Compared to linear regulated power supplies, switching power supplies offer numerous advantages, including lower power consumption, higher efficiency, smaller size, lighter weight, and wider voltage regulation range. They are widely used in computers and peripherals, communications, automatic control, and home appliances. However, a significant drawback of switching power supplies is their ability to generate strong electromagnetic interference (EMI). EMI signals have a wide frequency range and a certain amplitude; after conduction and radiation, they pollute the electromagnetic environment and interfere with communication equipment and electronic products. If not handled properly, the switching power supply itself can become a source of interference. Currently, electromagnetic compatibility (EMC) of electronic products is receiving increasing attention. Suppressing EMI from switching power supplies and improving the quality of electronic products to meet EMC standards has become a growing concern for electronic product designers. This article discusses electromagnetic compatibility issues in the design of high-frequency switching power supplies.
1. Composition and working principle of switching power supplies
1.1 Composition
A switching power supply consists of the following parts:
1) The main circuit includes an input filter, rectification and filtering, an inverter, and an output rectification and filtering;
2) Control and protection circuits;
3) In addition to providing the various parameters required by the protection circuit, the detection and display circuit also provides various display data;
4) Auxiliary power supply.
1.2 Switching Power Supply Principle
In a switching power supply circuit, the switch K is repeatedly turned on and off at certain time intervals. When K is on, the input power supply Vin supplies power to the load RL through K and the filter circuit. When K is off, the input power supply Vin interrupts its energy supply. It is evident that the energy supply from the input power supply to the load is intermittent. To ensure a continuous energy supply to the load, the switching power supply must have an energy storage device to store a portion of the energy when the switch is on and release it to the load when the switch is off. A circuit consisting of an energy storage inductor L, a filter capacitor C2, and a freewheeling diode D has this function. The average voltage VAB between A and B can be expressed by equation (1).
VAB = Vinton/T = DVin (1)
In the formula: ton is the conduction time of K;
T represents the K working cycle;
D is the duty cycle, D = ton/T.
As shown in equation (1), changing D will change VAB. Therefore, adjusting D according to changes in load and input power supply voltage will keep the output voltage Vo constant. This control method is called Time Ratio Control (TRC). According to the TRC principle, it has three modes:
1) Pulse Width Modulation (PWM) has a constant switching period, and changes the duty cycle by altering the pulse width;
2) Pulse Frequency Modulation (PFM): The conduction pulse width is constant, and the duty cycle is changed by altering the switching frequency.
3) The hybrid modulation method combines the two methods above. Both the conduction pulse width and the switching frequency are not fixed and can be changed.
2. Mechanism of electromagnetic interference generated by switching power supplies
The reason why switching power supplies are a strong source of electromagnetic interference is due to the high-frequency switching devices, output rectifier diodes, pulse transformers, and filter inductors.
2.1 Switching transistors and rectifier transistors
The dv/dt and di/dt generated when switching transistors and rectifiers are switched on and off at high frequencies are pulses with large amplitude, wide frequency band and rich harmonics, which is a strong source of interference.
2.2 High-frequency transformer
The load of the switching transistor is the primary coil of a high-frequency transformer. At the moment the switching transistor is turned on, the primary coil generates a large inrush current and a high surge voltage spike. At the moment the switching transistor is turned off, due to the leakage flux of the primary coil, some energy is not transferred to the secondary coil. Instead, it forms a damped oscillation with a spike through the capacitor and resistor in the collector circuit. This oscillation is superimposed on the turn-off voltage, forming a turn-off voltage spike. This generates the same magnetizing inrush current transient as when the primary coil is turned on. This noise will be conducted to the input and output terminals, forming conducted interference. In severe cases, it may even break down the switching transistor.
Furthermore, the high-frequency switching current loop formed by the primary coil of the high-frequency transformer, the switching transistor, and the filter capacitor may generate significant spatial radiation, resulting in radiated interference. If the capacitor's filtering capacity is insufficient or its high-frequency characteristics are poor, the high-frequency impedance on the capacitor will cause the high-frequency current to be conducted into the AC power supply in differential mode, forming conducted interference. It is important to note that in the electromagnetic interference generated by the diode rectifier circuit, the |di/dt| of the reverse recovery current of the rectifier diode is much larger than that of the freewheeling diode. As an electromagnetic interference source, the interference intensity and bandwidth generated by the reverse recovery current of the rectifier diode are high. However, the voltage jump generated by the rectifier diode is much smaller than the voltage jump generated when the power switch is turned on and off. Therefore, it is acceptable to disregard the effects of |dv/dt| and |di/dt| generated by the rectifier diode and study the rectifier circuit as part of the electromagnetic interference coupling path.
