الفهرس | Only 14 pages are availabe for public view |
Abstract The accuracy of a digital filter is limited by the finite wordlen,-th used in its implementation. These problems are investigated extensively, and techniques are developed to treat the effects of the three sources of error: input quantization, coefficient inaccuracy, and power limitations. A statistical method is described for designing a digital filter in the presence of input quantization. The design criteria are based mainly on the minimization of the mean-square error between the desired output and the actual output. The design procedure is applied to an illustrative example. Special attention is directed to the problems of coefficient accuracy and of rounding effect. A new sensitivity analysis technique is presented to analyze the errors in pole positions and still keep the filter stable. Two improved realization are proposed which lead to a digital filter that is less sensitive to parameter quantization than are the standard realization.The first-, technique uses the dipole method of compensation. The dipole is added near z=+l in such a way that it does not affect the bandwidth and the transient overshoot of the designed system. An expression relating the steady state |