Protective Earth in EMC: Why the Safety Ground Wire Often Increases Conducted Emissions

Protective earth is mandatory for safety, but in many power electronic systems it forms the dominant return path for common-mode current, which can increase conducted emissions if not carefully managed.

Protective earth (PE), commonly recognised as the green-and-yellow safety earth conductor, is essential for electrical safety and regulatory compliance. However, its role in electromagnetic compatibility (EMC) is often misunderstood. Many engineers assume that bonding equipment to earth will automatically reduce noise by “grounding” the system. In reality, when analysed from a conducted emissions perspective, the protective earth wire frequently provides an efficient return path for common-mode currents and can increase measured EMI levels. Understanding how and why this occurs is critical when designing power electronics equipment, particularly switched-mode power supplies, motor drives, and other mains-connected products.

Before diving into the technical details, it is helpful to briefly describe the test set-up. In this article, we use a typical AC-DC switched-mode power supply as an example in a conducted emissions measurement configuration. The test set-up follows standard practice and is installed in our office laboratory environment, which provides reliable accuracy for EMC investigations. Readers interested in the full details of the conducted emissions set-up can refer to this blog.

The LISN used in this demonstration is Tekbox TBL5008-2B. The dual measurement outputs allow the use of LISN Mate (TBLM2), which is a useful tool for mode separation. Multiple RF current probes are also used to visualise common-mode current flow. Both spectrum analysers and oscilloscopes are used throughout the measurements.

The device under test (DUT) was selected because it is a low-cost power supply that fails EMC conducted emissions limits. For demonstration purposes, most of the input filter components — primarily the common-mode choke and Y-capacitors — were removed. In products of this type, the main switching devices are often thermally coupled to the metal enclosure. Through parasitic capacitance, switching noise can therefore be injected into the metal case. From a safety standpoint, the metal enclosure must be connected to the PE conductor. Figure 1(a) shows the test set-up and Figure 1(b) shows the internal construction of the DUT.

Figure 1 Conducted emissions test set-up and the device under test

A short educational note is useful here. In conducted emissions testing, the LISN is the most important element of the measurement set-up. Figure 2 shows a simplified LISN model typically used for commercial conducted emissions testing. In this example, a table-top DUT is used, and therefore the DUT is placed on an 80 cm high insulated support, with no bonding straps connected to the test ground plane. For floor-standing equipment, additional earth straps are sometimes required, which can significantly influence common-mode current flow, particularly at lower frequencies.

To help illustrate current flow, the reactance values of the LISN inductors and capacitors are shown at 150 kHz and 30 MHz. The reader can clearly see that the LISN L-C network blocks high-frequency current from returning to the mains supply. The key element to focus on is the 50 Ω resistor in parallel with the 1 kΩ resistor, which effectively forms a low-impedance path close to 50 Ω inside the LISN.

For AC mains LISNs, one 50 Ω impedance represents the RF input impedance of the receiver or spectrum analyser. The second 50 Ω impedance depends on the LISN design. If the LISN has a single measurement port with an internal switch between Live and Neutral, the second 50 Ω termination is automatically switched in. In LISNs with dual measurement ports, as used in this demonstration, a 50 Ω termination must be manually connected to the unused port. This is extremely important, as illustrated in the schematic.

In the schematic, the mains cable is modelled as a simple R-L network. In a typical conducted emissions test, the cable length is maintained at approximately one metre. Although a more accurate lumped L-C-R model could be used, a simplified R-L representation is sufficient for this discussion.

Figure 2 Equivalent circuit of the conducted emission set-up

Figure 3 illustrates the differential-mode noise loop in this configuration. This loop can be drawn with confidence because current naturally flows through the lowest impedance path. Effectively, differential-mode voltage is developed across an impedance of approximately 100 Ω, representing two 50 Ω resistances in series. Under normal conditions, the PE conductor plays little role in differential-mode current flow, and theoretically there should be no differential-mode current circulating in the green-and-yellow conductor.

Figure 3 The differential mode current loop in the conducted emission set-up

We now move to the more complex topic of common-mode currents.

For readers with an EMC or RF background, it is well understood that current is a by-product of energy transfer. As energy moves from one location to another, it creates a propagating voltage wavefront accompanied by forward and return currents. Importantly, energy propagates through electromagnetic fields in space, not inside conductors, as is often assumed. As a result, forward and return currents may flow on any available conductive structure.

If return current flows through earth or chassis instead of through the intended return conductor — such as Neutral in mains systems or 0 V in DC systems — this is defined as common-mode current. In AC mains powered systems, this means current flowing on Live and Neutral that is common with respect to earth. In automotive or vehicle systems, currents flowing on both supply and return wires relative to chassis would also be considered common-mode currents.

Since the objective of this article is to understand the role of the PE conductor, we first examine a scenario where the PE wire is disconnected from the DUT. In this condition, the common-mode current path is relatively straightforward and is illustrated in Figure 4. The noise sources are modelled as voltage sources present on both Live and Neutral conductors.