2.3 The influence of stray parameters on the characteristics of the coupling channel
In the conducted interference band (<30MHz), most coupling paths of switching power supply interference can be described using circuit networks. However, every actual component in a switching power supply, such as resistors, capacitors, inductors, and even switching transistors and diodes, contains stray parameters. Furthermore, the wider the frequency band under study, the higher the order of the equivalent circuit. Therefore, the equivalent circuit of a switching power supply, including the stray parameters of each component and the coupling between components, becomes much more complex. At high frequencies, stray parameters significantly affect the characteristics of the coupling paths, and the presence of distributed capacitance becomes a channel for electromagnetic interference. Additionally, when the switching transistor has high power, a heatsink is generally required at the collector. The distributed capacitance between the heatsink and the switching transistor cannot be ignored at high frequencies; it can generate radiated interference into space and common-mode interference conducted through power lines.
3 Electromagnetic compatibility design
3.1 Design of Input Filter
Noise generated by switching power supplies includes common-mode noise and differential-mode noise. Common-mode interference is caused by the potential difference between the current-carrying conductor and ground, characterized by noise voltages on both lines being at the same potential and in the same direction. Differential-mode interference, on the other hand, is caused by the potential difference between the current-carrying conductors, characterized by noise voltages on both lines being at the same potential but in opposite directions. Typically, both components of the interference voltage on the line exist simultaneously. Therefore, a filter should be added at the power input. The filter impedance should be mismatched with the power supply impedance; the greater the mismatch, the more ideal the attenuation and the better the insertion loss characteristics. That is, if the noise source has low internal resistance, the input impedance of the EMI filter should be high (e.g., a series inductor with a large inductance); if the noise source has high internal resistance, the input impedance of the EMI filter should be low (e.g., a parallel capacitor with a large capacitance). Due to the impedance imbalance of the line, the two components can transform into each other during transmission, making the situation quite complex. A typical EMI filter includes suppression circuits for both common-mode and differential-mode noise.
Differential mode suppression inductors L1 and L2: 100–130 μH;
Common-mode rejection capacitors Cy1 and Cy2 < 10000pF;
Common-mode rejection inductance L 15～25mH.
As defined by insertion loss, when no filter is connected, the signal source output voltage is V1. When a filter is connected, the voltage of the signal source measured at the filter output is V2. If the output impedance of the signal source and the input impedance of the receiver are equal, both being 50Ω, then the insertion loss of the filter is...
During the design process, the resonant frequencies of the common-mode filter circuit and the differential-mode filter circuit must be significantly lower than the operating frequency of the switching power supply, generally below 10kHz.
In practical applications, since the common-mode and differential-mode components generated by the equipment are different, the filter circuit can be adjusted by adding or removing filter components as needed. Satisfactory results can generally only be obtained after EMI testing. When installing the filter circuit, it is essential to ensure good grounding and proper isolation between the input and output terminals; otherwise, the filtering effect will be ineffective.
3.2 Measures to suppress radiated EMI
To reduce radiated interference, voltage buffer circuits can be used, such as connecting an RCD buffer circuit in parallel across the switching transistor, or current buffer circuits, such as connecting an inductor of 20–80 μH in series with the collector of the switching transistor.
The collector of a power switching transistor is a strong source of interference. The heatsink should be connected to the collector to ensure that current generated between the collector and the heatsink due to distributed capacitance flows into the main circuit. To reduce distributed capacitance between the heatsink and the chassis, the heatsink should be kept as far away from the chassis as possible; if possible, a shielded heatsink should be used. Rectifier diodes should be selected with low recovery charge and short reverse recovery time, such as Schottky diodes, preferably those with soft reverse recovery characteristics. Additionally, placing ferrite beads and connecting a parallel RC snubber network across the Schottky diode can reduce interference. The values of resistors and capacitors can be several Ω or several thousand pF, and capacitor leads should be as short as possible to reduce lead inductance.
The larger the load current, the longer the reverse recovery time of the diode, and the greater the impact of the short-circuit current spike. Using multiple diodes in parallel to share the load can reduce the impact of short-circuit current spikes.
