Neuromorphic Circuit Parasitic Capacitance Area Reduction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Classic neuromorphic circuits based on CMOS technology face issues with high power consumption and circuit area due to the use of large capacitors or digital solutions for implementing dynamic neurons and synaptic coefficients, which become exacerbated with increasing neural network complexity.

Innovation Solution

A neuromorphic circuit design utilizing resistive random access memory (RRAM) with ferro-tunnel junction elements and parasitic capacitance, where the common conductive line is charged by the sum of currents through resistive elements, and a comparator detects voltage exceeding a threshold to generate an output voltage, reducing the need for large capacitors and optimizing energy usage.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If large capacitors are used to implement dynamic neurons and synaptic coefficients in conventional CMOS neuromorphic circuits, then the circuit can achieve the required dynamic behavior and memory functionality, but the circuit area and power consumption increase excessively

Engineering Contradiction:
Improvedynamic neuron behaviorVSAvoidcircuit area
Core Design Contradiction:
ReliabilityVSArea of stationary object

Solution Approach 1:

The patent extracts the capacitive functionality from separate large capacitor components and integrates it into the resistive memory element itself. The resistive memory element's parasitic capacitance is utilized to provide the necessary dynamic behavior, eliminating the need for external large capacitors and thereby reducing circuit area.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The resistive memory element serves multiple functions simultaneously: it acts as both the synaptic coefficient storage element and the capacitor for dynamic neuron behavior. This multi-functionality eliminates the need for separate dedicated capacitor components, reducing overall circuit area while maintaining required functionality.

Inventive Principle:
Principle #6Universality (Multi-functionality)

2Reliability

If large capacitors are used to implement dynamic neurons and synaptic coefficients in conventional CMOS neuromorphic circuits, then the circuit can achieve the required dynamic behavior and memory functionality, but the power consumption increases excessively

Engineering Contradiction:
Improvedynamic neuron behaviorVSAvoidpower consumption
Core Design Contradiction:
ReliabilityVSUse of energy by moving object

Solution Approach 1:

The patent extracts the capacitive functionality from separate large capacitor components and integrates it into the resistive memory element itself. The resistive memory element's parasitic capacitance is utilized to provide the necessary dynamic behavior, eliminating the need for external large capacitors and thereby reducing circuit area.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The resistive memory element's parasitic capacitance is utilized to provide the necessary dynamic behavior, eliminating the need for external large capacitors and thereby reducing circuit area.

Inventive Principle:
Principle #25Self-service

3Measurement precision

If digital solutions are used to store synaptic coefficients and implement neurons, then the circuit can achieve precise storage and computation, but the circuit area and power consumption increase

Engineering Contradiction:
Improvesynaptic coefficient storage precisionVSAvoidcircuit area
Core Design Contradiction:
Measurement precisionVSArea of stationary object

Solution Approach 1:

The resistive memory element's parasitic capacitance is utilized to provide the necessary dynamic behavior, eliminating the need for external large capacitors and thereby reducing circuit area.

Inventive Principle:
Principle #25Self-service

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

This approach reduces energy consumption and circuit area by leveraging the capacitive effects of resistive elements, enabling efficient operation of neuromorphic circuits with reduced complexity and power requirements.

Implementation Method 1

each comprising a first resistive memory element having an associated parasitic capacitance

Methodology Applied
Scientific EffectParasitic capacitance: Capacitance

Implementation Method 2

an output device configured to compare a first voltage on the first common conductive line to a threshold voltage and to generate an output voltage of the neural circuit based on the comparison

Methodology Applied
Scientific EffectVoltage comparison:

Implementation Method 3

the first resistive memory elements are ferrotunnel junction elements

Methodology Applied
Scientific EffectQuantum tunneling:

Data Source

PatentEP4354352A1Neuromorphic circuit
Publication Date: 2024.04.17 COMMISSARIAT A LENERGIE ATOMIQUE ET AUX ENERGIES ALTERNATIVES
  • EP4354352A1 patent drawingFigure 1~2
  • EP4354352A1 patent drawingFigure 3~4
  • EP4354352A1 patent drawingFigure 5~6

AI summary

The present description relates to a neural circuit (200) comprising: - a first plurality of synapses, each comprising a first resistive memory element having an associated parasitic capacitance, each of the first resistive memory elements being configured to supply a current to a first common conductive line (SL); and - an output device (206) configured to compare a first voltage on the first common conductive line to a threshold voltage (Vth) and to generate an output voltage (Vout) of the neural circuit on the basis of the comparison.