Common-Mode Chokes: From Working Principles to Important Performance Parameters

Noise is classified as either common-mode or differential-mode, depending on how it’s conducted. If we don’t identify the noise mode correctly, we risk making the noise situation worse by adding inappropriate noise-suppression components to the circuit. In this article, we’ll discuss a solution for common-mode noise: the common-mode choke (CMC), which attenuates common-mode signals while allowing differential signals to pass through with ideally no attenuation.

A previous article introduced the issue of common-mode noise in high-speed, differential digital interfaces like USB, HDMI, and Ethernet. CMCs are useful in these and other differential data transmission applications. They’re also used to filter out the noise of power lines in switch-mode power supplies and AC/DC rectifiers. Figure 1 shows two different applications for CMCs.

 

Figure 1. Applications of CMCs in power line and signal line filtering. Image used courtesy of Murata

 

How Does a CMC Work?

A common-mode choke consists of two windings wound around a magnetic core. Figure 2 shows a CMC for power line filtering applications.

 

Figure 2. A CMC for power applications. Image used courtesy of Octopart

 

The directions of the windings with respect to one another play a key role in the operation of a CMC. They’re chosen so that the device presents a high impedance to common-mode signals while passing differential signals largely unaffected.

We can understand the operation of a CMC by using Faraday’s Law and the right-hand rule. Consider the currents in Figure 3, which shows a CMC when a differential signal is applied.

 

Figure 3. CMC excited by a differential signal. Image used courtesy of Pulse Electronics

 

With a differential signal, the two windings produce magnetic fluxes that are equal in magnitude but opposite in direction. Since the magnetic fluxes cancel each other out, the filter should have a negligible impact on the signal, allowing it to to pass with minimal attenuation. Due to this flux cancellation, a differential signal can’t drive the core of a CMC into saturation.

Figure 4 shows how the CMC responds to a common-mode signal.

 

Figure 4. CMC excited by a common-mode signal. Image used courtesy of Pulse Electronics

 

With a common-mode signal, the magnetic fluxes from the two windings are in the same direction, producing a large inductive impedance. By presenting a high impedance, the device effectively suppresses high-frequency common-mode noise in the lines.

 

Calculating Differential and Common-Mode Impedance

To quantify the above qualitative description, let’s calculate the differential and common-mode impedances of a CMC. Figure 5, which consists of two coupled inductors, represents the simplest circuit model we can use. For a differential impedance calculation, we apply a differential signal and connect the outputs to ground.

 

Figure 5. A simple circuit model of a CMC for calculating differential impedance. Image used courtesy of Steve Arar

 

In the above model, R represents the copper losses of the windings. The impedance of each winding is:

$$Z_{dm} ~=~ \frac{V_s}{I_1}~=~ jL \omega ~-~ jM \omega ~+~ R ~\approx~ R$$

Equation 1.

 

where it’s assumed that maximum coupling exists between the windings (L₁ = L₂ = M). This means that the filter ideally presents a small resistive impedance to a differential signal. As we’ll discuss later on in the article, this impedance should be kept as low as possible.

Figure 6 models a common-mode excitation in the same circuit.

 

Figure 6. A simplified model of a CMC for calculating common-mode impedance. Image used courtesy of Steve Arar

 

In this case, the impedance of each winding is:

$$Z_{cm} ~=~ \frac{V_s}{I_1}~=~ jL \omega ~+~ jM \omega ~+~ R ~\approx~ j \omega(2L)$$

Equation 2.

 

where L⍵ ≫ R. The common-mode input impedance is thus very high, especially at frequencies where a strong coupling exists between the two windings. If the CMC uses a magnetic core, the coupling—and, by extension, the common-mode impedance—will be higher at lower frequencies where the core is more effective in boosting the inductance.

 

What If We Used Two Uncoupled Inductors?

