An improved design method for even-degree optimum allpole filters with equiripple passband responses

Abstract

The maximum power transfer between source and load is extremely important in analog signal processing. The main disadvantage of signal processing using the even-degree Chebyshev filter is the impossibility of LC ladder network realization between two equal resistors. Therefore, either the utilization of the ideal transformer or the Chebyshev filter of odd-degree is required to achieve the maximum power transfer. To overcome this limitation, the class of even-degree polynomials with the equiripple behavior, referred to as scaled modified Chebyshev polynomials, can be used as the filter's characteristic function. Passband ripples of obtained filters can be controlled through an embedded parameter, while normalized half-power frequency is kept equal to unity. Furthermore, these filters can be optimized for the minimum amount of reflected power at the input terminals. In this paper, the optimum modified Chebyshev filters are first compared to the classic Chebyshev filters to show that both filters exhibit similar performances. Second, proposed filters are compared to the Least Square Monotonic (LSM) filters that are optimized by minimizing the area under the characteristic function in the passband. Results of comparison reveal that the proposed even-degree scaled modified Chebyshev filters outperform the LSM filters, as is already proved for the odd-degree filters.