PCB Layout Parasitic Impedance Estimation via Machine Learning

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current circuit board layout tools inaccurately predict parasitic impedances, leading to performance discrepancies between simulation and measurement, which prolongs the design cycle and increases manufacturing costs due to iterative revisions.

Innovation Solution

The implementation of a machine learning system, specifically a graphical neural network (GNN), that analyzes PCB layouts to estimate parasitic impedances more accurately, allowing for better prediction of circuit board performance before manufacturing.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If current circuit board layout tools are used to estimate parasitic impedances, then the design process is simple and fast, but the prediction accuracy of parasitic impedances is insufficient

Engineering Contradiction:
Improveprediction accuracy of parasitic impedancesVSAvoidcomplexity of layout tool
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent introduces machine learning models as an intermediary between the PCB layout and parasitic impedance estimation. The machine learning model is trained on simulation data to serve as a mediator that translates layout features into accurate parasitic impedance predictions, resolving the contradiction between simplicity and accuracy.

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent applies preliminary action by pre-training machine learning models using electromagnetic simulation data before actual PCB design work. This pre-computation of training data from detailed simulations allows the model to learn accurate parasitic impedance relationships in advance, providing fast and accurate predictions during subsequent design iterations without requiring repeated full simulations.

Inventive Principle:
Principle #10Preliminary action

2Measurement precision

If iterative PCB designs are manufactured to improve performance accuracy, then the prediction accuracy improves, but the time and cost increase

Engineering Contradiction:
Improveperformance prediction accuracyVSAvoiddesign cycle time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent creates a virtual copy of the complex electromagnetic simulation process through machine learning models. Instead of repeatedly manufacturing physical PCBs for testing, the trained model serves as a virtual replica that can quickly predict performance outcomes for different layout iterations, eliminating the need for multiple physical prototypes and reducing design cycle time.

Inventive Principle:
Principle #26Copying

Solution Approach 2:

The patent replaces the mechanical/physical iteration process of manufacturing and testing PCBs with a computational machine learning system. The machine learning model substitutes the physical trial-and-error process, providing fast virtual predictions that eliminate the time-consuming cycle of manufacturing multiple PCB revisions for performance validation.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

3Measurement precision

If machine learning models are trained with simulation data, then the prediction accuracy improves, but the training time and computational resources increase

Engineering Contradiction:
Improveparasitic impedance estimation accuracyVSAvoidmodel training time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent applies preliminary action by performing the computationally intensive simulation work once during the data preparation phase to create training datasets. This upfront investment in generating training data from simulations enables the machine learning model to learn accurate relationships, after which the model provides fast predictions without requiring repeated full simulations for each design iteration.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4345678A1Machine learning tool for layout design of printed circuit board
Publication Date: 2024.04.03 HONEYWELL INTERNATIONAL INC
  • EP4345678A1 patent drawingFigure 1
  • EP4345678A1 patent drawingFigure 2
  • EP4345678A1 patent drawingFigure 3

AI summary

This disclosure is directed to a system and method for applying machine learning model (300) to a PCB layout (310) to obtain an estimate of the parasitic values of the board (320). The machine learning model (300) may be trained with various PCB layouts (310) to accurately determine an estimated parasitic impedance (329). Determine regions of the board having a high parasitic impedance value, alerting the user to reconfiguring the region so to achieve a desired range of values.