Key Designs of Switching Power Supplies that Engineers Must Know (Part 4)

1. Analysis of electromagnetic compatibility technology of switching power supply

1 Introduction

Electromagnetic compatibility is an emerging interdisciplinary and comprehensive applied discipline. As an edge technology, it is based on the basic theories of electrical and radio technology and involves many new technical fields, such as microwave technology, microelectronics technology, computer technology, communication and network technology, and new materials. The application scope of electromagnetic compatibility technology is very wide. Almost all modern industrial fields, such as electricity, communication, transportation, aerospace, military industry, computer and medical treatment, must solve electromagnetic compatibility problems. The hot topics of its research mainly include: the characteristics of electromagnetic interference sources and their transmission characteristics, the harmful effects of electromagnetic interference, the suppression technology of electromagnetic interference, the utilization and management of electromagnetic spectrum, electromagnetic compatibility standards and specifications, the measurement and test technology of electromagnetic compatibility, electromagnetic leakage and electrostatic discharge, etc.

The English name of electromagnetic compatibility is Electromagnetic Compatibility, or EMC for short. The so-called electromagnetic compatibility refers to the coexistence state in which devices (subsystems, systems) can perform their respective functions together in a common electromagnetic environment. This contains two meanings, namely, the electromagnetic radiation generated during its operation must be limited to a certain level, and it must have a certain anti-interference ability. This is the compatibility problem that must be solved in the development of equipment. The frequency range involved in electromagnetic compatibility technology is as wide as 0 GHz ~400GHz. In addition to traditional equipment, the research objects also involve chip level, all kinds of ships, space shuttles, intercontinental missiles and even the electromagnetic environment of the entire earth.

The three elements of electromagnetic compatibility are interference source (disturbance source), coupling path and sensitive body. Cutting off any of the above can solve the electromagnetic compatibility problem. The commonly used methods to solve electromagnetic compatibility are shielding, grounding and filtering.

2 Electromagnetic compatibility technical terms

(1) Electromagnetic compatibility

Electromagnetic compatibility refers to the ability of a device or system to work normally in its electromagnetic environment without causing unbearable electromagnetic interference to anything in the environment.

(2) Electromagnetic disturbance

Electromagnetic disturbance refers to any electromagnetic phenomenon that may cause the performance of equipment, equipment or systems to degrade, or cause damage to living or inanimate matter. Electromagnetic disturbance can cause the performance of equipment, transmission channels or systems to degrade. Its main elements include natural and man-made disturbance sources, coupling through common ground impedance/internal resistance, electromagnetic disturbance conducted along power lines, and radiated interference. The paths through which electronic systems are disturbed are: through the power supply, through signal lines or control cables, field penetration, and direct entry through antennas; through cable coupling, conducted interference from other devices; internal field coupling of electronic systems; radiated interference from other devices; external coupling of electronic equipment to internal fields; broadband transmitter antenna systems; external environmental fields, etc.

(3) Electromagnetic environment

The electromagnetic environment is a time-varying electromagnetic phenomenon that obviously does not transmit information, but may be superimposed or combined with useful signals.

(4) Electromagnetic radiation

Electromagnetic radiation refers to the phenomenon of electromagnetic waves being emitted from a source into space. The meaning of the term "electromagnetic radiation" is sometimes extended to include electromagnetic induction. RFI/EMI can be radiated through any openings, vents, ports, cables, measurement holes, door frames, hatches, drawers and panels in the housing of any equipment, as well as non-ideal connection surfaces of the housing. RFI/EMI can also be radiated by wires and cables entering sensitive equipment. Any good radiator of electromagnetic energy can also be a good receiver.

(5) Pulse

A pulse is a physical quantity that changes suddenly in a short period of time and then quickly returns to its initial value.

(6) Common mode interference and differential mode interference

There are two types of interference on the power line: common mode interference and differential mode interference. Common mode interference exists between any phase of the power supply and the earth or between the wire and the earth. Common mode interference is sometimes called longitudinal mode interference, asymmetric interference or ground interference. This is interference between the current-carrying conductor and the earth. Differential mode interference exists between the power phase line and the neutral line and between the phase lines. Differential mode interference is also called normal mode interference, transverse mode interference or symmetrical interference. This is interference between current-carrying conductors. Common mode interference indicates that the interference is coupled into the circuit by radiation or crosstalk, while differential mode interference indicates that the interference comes from the same power circuit. Usually these two types of interference exist at the same time. Due to the imbalance of line impedance, the two types of interference will also transform into each other during transmission, so the situation is very complicated. After the interference is transmitted over a long distance, the attenuation of the differential mode component is greater than that of the common mode, because the impedance between the lines is different from the line-to-ground impedance. For the same reason, common mode interference will also radiate to the adjacent space during line transmission, while differential mode will not, so common mode interference is more likely to cause electromagnetic interference than differential mode. Different interference modes require different interference suppression methods to be effective. A simple way to determine the interference method is to use a current probe. The current probe first surrounds each wire separately to obtain the induction value of a single wire, and then surrounds two wires (one of which is the ground wire) to detect their induction conditions. If the induction value is increased, the interference current in the line is common mode; otherwise, it is differential mode.

(7) Immunity level and sensitivity level

The immunity level refers to the maximum disturbance level when a given electromagnetic disturbance is applied to a device, equipment or system and it can still work normally and maintain the required performance level. In other words, when this level is exceeded, the performance of the device, equipment or system will be degraded. The sensitivity level refers to the level at which performance degradation just begins to occur. Therefore, for a device, equipment or system, the immunity level and the sensitivity level are the same value.

(8) Immunity margin

The immunity margin refers to the interpolation between the immunity level limit and the electromagnetic compatibility level of equipment, devices or systems.

