Yield-Performance Tradeoff Visualization in Circuit Design
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In semiconductor circuit design, determining tradeoffs between yield and performance is complex due to the exponential increase in design choices with multiple variables, leading to information overload and reliance on intuition, especially when manufacturing yield and performance specifications need to be balanced for profitability.
Innovation Solution
A system and method that generate and filter specification candidate vectors with associated performance values, using statistical estimators to calculate parameters like yield and process capability, and display these using graphical representations to simplify the evaluation of tradeoffs between yield and performance specifications.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple performance specifications are analyzed using traditional spreadsheet methods, then comprehensive tradeoff evaluation is possible, but the analysis becomes tedious and time-consuming due to information overload
Solution Approach 1:
The patent segments the complex tradeoff analysis into distinct modules: specification candidate generation, performance vector calculation, statistical parameter estimation, and filtering. This modular approach breaks down the overwhelming comprehensive analysis into manageable sequential steps, reducing analysis time while maintaining evaluation accuracy.
Solution Approach 2:
The patent introduces intermediary computational structures (performance vectors, specification candidate vectors, and statistical estimators) that mediate between raw simulation data and final tradeoff conclusions. These intermediaries automatically process and organize data, eliminating the manual spreadsheet work that causes time loss.
2Adaptability or versatility
If the number of design variables increases to achieve better optimization, then design flexibility improves, but the number of choices increases exponentially leading to information overload
Solution Approach 1:
The patent implements dynamic filtering that adaptively reduces the solution space based on statistical parameters. As design variables increase, the system dynamically identifies and eliminates dominated specification candidates, maintaining manageable complexity while preserving design flexibility through automated pruning of inferior options.
Solution Approach 2:
The patent transforms the evaluation from considering raw specification values to considering statistical parameters (yield, process capability). This parameter transformation compresses the exponential design space into a manageable set of statistical tradeoff candidates, reducing evaluation complexity while maintaining adaptability.
3Manufacturing precision
If performance specifications are tightened to improve product performance, then product quality increases, but manufacturing yield decreases
Solution Approach 1:
The patent implements feedback through statistical parameter calculation that quantifies the yield-performance relationship. By calculating yield and process capability for different specification candidates, the system provides feedback on the consequences of tightening specifications, enabling informed decisions that balance quality and productivity.
Solution Approach 2:
The patent applies partial action by identifying the minimum necessary specification tightening required to achieve acceptable performance levels. Rather than uniformly tightening all specifications, the system selectively adjusts only those specifications that provide meaningful performance improvement while maintaining acceptable yield, avoiding excessive action that would unnecessarily reduce productivity.
Data Source
AI summary
A system that includes a candidate generator that generates candidate vectors having as components performance specifications of an electrical circuit design. The system also includes a performance estimator that generates performance vectors of the electrical circuit design, the performance vectors having as components performance values of the electrical circuit design. The candidate vectors and the performance vectors are input into a statistical estimator that calculates a statistical parameter, for example, yield, for each candidate vector in accordance with the performance vectors. The system further includes a filter that receives the candidate vectors and their respective statistical parameters, and outputs a filtered candidate vector with its corresponding filtered statistical parameter. A display system visually represents the filtered candidate vector and its corresponding filtered statistical parameter.