Device Mismatch Variation Contributions in Circuit Design
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for modeling and analyzing mismatch variations in integrated circuits are computationally expensive and do not effectively identify which devices have the highest impact on circuit performance, leading to inefficiencies in design optimization and yield improvement.
Innovation Solution
A new metric called the 'ordered' metric is introduced to allocate variance contributions of mismatch parameters, combined with orthogonal matching pursuit (OMP) to estimate device contributions efficiently, reducing the number of required simulations and enabling accurate mismatch contribution analysis with fewer samples.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional OFAT sensitivity analysis is used to analyze mismatch variations, then sensitivity coefficients for each mismatch parameter can be obtained, but the computational expense becomes significant or prohibitive
Solution Approach 1:
The patent segments the computational problem by dividing the circuit into individual devices and further segmenting the analysis into device-level contributions versus circuit-level performance. This allows the sensitivity analysis to focus on identifying which specific devices contribute most to performance variation, rather than analyzing all devices equally. The segmentation enables a hierarchical approach where computational resources are allocated to the most impactful devices.
Solution Approach 2:
The patent extracts and isolates the most significant mismatch parameters and devices that contribute to performance variation. By using the ordered metric and OMP, the method identifies and extracts only the critical subset of devices and parameters that have the highest impact, rather than analyzing all parameters equally. This extraction reduces the computational burden by focusing only on the essential contributors.
2Productivity
If simple linear sensitivity analysis is used, then sensitivity coefficients can be obtained quickly, but insufficient information is provided for designers to fully optimize the design
Solution Approach 1:
The patent implements a feedback mechanism where the ordered metric calculation provides designers with specific information about which devices contribute most to performance variation. This feedback loop allows designers to identify critical devices and make targeted design decisions. The method continuously refines the understanding of device contributions by allocating variance in an ordered sequence, providing actionable insights for design optimization.
Solution Approach 2:
The patent changes the approach from traditional sensitivity coefficients to an ordered metric that ranks devices by their contribution to performance variation. This parameter change transforms the output from generic sensitivity coefficients to device-specific contribution rankings. The ordered metric provides a more informative representation that directly indicates which devices should be prioritized for design optimization.
3Measurement precision
If multivariate linear regression is used to analyze mismatch contributions, then comprehensive variance analysis can be performed, but the computational expense is significant
Solution Approach 1:
The patent applies partial action by performing variance analysis only on the most significant devices rather than all devices in the circuit. The ordered metric allows the analysis to focus on the top contributing devices, performing comprehensive variance analysis only where needed. This partial approach maintains accuracy for the critical devices while reducing overall computational complexity.
Solution Approach 2:
The patent performs preliminary sorting and ranking of devices using the ordered metric before conducting detailed variance analysis. This preliminary action identifies which devices warrant detailed analysis based on their contribution ranking. By pre-screening devices and focusing detailed analysis only on the top contributors, the method reduces computational complexity while maintaining accuracy for the most impactful devices.
Data Source
AI summary
A system, method, and computer program product for computing device mismatch variation contributions to circuit performance variation. Embodiments estimate which individual devices in a simulated circuit design have the largest impact on circuit performance, while requiring far fewer simulations than traditional multivariate linear regressions. An ordered metric allocates output variance contributions for each input mismatch parameter in a linear model. The embodiments summarize the output variance in each device, and rank the mismatch contributions based on the summarized contributions. Additional sensitivity analysis can derive a final accurate linear contribution. Embodiments can reduce required simulations by a factor of ten.