# What are the main considerations when choosing a topology? Summarize how to choose a topology

An important factor that determines topology selection is the input voltage and output/input ratio. Figure 1 shows the relatively suitable voltage range for commonly used isolated topologies. Topology selection is also related to output power, output voltage channels, output voltage adjustment range, etc. Under normal circumstances, you can apply multiple topologies for a given occasion. It is impossible to say that a certain topology is suitable for a certain application, because the product design also has the designer’s experience of a certain topology, whether the components are easily available, and the cost Requirements, requirements for technical personnel, debugging equipment and personnel quality, production process equipment, batch size, military or civilian products, etc. are related to factors.So choose a good one

01. Summary

An important factor that determines topology selection is the input voltage and output/input ratio. Figure 1 shows the relatively suitable voltage range for commonly used isolated topologies. Topology selection is also related to output power, output voltage channels, output voltage adjustment range, etc. Under normal circumstances, you can apply multiple topologies for a given occasion. It is impossible to say that a certain topology is suitable for a certain application, because the product design also has the designer’s experience of a certain topology, whether the components are easily available, and the cost Requirements, requirements for technical personnel, debugging equipment and personnel quality, production process equipment, batch size, military or civilian products, etc. are related to factors. Therefore, to choose a good topology, you must be familiar with the strengths and weaknesses of each topology and the application field of the topology. If you choose a topology arbitrarily, you may announce the failure of the new power supply design from the beginning.

Figure 1: Application voltage range of various isolation topologies

02, input and output

If the output and input share a common ground, a non-isolated Buck, Boost common ground converter can be used. These circuits have a simple structure and few components. If the input voltage is very high, generally the output needs to be isolated from the input for safety reasons.

Before choosing a topology, you should first know whether the output voltage is higher or lower than the input voltage within the input voltage range? For example, Buck converters can only be used where the output voltage is lower than the input voltage, so the output voltage should be lower than the input voltage at all times. If you require 24V input and 15V output, Buck topology can be used; but if the input 24V is from 8V to 80V, you cannot use Buck converters because Buck converters cannot convert 8V to 15V. If the output voltage is always higher than the input voltage, the Boost topology must be used.

If the ratio of the output voltage to the input voltage is too large (or too small), there is a limit. For example, if the input voltage is 400V, and the output voltage is required to be 48V or a Buck converter is used, the voltage ratio is too large. Although the output voltage is always lower than the input voltage, it is so large Although the voltage ratio does not exceed the small duty cycle range of the control chip, it limits the switching frequency. Moreover, the peak current of the power device is large, and the selection of the power device is difficult. If an isolated topology is used, the appropriate duty cycle can be adjusted through the turns ratio. Achieve a better price-performance ratio.

03. Practical limitations of switching frequency and duty cycle

1) Switching frequency

When designing the converter, first choose the switching frequency. The main purpose of increasing the frequency is to reduce the size and weight of the power supply. However, it is the magnetic component that accounts for the large volume and weight of the power supply. Magnetic components in modern switching power supplies account for the volume (20% to 30%), weight (30% to 40%), and loss of 20% to 30% of the switching power supply. According to the law of electromagnetic induction:

Where:

U-voltage applied by the transformer;
N-the number of turns of the coil;
A-the cross-sectional area of ​​the magnetic core;
ΔB-the amount of change in magnetic flux density;
f-transformer operating frequency.

At lower frequencies, ΔB is limited by the saturation of the magnetic material. It can be seen from the above formula that when U is constant, the volume of the magnetic core should be reduced, and the product of the number of turns and the cross-sectional area of ​​the magnetic core is inversely proportional to the frequency. Increasing the frequency is the main measure to reduce the volume of the power supply. This is the main research topic of countless scientific and technological workers since the advent of switching power supply.

But can the frequency of the switching power supply be increased without limitation? No. There are mainly two limiting factors: It is the loss of magnetic materials. Ferrite is generally used at high frequency, and its unit volume loss is expressed as:

In the formula, η-coefficient of different materials; f-working frequency; Bm-working magnetic induction amplitude. α and β are the frequency greater than 1 and the magnetic induction loss index, respectively. Generally α=1.2～1.7; β=2～2.7. Increase the frequency and increase the loss. In order to reduce the loss, at high frequency, reduce the magnetic induction Bm so that the loss is not too large, which violates the purpose of reducing the volume. Otherwise, the loss is too large and the efficiency is reduced. Furthermore, the greater the processing power of the magnetic core, the larger the volume, the worse the heat dissipation conditions, and the high-power magnetic core also limits the switching frequency.