3 Electromagnetic compatibility of switching power supplies

The reasons for the electromagnetic compatibility problems caused by the switching power supply are quite complicated because it works in the switching state of high voltage and large current. From the perspective of the electromagnetic properties of the whole machine, there are mainly common impedance coupling, line coupling, electric field coupling, magnetic field coupling and electromagnetic wave coupling. Common impedance coupling is mainly the common impedance between the disturbance source and the disturbed body in electricity, through which the disturbance signal enters the disturbed body. Line coupling is mainly the mutual coupling caused by the parallel wiring of the wires or PCB lines that generate the disturbance voltage and disturbance current. Electric field coupling is mainly due to the field coupling generated by the induced electric field on the disturbed body due to the existence of potential difference. Magnetic field coupling mainly refers to the coupling of the low-frequency magnetic field generated near the high-current pulse power line to the disturbed object. Electromagnetic field coupling is mainly due to the high-frequency electromagnetic waves generated by the pulsating voltage or current radiating outward through space, and the coupling to the corresponding disturbed body. In fact, each coupling method cannot be strictly distinguished, but the emphasis is different.

In a switching power supply, the main power switch tube works at a high voltage and in a high-frequency switching mode. The switching voltage and switching current are close to square waves. From the spectrum analysis, it is known that the square wave signal contains abundant high-order harmonics. The spectrum of the high-order harmonics can reach more than 1000 times the square wave frequency. At the same time, due to the non-ideal working state of the leakage inductance and distributed capacitance of the power transformer and the main power switch device, high-frequency and high-voltage peak harmonic oscillations are often generated when the high-frequency is turned on or off. The high-order harmonics generated by the harmonic oscillation are transmitted to the internal circuit through the distributed capacitance between the switch tube and the heat sink or radiated to the space through the heat sink and the transformer. The switching diode used for rectification and freewheeling is also an important cause of high-frequency interference. Because the rectification and freewheeling diodes work in a high-frequency switching state, the parasitic inductance of the diode leads, the existence of junction capacitance and the influence of reverse recovery current make them work at a very high voltage and current change rate and generate high-frequency oscillations. The rectification and freewheeling diodes are generally close to the power output line, and the high-frequency interference they generate is most likely to be transmitted through the DC output line. In order to improve the power factor, the switching power supply adopts active power factor correction circuit. At the same time, in order to improve the efficiency and reliability of the circuit and reduce the electrical stress of the power device, soft switching technology is widely used. Among them, zero voltage, zero current or zero voltage/zero current switching technology is the most widely used. This technology greatly reduces the electromagnetic interference generated by the switching device. However, most of the soft switching lossless absorption circuits use L and C for energy transfer and use the unidirectional conductivity of the diode to achieve unidirectional energy conversion. Therefore, the diode in the resonant circuit becomes a major source of electromagnetic interference.

Switching power supplies generally use energy storage inductors and capacitors to form L and C filter circuits to filter differential and common mode interference signals. Due to the distributed capacitance of the inductor coil, the self-resonant frequency of the inductor coil is reduced, so that a large amount of high-frequency interference signals pass through the inductor coil and propagate outward along the AC power line or DC output line. As the frequency of the interference signal increases, the lead inductance of the filter capacitor causes the capacitance and filtering effect to continue to decrease, and even causes the capacitor parameters to change, which is also a cause of electromagnetic interference.

4 Solutions to electromagnetic compatibility

From the three elements of electromagnetic compatibility, to solve the electromagnetic compatibility problem of switching power supply, we can start from three aspects: first, reduce the disturbance signal generated by the disturbance source; second, cut off the propagation path of the disturbance signal; third, enhance the anti-disturbance ability of the disturbed body. When solving the internal compatibility of the switching power supply, the above three methods can be used in combination, with the cost-effectiveness and the ease of implementation as the premise. Therefore, the external disturbance generated by the switching power supply, such as power line harmonic current, power line conduction disturbance, electromagnetic field radiation disturbance, etc., can only be solved by reducing the disturbance source. On the one hand, the design of input/output filter circuit can be enhanced, the performance of APFC circuit can be improved, the voltage and current change rate of the switch tube and the rectifier and freewheeling diode can be reduced, and various soft switching circuit topologies and control methods can be adopted; on the other hand, the shielding effect of the casing can be strengthened, the gap leakage of the casing can be improved, and good grounding treatment can be performed. For the external anti-disturbance ability (such as surge and lightning strike), the lightning protection ability of the AC input and DC output ports should be optimized. Usually, for the combined lightning waveform of 1.2/50µs open-circuit voltage and 8/20µs short-circuit current, due to the small energy, a combination of zinc oxide varistor and gas square tube is usually used to solve it. For electrostatic discharge, TVS tubes and corresponding grounding protection are usually used in the small signal circuits of communication ports and control ports, and the electrical distance between the small signal circuit and the housing is increased to solve it, or devices with anti-static interference are selected. Fast transient signals contain a wide spectrum and are easily transmitted into the control circuit in a common mode. The same method as anti-static is used to reduce the distributed capacitance of the common mode inductor and strengthen the common mode signal filtering of the input circuit (adding common mode capacitors or insertion loss type ferrite magnetic rings, etc.) to improve the anti-interference performance of the system.

