Monte Carlo Corner Extraction for Circuit Yield Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing corner-based design methods for electrical circuits fail to effectively capture and represent variations in performance and yield, particularly ignoring local process variations and lacking a reliable means to design against specific target yields or performance metrics, leading to inferior circuit designs.
Innovation Solution
A method that iteratively identifies corner points by simulating electrical circuits across a distribution of process and environmental points, using Monte Carlo sampling and line searches to optimize design variables, ensuring the circuit meets pre-determined yield and performance criteria, and stores the design point for improved circuit design.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If Monte Carlo sampling is used to estimate yield accurately, then measurement precision is improved, but loss of time increases significantly
Solution Approach 1:
The patent segments the continuous process variable space into discrete corner regions. Instead of sampling the entire continuous space with Monte Carlo methods, the algorithm identifies and analyzes specific corner regions that dominate yield behavior. This segmentation reduces the effective search space from infinite to a manageable set of discrete corners, achieving accurate yield estimation without exhaustive sampling.
Solution Approach 2:
The patent performs preliminary corner extraction and identification before conducting detailed yield analysis. By pre-identifying which corners are most influential to yield performance, the method avoids performing full Monte Carlo simulations across all possible process variations. This preliminary action filters out non-critical regions, reducing overall computation time while maintaining accuracy.
2Productivity
If traditional corner-based design methods are used, then productivity is improved through faster design iterations, but measurement precision deteriorates due to inability to capture local process variations
Solution Approach 1:
The patent applies local quality by focusing computational resources on specific corner regions that are most critical to yield, rather than treating all process variations uniformly. The algorithm identifies local corners where performance metrics approach specification boundaries and concentrates analysis there. This localized approach captures local process variations accurately while avoiding wasted computation in regions that don't affect yield, thus maintaining both speed and precision.
3Reliability
If comprehensive Monte Carlo sampling is performed to capture all process variations, then reliability is improved, but device complexity increases due to computational requirements
Solution Approach 1:
The patent extracts and isolates the most influential corner regions from the complete process variable space. By taking out only the critical corners that dominate yield behavior and analyzing those in detail, the method achieves reliable yield prediction without needing to evaluate the entire complex multidimensional space. This extraction reduces computational complexity while preserving reliability.
4Manufacturing precision
If design variables are optimized against all possible process corners, then manufacturing precision is improved, but loss of time increases due to excessive simulation requirements
Solution Approach 1:
The patent applies partial action by optimizing design variables against only the most critical corner regions rather than all possible corners. The algorithm identifies a subset of corners that have the greatest impact on yield and performs optimization focused on those regions. This partial approach achieves sufficient manufacturing precision for robust design while dramatically reducing optimization time compared to exhaustive corner analysis.
Data Source
AI summary
For application to analog, mixed-signal, and custom digital circuits, a system and method to extract circuit-specific process/environmental corners that is yield-aware and/or specification-aware. Simulation data from previous Monte Carlo-based verification actions can be re-used.