Quantized Probability Circuit Design Principles Applied to Linear Circuits

Abstract

A circuit design method for linear circuits is needed which will adequately predict circuit per formance as a function of component tolerances. This paper describes a method which weights the probable component variation from its nominal value into one of three groups; the group assignment depends on how seriously the component variation affects over-all performance This technique, identified as the quantized probability design (QPD) method, is compared with the absolute worst case (AWC), the Taylor worst case (TWC), and the uniform probability (UP) methods. The QPD procedure is given, based on the circuit performance equation. Two linear circuit applications are presented and analyzed whic show the effect each component will have on circuit performance. A comparison of design methods shows that the quantized probability design predicts less amplifier gain variation. Actual experience has shown that a closer correlation exists with the quantized probability design method, and therefore its use for linear circuits is recommended.