Adaptive Mixed-Signal Circuit Simulation Using Regional Solvers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional mixed-signal simulation techniques, such as SPICE and FastSpice, face limitations in handling large designs, requiring trade-offs between accuracy and speed, and fail to accurately simulate complex interactions between analog and digital circuits, leading to potential silicon malfunctions and increased costs due to inadequate verification.
Innovation Solution
The method involves adaptively applying multiple simulation engines to different regions of a mixed-signal IC design, performing adaptive DC and transient analyses, and selectively solving regions with significant state changes, using linear or non-linear solvers based on the amount of variance, to optimize accuracy and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional SPICE simulation is used to ensure full circuit accuracy, then simulation accuracy is improved, but simulation time increases super-linearly and capacity is limited
Solution Approach 1:
The circuit design is divided into multiple partitions or regions, allowing different simulation techniques to be applied to different parts. This segmentation enables the system to handle large designs by breaking them into manageable pieces that can be processed in parallel or sequentially with appropriate simulation engines.
Solution Approach 2:
Different simulation engines are selectively applied to different regions based on their specific characteristics and requirements. High-accuracy SPICE-like simulation is used for critical analog regions, while faster digital simulation engines handle digital portions, optimizing the balance between accuracy and speed for each local region.
2Quantity of substance
If conventional SPICE simulation is used to handle large designs, then simulation capacity is improved, but simulation time increases super-linearly
Solution Approach 1:
The design is partitioned into multiple regions that can be simulated independently or in parallel, allowing the system to scale to larger designs without super-linear time increases. Each partition can be processed with appropriate simulation depth and accuracy.
Solution Approach 2:
The simulation system dynamically adapts its approach based on the specific characteristics of each region and the simulation progress. Simulation engines are selected and configured in real-time based on circuit activity, state changes, and regional importance, enabling efficient handling of large designs.
3Productivity
If simplified simulation techniques are used to increase simulation speed, then productivity is improved, but accuracy deteriorates and Kirchhoff's laws are not correctly solved
Solution Approach 1:
Different regions are simulated with different levels of accuracy appropriate to their function. Critical analog regions use full SPICE-like accuracy with proper Kirchhoff's law enforcement, while digital regions use faster approximation methods where full accuracy is less critical.
Solution Approach 2:
The system uses an adaptive controller that acts as an intermediary, selecting and coordinating multiple simulation engines. This controller ensures that regions requiring high accuracy are properly simulated while maintaining overall simulation efficiency through intelligent engine selection and coordination.
4Productivity
If manual partitioning and hybrid approaches are used, then some optimization is achieved, but designer judgment introduces bias and design error escapes
Solution Approach 1:
The simulation system automatically performs intelligent partitioning and engine selection without requiring manual designer intervention. The adaptive controller analyzes circuit characteristics, state changes, and regional importance to automatically determine the optimal simulation approach for each region, eliminating human bias while maintaining efficiency.
Solution Approach 2:
The system uses feedback from circuit state changes, simulation progress, and regional characteristics to dynamically adjust simulation parameters and engine selection. This automated feedback mechanism ensures consistent, unbiased decisions about where to apply different simulation techniques, improving both reliability and efficiency.
Data Source
AI summary
Some embodiments simulate a mixed-signal circuit design by adaptively applying multiple simulation engines at various regions of the design at various stages of the simulation. Some embodiments partition the mixed-signal design into multiple regions. Some embodiments classify the regions at different time steps of transient analysis. The regions are classified to indicate whether a region is active or inactive at each such time step. Then when analyzing the active regions, some embodiments adaptively apply different solvers to at least two of the active regions based on criteria associated with the active regions. Additionally, some embodiments perform an adaptive bi-direction analysis of the regions. In this manner, some embodiments optimize the circuit simulation by adaptively simulating the design using different solvers that employ greater accuracy where required and greater efficiency when less accuracy is required, thus allowing the simulation to occur with greater overall accuracy, efficiency, and capacity.