Figure 2: Buck converter power tube current and voltage waveforms

Secondly, the switching loss of power devices is limited. Take the Buck converter as an example to illustrate the switching loss. Figure 2 is a typical current and voltage waveform diagram of the power tube of a continuous Buck converter. It can be seen that when the transistor is turned on, the collector voltage starts to drop when the collector current rises to a large value. When it is turned off, the collector voltage first rises to a large collector current before it starts to drop. It is assumed that the voltage and current rise and fall are linear.The switching loss can be obtained as

Where tr = tri trv-the sum of the current rise time and the voltage fall time when it is turned on; td = tdi tdv-the sum of the voltage rise time and the current fall time when it is turned off. General tr td

If the current is intermittent, only the breaking loss is concerned, and the switching loss is:

It can be seen that switching loss is proportional to frequency and switching time. Discontinuity seems to be half the loss of continuous switching, but it should be noted that at the same output power, the power tube current is at least twice that of continuous current. In addition to the increase in the device current rating and the increase in cost, the conduction voltage drop loss also increases. . The filter Inductor core works in the state of a forward transformer, and the high-frequency loss of the core and coil will also be greatly increased. Although the switching loss can be reduced through soft switching technology, please note that soft switching always uses LC resonance. The resonance current (or voltage) is very large. The resonance current passes through the transistor, the inductor L and the capacitor C. These components are also lossy. . Sometimes the efficiency is only improved by 1 to 2%, but the circuit is complicated, the number of components increases, the cost increases, and sometimes the gain is not worth the loss.

At present, the power of MOSFET switching power supply is below 5kW, and the operating frequency is generally below 200kHz. BJT is up to 50kHz. The high 30kHz of the IGBT is used for 3kW or more. With MOSFET and IGBT (BJT) combination, the tube height does not exceed 100kHz. The conversion power is tens of watts, and of course the operating frequency can be increased.

In addition, the greater the conversion power, the greater the current and voltage. If the current rise and fall rates of the high-power tube and the low-power tube are the same, the high-power tube needs a longer switching time. What’s more, the chip area of ​​high-power devices is large, and the switching time is also increased in order to avoid current concentration to reduce the current rise and fall rate during switching. It can be seen that the greater the conversion power, the lower the allowable switching frequency.

If you heard that his switching power supply has a working frequency of up to several MHz, you have to ask how much his conversion power is?

2) Duty cycle

The conversion ratio (ratio of output voltage to input voltage) of a switching converter is too large or too small, there are limits. First, the duty cycle of the converter (the ratio of the on-time of the switch to the switching period) is limited by the large and small values ​​of the control chip. In some topologies, the duty cycle cannot be greater than 0.5. In short, general-purpose PWM control IC chips usually do not guarantee that the duty cycle can be greater than 0.85; some chips do not guarantee that the duty cycle can be below 0.05 at a reasonable operating frequency to quickly drive the gate of the MOSFET with a small loss.

For example, if the switching frequency is 250kHz and the period is 4?s, if the duty cycle is 0.1, the on-time of the MOSFET is only 0.4?s. If the on-time of the MOSFET is 0.1?s, the off-time is also 0.1?s, Almost most of the on-time is “eaten” by the transition time, and the loss increases. This is one of the reasons why the higher the conversion power, the lower the operating frequency.

Regardless of control ICs and high-current gate drives, etc., as long as the duty cycle is not designed to be smaller than 0.1 and larger than 0.8 (0.45 for the 0.5 limit converter), there is no need to worry.

If the topology used has a transformer, the transformation ratio can adjust the duty cycle. But the transformation ratio also has limitations. If the transformation ratio is too large or too small, the size of the primary and secondary wires will differ too much, and the winding of the coil will be difficult. Generally, the primary to secondary turns ratio is as large as 10:1 and as small as 1:10. If you need to obtain a high voltage from a very low voltage, do you consider using a two-stage converter or a secondary voltage doubler circuit to increase the voltage.

04. How many outputs?

The question next to the duty cycle is how much output. For example, if it is not 1 output, Buck is not suitable. In some cases, you can add a subsequent regulator to get another voltage. A practical example is to use a Buck converter to generate a 5V output, and then a linear regulator (or another switch) generates a 3.3V output from the 5V input. However, the relevant transient, noise, and loss should meet the requirements.

In the worst case, design multiple independent converters instead of using complex magnetic components with many coils. Before starting the design, you have to consider that if you use a multi-output converter, you may save a few dollars of control IC, but you may spend dozens of dollars to make that complicated multi-coil magnetic component. Before designing, you should first weigh the magnetic components, circuit components, and additional costs, and don’t just talk about the facts.

05, isolation

Before designing, it is necessary to know in advance whether the secondary and primary need to be isolated. If the input is powered by the grid or high voltage, as a commodity, there are safety regulations (and EMI issues) that require isolation. A typical example is that the input and output have 500V AC withstand voltage requirements. After you know the security requirements, some topologies, such as Buck, Boost, etc. without isolation, will be excluded.