In this configuration, current is injected into the test ground plane through parasitic capacitance between the DUT and the ground plane. For consistency in the explanation, this capacitively coupled current is treated as the forward current, while the return current flows through the Live and Neutral conductors. As shown in the diagram, common-mode current splits at the LISN connection point (Point A) and flows through both Live and Neutral conductors. From a common-mode perspective, the current encounters two 50 Ω impedances in parallel.

Figure 4 Common mode current loop when the PE wire is disconnected from the DUT earthing point

This behaviour can be verified using measurement techniques. Two matched RF current probes (Tekbox TBCP2-500) are placed around the mains cable, one close to the DUT, the other next to the LISN mains. Since the PE conductor is disconnected, this pair of matched probes measure only the common-mode current flowing through Live and Neutral. A surface current probe (TBSCP1-5M300) is then placed on the test ground plane to capture current flowing through the ground plane.

To observe current direction, an oscilloscope is used because it can capture real-time multi-channel waveforms. If two measured waveforms are in phase, the currents flow in the same direction. If they are out of phase, the currents flow in opposite directions.

Figure 5 Multiple current probes are used to capture the common mode currents in the test set-up

Figure 6 shows the captured waveforms. Probes 1 and 2 produce nearly identical results, as expected, since they measure current on the same cable at different locations. The more interesting observation is from Probe 3 (Channel 4), which measures ground plane current. Although the waveform shows partial phase opposition relative to Probes 1 and 2, it is not perfectly 180 degrees out of phase. This is likely due to the difficulty of accurately measuring distributed RF current on a large ground plane. Nevertheless, the overall waveform shape supports the theoretical current flow model.

It is also important to note that because the transfer impedances of the probes differ, the measured voltages displayed in millivolts do not represent absolute current magnitude. However, it is reasonable to assume that the currents should be approximately equal in magnitude and opposite in phase.

Figure 6 Measured currents using RF current probes

So far, this discussion has excluded the PE conductor and presents a relatively straightforward scenario. We now reconnect the PE conductor to the DUT enclosure. At this point, the forward current path becomes dependent on the impedance of the green-and-yellow conductor and associated parasitic impedances, both of which vary with frequency.

Figure 7 The PE wire offers another path

From Figure 7, it is clear that, particularly at lower frequencies, the PE conductor provides a significantly lower impedance path for common-mode current (Path 1) compared to the capacitive path through the insulation support (Path 2). This results in increased forward current, which is consistent with experimental observations (as shown in Figure 8).

Figure 8 Common mode current loop when the PE wire is introduced to the system

Figure 9 shows that conducted emission levels increase when the PE conductor is connected. Here we only show the comparison of quasi-peak results.

Figure 9 The conducted emissions increased when the PE wire is connected to(Green)

This leads to an important question. Does the PE conductor simply carry the forward current, while Live and Neutral carry the return current? What role does the test ground plane play?

At higher frequencies, parasitic capacitance between the DUT and the ground plane — largely determined by the insulation support height — allows current to flow through Path 2 (Figure 7) as well. The question then becomes whether this current acts as forward current (same direction as PE current) or return current.

To investigate this, current measurements were performed in the following configurations using multiple RF current probes and an oscilloscope.

Figure 10 Current probes measurement on CM currents

Shown in Figure 11, the results show:

1. The current flowing through the PE conductor is opposite in phase to the current flowing through Live and Neutral.

2. The amplitude of current in the PE conductor is slightly higher than the CM currents in Live and Neutral.

3. Mathematical summation of currents measured on Live, Neutral, and PE demonstrates a net current component that is in phase with the PE conductor.

Figure 11 Currents from the measurement results

Based on the measurement results, the common-mode current path can be illustrated as shown in Figure 12. The arrows indicate the direction of current flow on the test ground plane. To verify this behaviour, we measured the ground plane current using a surface current probe and compared it with the net common-mode current flowing on the AC mains cable (L, N, and PE combined). The measurement results shown in Figure 13 support this current-flow assumption.

Figure 12 Common mode current loop when the PE is connected to the DUT

Figure 13 measured results

In real systems, current distribution is often significantly more complex. The presence of input EMI filters, enclosure bonding arrangements, and cable routing can all influence common-mode current behaviour (i.e the common mode current directions can change). In Figure 14, for example, when the product is properly designed with effective filtering, the presence or absence of the PE wire does not result in a significant difference in the conducted emissions performance. Nevertheless, this study clearly demonstrates that introducing a PE conductor can substantially change common-mode current paths and, in many cases, increase conducted emissions.

Figure 14 When the product is properly designed with effective filtering, the presence or absence of the PE wire does not result in a significant difference in the conducted emissions performance.

Conclusion

While protective earth is essential for electrical safety, it should not automatically be assumed to improve EMC performance. In many power electronic systems, the PE conductor forms a low-impedance path for common-mode current and can increase conducted emissions if not carefully considered. Understanding the interaction between safety earthing, parasitic coupling, and measurement system impedance is therefore critical when designing EMC-compliant equipment.