Selection of Common-Mode Chokes (Inductors)

Before introducing common-mode chokes, let's first discuss chokes in general. A choke is a low-impedance coil designed to attenuate high-frequency currents in a circuit. To enhance its inductance, chokes often incorporate a core made of soft magnetic material. A common-mode choke consists of multiple identical coils through which currents flow in opposite directions, resulting in magnetic field cancellation within the choke's core. Common-mode chokes are frequently employed to suppress radiated interference because such interfering currents flow in opposite directions in the different coils, thereby improving the system's electromagnetic compatibility (EMC). For such currents, the inductance of a common-mode choke is very high. A circuit diagram of a common-mode choke is shown in Figure 1.

Common-mode and differential-mode signals are relative terms. Common-mode signals, also known as common-mode noise or ground noise, refer to the noise on each conductor relative to ground. In the context of an input filter for a switching power supply, this refers to the electrical signals on the neutral and live wires relative to the earth ground. Although neither the neutral nor the live wire is directly connected to the earth ground, they can be connected through parasitic or stray capacitances or parasitic inductances on the circuit board. Differential-mode signals, on the other hand, represent the difference in signals between the two conductors and can also be referred to as the line differential.

Given two signals, V1 and V2:

Characteristics of common-mode signals: Signals of equal amplitude and the same phase.

Characteristics of differential-mode signals: Signals of equal amplitude but opposite phases.

For a switching power supply, if the bulk storage and filtering capacitor after the rectifier bridge is ideal (i.e., with zero equivalent series resistance, ignoring all parasitic parameters of the capacitor), all possible differential-mode noise sources entering the power supply would be completely bypassed or decoupled by this capacitor. However, the equivalent series resistance of large-capacity capacitors is not zero. Therefore, the equivalent series resistance of the input capacitor constitutes the primary component of the impedance Zdm as seen from the differential-mode noise generator. In addition to carrying the operating current flowing from the power line, the input capacitor must also supply the high-frequency pulse current required by the switching transistor. Regardless, current flowing through a resistor inevitably generates a voltage drop, such as across the capacitor's equivalent series resistance. Consequently, high-frequency voltage ripple appears across the input filter capacitor, originating from differential-mode currents. This ripple is essentially a voltage source (caused by the equivalent series resistance). Theoretically, when the rectifier bridge is conducting, this high-frequency ripple noise should only appear on the input side of the rectifier bridge. In reality, when the rectifier bridge is off, the noise leaks through the parasitic capacitance of the rectifier bridge diodes.

There are various incidental paths for high-frequency currents to flow into the chassis. When the drain of the main switching transistor in a switching power supply undergoes high-to-low transitions, current flows through the parasitic capacitance between the switching transistor and the heat sink (which is connected to the chassis or is the chassis itself). When the AC mains current keeps the rectifier bridge conducting, the noise injected into the chassis encounters nearly equal impedance, resulting in equal amounts flowing into the neutral and live wires. This is pure common-mode noise.

It is known that common-mode signals are of equal amplitude and the same phase, typically originating from the power grid. These signals can interfere with the normal operation of the circuit board and also radiate as electromagnetic waves to disturb the surrounding environment.

Since inductors are used to suppress common-mode signals, this must be related to magnetic fields. Let's first introduce the direction of the magnetic field generated by a solenoid (for project applications, in some scenarios such as suppressing common-mode signals, quantitative calculations of the magnetic field and magnetic flux generated by the inductor may not be necessary. For those interested, a recommended reference book is "Magnetic Components in Switching Power Supplies" by Professor Zhao Xiuke). The method to determine the direction of the magnetic field generated by a solenoid is to grasp the solenoid with your right hand, with your fingers pointing in the direction of the current, and your thumb will point in the direction of the magnetic field. Next, an important term is introduced: magnetic flux. The total amount of magnetic field lines passing perpendicularly through a section is called the magnetic flux through that section, abbreviated as magnetic flux. Magnetic field lines are generated by the solenoid and actually exist, though they are invisible and intangible. Magnetic field lines form a closed loop, and for a solenoid, they all pass through the interior of the solenoid. The magnetic field lines are proportional to the magnetic induction intensity B. Figure 3 shows a schematic diagram of the magnetic field lines generated by a solenoid.

Magnetic flux is denoted by F, is a scalar quantity, and has units of webers (Wb). The relationship between magnetic flux, magnetic induction intensity B, and cross-sectional area A is:

F = BA

From this relationship, it can be seen that the more magnetic field lines passing through the cross-section, the greater the magnetic flux. For a coil wound around a magnetic core with current i flowing through it, the inductance L of the coil can be expressed as:

L = NF / i

where N is the number of turns of the coil.