06, EMI

Think of EMI at the beginning of the design. Don’t wait until the design is complete before considering EMI. Some topologies may have many successfully avoided EMI problems. If it is a non-isolated system, because the third wire is not involved in the system, such as using battery power alone, there will be no common mode noise, which makes it easy for you to filter.

In addition, some topologies are more noisy than others. The difference is that some topologies are disconnected from the input for part of each cycle, causing the input current to be interrupted. If the input current is continuous, there will be no steep rising and falling edges, the current will not be zero, and it will be easy to filter.

A Buck converter is an example of intermittent input current, because when the switch is open, the input current is zero. The inductance of the Boost converter is always connected to the input loop, but whether the input current is continuous depends on whether the Boost works intermittently or continuously.

The author recommends that high-power power supplies do not use topologies with intermittent input current, because those topologies usually require expensive magnetic components.

07. BJT, MOSFET or IGBT?

Topology selection is related to the power devices that can be used. Currently available power devices include bipolar (BJT) power tubes, MOSFETs and IGBTs. The voltage rating of the bipolar tube can exceed 1.5kV, commonly used below 1kV, the current is from a few mA to hundreds of A; MOSFET is below 1kV, commonly used below 500V, and the current is several A to hundreds of A; the IGBT voltage rating is above 500V, which can be Up to several kV, current tens of A to several kA.

Different devices have different driving requirements: bipolar transistors are current driven, and high-power high-voltage tubes have low current gains, and are often used in single-switch topologies. In the low to medium power range, 90% of them choose MOSFETs except for special reasons.

One of the reasons is cost. If the output of the product is large, the bipolar tube is still cheaper than the MOSFET. However, the use of a bipolar power tube means that the switching frequency is lower than that of a MOSFET, so the volume of the magnetic element is relatively large. Is this still cost-effective? You have to study the cost carefully.

When the input voltage is high (380V), or push-pull topology and transient voltage requires more than double the voltage, you may feel embarrassed to choose a power tube. If you use a bipolar tube, you can buy a 1500V bipolar tube. MOSFET can be purchased with a high voltage of 1000V, and the on-resistance is larger than that of BJT. Of course, you may consider using IGBTs. Unfortunately, although the IGBT drive is like a MOSFET, its switching speed is similar to that of a bipolar tube, which has serious tailing problems.

It can be seen that the low voltage (500V) is basically the world of MOSFET, and the switching frequency of low power (hundreds of watts) is hundreds of kHz. The IGBT rating is generally above 500V, and the current is more than tens of A. It is mainly used for speed regulation and basically replaces the high-voltage Darlington bipolar tube. The operating frequency is as high as 30kHz, usually around 20kHz. Because of the large turn-on voltage drop, it is not used below 100V.

Figure 3: Improve power switching frequency (a) IGBT and MOSFET in parallel (b) BJT and MOSFET in series

In order to increase the switching speed of IGBT or BJT, MOSFET and BJT or IGBT can also be combined into a composite tube. In Figure 3(b), the U(BR) CBO/70A BJT is connected in series with the 50V/60A MOSFET, which is used in the three-phase 380V rectifier inductor filter input (510V) double-ended forward 3kW communication power supply. The power MOSFET is first driven when it is turned on. At this time, the BJT works in a common base configuration, and the emitter input current, or the drain voltage of the MOSFET drops, the BJT emitter junction is forward biased, and the base current is generated, which leads to the collector current. A positive feedback is formed by a proportional drive circuit, so that the BJT is saturated and turned on. When turning off, the MOSFET is turned off first, and the emitter junction is reverse biased, making the BJT turn off quickly. Common base frequency characteristic is β times of common emitter. Improved shutdown speed. The low-voltage MOSFET on-resistance is only in the order of mΩ, and the conduction loss is very small. The actual circuit operating frequency is 50kHz.

Parallel connection of MOSFET and IGBT also utilizes the switching characteristics of MOSFET. To achieve this goal, the drive of MOSFET and IGBT should be designed like this: When turning on, the PWM signal can drive the MOSFET to turn on at the same time or first, and then turn on the IGBT. The IGBT turns on at zero voltage. When turning off, the IGBT is turned off first, and the IGBT is turned off at zero voltage; the MOSFET is turned off after a certain delay. MOSFET bears the switching loss; during the turn-on period, the high-voltage MOSFET conduction voltage drop is greater than that of the IGBT, and most of the current flows through the IGBT, allowing the IGBT to bear the conduction loss. A practical example of this combination operates at a frequency of 50kHz and a 3kW half-bridge topology.

08, continuous or intermittent

Inductance (including flyback transformer) and current (ampere turns) are continuous or intermittent: In a discontinuous mode converter, the inductor current is zero at certain moments in the cycle. Current (ampere-turn) continuous is to have enough inductance to maintain a small load current ILmin (including dummy load), and there should be current flowing in the inductance at any time of the cycle.which is

Among them, T-switching period; D=Ton/T-duty ratio; Ton-transistor on time. We assume that the forward voltage drop of the rectifier is small compared to the output voltage. If the small load current is zero, you must enter the discontinuous mode.