Though we could use uncoupled inductors to suppress both differential and common-mode noise, CMCs have some important advantages over separate inductors. For example, due to the flux cancellation during differential-mode excitation, the core of a CMC doesn’t saturate during normal operation. This is true even when a relatively large current flows through the circuit. It’s therefore easier to use a CMC for noise suppression on lines with large current flows, such as AC/DC power-supply lines.

In high-speed digital interfaces, the fact that CMCs are invisible to differential signals also represents an advantage. Figure 7 illustrates the difference between using uncoupled and coupled inductors to filter noise out of a differential signal.

 

Figure 7. A differential signal before and after filtering with uncoupled inductors (a) and a CMC (b). Image used courtesy of Murata

 

In Figure 7(a), two separate inductors are used. The filtering effect smooths out the edges and distorts the signal. This increase in the rise time can be detrimental to the signal integrity and lead to intersymbol interference. By contrast, the ideal CMC shown in Figure 7(b) doesn’t slow the edges down at all.

 

Variation of Impedance With Frequency

In Figures 5 and 6, we used two coupled inductors to model the CMC. This simplified circuit model ignores the parasitic capacitances of the windings. By taking the intra-winding capacitances into account, we obtain the more elaborate model in Figure 8.

 

Figure 8. Equivalent circuit model of a CMC for a common-mode excitation. Image used courtesy of Abracon

 

The winding capacitance plays a key role in the frequency response of the CMC. Having a parallel RLC circuit at the heart of the equivalent model means that there’s a resonant frequency at which the parallel LC circuit behaves like an open circuit. At this frequency, the impedance of the parallel RLC circuit is at its maximum and is equal to R_ac.

Below the resonant frequency, the circuit behaves inductively. However, as we move to frequencies above the resonant frequency, the circuit's behavior becomes capacitive. Figure 9 shows how the impedances of several different CMCs from Pulse Electronics change with frequency.

 

Figure 9. Common-mode impedance of several CMCs. Image used courtesy of Pulse Electronics

 

If the manufacturer doesn’t provide a model for the CMC, we can use lab measurements to estimate parameters for the model in Figure 8. This model can then be used to simulate the CMC’s effect on common-mode noise.

 

Selecting a Common-Mode Choke

When selecting a common-mode choke, we should consider both its common-mode and differential-mode impedances. The differential impedance should be as low as possible so that the desired signal can pass unaffected. The filter's ability to suppress the noise depends directly on its common-mode impedance, however—the higher the common-mode impedance, the better the noise suppression.

We also need to consider how the common-mode impedance varies with frequency to make sure that it’s acceptably high in the frequency range of interest. Note that a higher common-mode impedance typically corresponds to a larger component size, which can be harder to fit in dense PCB designs.

Figure 10 shows the differential and common-mode impedances of two CMCs from Murata.

 

Figure 10. CMC datasheets provide differential and common-mode impedance vs. frequency. Image used courtesy of Murata

 

In the above figure, the common-mode impedance of the DLMNSN900HY2 is greater than 2 kΩ at around 900 MHz. Its differential impedance at the same frequency is about 200 Ω. The common-mode impedance of the DLM0NSN500HY2 peaks above 1 kΩ at approximately 1,000 MHz, where its differential impedance is only about 100 Ω.

Some datasheets also plot differential-mode insertion loss across frequency. The insertion loss curves for the above Murata devices are shown in Figure 11.

 

Figure 11. Insertion loss of two CMCs from Murata. Image used courtesy of Murata

 

Two Different Winding Methods: Bifilar and Sectional

A CMC can use either bifilar or sectional windings. Both are illustrated in Figure 12.

 

Figure 12. Bifilar (left) and sectional (right) windings used in CMCs. Image used courtesy of Würth Elektronik

 

Because they have a lower leakage inductance and exhibit smaller attenuation for differential signals, bifilar-wound components are commonly used in high-speed differential signaling applications. Sectional-wound CMCs have a larger separation between the windings, making them more suitable for high voltages. However, this separation leads to a higher leakage inductance and a higher differential impedance. To learn more about the pros and cons of these windings, check out this helpful white paper from Würth Elektronik.