To reduce the internal interference of the switching power supply, realize its own electromagnetic compatibility, and improve the stability and reliability of the switching power supply, we should start from the following aspects: ① Pay attention to the correct partition of the PCB wiring of the digital circuit and the module circuit; ② Decoupling of the power supply of the digital circuit and the analog circuit; ③ Single-point grounding of the digital circuit and the analog circuit, the single-point grounding of the large current circuit and the small current, especially the current and voltage sampling circuit to reduce the common resistance interference, reduce the influence of the ground loop, pay attention to the spacing between adjacent lines and the signal properties when wiring, avoid crosstalk, reduce the area surrounded by the output rectifier circuit and the freewheeling diode circuit and the tributary filter circuit, reduce the leakage of the transformer, the distributed capacitance of the filter inductor, and use filter capacitors with high resonant frequency.

5 Filter Structure

滤波是一种抑制传导干扰的方法。例如，在电源输入端接上滤波器，可以抑制来自电网的噪声对电源本身的侵害，也可以抑制由开关电源产生并向电网反馈的干扰。电源滤波器作为抑制电源线传导干扰的重要单元，在设备或系统的电磁兼容设计中具有极其重要的作用。它不仅可以抑制传输线上的传导干扰，同时对传输线上的辐射发射也具有显著的抑制效果。在滤波电路中，选用穿心电容、三端电容、铁氧体磁环，能够改善电路的滤波特性。进行适当的设计或选择合适的滤波器，并正确的安装滤波器是抗干扰技术的重要组成部分。在交流电输入端加装的电源滤波器电路如图1所示。图中Ld、Cd用于抑制差模噪声，一般取Ld为100 mH -700mH，Cd取1µF -10µF。Lc、Cc用于抑制共模噪声，可根据实际情况加以调整。

All power supply filters must be grounded (except those that the manufacturer specifically states are not allowed to be grounded), because the common mode bypass capacitor of the filter must be grounded to work. The general grounding method is to connect the filter housing to the grounding point of the equipment with a thicker wire in addition to connecting the filter to the metal housing. The lower the grounding impedance, the better the filtering effect.

The filter should be installed as close to the power inlet as possible. The input and output ends of the filter should be as far apart as possible to avoid direct coupling of interference signals from the input end to the output end.

If you add an output filter to the output end of the power supply, install a high-frequency capacitor, increase the inductance of the output filter inductor and the capacity of the filter capacitor, you can suppress differential mode noise. If you connect multiple capacitors in parallel, the effect will be better.

The structure of several filters is shown in Figure 2. In Figure 2 (a), the impedance Z = 1/(ωC1), and ceramic capacitors and polyester film capacitors are connected in parallel in the high-frequency region, which has a better filtering effect. In Figure 2 (b), the noise can be bypassed to the ground through the capacitor. When this filter is connected, the ground impedance should be as small as possible. In Figure 2 (c), C1 and C2 have a good filtering effect on asymmetric noise, and C3 has a good filtering effect on symmetrical noise. When connected, the leads and ground wires of the capacitors should be as short as possible. Figure 2 (d) is a commonly used noise filtering circuit. L1 and L2 present high impedance to noise, while C1 presents low impedance to noise. When L1 and L2 use a common-mode inductor structure, they have a good filtering effect on both symmetrical and asymmetric noise. Figure 2 (e) is suitable for filtering common-mode noise. It should be noted that its ground impedance should also be as small as possible.

Figure 3 is a filter circuit that is effective for both common mode noise and differential mode noise. Among them, L1, L2, and C1 are the differential mode noise suppression circuit, and L3, C2, and C3 constitute the common mode noise suppression circuit. The cores of L1 and L2 should be made of materials that are not easy to be magnetically saturated and have excellent MF characteristics. C1 uses a ceramic capacitor or a polyester film capacitor with sufficient withstand voltage. Its capacity is generally 0.22µF -0.47µF. L3 is a common mode inductor with high impedance and good suppression effect on common mode noise.

6 EMI filter selection and installation

The four capacitors in the switching power supply EMI filter use two different subscripts "x" and "y", which not only explain their role in the filter network, but also indicate their safety level in the filter network. Whether selecting or designing EMI filters, the safety level of Cx and Cy must be carefully considered. In practical applications, the Cx capacitor is connected between L and N of the single-phase power line. In addition to the rated power supply voltage, it also superimposes the EMI signal peak voltage between L and N. Therefore, according to the application of the EMI filter and the possible EMI signal peak, the Cx capacitor with the appropriate safety level should be correctly selected. The Cy capacitor is connected between the power supply lines L, N and the metal casing (E). For a 220V, 50Hz power supply, in addition to meeting the 250V peak voltage withstand voltage requirement, it is also required that this capacitor has sufficient safety margin in terms of electrical and mechanical performance to avoid possible breakdown short circuit phenomena.

EMI filters are mutually heterogeneous, that is, the load can be connected to the power supply or the load. In practical applications, in order to effectively suppress EMI signals, the network structure and parameters of the filter must be selected according to the EMI signal source impedance and load impedance to be connected at both ends of the filter. When the impedances at both ends of the EMI filter are in a mismatched state, that is, when Zs≠Zin and ZL≠Zout in Figure 4, the EMI signal will be reflected at its input and output ends, increasing the attenuation of the EMI signal. The relationship between the attenuation A of the signal and the reflection Γ is: A=–10Lg(1-|Γ|2).

When using a switching power supply filter, pay attention to the power frequency of the filter at rated current. When installing the filter, pay special attention to the spacing between the filter's input wire and output wire. Do not bundle them together for routing, otherwise the EMI signal will easily couple from the input wire to the output wire, which will greatly reduce the filter's suppression effect.

7 Conclusion

In the design of switching power supplies, in order to avoid detours and save time, the requirements for anti-interference should be fully considered and met to avoid taking remedial anti-interference measures after the design is completed.

