This is fundamentally different from the power transformer. For example, the volume of the output inductive filter that is sufficient to make the ripple factor of the 100A power chip VTM lower than 0.1-0.05% will be much larger than the volume of the 3.5MHz, 100A power transformer in the VTM.

Can we try to change the electrical and physical properties of components such as inductive filters so that they can easily meet the size restrictions imposed by the power system?

After several years of research, we have developed a new type of high-current output inductive filter NIF. Its distinctive feature is that NIF is not an energy storage element, which is the most important and essential change. Therefore, its volume is not proportional to the square of the output current, but only proportional to the output current, that is, V=kI. This means that under the conditions of the same inductance L and rated output current I, the volume ratio of NIF to the traditional output inductive filter is inversely proportional to the output current. That is:

(Vn/V) = h (1/I) ﹤ ﹤ 1

Where: Vn, V, h, I are the volume of NIF, the volume of the traditional output inductor filter, the proportional coefficient, and the output current. The inductance, operating frequency, and rated output current of the two are the same.

This means that, compared with the traditional output inductive filter, the size of the NIF will be greatly reduced, and the larger the current, the greater the reduction.

The volume of NIF is inversely proportional to the operating frequency, and directly proportional to the inductance L and the output current I, so it has a very high inductance/volume ratio L/V. This high-quality feature is due to a novel idea and unique design method. For example, a NIF suitable for a 3.5MHz operating frequency and an output current of 100A, its inductance is enough to make the output ripple factor of a 100A VTM power chip less than 0.1%-0.05%, but its volume is only equivalent to the volume of the power transformer in this power chip VTM. If we integrate it in the VTM, because it is very small, it will not reduce the power density of the power chip too much.

The extremely high inductance/volume ratio L/V of NIF enables balanced development of all components of the power supply, which will enable high power density and high frequency DC/DC switching power supply modules or power chips to achieve very low ripple factors under all load conditions, thus achieving the two basic goals that switching power supplies have been striving to achieve for decades.

It is also important to mention that since NIF is not an energy storage element, it is not a limiting factor in improving response speed, which is also one of the unparalleled advantages of NIF.

7. About Response Speed

Finally, it is necessary to analyze the relationship between the ripple coefficient index and the response speed of the power supply. We know that there is a contradiction between the dynamic index and the static index of a control system. For example, an ideal inductive voltage divider (transformer) or a resistive voltage divider has an extremely fast response speed. However, their harmonic suppression capability is equal to zero. Similarly, the transformer in the power chip VTM is close to an ideal transformer, so it has a very fast response speed, but its ripple suppression capability is not strong. As for whether to mainly meet the static index or the dynamic index, this should be determined by comprehensive consideration based on the needs of the user. For example, first design the filter according to the requirements to achieve a ripple coefficient of less than 0.05% when the output is 100A. At this time, the response speed may not be enough. We can use appropriate closed-loop control. If the control system has sufficient open-loop gain and a reasonable open-loop frequency characteristic function, it can generally achieve the predetermined closed-loop response speed.

8. Outlook on the development of switching power supplies

In terms of high frequency: there is definitely a limit to the increase in frequency. Basic circuit theory tells us that the time of a cycle should be much longer than the action time of a switch, otherwise the handling of the transition process will become increasingly difficult, and the release of a switch's energy and the transfer of energy (or charge) in the energy storage element in the circuit all require time, and the greater the energy, the longer it takes. In addition, there are more problems such as the increasing influence of parasitic parameters on high-frequency work and the increasing complexity of control circuits. Energy cannot change suddenly, and we cannot "make a perpetual motion machine."

In terms of miniaturization: Soft-switching high-frequency DC/DC modules with a power density of 120-180W/in3 are still the best mainstream products for module power supplies in the world today. The power density of the power chip in the newly emerged factored power architecture has even reached 1000W/in3, 3.5MHz. It is much more difficult to increase its operating frequency again than from 20KHz to hundreds of KHz. This seems to indicate that under the current conditions of microelectronics technology, it may be close to the limit of frequency use. If it exceeds this range, the difficulty of power supply manufacturing will increase significantly, and there may be problems with whether it is reasonable. In addition, due to the uneven development of various technologies within the DC power supply system, the fastest developing one is rectifier technology, while the distribution technology is relatively slow. Taking the communication power supply system as an example, the power density of the rectifier, the core component of the primary power supply, has been continuously improved, which has promoted the continuous improvement of the power density of the communication DC power supply. However, since the density of distribution devices and batteries has basically remained stable, this has also restricted the improvement rate of the power density of the whole system to a certain extent.

In other aspects such as components, control technology, manufacturing processes, integration technology, etc.: Under the current conditions of electronic technology, except for new material properties that have not yet been discovered, such as room-temperature superconductivity, we have successfully solved many problems: such as power semiconductor devices, materials for high-frequency magnetic components, power transformers, new capacitors and inductors, resonance technology and soft switching, synchronous rectification technology, distributed power supply structure, PFC converter, full digital control, electromagnetic compatibility, design and testing technology, integration of control systems, etc.

Based on this, some scholars believe that: according to the "theory of creative problem solving", a theory that describes the evolutionary laws of technological system development, generally speaking, the life cycle of technology includes four stages: infancy, growth, maturity and decline. Various signs indicate that the core technology of DC power supply - switching power supply technology has basically begun to enter the maturity stage: efficiency improvement has become slow and difficult, and the power loss cannot be greatly reduced, which limits the further improvement of power density... In the next few years or even more than ten years, DC power supply products will enter a stage of slow development until one day, a new power conversion technology appears, and DC power supply products will see another step-by-step development, just like the switching voltage regulation technology replacing the linear voltage regulation technology, which has brought revolutionary changes to the power supply.

I think this inference is roughly correct. But what can we do before this new power conversion technology appears?

At present, the standards and specifications of power supply manufacturing are varied, which is the product of the backward technology era, which is not conducive to the advancement of technology and the use of users. Today, when power supply technology has entered maturity, we should strive to unify the manufacturing standards of power supply into an advanced model.

I highly appreciate and support the flexible power supply structure of the factorized power architecture. Based on this chip-based idea, we can make corresponding functional (power) chips according to the various functional components that make up the power supply. In this way, we can basically use no external components, and through series, parallel and feedback, we can form a power supply with any output voltage, current and power, various specifications and indicators.

These functional (power) chips are the most basic units that make up the power supply. They should be of the best quality and do not require any external components. The manufacturing of chips is standardized and has various specifications.

This may be the best option for us at this stage of development.

The above article only raises the question. In the following articles, I will discuss and evaluate more specific issues that must be involved in a comprehensive and detailed manner, and try to propose ideas and methods to solve these problems. At the same time, I also welcome colleagues in the power supply industry to participate in our discussion and comments.