SLVAFJ9 March   2023 TPSF12C1 , TPSF12C1-Q1 , TPSF12C3 , TPSF12C3-Q1

 

  1.   Abstract
  2. 1 Table of Contents
  3.   Trademarks
  4. 2 Introduction
  5. 3 EMI Frequency Ranges
  6. 4 Passive EMI Filters for High-Power, Grid-Tied Applications
  7. 5 Active EMI Filters
  8. 6 Generalized AEF Circuits
  9. 7 Selection of the CM Active Filter Circuit
  10. 8 The Concept of Capacitive Amplification
  11. 9 Practical AEF Implementations
  12. 10Practical Results
    1. 10.1 Low-Voltage Testing
    2. 10.2 High-Voltage Testing
  13. 11Summary
  14. 12References

6 Generalized AEF Circuits

Figure 6-1 illustrates six active-filter configurations generalized according to the sensed noise parameter (voltage or current), the means by which the cancellation signal is injected (voltage or current), and the active control technique (FB or FF).

  • Voltage sense (VS) or current sense (CS)
  • Voltage injection (VI) or current injection (CI)
  • FB control or an FF control structure
Figure 6-1 Basic Active Filter Structures in Their Single-Phase Equivalents – Four FB and Two FF Circuits – Categorized According to their Control, Sensing and Injection Techniques: FB-CSVI (a); FB-CSCI (b); FB-VSVI (c); FB-VSCI (d); FF-VSVI (e); and FF-CSCI (f)

The terms iS and ZS in Figure 6-1 designate the Norton-equivalent noise current source and parallel source impedance of the power stage. ZL is the load impedance at the noise-receiving end (or the EMI victim); for example, the LISN for EMI measurement. G represents the gain of the active circuit. Adding different passive elements in place of ZS and ZL will form different hybrid circuits.

From a control standpoint, FB designs sense the residual disturbance at the EMI victim, invert the signal, amplify it with a high gain G, and inject a cancellation signal back into the system to drive the sensed parameter to zero over the required frequency range. In contrast, FF designs sense the disturbance at the EMI source, invert the signal, amplify it with unity gain and inject it back at the EMI victim. The amplifier unity-gain setting for FF must be highly accurate such that the EMI and anti-EMI signals cancel, making FF designs more difficult to design.

In terms of noise sensing, the VS and CS elements are typically capacitors and CS transformers (or auxiliary windings on existing magnetics), respectively. In terms of noise cancellation, VI designs use a controlled series voltage source to impede the flow of noise current to the LISN, whereas CI designs involve a controlled shunt current source to reroute the flow of the noise current produced by the noise source to prevent it from flowing in and being measured by the LISN. VI and CI designs effectively create a voltage divider and current divider, respectively, with the load. In general, transformers can embody the series element, whereas capacitors implement a shunt conduction path.

Table 6-1 summarizes the salient characteristics of the AEF circuits embodied in Figure 6-1, including expressions for insertion loss and circuit conditions for high attenuation [4]. YS and YL represent the admittance of the noise source and load, respectively, for a FB-VSCI design.

Table 6-1 AEF Circuits From Figure 5-1 Categorized by Topology (Control, Sensing and Injection Techniques)
AEF Topology Control (FB/FF) Sensing (VS/CS) Injection (VI/CI) Insertion Loss (IL) High-Attenuation Condition
a FB-CSVI Feedback Current Voltage 1 + G Z S + Z L G 1 >> Z S + Z L
b FB-CSCI Feedback Current Current 1 + Z S Z S + Z L ⋅ G Z S >> Z L
c FB-VSVI Feedback Voltage Voltage 1 + Z L Z S + Z L ⋅ G Z S << Z L
d FB-VSCI Feedback Voltage Current 1 + G Y S + Y L G >> Y S + Y L
e FF-VSVI Feedforward Voltage Voltage 1 1 - G ⋅ 1 - Z S Z S + Z L ⋅ G G = 1 , Z S << Z L
f FF-CSCI Feedforward Current Current 1 1 - G ⋅ 1 - Z L Z S + Z L ⋅ G G = 1 , Z S >> Z L

IL = iL,w/oAEF / iL,w/AEF is the quotient of the filter output current without and with AEF installed, normally measured with 50-Ω source and load impedances, and correlates to the achievable attenuation of EMI. As indicated in Table 6-1, each AEF topology requires a specific impedance behavior to achieve high attenuation.