Frequency-Domain DUT Parameter Estimation With Adaptive Convergence
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for estimating device under test (DUT) model parameters in the frequency domain, such as non-linear least squares algorithms, face challenges with convergence speed and accuracy under varying conditions like different ICs, temperatures, and DUT age.
Innovation Solution
A parameter convergence model is employed that adjusts a regularization parameter based on a cost function improvement ratio, iteratively converging to a target tolerance, using a frequency domain estimation approach to obtain accurate DUT model parameters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If non-linear least squares algorithm is used to estimate DUT model parameters, then parameter estimation can be performed in frequency domain, but convergence is time-consuming and results are inaccurate for different conditions
Solution Approach 1:
The patent applies preliminary action by using a grid search algorithm to obtain initial parameter estimates before performing refined optimization. This preliminary estimation step provides a good starting point that accelerates convergence of subsequent optimization algorithms and avoids getting trapped in local minima, thereby reducing overall convergence time while maintaining accuracy across varying conditions.
Solution Approach 2:
The patent introduces an intermediary approach by combining multiple estimation methods (grid search, least squares, and gradient-based optimization) in a sequential manner. Each method acts as an intermediary step that prepares the parameters for the next more sophisticated method, ensuring both global optimality and local precision without requiring any single algorithm to work alone.
2Adaptability or versatility
If non-linear least squares algorithm is used to estimate DUT model parameters, then parameter estimation can be performed, but convergence accuracy deteriorates under varying conditions such as different ICs, temperatures, and DUT age
Solution Approach 1:
The patent applies dynamics by implementing an adaptive parameter estimation process that adjusts its strategy based on the specific conditions being measured. The system dynamically selects and combines estimation methods (grid search for global exploration, least squares for precision, gradient methods for optimization) depending on the initial parameter values and measurement conditions, ensuring robust accuracy across different ICs, temperatures, and device ages.
Solution Approach 2:
The patent utilizes parameter changes by systematically varying estimation parameters such as regularization coefficients, optimization step sizes, and search grid resolutions based on the specific measurement conditions. This allows the estimation algorithm to adapt its behavior to different device states and environmental conditions, maintaining high accuracy across varying operating parameters.
3Measurement precision
If iterative parameter optimization is performed to improve accuracy, then measurement precision improves, but convergence speed decreases
Solution Approach 1:
The patent applies partial action by implementing a multi-stage optimization process where each stage performs a limited number of iterations focused on specific aspects of parameter refinement. Rather than performing exhaustive optimization in a single long process, the system performs multiple partial optimization passes with different algorithms, each contributing to the final accuracy while keeping individual computation times short.
Data Source
AI summary
A circuit for determining device under test (DUT) model parameters is described. The circuit includes a parameter estimator circuit configured to: obtain initial values for DUT model parameters based on sense signal samples; execute a parameter convergence model having a regularization parameter and a cost function that accounts for error residuals; and obtain final values for the DUT model parameters by adjusting the regularization parameter in iterations of the parameter convergence model as a function of cost function improvement until the parameter convergence model converges to within a target tolerance.