2. Design of flyback switching power supply based on UC3845

introduction

Flyback switching power supply is widely used in automatic control and intelligent instrument power supply due to its simple structure and few components. The regulation part of the switching power supply usually adopts pulse width modulation (PWM) technology, that is, when the main converter cycle remains unchanged, the duty cycle of the power MOSFET tube is adjusted according to the change of input voltage or load, so as to stabilize the output voltage. There are many methods of pulse width modulation. This article introduces a high-performance fixed frequency current type pulse width integrated control chip UC3845. This chip is designed for offline DC to DC converter applications. Its main features are internal oscillator, high-precision error comparator, cycle-by-cycle current sampling comparison, small startup current, large current totem pole output, etc. It is an ideal device for driving MOSFET.

1 Introduction to UC3845

UC3845 chip is a plastic surface mount component with SO8 or SO14 pins. It is designed for low voltage applications. Its undervoltage lockout threshold is 8.5V (on), 7.6V (off); current mode operation up to 500 kHz output switching frequency; maximum duty cycle in flyback applications is 0.5; output dead time adjustable from 50% to 70%; automatic feedforward compensation; latched pulse width modulation for cycle-by-cycle current limiting; internal fine-tuned reference source; with undervoltage lockout; high current totem pole output; input undervoltage lockout with hysteresis; low startup and operating current.

The chip pin diagram and pin functions are shown in Figure 1.

Figure 1 UC3845 chip pin diagram

Pin 1: Output/compensation, the output of the internal error amplifier. Usually a feedback network is connected between this pin and pin 2 to determine the gain and frequency response of the error amplifier.

Pin 2: Voltage feedback input. This pin is compared with the reference voltage (2.5 V) at the same-direction input of the internal error amplifier to adjust the pulse width.

Pin 3: Current sampling input terminal.

Pin 4: The common end of the external capacitor C and resistor R of the RT/CT oscillator. Connected to Vref through a resistor and grounded through a resistor.

5th leg: Grounded.

Pin 6: Totem pole PWM output, driving capability is ±1A.

Pin 7: Positive power pin.

Pin 8: V ref, 5V reference voltage, output current up to 50mA.

2 Design Methodology

As shown in Figure 2, it is a circuit diagram of a flyback switching power supply based on UC3845, and the dotted box is a simplified block diagram of the UC3845.

1) Selection of starting voltage and capacitor

The AC power supply 115VAC becomes a DC high voltage Udc with very small ripple after rectification and filtering. This voltage can often obtain a maximum voltage Udcmax and a minimum voltage Udcmin according to the AC power supply range.

When the DC input voltage is greater than 144V, UC3845 should start working, and the starting resistance should be determined by the line DC voltage and the current required for starting.

According to the parameter analysis of UC3845, when the startup voltage is lower than 8.5V, the entire circuit of UC3845 consumes only 1mA of current, that is, the typical startup voltage of UC3845 is 8.5V and the current is 1mA. Add the peripheral circuit loss of about 0.5mA, that is, the entire circuit loss is about 1.5mA. When the input DC voltage is the minimum voltage Ddcmmn, the calculation of the startup resistance Rin is as follows:

Figure 2 Circuit diagram of flyback switching power supply based on UC3845

After the startup process is completed, the current consumption of UC3845 will increase to about 100mA as the MOSFET tube is turned on. This current is provided by the charge stored in the startup capacitor at startup. At this time, the voltage on the startup capacitor will drop to above 7.6V. To keep UC3845fj~ working, the feedback winding L should be able to provide the feeding voltage in time. If the voltage is lower than 7.6V, the undervoltage comparator will be activated and the PWM output will be low. The self-feeding time is determined by the switching cycle of UC3845. The oscillation frequency of UC3845 is 54kHz. The capacity of the startup capacitor can be calculated by the following formula:

2) Calculation of the number of turns of the feedback winding

Ns=Np(Vcc+0.8)(1-Dmax)/(UdcminiDmax), where NP is the number of primary turns of the transformer.

3) Filtering

In order to filter out the high-frequency signal at the power supply end, a ceramic capacitor is connected to the ground from Vcc. When wiring the PCB, care should be taken to ensure that no inductance components are involved to avoid interference and cause circuit instability.

4) Duty cycle D

UC3845 will adjust its duty cycle according to the change of input voltage. According to the parameter requirements of UC3845, the maximum duty cycle Dmax of UC3845 corresponding to the lowest DC voltage input is set to 0.5.

When the input DC voltage is between 144V and 177V, the duty cycle range of UC3845 is:

5) Modulation frequency f

The frequency of the oscillator OSC is determined by the selected values ​​of the timing components RT and CT. The capacitor CT is charged to 2.8V by the reference voltage V ref (=5V) through the resistor R, and then discharged to 1.2V by the internal current sink, forming a sawtooth pulse signal, as shown in Figure 3. Regardless of whether it is large RT and small CT or large CT and small RT, the latch sets the input square wave to a low level when the oscillator is charged, and the input square wave is high when it is discharged.

When the latch is set to input square wave at high level, the NOR gate output is always low level, blocking PWM. This period of time is determined by the discharge time of the internal oscillator OSC. At the same time as the latch is set to input square wave falling edge, if the other three input signals of the NOR gate input invalid level, the NOR gate output is high level, and the MOSFET is turned on.

The other three input signals are: one is the output of the current sampling comparator, one is the output of the error amplifier, and one is the output of the input undervoltage comparator.

In order to filter out the high-frequency signal at the reference end, a ceramic capacitor is connected to the ground from V. When wiring the PCB, care should be taken to ensure that no inductance components are involved to avoid interference and cause circuit oscillation.

OSC oscillation frequency f=1.8/(RtCt), when RT=33kf2, CT=1000PF, f=-54kHz.