Switching power supplies must be shielded and employ a modular, fully sealed structure, typically using galvanized steel sheets with a thickness of 1mm or more. The shielding layer must be properly grounded. Adding a shielding layer between the primary and secondary windings of the high-frequency pulse transformer and grounding it can suppress the electric field coupling of interference. Adding shielding covers to magnetic components such as the high-frequency pulse transformer and output filter inductor can confine magnetic lines of force within a shield with low magnetic reluctance.
For example, improvements were made to a switching power supply whose radiated interference exceeded the standard limit by 20 dB using the following measures that are easily implemented in the laboratory:
1) Connect a 470pF capacitor in parallel across all rectifier diodes;
2) Connect a 50pF capacitor in parallel at the input terminal of the gate (G) of the switching transistor to form an RC low-pass filter with the existing 39Ω resistor;
3) Connect a 0.01μF capacitor in parallel with each output filter capacitor (electrolytic capacitor);
4) Place a small ferrite bead on the pin of the rectifier diode;
5) Improve the grounding of the shield.
After the above improvements, the power supply can pass the limit requirements of the radiated interference test.
3.3 Solutions to Conducted Interference
Conducted interference from switching power supplies propagates outward through the input power lines, including both differential-mode and common-mode interference. The test frequency range for conducted interference is 0.15–30 MHz.
Power port frequency range/MHz Quasi-peak dB/μV Average dB/μV
Grade A: 0.15–0.57966
0.5～307360
Grade B: 0.15–0.56656
0.5～55646
5～306050
Within the frequency range of 0.15–1 MHz, interference mainly exists in the form of common mode. Within the frequency range of 1–10 MHz, interference is characterized by a coexistence of differential and common modes. Above 10 MHz, interference is primarily in the form of common-mode interference. Differential-mode interference is mainly caused by the switching transistor operating in a switching state. When the transistor is turned on, the current flowing through the power line increases linearly, and when the transistor is turned off, the current abruptly drops to zero. Therefore, the current flowing through the power line is a high-frequency triangular pulsating current containing abundant high-frequency harmonic components. As the frequency increases, the amplitude of these harmonic components decreases, thus differential-mode interference decreases with increasing frequency. The output circuit filter is shown in Figure 8. Capacitor C5 and inductor L3 form a low-pass filter, and differential-mode conducted interference mainly exists in the low-frequency range.
The main reason for common-mode interference is the distributed capacitance between the power supply and the ground (protective ground). High-frequency harmonic components of the square wave voltage in the circuit are transmitted to the ground through this distributed capacitance, forming a loop with the power line and generating common-mode interference. As shown in Figure 8, L and N are the power inputs. C1, C2, C3, C4, C5, L1, and L2 form the input EMI filter. DB1 is the rectifier bridge, and VT2 is the switching transistor. When the switching transistor is mounted on the heatsink, its drain (D) is connected to the heatsink, forming a coupling capacitor, as shown by C7 in Figure 8. VT2 operates in switching mode, and the voltage at its drain is a high-frequency square wave. The frequency of this square wave is the switching frequency of the switching transistor. The harmonics in the square wave will form a loop through the coupling capacitor and the L and N power lines, generating common-mode interference. The distributed capacitance between the power supply and ground is relatively dispersed and difficult to estimate. However, as shown in Figure 8, the coupling capacitance between the drain (D) of VT2 and the heatsink has the greatest effect. The voltage from DB1 to inductor L3 is 100Hz, while the voltage from L3 to the connection between VD1 and the drain of VT2 is a square wave voltage containing a large number of high-order harmonics. Secondly, L3 also has a significant impact, but L3 is far from the chassis, and its distributed capacitance is much smaller than the coupling capacitance between the switching transistor and the heatsink. Therefore, we mainly consider the coupling capacitance between the switching transistor and the heatsink.
3.4 Application of Grounding Technology
"Grounding" has two meanings: signal grounding inside the equipment and grounding the equipment to the earth. These two concepts are different and have different purposes. The classic definition of "ground" is "an equipotential point or plane that serves as a reference for a circuit or system".
3.4.1 Signal grounding of the equipment
The signal grounding of a device may use a single point or a piece of metal within the device as the grounding reference point for the signal, providing a common reference potential for all signals within the device. This section introduces floating ground and hybrid grounding, as well as single-point grounding and multi-point grounding.
