“The EMI of switching regulators is divided into electromagnetic radiation and conducted radiation (CE). This article focuses on conducted emissions, which can be further divided into two categories: common mode (CM) noise and differential mode (DM) noise. Why distinguish CM-DM? EMI suppression techniques that work for CM noise are not necessarily effective for DM noise and vice versa, so identifying the source of conducted emissions can save time and money spent on noise suppression. This article describes a practical method to separate CM and DM emissions from a switching regulator controlled by the LTC7818.Knowing where CM noise and DM noise appear in the CE spectrum allows power supply designers to effectively apply EMI suppression techniques, which
By: Analog Devices – Ling Jiang, Applications Engineer; Frank Wang, EMI Engineer;
Keith Szolusha, Application Director; Kurk Mathews, Senior Application Manager
The EMI of switching regulators is divided into electromagnetic radiation and conducted radiation (CE). This article focuses on conducted emissions, which can be further divided into two categories: common mode (CM) noise and differential mode (DM) noise. Why distinguish CM-DM? EMI suppression techniques that work for CM noise are not necessarily effective for DM noise and vice versa, so identifying the source of conducted emissions can save time and money spent on noise suppression. This article describes a practical method to separate CM and DM emissions from a switching regulator controlled by the LTC7818. Knowing where CM noise and DM noise appear in the CE spectrum allows power supply designers to effectively apply EMI suppression techniques, which can save design time and BOM costs in the long run.
Figure 1. CM and DM noise paths in a buck converter
Figure 1 shows the CM noise and DM noise paths of a typical buck converter. DM noise is generated between the power and return lines, while CM noise is generated through the stray capacitance CSTRAYGenerated between power lines and ground planes such as copper test benches. The LISN for CE measurement is located between the power supply and the buck converter. The LISN itself cannot be used to directly measure CM and DM noise, but it does measure power and return power line noise—V1 and V2 in Figure 1, respectively. These voltages are measured across 50Ω resistors. According to the definition of CM and DM noise, as shown in Figure 1, V1 and V2 can be expressed as CM voltage (VCM) and DM voltage (VDM) and the difference. Therefore, the average of V1 and V2 is VCMand half the difference between V1 and V2 is VDM.
Measure CM noise and DM noise
A T-type power combiner is a passive device that combines two input signals into one port output. The 0° synthesizer produces the vector sum of the input signals at the output port, while the 180° synthesizer produces the vector difference of the input signals1. Therefore, a 0° synthesizer can be used to generate VCMthe 180° synthesizer produces VDM.
The two synthesizers ZFSC-2-1W+ (0°) and ZFSCJ-2-1+ (180°) shown in Figure 2 from Mini-Circuits are used to measure V from 1 MHz to 108 MHzCMand VDM. For these devices, measurement error increases at frequencies below 1 MHz. For lower frequency measurements, other synthesizers such as ZMSC-2-1+ (0°) and ZMSCJ-2-2 (180°) should be used.
Figure 2. 0° and 180° synthesizers
Figure 3. Used to measure (a) VCMand (b) VDMexperimental setup
Figure 4. Test Setup for Measuring CM Noise and DM Noise
The test setup is shown in Figure 3. A power combiner has been added to the standard CE test setup. The outputs of the LISN for the power and return lines are connected to the input port 1 and input port 2 of the synthesizer, respectively. The output voltage of the 0° synthesizer is VS_CM = V1 + V2; the output voltage of the 180° combiner is V S_DM = V1 C V2.
The output signal of the synthesizer VS_CMand VS_DMmust be processed in the test receiver to generate VCMand VDM. First, the power combiner has specified the insertion loss to be compensated in the receiver. Second, since VCM = 0.5VS_CMand VDM = 0.5VS_DM, so the test receiver subtracts an additional 6 dBμV from the received signal. After compensating for these two factors, the measured CM noise and DM noise are read out in the test receiver.
Experimental Verification of CM Noise and DM Noise Measurements
This method is verified using a standard demo board with dual buck converters. The demo board switches at 2.2 MHz, VIN = 12 V, VOUT1 = 3.3 V, IOUT1 = 10A, VOUT2 = 5 V, IOUT2 = 10A. Figure 4 shows the test setup in the EMI chamber.
Figures 5 and 6 show the test results. In Figure 5, the higher EMI curve represents the total voltage method CE measured using the standard CISPR 25 setting, while the lower radiated curve represents the separated CM noise measured with the addition of a 0° combiner. In Figure 6, the higher radiated curve represents the total CE, while the lower EMI curve represents the measured discrete DM noise after adding a 180° combiner. These test results are in line with theoretical analysis, showing that DM noise dominates in the lower frequency range, while CM noise dominates in the higher frequency range.
Figure 5. Measured CM Noise vs. Total Noise
Figure 6. Measured DM Noise vs. Total Noise
The adjusted demo board is CISPR 25 Class 5 compliant
According to the measurement results, the total radiated noise exceeds the CISPR 25 Class 5 limit in the 30 MHz to 108 MHz range. By separating the CM and DM noise measurements, it was found that the high conducted emissions in this range appear to be caused by CM noise. There is little point in adding or enhancing DM EMI filters or otherwise reducing input ripple, as these suppression techniques do not reduce the problem-causing CM noise in that range.
Therefore, this demo board demonstrates a specific solution to CM noise. One of the sources of CM noise is high dV/dt signals in switching circuits. This noise level can be reduced by increasing the gate resistance to reduce dV/dt. As mentioned earlier, CM noise passes through the stray capacitance CSTRAYGo through LISN. CSTRAYThe smaller it is, the lower the CM noise detected in the LISN. To reduce CSTRAY, the copper area of the switch node on this demo board should be reduced. In addition, a CM EMI filter is added at the converter input to obtain high CM impedance, thereby reducing the CM noise entering the LISN. By implementing these measures, the noise in the 30 MHz to 108 MHz range is sufficiently reduced to comply with the CISPR 25 Class 5 standard, as shown in Figure 7.
Figure 7. Overall Noise Improvement
This paper presents a practical method for measuring and separating CM noise and DM noise from total conducted emissions, and it is validated with test results. If designers can separate CM and DM noise, they can implement CM- or DM-specific mitigation solutions to effectively suppress noise. In conclusion, this approach helps to quickly find the root cause of EMI failures and saves time in EMI design.
“AN-10-006: Understanding Power Splitters.” Mini-Circuits, April 2015.
About the Author
Ling Jiang graduated from the University of Tennessee Knoxville in 2018 with a Ph.D. in Electrical Engineering. After graduation, she joined Analog Devices’ Power Products Group in Santa Clara, California, USA. Ling is an applications engineer supporting controllers and µModule devices for automotive, data center, industrial, and other applications. Contact information:[email protected]