6) Current sampling comparison

In Figure 2, when the MOSFET is turned on, Udc = Ldi/dt, and the transformer inductance current increases linearly with a slope of Udc/L, where L is the primary inductance of the transformer. A non-inductive sampling resistor Rs is connected in series between the source and ground of the MOSFET to convert the primary current of the transformer into a sampling voltage Ud = RSi. Under the same output power, the smaller the input DC voltage, the greater the primary current of the transformer, and the greater the current passing through the MOSFET. In order to protect the MOSFET from damage, the inductance peak current Ip=2P/(UdcminUmax) needs to be calculated. The maximum peak current Icmax of the selected power MOSFET should be greater than 1.3Ip.

The sampled voltage Ud is filtered by RC and sent to the 3rd pin of UC3845 . When the voltage exceeds 1V, the comparator outputs a high level, which is sent to the reset terminal of the RS latch, and the PWM output is a low level, which reduces the duty cycle of the PWM and limits the peak current of the inductor.

Non-inductive sampling resistor. The resistance value is: Rs=l/Ip, power 1W. The time constant of the RC filter is close to the duration of the sharp pulse, otherwise it will cause instability in the power supply output. Take R=lk, C=470PF.

7) Error comparator

Vref is divided into 2.5V by resistors and connected to the positive terminal of the error comparator, while the negative terminal (pin 2) is connected to the external monitoring voltage input. The output of the error comparator (pin 1) is used for compensation of the external loop, as shown in Figure 2. The output voltage is offset (=1.4V) due to two diode voltage drops and is divided into three before being connected to the negative input of the sampling comparator. An RC network is connected between pins 2 and 1 for loop compensation. Take R 11=150kQ, C11=100PF.

The external monitoring voltage input terminal (pin 2) can be used to introduce a voltage feedback link into the output circuit. If the stability of the main output circuit 5V is not required, the feed voltage can be introduced to monitor the overvoltage of the output circuit. Vcc is connected to the monitoring voltage input terminal of UC3845~b through a resistor divider. When the output circuit voltage rises for some reason, the external monitoring voltage input terminal is greater than 2.5V, the error comparator output is less than 2.5V, and the input voltage is compared with the current sampling. The PWM output is low level, which reduces the duty cycle of PWM and the output circuit voltage. If there is a requirement for the accuracy of the main circuit output 5V voltage. The feedback circuit should be composed of the optocoupler PC817, TL431 and the resistor-capacitor network connected to it. The control principle is as follows:

The 5V output voltage of the main circuit is divided by resistors to obtain a sampling voltage, which is compared with the 2.5V reference voltage provided by TL431. When the output voltage is normal (5V), the sampling voltage is equal to the 2.5V reference voltage provided by TL431. The K-pole potential of TL431 remains unchanged, the current flowing through the optocoupler diode remains unchanged, the current flowing through the optocoupler remains unchanged, the input voltage of pin 2 of UC3845 remains unchanged, the potential of pin 1 is stable, the duty cycle of the PWM drive output at pin 6 remains unchanged, and the output voltage remains stable at the set value.

When the output 5V voltage is too high for some reason, the voltage divided by the resistor will be greater than 2.5V, then the K-pole potential of TL431 will drop, the current flowing through the optocoupler diode will increase, and the current flowing through the optocoupler will increase. The input voltage of pin 2 of UC3845 will rise to greater than 2.5V, the potential of pin 1 will drop, the duty cycle of the output drive pulse PWM of pin 6 will decrease, and the output voltage will decrease, thus completing the function of output voltage feedback stabilization of the main circuit.

3 Conclusion

Practice has proved that the flyback switching power supply based on UC3845 has the characteristics of wide input voltage range, high output voltage accuracy, high load adjustment efficiency, etc. This power supply is applied to network electrical measuring instruments and has achieved good results, which has certain promotion value.

3. Application of Surge Current Suppression Module in Switching Power Supply

1 Power-on surge current

At present, considering factors such as volume and cost, most AC/DC converter input rectifier filters use capacitor input filtering, and the circuit principle is shown in Figure 1. Since the voltage on the capacitor cannot jump, at the beginning of the rectifier power-on, the filter capacitor voltage is almost zero, which is equivalent to a short circuit at the rectifier output. If the power is turned on in the most unfavorable situation (the instantaneous value of the voltage at power-on is the peak value of the power supply voltage), an input surge current far higher than the normal working current of the rectifier will be generated, as shown in Figure 2. When the filter capacitor is 470μF and the internal resistance of the power supply is small, the first current peak will exceed 100A, which is 10 times the normal working current peak.

Inrush current will cause the power supply voltage waveform to collapse, making the power supply quality worse, and even affecting the operation of other electrical equipment and causing the protection circuit to operate; because the surge current impacts the input fuse of the rectifier, it will be blown under the surge current impact of several power-on processes rather than overload. In order to avoid this kind of phenomenon, a fuse with a higher rated current has to be selected, but the fuse will not be blown when overloaded, and it will not play the role of protecting the rectifier and the power circuit; excessive power-on surge current causes irreversible damage to the rectifier and filter capacitor. Therefore, the input surge current of the rectifier with capacitor filtering must be limited.