1) Floating Ground: The purpose of floating ground is to isolate circuits or equipment from the common grounding system or common conductors that may cause circulating currents. Floating ground also facilitates the coordination of circuits at different potentials. Methods for implementing floating ground for circuits or equipment include transformer isolation and opto-isolation. The biggest advantage of floating ground is its good anti-interference performance. The disadvantage of floating ground is that, because it is not connected to the common ground, static electricity can easily accumulate between them. When the charge accumulates to a certain level, it may cause a violent electrostatic discharge, becoming a highly destructive source of interference. A compromise is to connect a high-resistance bleeder resistor between the floating ground and the common ground to release the accumulated charge. Care should be taken to control the impedance of the bleeder resistor; too low an impedance will affect the compliance of the equipment's leakage current.
2) Hybrid Grounding: Hybrid grounding allows the grounding system to exhibit different characteristics at low and high frequencies, which is necessary in broadband sensitive circuits. Capacitors have high impedance to low frequencies and DC, thus preventing ground loops between modules. When separating DC ground and RF ground, the DC ground of each subsystem is connected to the RF ground through a 10–100 nF capacitor. These two grounds should be connected at a low impedance point, which should be selected at the location where the highest switching speed di/dt signal exists.
3.4.2 Equipment grounded
In engineering practice, in addition to carefully considering the internal signal grounding of equipment, the equipment's signal ground, chassis, and earth are usually connected together, with earth serving as the equipment's grounding reference point. The purpose of connecting the equipment to earth is:
1) Ensure the personal safety of equipment operators;
2) To discharge the charge accumulated on the chassis, preventing the chassis potential from rising due to charge accumulation and causing instability in circuit operation;
3) Prevent the equipment's potential from changing due to external electromagnetic environment, which could cause instability in the equipment's operation.
Therefore, grounding equipment is not only a consideration for personnel and equipment safety, but also an important means of suppressing harassment.
3.5 Shielding Technology
Another method to suppress electromagnetic interference radiation from switching power supplies is shielding. The purpose is to cut off the propagation path of electromagnetic waves. Solving electromagnetic interference problems using electromagnetic shielding will not affect the normal operation of the circuit. It uses materials with good conductivity to shield the electric field and materials with high magnetic permeability to shield the magnetic field. To prevent magnetic field leakage from the pulse transformer, a closed loop can be used to form magnetic shielding. In addition, the entire switching power supply should be electric field shielded. Shielding should consider heat dissipation and ventilation. The ventilation holes on the shielding shell are best to be circular and multi-hole. The number of holes can be large while ensuring sufficient ventilation, and the size of each hole should be as small as possible. Joints should be welded to ensure the continuity of the electromagnetic path. If screws are used for fixing, the screw spacing should be short. Filtering measures should be taken at the input and output points of the shielding shell; otherwise, these will become interference transmitting antennas, severely reducing the shielding effect. If shielding the electric field, the shielding shell must be grounded; otherwise, it will not achieve the shielding effect. If shielding the magnetic field, the shielding shell does not need to be grounded. The shell of non-embedded external switching power supplies must be electric field shielded; otherwise, it will be difficult to pass radiated interference tests. For switching power supplies, the main focus is on effective shielding of the chassis, high-frequency transformer, switching transistors, and rectifier diodes, employing opto-isolation technology. Power switching transistors and output diodes typically have significant power losses, requiring heat sinks or direct mounting on the power supply baseplate for heat dissipation. During component mounting, insulating sheets with good thermal conductivity are used, creating distributed capacitance between the components and the baseplate/heat sink. Since the power supply baseplate is the AC ground, electromagnetic interference is coupled to the AC input terminal through this distributed capacitance, generating common-mode interference. This can be addressed by sandwiching a shielding sheet between two layers of insulating sheets and connecting the shielding sheet to DC ground, thus cutting off the path of radio frequency interference propagating to the input power grid. To suppress the radiated electromagnetic interference from the switching power supply on other electronic devices, the shielding cover can be fabricated using methods for magnetic field shielding. Connecting the entire shielding cover to the system chassis and ground as a single unit effectively shields the electromagnetic field. Connecting certain parts of the power supply to ground can also help suppress interference. For example, grounding the electrostatic shielding layer can suppress interference from changing electric fields; while conductors used for electromagnetic shielding can theoretically be ungrounded, ungrounded shielding conductors often enhance electrostatic coupling, producing a so-called "negative electrostatic shielding" effect. Therefore, grounding is still preferable, allowing the electromagnetic shielding to simultaneously function as electrostatic shielding. The circuit's common reference point is connected to the earth, providing a stable reference potential for the signal loop. Therefore, after the safety protective ground wire, shielding ground wire, and common reference ground wire in the system each form a grounding busbar, they are all ultimately connected to the earth.
