Elsevier

Microelectronics Journal

Volume 37, Issue 8, August 2006, Pages 792-799
Microelectronics Journal

Automated synthesis of passive filter circuits including parasitic effects by genetic programming

https://doi.org/10.1016/j.mejo.2005.12.012Get rights and content

Abstract

In this paper, we propose a genetic programing method to synthesize passive filter circuits including parasitic effects, which are very common in high-frequency application. This approach allows circuit topology and component values to be evolved simultaneously; therefore, novel circuits different from those generated by traditional methods can be explored. Experimental results show the proposed method can effectively generate not only compliant but also efficient solutions of such problems where the traditional approaches fail.

Introduction

Traditional passive filter synthesis methods treat circuit components as ideal ones. Unfortunately, this assumption is no longer suitable for high-frequency application because of parasitic effects. For a given specification, a compliant circuit resulting from traditional design methods may fail to satisfy the same specification when the parasitic effects are taken into account. The authors of [1] have proposed a genetic algorithm to synthesize analog circuits including parasitic effects. However, the genetic algorithm approach deals with only the evolution of component values, and the circuit topology should be predetermined by other methods. Some methods have been proposed to deal with the evolution of circuit topology as well as the component values and shown their capability of generating a variety of circuit topologies [2], [3], [4], [5], [6], [7], [8], [9], [10], [11]. In our previous work, we proposed a novel tree representation of RLC circuits and a genetic programing (GP) method to synthesize passive filter circuits [12], [13]. This approach allows circuit topology as well as component values to be evolved simultaneously. The proposed method is restricted to series–parallel circuit topology. However, this restriction makes it more efficient than those general-topology methods in passive filter synthesis if the target circuits are confined to series–parallel topology, which is commonly used in practice owing to less component value sensitivity and reduced circuit complexity [14]. Furthermore, for a given specification our method can automatically find efficient circuit (in terms of circuit complexity) without predetermining the number of the circuit components required. In this paper, we extend our previous work to take the parasitic effects into consideration. By exploring the possible solution space consisting of various circuit topologies and component values, the GP approach can generate circuits different from those generated by traditional methods. The experimental results show the extended GP method can effectively generate compliant circuits including the parasitic effects, while the traditional methods fail. In addition, the GP-evolved circuits are more efficient than the traditional ones.

The remainder of the paper is organized as follows. Section 2 introduces the tree representation of RLC circuits and a circuit analysis algorithm based on the tree representation. Section 3 introduces the GP method to synthesize passive filter circuits. Section 4 describes how to model the parasitic effects and how to incorporate them into the GP method. Experimental results are given in Section 5, and the conclusion is given in Section 6.

Section snippets

Tree representation of RLC circuits

We find that a binary tree structure can be used to represent series–parallel RLC circuits. An RLC circuit and the corresponding tree representation are shown in Fig. 1(a) and (b), respectively. Adopting the terminology commonly used in GP, we divide the nodes of the tree into two sets, the terminal set and the function set [15]. The non-terminal nodes belong to the function set including series and parallel nodes, which are denoted by + and //, respectively. The terminal nodes belong to the

Genetic programing

The main idea of GP is based on the evolutionary process observed in nature. To begin with, an initial population composed of a number of solutions (called individuals) is randomly generated. Each individual, in circuit synthesis problems, is a circuit consisting of circuit topology, component type, and component values. New potential solutions are created by genetic operations, such as crossover and mutation. Each individual is evaluated and assigned a scalar value (called fitness) reflecting

Parasitic effects

When operating in high frequencies, passive components are far from being ideal, but contain parasitic effects, which can be modeled as the equivalent circuits as shown in Fig. 5 [16]. In addition, circuit interconnect nodes are also non-ideal and can be modeled as capacitance between the interconnect node and the ground node. Fig. 6(a) illustrates how the ideal terminal node (i.e. L or C) is incorporated with the component parasitic effect subtree to create the equivalent subtree including the

Experimental results

Two low-pass filter design cases were considered in our experiment.

Conclusion

The proposed GP method has shown its capability of generating efficient passive filter circuits including parasitic effects, and it can include more complex parasitic effects as long as the effects can be modeled by parallel or series passive components. It should be mentioned that the proposed method is limited to the design cases where the parasitic effects are known. However, since the previously published work, such as [17], shows that genetic approaches can be applied to circuit extraction

References (17)

  • D.H. Horrocks, Y.M.A. Khalifa, Genetic algorithm design of electronic analogue circuits including parasitic effects,...
  • D. Andre et al.

    On the theory of designing circuits using genetic programming and a minimum of domain knowledge

    Proc. IEEE Int. Conf. Evol. Comput.

    (1998)
  • T.R. Dastidar et al.

    A synthesis system for analog circuits based on evolutionary search and topological reuse

    IEEE Trans. Evol. Comput.

    (2005)
  • J.B. Grimbleby

    Automatic analogue circuit synthesis using genetic algorithms

    IEE Proc. Circuits Devices Syst.

    (2000)
  • J.R. Koza et al.

    Automated WYWIWYG design of both the topology and component values of analog electrical circuits using genetic programming,

  • J.R. Koza et al.

    Automated synthesis of analog electrical circuits by means of genetic programming

    IEEE Trans. Evol. Comput.

    (1997)
  • J.R. Koza et al.

    Genetic Programming III: Darwinian Invention and Problem Solving

    (1999)
  • J.D. Lohn et al.

    A circuit representation technique for automated circuit design

    IEEE Trans. Evol. Comput.

    (1999)
There are more references available in the full text version of this article.

Cited by (2)

View full text