2 Power-on surge current limitation

The most effective way to limit the power-on surge current is to add a negative temperature coefficient thermistor (NTC) between the rectifier and the filter capacitor, or on the input side of the rectifier, as shown in Figure 3. The negative temperature coefficient thermistor has a high resistance at room temperature to limit the power-on surge current. After power-on, the NTC generates heat due to the current flowing through it, which reduces its resistance value to reduce the loss on the NTC. Although this method is simple, the problem is that the performance of limiting the power-on surge current is affected by the ambient temperature and the initial temperature of the NTC. When the ambient temperature is high or the power-on time interval is very short, the NTC cannot play a role in limiting the power-on surge current. Therefore, this method of limiting the power-on surge current is only used for low-cost microcomputer power supplies or other low-cost power supplies. On color TVs and monitors, a series current limiting resistor is used to limit the power-on surge current, and the circuit is shown in Figure 4. The most common application is color TVs. The advantages of this method are simplicity, high reliability, and allowance for operation in a wide range of ambient temperatures. Its disadvantage is that there is loss on the current limiting resistor, which reduces the power supply efficiency. In fact, after the rectifier is powered on and in steady-state operation, the current limiting function of this current limiting resistor has been completed, and it only plays a negative role in consuming power and generating heat. Therefore, in a switching power supply with a large power, a mechanical contact or an electronic contact is used to short-circuit the current limiting resistor after a certain delay after power-on, as shown in Figure 5. This method of limiting the power-on surge current has good performance, but the circuit is complex and occupies a large volume. In order to make the application of this method of suppressing the power-on surge current as convenient as just connecting a current limiting resistor in series, this article introduces a switching power supply power-on surge current suppression module.

3 Power-on surge suppression module

3.1 Power-on surge current suppression module with current limiting resistor

The power electronic switch (which can be MOSFET or SCR) and the control circuit are encapsulated in a relatively small module (such as 25mm×20mm×11mm for less than 400W), and 3 to 4 pins are led out. The external circuit is shown in Figure 6 (a). For the first period of time after the rectifier is powered on, the external current limiting resistor suppresses the power-on surge current. After the power-on surge current ends, the module is turned on to short-circuit the current limiting resistor. The input current waveform of the power-on process is shown in Figure 6 (b). Obviously, the peak of the power-on surge current is effectively suppressed. This power-on surge current suppression module requires an external current limiting resistor, which is very inconvenient to use. How to save the external resistor will be what the power supply designer hopes for.

3.2 Power-on surge current suppression module without current limiting resistor

Someone proposed a power-on surge current suppression circuit without a current-limiting resistor as shown in Figure 7 (a), and its power-on current waveform is shown in Figure 7 (b). The idea is to design the circuit into a linear constant current circuit. The actual circuit will have self-oscillation due to the high gain of the two-pole amplifier, but it does not affect the operation of the circuit. In principle, this circuit is feasible, but when it is used, there are the following problems that are difficult to solve: For example, the power-on current of a 400W switching power supply with a 220V input needs to be at least 4A. If the power-on happens to be the peak value of the grid voltage, the circuit will withstand a power of 4×220×=1248W. Not only does it far exceed the 125W rated dissipation power of IRF840, it also far exceeds the 150W rated dissipation power of IRFP450 and IRFP460. Even the linear MOSFET of APT has a rated dissipation power of only 450W. Therefore, the result of using IRF840 or IRFP450 is that the MOSFET can only withstand a limited number of power-on processes before it may suffer thermal breakdown. In terms of cost, the price of IRF840 is acceptable, while the price of IRFP450 and IRFP460 is unacceptable, and APT's linear MOSFET is even less acceptable.

To truly realize a power-on surge current suppression module without a current-limiting resistor, it is necessary to solve the power loss problem of the power device during the power-on process. The basic idea of ​​another power-on surge current suppression module introduced by the author is to make the power device work in a switching state, thereby solving the high power loss problem of the power device during the power-on process, and the circuit is simple. The circuit is shown in Figure 8 (a) and Figure 8 (b), and the power-on current waveform is shown in Figure 8 (c).

3.3 Test Results

When module A is used in a 400W switching power supply, the shell temperature rise is no more than 40°C, frequent repeated power-on with an interval of 20ms is allowed, the maximum peak current is no more than 20A, and the overall dimensions are 25mm×20mm×11mm or 35mm×25mm×11mm.

The rated temperature rise of B module and C module for 800W is not more than 40℃, the repeated power-on time interval is unlimited, the power-on peak current is 3 to 5 times the peak current during normal operation, and the overall dimensions are 35mm×30mm×11mm or 50mm×30mm×12mm.

The aluminum substrate of the module is attached to the radiator, and the module temperature is no higher than 5℃ of the radiator.

4 Conclusion

The advent of the switching power supply power-on surge current suppression module has brought great convenience to the switching power supply designers due to its simple external circuit and small size, especially the solution without current limiting resistor, which has not been reported at home and abroad. At the same time, the author will also launch the power-on surge current suppression module for other impact loads (such as AC motors and various lamps, etc.).

4. Methods for suppressing electromagnetic interference in switching power supplies

introduction

With the continuous development and maturity of switching power supply technology, the demand for switching power supplies in various application fields is also growing. However, switching power supplies have serious electromagnetic interference (EMI) problems. It not only pollutes the power grid and directly affects the normal operation of other electrical appliances, but also breaks into space as radiation interference, causing electromagnetic pollution to space. Therefore, the electromagnetic compatibility (EMC) problem of switching power supplies arises. Electromagnetic compatibility refers to the ability of a device or system to work normally in its electromagnetic environment and not cause unbearable electromagnetic disturbance to anything in the environment.

The electromagnetic interference of switching power supplies can be divided into two categories: conducted interference and radiated interference. Conducted interference is transmitted through the AC power supply with a frequency below 30 MHz. Radiated interference is transmitted through the air with a frequency above 30 MHz.

This article analyzes the problem of excessive electromagnetic interference in a desktop 180W plastic case switching power supply (the load is a 12V/15A semiconductor refrigeration refrigerator, and the power supply size is 205mm×90mm×62mm) from the principle and explores the solution.