In the actual power supply design, the general power supply has no-load requirements, and the inductance is not allowed to be too large. It must be intermittent at light load. In this case, a dummy load is sometimes set, and the dummy load is disconnected when the load current exceeds. Otherwise, it may cause the stability problem of the closed-loop control, and the feedback compensation network should be carefully designed.

Synchronous rectification is an exception. Synchronous rectification in converter applications is always in continuous mode, and there is no requirement for small inductance.

09. Synchronous rectification

In many low output voltage applications today, converter efficiency is (almost) more important than cost. From the user’s point of view, the more expensive but high-efficiency converter is actually cheaper. If the power efficiency of a computer is low, the real computing time is often very small, and the standby time is very long, which will cost more electricity.

If efficiency is important, consider using synchronous rectification technology. That is, MOSFET is used for output rectification. Many IC driver chips available today can drive both FETs and synchronous rectifiers.

Another reason for using synchronous rectification is that it converts a converter that works in discontinuous current mode to a continuous current mode. This is because even if there is no load, current can flow in both directions (because the MOSFET can be turned on in both directions). The use of synchronous rectification relieves your worries about mode changes (mode changes may cause instability of the converter) and guarantees continuous small inductance requirements.

Figure 4 (a): Diode rectifier converter and (b): Synchronous rectifier converter

A problem with synchronous rectification is worth mentioning here. The main switch tube is turned off before the synchronous rectification is turned on, and vice versa. If this treatment is neglected, a punch-through phenomenon will occur, that is, the input (or output) voltage will be directly short-circuited to the ground, which will cause high losses and may lead to failure. During the off time of the two MOSFETs, the inductor current is still flowing. Normally, the MOSFET body diode should not flow current, because this diode has a long recovery time. If it is assumed that current flows through the body diode when the MOSFET is turned off, when the body diode recovers, it acts as a short circuit in reverse recovery. Therefore, once the input (or output) path to ground occurs, the converter may fail, as shown in Figure 4. (B) Shown.

To solve this problem, a Schottky diode can be used in parallel with the body diode of the MOSFET to allow it to flow current when the FET is off. (Because the forward voltage drop of Schottky is lower than that of body diode, Schottky almost flows all current, and the reverse recovery time of body diode is related to the forward current before turning off, so it can be ignored at this time)

10. Voltage type and current type control

Switching power supply design should consider in advance whether to use voltage-type or current-type control, which is a control problem. Almost every topology can use either. Current-type control can limit the current cycle by cycle, and overcurrent protection becomes easy to implement. At the same time, push-pull or full-bridge converters can overcome the magnetic deviation of the output transformer. But if the current is very large, the current type requires a detection resistor (which consumes a lot of power) or a transformer (which costs a lot of money) to detect the current, which may affect your choice. However, such over-current protection detection is a smooth ride. However, if you use the current control type for a half-bridge converter, it may cause the voltage divider capacitor voltage to be unbalanced. So for high-power output, you should consider which one is better.

11. Conclusion

Fortunately, before you design a power supply, you should know in advance the system your power supply will work on. Learn more about the power requirements and limitations of this system. A thorough understanding of the system can greatly reduce costs and reduce design time.

In actual operation, you can make a list from the specifications required by the converter and consider them one by one. You will find that the topologies you can choose are limited to one or two based on these specifications, and it is easy to choose topologies based on cost and size. Under normal circumstances, the topology can be selected based on the above considerations:

① Step-up or step-down: Is the output voltage always higher or lower than the input voltage? If not, you cannot use Buck or Buck/Boost.

② Duty cycle: Is the ratio of output voltage to input voltage greater than 5? If so, you may need a transformer. Calculate the duty cycle to ensure that it is not too large or too small.

③ How many sets of output voltages are needed? If it is greater than 1, unless a follow-up regulator is added, a transformer is generally required. If there are too many output groups, it is recommended to use several converters. ④ Is isolation required? How much voltage? Isolation requires a transformer.

⑤ What are the EMI requirements? If the requirements are strict, it is recommended not to use a topology with intermittent input current like Buck, but to choose a continuous current working mode.

⑥ Is cost extremely important? BJT can be selected for low power and high voltage. If the input voltage is higher than 500V, consider choosing IGBT. On the contrary, use MOSFET.

⑦ Is the power supply no-load required? If required, choose intermittent mode, unless question 8 is used. Dummy load can also be added.

⑧ Can synchronous rectification be used? This allows the converter current to be continuous, independent of the load.

⑨ Is the output current large? If it is, the voltage type should be used instead of the current type.