Search

A new approach to the synthesis of element simulation type bandpass switched-capacitor (SC) filters associated with parasitic capacitances is presented. The synthesis of a parametric bandpass LCR prototype is considered as a starting point of the method. The proposed approach allows one to design two-phase controlled SC filters with a low number of operational amplifiers. A predistortion technique is used for elimination of the influence of the parasitics on the circuit characteristics. A theoretical discussion as well as the results of computer simulation on a twelfth-order parametric bandpass SC filter is included. Index Terms - Element simulation method, parasitic capacitance, predistortion technique.

A simple and low-cost analog neural network architecture for solving linear programming (LP) problems is described. The universal and flexible network architecture, which may be easily implemented using very-large-scale-integration technology, is based on nonlinear differential equation representations of LP problems. Its main components are a single neuron, adaptive synaptic weights which are modified by a learning algorithm and a pre-processing circuit.

To overcome the difficulty of discretization due to a very small penetration depth, the 3-D eddy current field problem is solved by the longitudinal component method together with a discrete Fourier transform (DFT) in cylindrica coordinates. Galerkin's FEM is applied to calculate the 3-D heat transfer in a conducting and permeable moving shell. Numerical results are presented.

This paper concerns the application aspect of the two-dimensional (2-D) LMS adaptive filter in image processing. It is a recent research interest to extend some successful adaptive filtering techniques to the 2-D case. The LMS adaptive filter (1) has drawn attention because of its simplicity, which is significant in 2-D applications. On the basis of analyzing the strong points and weakness of the 2-D LMS adaptive filter in image processing, an adaptive scheme is proposed whereby a local-mean estimation substructure is incorporated to overcome the problem of the 2-D LMS adaptive filter in dealing with signals having a nonzero mean with, especially, abrupt variations (edges), which is a typical feature of practical images. The advantages of this scheme over the direct use of the 2-D LMS are considered. As a direct application, this scheme is used for adaptive cancellation of an artifact disturbance in an image. An implementation method of the local-mean estimation substructure with the edge-preserving property is introduced. With this local-mean estimator, the proposed scheme is applied for enhancement of images disturbed by wide-band noises. All of the application examples presented in this paper illustrate some useful specialties of the 2-D LMS adaptive filter in image processing and the improved performance of the proposed scheme.

The multiprocessor realization of two-dimensional denominator-separable digital filters is presented. A configuration, based on a structure proposed by M.Y. Dabbagh and W.E. Alexander, is obtained by keeping the throughput delay a minimum while optimizing the parallelism and pipelining. Suitable for very large scale integration implementation, the structure is comparable to the symmetric fan filters presented by the two authors in another paper.

The quantization effects for the systolic structure of a 2-dimensional IIR digital filter proposed by Sid-Ahmed are analyzed. Results show that the combined output error in the systolic realization is almost the same as in the direct form realization, but the storage error additionally occurs in the systolic realization. Lastly, the combined error at the output of each processing element and the total output error of the filter are given.

Low sensitivity, low roundoff noise and freedom from overflow oscillations are often the most desired properties in the realization of recursive digital filters. If a is a pole of a given transfer function, the complex first-order transfer function square root of 1 - absolute value of a square/(z - a) represents a basic building block in various favorable realizations. In this brief, the realization of this basic building block and its properties are investigated. It is shown that the structure is automatically scaled (normalized) and that it is a minimum norm structure (MNS) possessing e.g., the low-sensitivity and low-noise properties. So the structure is referred to as the normalized MNS. An elegant noise analysis of this structure is also given. As a direct application of the normalized MNS, a simple method is given to derive efficient structures for second-order filters. A particular second-order structure is suggested for the very low sensitivity realization of classical filters.

This paper presents a semianalytical approach for calculating the electromagnetic field due to a time-varying, three-dimensional incident field which is caused by a line current situated in the cavity of a finitely long circular cylindrical conducting and permeable shell of finite thickness. The 3-D Helmholtz equations are solved using the Fourier coefficients of the longitudinal field components. The Fourier coefficients of the transversal components are expressed in terms of the Fourier coefficients of the longitudinal components. To deal with the complicated incident field and the finite length of the shell a Fourier expansion of the field is made in the azimuthal and in the axial direction.

Results of theoretical and experimental investigations of the element simulation type switched-capacitor (SC) filters based on Lee--Chang imitators are discussed. The main characteristics of the filters, including frequency response, noise, and nonlinear distortion, are under consideration. Practically oriented methods for estimation of the noise and harmonics are proposed. The application of these methods is illustrated by examples of low-pass and bandpass SC-filter design. Index Terms--Continuant, high-order SC filter, noise analysis, nonlineaer distortion analysis, parasitic capacitor.