1 180 W switching power supply circuit structure analysis and electromagnetic interference test

1.1 Analysis of main circuit and structural layout

The circuit principle of the switching power supply is shown in Figure 1.

The power factor of the capacitor filter rectifier is low, the conduction time of the rectifier diode is short, the peak value of the instantaneous value of the filter capacitor charging current is large, and the current waveform after rectification is pulsating, generating high harmonic current.

In the half-bridge circuit, S1, S2, D3, D4 and transformer T1, which are turned on and off at high frequency, are the main interference sources of the switching power supply, generating high-frequency and high-voltage peak harmonic oscillations. The higher-order harmonics generated by the harmonic oscillations are transmitted into the internal circuit through the distributed capacitance between the switching tube and the radiator, or radiated into space through the radiator and transformer.

The internal layout of the switching power supply is shown in Figure 2. The left side is the AC power input and DC output. There are ventilation holes on the upper and lower sides of the left side. The fan is on the right side. The heat is dissipated by exhausting air outward to ensure that the heat in the plastic shell is discharged in time and avoid heat accumulation in the plastic shell. The advantage of this layout is that the ventilation path is relatively smooth, but there are also disadvantages - the input and output interfaces are installed close to each other, which is easy to cause spatial coupling between them, forming radiation interference.

1.2 Electromagnetic interference test

Table 1 lists the measured values ​​of harmonic currents from order 7 to order 21, among which harmonic currents of order 11, 15 and 17 all exceed the standard.

The radiated disturbance prediction results exceed the limit at 30-50MHz and 100MHz, as shown in Figure 4.

2. Suppression of electromagnetic interference

2.1 Suppression of harmonic current

Power factor correction can solve the problem of excessive harmonic current. Active power factor correction uses Boost PFC circuit to increase the power factor to above 0.99, making the harmonic current very small, but the circuit is complex and the cost is not low. In addition, the switching noise of the switch tube and high-voltage rectifier diode in the circuit will become a new source of interference, making it more difficult for the whole machine to meet the EMI standard.

Considering that within the range of AC input voltage (AC 220~250V), under the condition of meeting the voltage regulation rate, appropriately reducing the filter capacitor, the input series resistance can reduce the peak value of the instantaneous value of the filter capacitor charging current to a certain extent, meet the harmonic current limit, and the power loss is within an acceptable range. The power efficiency of the whole machine does not decrease much, which is also a good method. The measured harmonic current values ​​after using this method are listed in Table 2.

2.2 Suppression of conducted disturbances

The main source of conducted noise is the alternating operation of the power switch tubes S1 and S2 in the half-bridge at a frequency of 25 kHz. The collector-emitter voltage Uce and emitter current of the power switch tube have waveforms close to rectangular waves. Fourier analysis shows that the rectangular wave pulse has a fairly wide frequency bandwidth, contains abundant high-order harmonics, and the spectrum amplitude of the pulse waveform is higher in the low-frequency band. In addition, the current mutation caused by the leakage inductance of the high-frequency transformer winding during the cut-off period of the power switch tube will also produce spike interference.

The input filter is a low-impedance channel (i.e., low-pass filter) designed for the electromagnetic disturbance level of the converter and the external electromagnetic disturbance source to suppress or remove the electromagnetic disturbance and achieve the purpose of electromagnetic compatibility.

As shown in Figure 5, the input filter is a low-pass filter circuit composed of inductors (LFI, LF2), CY capacitors (C4, C5) and Cx capacitors (C1, C2, C3). It has a greater attenuation for higher frequency noise signals. C1, C2, C3 are capacitors for filtering common mode interference, C4, C5 are capacitors for filtering differential mode interference, and LF1, LF2 are common mode coils.

图3中低频传导干扰（O.15～lMHz范围）超标，共模噪声的主要骚扰源是功率开关管，低频传导干扰抑制以增加共模电感的电感量为主，当共模电感从原设计的15mH增加到24mH时，低频传导干扰最大处下降30dB，得到了显著改善。如图6所示。

The input filter has a significant effect on noise suppression below 20MHz. The ideal input filter is a low-pass filter, but in reality it is a band-stop filter.

When the switching power supply frequency increases, the required common mode inductance can be greatly reduced, and the volume of the common mode inductor is also reduced. However, the radiation noise component of the switching power supply in the frequency band above 20MHz increases, which makes it difficult to meet the radiation interference standard. The relationship between switching frequency and common mode inductance is listed in Table 3.

Due to the parasitic capacitance of the common-mode inductor coil, the high-frequency noise component radiates interference outward through the parasitic capacitance. Therefore, it is not easy to achieve good high-frequency filtering effect by using a single large-inductance common-mode inductor. Generally, two common-mode inductors with the same inductance are used to effectively suppress high-frequency noise, and there will be a difference of more than 6dB.

The high-frequency impedance frequency characteristic of Cx capacitor is an important parameter related to the electromagnetic disturbance suppression effect. When used at high frequencies, the capacitor is equivalent to a circuit of r (equivalent series resistance) + c + L (equivalent series inductance). Due to the inherent inductance of the capacitor itself (i.e., equivalent series inductance), the capacitor reactance is capacitive in the low frequency range and inductive in the high frequency range, and the ability to suppress disturbance is significantly reduced. The smaller the inherent lead inductance of the capacitor and the larger the high-frequency internal impedance of the disturbance source, the better the effect of suppressing disturbance.

First, we should start from the mechanism of electromagnetic interference source, find the location of radiation interference source, and fundamentally reduce the level of radiation interference noise. When the output voltage is relatively low, the interference of output rectifier and smoothing circuit may be relatively large.

