Capacitor Synapse Metal Shielding for Parasitic Coupling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Capacitor-based synapse networks in neural networks face issues with unwanted coupling capacitance, leading to interference and errors in decision-making processes due to parasitic capacitive coupling between output nodes.
Innovation Solution
Implementing a capacitor-based synapse network with metal shielding between output terminals to prevent capacitive coupling, using high-k dielectric materials and metal shielding layers to decouple adjacent output signals effectively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Use of energy by moving object
If capacitor-based synapse networks are used to implement neural networks, then energy efficiency is improved, but unwanted coupling capacitance between output nodes causes interference and errors
Solution Approach 1:
A metal shielding layer is introduced as an intermediary component between adjacent output nodes of capacitor-based synapse networks. This shielding layer acts as a mediator that blocks the parasitic capacitive coupling between outputs, preventing interference while maintaining the energy-efficient capacitor-based architecture. The shielding layer is typically connected to ground potential to effectively decouple adjacent signal lines.
Solution Approach 2:
The patent addresses the harmful effect of parasitic capacitance by using metal shielding that creates controlled capacitance paths to ground. The harmful coupling between signal lines is converted into beneficial isolation, where the shielding layer's capacitance to ground provides a preferred path for stray electric fields, thereby protecting the signal integrity between adjacent outputs.
2Object-affected harmful factors
If metal shielding is added between output terminals, then interference from capacitive coupling is reduced, but device complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the output region into isolated zones using metal shielding layers. Each output node is segmented from its neighbors by these shielding structures, creating electrically independent regions. This segmentation approach systematically addresses the coupling problem without requiring complete redesign of the entire network architecture.
Solution Approach 2:
The patent modifies the physical and electrical parameters of the output region by introducing metal shielding layers with specific materials, thicknesses, and grounding configurations. These parameter changes optimize the shielding effectiveness while controlling the added complexity, allowing designers to adjust shielding properties based on specific performance requirements.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The metal shielding effectively isolates output signals, reducing interference and enhancing the accuracy of neural network decisions by eliminating parasitic capacitive coupling effects.
Implementation Method 1
unwanted coupling capacitance, leading to interference and errors in decision-making processes due to parasitic capacitive coupling between output nodes
Implementation Method 2
metal shielding disposed between the first output terminal and the second output terminal
Data Source
AI summary
A neural network device comprises a first plurality of synapse network capacitors, wherein the synapse network capacitors of the first plurality of synapse network capacitors share a first output terminal. The neural network device further comprises a second plurality of synapse network capacitors, wherein the synapse network capacitors of the second plurality of synapse network capacitors share a second output terminal. Still further, the neural network device comprises a metal shielding disposed between the first output terminal and the second output terminal. The neural network device may be used as part of an artificial intelligence system.