Worst-Case Distance Statistical Corner Extraction for Circuit Yield
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for characterizing statistical variations in circuit design are less effective for high-yield applications, particularly in analog circuits, as they require a large number of sample points and are inefficient in improving performance and yield.
Innovation Solution
The method involves using worst-case-distance calculations to determine design parameter adjustments that improve performance by identifying and scaling statistical corners to meet specified tolerances, reducing the need for extensive sample points and enhancing yield.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If current approaches to characterizing statistical variations are used, then the circuit design can be modeled with statistical parameters, but the method requires a large number of sample points and is inefficient for high-yield applications
Solution Approach 1:
The patent changes the approach from using numerous sample points to using worst-case distance calculations. By transforming the statistical characterization into a parameter-based method that identifies corners in the statistical parameter space, the system achieves high-yield applications without requiring extensive sampling, thus improving efficiency while maintaining reliability
Solution Approach 2:
The patent replaces the mechanical sampling process with a mathematical calculation approach. Instead of physically or computationally sampling numerous points to characterize statistical variations, the system uses analytical methods to calculate worst-case distances and identify critical corners, significantly reducing computational effort while achieving the same reliability goals
2Reliability
If current approaches are used to characterize statistical variations, then uncertainty can be modeled, but the method is less effective for high-yield applications like 3-sigma yield
Solution Approach 1:
The patent transforms the statistical parameter characterization into a worst-case distance-based approach. By defining statistical corners through worst-case distances in the statistical parameter space, the system achieves precise and effective characterization for high-yield applications, overcoming the limitations of conventional methods
3Reliability
If design parameters are adjusted to improve performance for statistical corners, then overall performance and yield improve, but the process requires identifying and scaling corners based on worst-case distance
Solution Approach 1:
The patent performs preliminary identification of statistical corners and calculation of worst-case distances before final design optimization. By pre-characterizing the statistical parameter space and identifying critical corners, the system simplifies the subsequent design adjustment process, making the overall complexity manageable while achieving improved yield
Data Source
AI summary
Certain circuit models include design parameters that reflect user choices and statistical parameters that reflect modeling uncertainty. For each performance goal (e.g., a one-sided performance goal), a closest point of failure in the statistical parameters is used to identify a statistical corner that characterizes a specified tolerance for that performance goal. Adjusting the design parameters to improve performance for these corners improves overall performance and corresponding yields.