From the above overview, it can be understood that for a coil wound around a magnetic core with a constant number of turns and current, the more magnetic field lines passing through the magnetic core, the greater the magnetic flux, and consequently, the greater the inductance. The inherent function of an inductor is to resist changes in the current flowing through it, which is essentially to resist changes in its magnetic flux. This is the basic principle behind using a common-mode choke to suppress common-mode currents.

The magnetic induction intensities generated by common-mode currents in a common-mode choke are B1 for current I1 and B2 for current I2. The two yellow arrows in Figure 3 represent the magnetic field lines generated by currents I1 and I2 in the ferrite core, respectively. It can be seen that the magnetic field lines generated by I1 and I2 add up, so do the magnetic fluxes, and consequently, the inductances add up. The greater the inductance, the stronger the suppression of the currents.

In one sentence, the suppression of common-mode currents by a common-mode choke can be explained as follows: When common-mode currents flow through a common-mode choke, the magnetic fluxes in the magnetic ring add up, resulting in a significant inductance that suppresses the common-mode currents.

When differential-mode currents flow through the two coils, the magnetic field lines in the ferrite magnetic ring are in opposite directions, causing the magnetic fluxes to cancel each other out, resulting in almost no inductance. Therefore, differential-mode signals can pass through with minimal attenuation (considering the inherent resistance of the inductor). Therefore, not only for the input filter of a switching power supply but also when routing differential signal lines, a common-mode choke can be added to suppress common-mode currents and prevent phenomena such as false triggering of the circuit.

Based on the requirements for rated current, DC resistance, and impedance at the rated frequency of a common-mode choke, the design can proceed step by step as follows:

The formula for calculating the minimum inductance value of a common-mode choke is:

where Xl is the impedance value at frequency f.

The inductance value of a choke is determined by dividing the load (in Ohms) by the angular frequency or higher frequency at which the signal begins to attenuate. For example, in a 50Ω load, if the signal begins to attenuate at or above 4000 Hz, an inductance of 1.99 mH (50 / (2π × 4000)) is required. The corresponding construction of the common-mode filter is as follows:

When selecting the frequency band to be filtered, the higher the common-mode impedance, the better. Therefore, when choosing a common-mode choke, it is necessary to refer to the device's datasheet, primarily based on the impedance-frequency curve.

After calculating the inductance, the subsequent design steps are similar to those for designing a regular inductor and will not be elaborated on here.

When winding your own inductor, note the following:

When selecting the magnetic core for a common-mode choke, factors such as shape, size, applicable frequency band, temperature rise, and price should be considered. Common magnetic core shapes include U-shaped, E-shaped, and toroidal.

Relatively speaking, toroidal cores are cheaper because only one is needed for fabrication. Other shapes of cores must come in pairs for use in common-mode chokes, and during molding, the pairing of the two cores must be considered, requiring an additional grinding process to achieve a higher permeability. This is not necessary for toroidal cores. Compared to other shapes, toroidal cores have a higher effective permeability because, no matter how they are assembled, there will always be an air gap phenomenon between paired cores, resulting in lower effective permeability than a single closed-form core. However, the winding cost of toroidal cores is higher because other shapes of cores have a matching bobbin that can be used for machine winding, while toroidal cores can only be wound manually or by machine (at a lower speed). Moreover, the small aperture of toroidal cores makes it difficult for machines to thread the wire, requiring manual winding, which is time-consuming, labor-intensive, and results in high processing costs and low efficiency. Installation is also inconvenient, and if a base is added, the cost will increase. Considering the overall performance, toroidal cores have better performance but also a higher price. Due to cost factors, toroidal cores are mostly used in high-power power supplies. Of course, due to their small size, they can also be used in low-power power supplies where size is a concern. For common-mode chokes whose main function is to filter out low-frequency noise, high-permeability manganese-zinc ferrite cores should be selected. Conversely, for high-frequency applications, nickel-zinc ferrite cores or powder cores should be selected. Magnetic cores suitable for high frequencies generally have distributed air gaps, resulting in relatively lower permeability; you cannot have both. However, unlike regular inductors, the function of a common-mode choke is to create a large insertion loss for noise signals to reduce noise interference. Although the effective permeability of manganese-zinc ferrite is very small at high frequencies, the core loss increases with frequency, providing significant impedance to high-frequency noise and thus weakening high-frequency interference, albeit with relatively poorer performance. Nevertheless, larger core losses can cause the core to heat up, and cores with smaller losses are also more expensive.