Circuit yield estimation method based on CAD (computer aided design) Monte Carlo analysis

Abstract

The invention discloses a circuit yield estimation method based on CAD Monte Carlo analysis. The method comprises steps as follows: a topological structure of an analyzed electronic circuit or integrated circuit, a performance parameter index range of the circuit and statistical distribution laws of a key element parameter and an element parameter are determined; an agent model between a circuit performance parameter and an element parameter nominal value is established for approximating a circuit; a random sampling combination sequence of circuit Monte Carlo analysis is generated according to the circuit element parameter distribution laws; the agent model of the circuit performance parameter is used for predicating the random sampling combination sequence, and circuit performance parameter predicted values and a confidence interval range of each performance parameter predicted value are obtained; the circuit performance parameter predicted values are modified and calculated; and the performance parameter predicted values are substituted into the a performance parameter index standard range for statistical analysis, and the yield of the circuit performance parameter is calculated. According to the method, the time cost is low, the circuit simulation times are few, the analysis efficiency is high, and the accuracy for estimating the yield is high.

Description

technical field [0001] The invention belongs to the technical field of computer-aided design of electronic circuits and integrated circuits, and in particular relates to a circuit yield estimation method based on Monte Carlo analysis of computer-aided design (CAD). Background technique [0002] Due to the manufacturing process and service conditions, the components in any electronic circuit will inevitably be affected by random disturbance factors, so that there will be random errors between the parameters of the components in the actual circuit and the design target values, resulting in The actual value of the component parameter is not necessarily equal to its nominal value, but just a random value within the tolerance range of its nominal value. In the actual electronic circuit design process, the nominal values ​​of component parameters are often used for analysis and design, so there must be differences between the designed circuit and the actual circuit performance. I...

AI Technical Summary

Problems solved by technology

[0016] The invention provides a circuit Monte Carlo analysis method based on computer-aided design, aiming to solve the traditional circuit Monte Carlo analysis based on actual circuit simulation, the number of circuit simulations is large, and the simulation cost is high. The circuit is becoming more and more complex, the time cost is unbearable, and the Monte Carlo analysis based entirely on the circuit proxy model is limited by the predictive ability of the proxy model. There are certain errors in the analysis results, and sometimes the error is large, which affects the analysis of the yield of the circuit.

Images