Serious + By reducing the loop area, the magnetic field radiation generated by the di/dt loop can be suppressed. The rectifier and freewheeling diodes work in a high-frequency switching state and are also a high-frequency interference source. The parasitic inductance of the diode leads, the existence of junction capacitance and the influence of reverse recovery current make it work at a very high voltage and current change rate, and produce high-frequency oscillation. The longer the diode reverse recovery time is, the greater the impact of the peak current is.

The leads of C4 and Cs and the ground leads should be as short as possible to minimize the ground impedance and allow the noise to bypass the capacitor to the ground. C4 and C5 have larger capacitances for better filtering effects. However, as the capacitance increases, the leakage current also increases. The leakage current value is an important indicator of electrical safety and must not exceed the specified value. The general leakage current limit is 3.5 mA. This desktop plastic case switching power supply is a handheld device with a maximum leakage current limit of 0.75 mA. The actual measured value is 0.55 mA.

The power input cable should be short, and the filter should be as close to the input port as possible to avoid coupling between the filter input and output, which would result in loss of filtering effect. The grounding should be as short and reliable as possible to reduce high-frequency impedance and effectively bypass interference. After several rectifications, a satisfactory result was obtained as shown in Figure 7.

2.3 Suppression of radiated disturbance

Radiated disturbance refers to electromagnetic interference radiated by any component, antenna, cable or connecting wire.

Usually, in the layout of circuit components, the input AC and output DC sockets (including leads) should be separated and kept away as much as possible. One-end input and the other-end output is a reasonable layout. However, considering the internal heat dissipation and ventilation of the power supply, the power supply adopts the heat dissipation structure of Figure 2. The unavoidable problem is that spatial coupling may occur between the input and output cables, and strong radiation will be generated when high-frequency conduction current passes through.

First, starting from the mechanism of electromagnetic disturbance source generation, find the location of radiation disturbance source, and fundamentally reduce the level of radiation disturbance noise generated by it. When the output voltage is relatively low, the interference of output rectifier and smoothing circuit may be more serious. By reducing the loop area, the magnetic field radiation generated by di/dt loop can be suppressed. The rectifier and freewheeling diodes work in high-frequency switching state and are also a high-frequency disturbance source. The parasitic inductance of the diode lead, the existence of junction capacitance and the influence of reverse recovery current make it work at a very high voltage and current change rate, and produce high-frequency oscillation. The longer the reverse recovery time of the diode, the greater the influence of the peak current.

Ferrite magnetic rings and beads are easy to use, cheap, and have a significant effect in suppressing electromagnetic interference. The equivalent circuit of a ferrite inductor is a series circuit consisting of an inductor L and a resistor R, where L and R are both functions of frequency. The resistance value increases with increasing frequency, thus forming a low-pass filter. At low frequencies, R is very small, L plays a major role, and electromagnetic interference is reflected and suppressed; at high frequencies, R increases, electromagnetic interference is absorbed and converted into heat energy, greatly attenuating high-frequency interference. Different ferrite suppression components have different optimal suppression frequency ranges. Generally, the higher the magnetic permeability, the lower the suppression frequency. In addition, the larger the volume of the ferrite, the better the suppression effect. When the volume is constant, a long and thin shape has a better suppression effect than a short and thick one, and the smaller the inner diameter, the better the suppression effect. Ferrite suppression components should be installed close to the interference source. For input and output circuits, they should be as close as possible to the entrance and exit of the shielding shell.

The rectifier diode uses a Schottky diode, and its anode is covered with a ferrite bead (φ3.5×φ1.3×3.5), and the DC output cable is surrounded by a ferrite magnet (φ13.5×φ7.5×7) for 2.5 turns and close to the exit. After the rectification, the maximum radiation interference dropped by about 10dB, but the margin at 40MHz and 100MHz was small, and the quasi-peak test had only a 5dB margin. Considering that the certification process is cumbersome and the cycle is long, and the error of 2 to 3dB is allowed between each certification and testing service center, the product prediction should be more than 6dB margin, as shown in Figure 8.

The use of ferrite beads and ferrite rings has greatly improved the suppression of interference source noise. If it still cannot meet the requirements, shielding measures have to be adopted, and 2mm thick aluminum plates are used to isolate the input and output to cut off the electromagnetic noise propagation path formed by spatial coupling. As a result, the radiated interference noise margin reached more than 12dB, and the noise suppression effect was quite obvious. Through the above measures, the conversion of the limit values ​​specified in the 3m method anechoic chamber and the 10m method anechoic chamber test: Since the standard GB9254 recognizes that ITE (information technology equipment) obtains the radiated interference limit value at a measurement distance of 10m, and most EMC testing service centers test in a 3m anechoic chamber, because the field strength is inversely proportional to the distance, the noise level measured in the 3m method is 10 dB lower than the noise level value in the 10m method.

Figures 4, 8 and 9 were measured in a 3m anechoic chamber, with a quasi-peak limit of 40dB for 30-230MHz and 47dB for 230-1000MHz. Figure 10 was measured in a 10m anechoic chamber, and the waveforms of the radiated noise are similar when compared with Figure 10. The noise level at only a few frequency points increases slightly.

3 Conclusion

After the above rectification, the electromagnetic compatibility of the l80W power supply was tested again and it fully met the design requirements. To solve the EMI problem in the early stage of power supply design, when the structure has not yet been finalized, there are many methods available, which is conducive to reducing costs.

In addition to the suppression measures mentioned above, there are other solutions, but the design solutions must take into account the power supply cost.

There are many and complex factors related to EMI. It is far from enough to just do the above points. Grounding technology, PCB layout and routing are also very important. The design of electromagnetic compatibility is a long and arduous task. We must continue to conduct research to make the electromagnetic compatibility level of our electronic products synchronized with the international level.