Electronic Circuit Parameter Optimization Across Operating Corners
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing circuit design processes are computationally expensive and inefficient due to the high number of simulations required to account for various environmental conditions, leading to slow and costly manual parameter tuning.
Innovation Solution
A system utilizing a surrogate model for efficient optimization of circuit design parameters, incorporating techniques like early dropping and coverage selection to reduce simulations, and incremental training based on simulation results, thereby managing optimization across environmental variations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If simulations are performed for all combinations of corners and Monte Carlo variations, then the circuit design reliability is improved, but the computational cost and time increase significantly
Solution Approach 1:
The patent uses a surrogate model (emulator) that copies the behavior of the full circuit simulation. This surrogate model is trained on a subset of simulation data and then used to predict circuit performance for all corner and Monte Carlo combinations, replacing the need to run full simulations for every case while maintaining prediction accuracy.
Solution Approach 2:
The patent performs preliminary simulations for a subset of corner and Monte Carlo combinations to train the surrogate model before using it for the remaining cases. This preliminary action creates a predictive model that can quickly evaluate all scenarios without requiring full simulations for each one.
2Manufacturing precision
If manual parameter tuning is performed to meet requirements, then the design precision is improved, but the productivity decreases
Solution Approach 1:
The patent implements an automated optimization system that self-adjusts circuit parameters without manual intervention. The system uses the surrogate model to evaluate different parameter sets and automatically identifies optimal values that meet all corner and Monte Carlo requirements, replacing the manual tuning process while maintaining precision.
Solution Approach 2:
The patent employs an iterative feedback loop where the surrogate model predicts circuit performance for given parameters, the system evaluates whether requirements are met, and parameters are adjusted accordingly. This automated feedback mechanism continuously refines parameter values until all requirements are satisfied, eliminating manual iteration while preserving tuning precision.
Data Source
AI summary
A system performs optimization of parameters of a circuit design. The system accesses a model configured to receive design variables of the circuit design and predict a measure of quality of the circuit design. For multiple coverage levels, the system generates samples representing a values of design parameters. For each sample, the system predicts the quality of the sample using the model. The system selects a subset of samples having a predicted quality that exceeds a target. The system performs simulations of the selected subset of samples. The system maintains a moving target and drops samples encountering worse results before all simulations finish. The system performs incremental training of the model based on results of the simulations of the selected subset of samples. The system also decides whether to enter the next coverage level based on the simulation results and the moving target.