Switching mode power supply noise source impedance measurement and EMI filter characterization

Abstract

The effectiveness of an EMI filter is closely related to the impedance of the noise source. Due to the time-varying nature of the noise source impedance of an offline power supply, the measurement of it cannot be done using conventional impedance measuring methods. Two new methods, the insertion loss method and the signal injection method, are proposed to perform such a measurement. The insertion loss method utilizes the EMI emission as a signal source and derives the source impedance by measuring the emission attenuation caused by an inserted impedance. The signal injection method injects a signal to the power supply and measures the response. The insertion loss method is verified experimentally both for common mode and differential mode, The signal injection method for common mode source impedance measurement is experimentally verified. That for differential mode source impedance measurement requires a faster equipment and is not fully verified.

An EMI filter is usually characterized in the manufacturer’s catalog by its attenuation in a standard system (50Ω source impedance and 50Ω load impedance). The effectiveness of the filter in a practical system may significantly deviate from the manufacturer’s data. To provide the users with practically useful information, a scheme to characterize the EMI filter, the impedance matrix approach is proposed. This approach takes parasitic effect into consideration and the parameters can be measured relatively easily. The approach is verified experimentally by applying a commercial EMI filter to a power supply and comparing the predicted attenuation with the measured one.