Capacitance Extraction Using Segmented 2D Tables
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current 2.5D capacitive analysis techniques in VLSI design are too simplistic and inaccurate for modern designs, particularly at the 7 nanometer scale, due to assumptions about wire densities and structures, leading to degraded performance and increased design cycle times.
Innovation Solution
The method involves calculating effective spacings between neighbor nets, segmenting the target net, identifying metal configurations, and using 2D capacitance tables to scale capacitance values, allowing for more accurate parasitic capacitance extraction and improved circuit performance without the complexity of 3D analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If 2.5D capacitive analysis techniques are used, then analysis speed is improved, but accuracy deteriorates due to gross assumptions about wire densities and structures
Solution Approach 1:
The patent divides the target net into multiple segments along its length, where each segment is assigned a uniform metal configuration based on the surrounding neighbor nets. This segmentation allows the method to capture local variations in wire density and structure without requiring full 3D analysis, thereby maintaining both speed and accuracy.
Solution Approach 2:
The patent applies different metal configuration characteristics to different segments of the target net based on the actual wire density and structure in each local region. By using effective spacing calculations that reflect local neighbor net configurations, the method achieves accurate capacitance extraction for each segment while maintaining overall computational efficiency.
2Measurement precision
If 3D capacitance analysis methods are used, then accuracy is improved, but analysis speed deteriorates
Solution Approach 1:
The patent creates a simplified 2.5D representation of the 3D wire structure by copying and organizing metal configuration data into discrete segments. Each segment captures the essential capacitive characteristics of its local 3D environment without requiring full 3D computational complexity, thus achieving near-3D accuracy with 2.5D speed.
Solution Approach 2:
The patent transforms the continuous 3D spatial problem into a discrete segmented representation by introducing effective spacing as a key parameter. This parameter change allows the method to capture the essential capacitive effects of 3D wire structures using simplified 2.5D calculations, significantly reducing computational complexity while maintaining accuracy.
3Ease of manufacture
If simplified density assumptions are made in 2.5D analysis, then computation is simplified and speed is improved, but manufacturing precision deteriorates
Solution Approach 1:
The patent introduces dynamic segmentation where the number and boundaries of segments are determined based on actual variations in neighbor net configurations along the target net. This dynamic approach allows the method to simplify computation in regions with uniform wire density while automatically increasing segmentation detail in regions with significant structural variations, thereby maintaining both computational efficiency and design accuracy.
Data Source
AI summary
An effective spacing is calculated for each physical spacing between two or more neighbor nets of a target net. Segment boundaries are determined based on the calculated effective spacing to define segments for the target net and one of the segments is selected. A metal configuration for the selected segment is identified and a table of capacitance per-unit-length is accessed for the identified metal configuration to return an above capacitance value, a below capacitance value, a left-side capacitance value, and a right-side capacitance value for the corresponding segment, the table comprising at least a two-dimensional (2D) table. The capacitance values are scaled based on a corresponding segment length determined from the calculated effective spacing. The selecting, identifying, accessing and scaling operations are repeated for each remaining segment of the target net. Optionally, the above, below, left, and right capacitance values for all segments of the target net are summed.