3.6 Component Layout and Printed Circuit Board Routing Technology
The radiated emissions from a switching power supply are proportional to the product of the current magnitude in the current path, the loop area of the path, and the square of the current frequency, i.e., radiated emissions E∝IAf². This relationship is based on the premise that the path size is much smaller than the wavelength of the frequency.
The above relationship shows that reducing the path area is key to reducing radiated interference, meaning that the components of the switching power supply should be arranged close together. In the primary circuit, the input capacitor, transistor, and transformer are required to be close to each other, and the wiring should be compact; in the secondary circuit, the diode, transformer, and output capacitor are required to be close to each other.
When designing printed circuit boards (PCBs), related components should be placed together as much as possible to avoid interference caused by excessively long traces due to components being too far apart. Furthermore, input and output signals should be placed near the lead ports to avoid interference caused by coupling paths. On the PCB, positive load conductors should be placed close together on both sides, and kept parallel, as the external magnetic fields generated by parallel, close-proximity positive load conductors tend to cancel each other out. Practice has shown that the component placement and wiring design on the PCB have a significant impact on the EMC performance of switching power supplies. In high-frequency switching power supplies, the PCB contains small signal control lines (±5V to ±15V), high-voltage power buses, high-frequency power switches, and magnetic components. How to rationally arrange the component positions within the limited space of the PCB directly affects the anti-interference capabilities of each component and the reliability of the circuit operation. Additionally, parallel routing of two printed signal lines should be avoided. If parallel routing is unavoidable, the following methods can be used to remedy the situation:
1) Add a ground wire between the two signal lines to provide shielding;
2) Maximize the distance between two parallel signal lines to reduce the influence of the electromagnetic field between them;
3) Ensure the current flowing through the two parallel signal lines is in opposite directions. Electromagnetic coupling between wirings occurs through electric and magnetic fields; therefore, during wiring, attention should be paid to suppressing electric and magnetic field coupling. Methods for suppressing electric fields include:
1) Maximize the distance between lines to minimize capacitive coupling;
2) Electrostatic shielding must be used, and the shielding layer must be grounded;
3) Reduce the input impedance of sensitive lines.
Methods for suppressing magnetic fields include:
1) Reduce the loop area of interference sources and sensitive circuits;
2) Increase the distance between lines to minimize the mutual inductance between the coupled interference source and the sensitive circuit;
3) It is best to route the interference source and the sensitive circuit at right angles to greatly reduce the coupling between the lines.
Furthermore, by analyzing the characteristic impedance of printed conductors, the placement, length, width, and layout of the conductors can be selected. The characteristic impedance of a single conductor consists of DC resistance R and self-inductance L. The shorter the conductor length l, the smaller the DC resistance R; increasing the width and thickness of the conductor can also reduce the DC resistance R. The shorter the conductor length l, the smaller the self-inductance L, and increasing the conductor width b can also reduce the self-inductance L. The characteristic impedance of multiple conductors, in addition to being composed of DC resistance R and self-inductance L, is also affected by mutual inductance M. Mutual inductance M is affected not only by the length and width of the conductors but also by the distance between the conductors; increasing the distance between two conductors can reduce mutual inductance. Considering these phenomena, when designing printed circuit boards, the impedance of power lines and ground lines should be minimized as much as possible. This is because power lines, ground lines, and other printed lines all have inductance. When the power current changes significantly, a large voltage drop will occur, and the ground voltage drop is a significant factor in forming common impedance interference. Therefore, ground lines should be shortened as much as possible, and the thickness of power lines and ground lines should also be increased. In double-sided printed circuit board design, in addition to making the power and ground lines as thick as possible, decoupling capacitors with good high-frequency characteristics should be installed between the ground and power lines.
4. Conclusion
To improve switching frequency and the quality of switching power supply products, electromagnetic compatibility (EMC) must be a primary consideration. This article proposes effective suppression measures based on an analysis of interference generation mechanisms and extensive practical experience. Many factors contribute to EMC in switching power supplies, and significant work remains to be done in suppressing it. During the design phase, efforts should focus on eliminating coupling and radiation between the interference source and the affected equipment, and cutting off the propagation path of EMC to minimize its impact on the switching power supply.
