Chip Thermal Sensor Fusion for Accurate Hotspot Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing thermal management solutions in semiconductor devices face challenges in accurately measuring temperatures in critical regions due to competing real estate interests and interference with local thermal equilibrium, leading to inaccurate thermal readings and potential overheating issues.
Innovation Solution
A predictive model using multiple thermal sensors located outside critical areas combines sensor readings and temperature gradients to accurately predict hotspot temperatures, integrating with a feedback loop for efficient thermal management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If thermal sensors are placed in critical regions to directly measure temperature, then measurement precision improves, but device complexity and interference with thermal equilibrium worsen
Solution Approach 1:
The patent uses temperature gradient measurements from sensors placed in non-critical regions as an intermediary to indirectly determine hotspot temperatures. Instead of placing sensors directly in critical regions, the system measures gradients in accessible areas and uses these as mediators to calculate target temperatures through the thermal diffusion equation, resolving the conflict between measurement accuracy and device complexity.
Solution Approach 2:
The system creates a virtual representation of the thermal field by measuring gradients at accessible locations and using mathematical modeling to copy or reconstruct the temperature distribution in critical regions. This allows temperature information from hard-to-reach areas to be obtained without physically placing sensors there, maintaining measurement precision while avoiding device complexity.
2Measurement precision
If thermal sensors are placed in critical regions, then temperature measurement accuracy improves, but local thermal equilibrium is disturbed
Solution Approach 1:
The patent employs temperature gradients in non-critical regions as intermediaries to infer hotspot temperatures without directly measuring them. By using the thermal gradient field as a mediator and applying the thermal diffusion equation, the system obtains accurate hotspot temperature data while leaving the critical thermal equilibrium undisturbed.
Solution Approach 2:
The patent replaces direct physical measurement (mechanical sensor placement) with a mathematical modeling approach. Instead of physically inserting sensors into critical regions that would disturb thermal equilibrium, the system substitutes a computational model based on thermal diffusion equations that uses gradient measurements from safe locations to calculate target temperatures, thereby maintaining thermal stability.
3Measurement precision
If multiple thermal sensors are deployed to improve measurement accuracy, then measurement precision improves, but device complexity increases
Solution Approach 1:
The patent replaces a complex network of multiple physical sensors with a simplified system using fewer sensors combined with mathematical modeling. By substituting direct measurement with thermal diffusion equation-based calculations, the system achieves high measurement precision using minimal sensors, thereby reducing device complexity while maintaining accuracy.
Solution Approach 2:
The system changes the measurement parameters from direct temperature readings at multiple locations to temperature gradient measurements combined with mathematical transformation. By measuring gradients at fewer locations and using the thermal diffusion equation to compute target temperatures, the system achieves equivalent or superior precision with reduced sensor count and lower device complexity.
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 model provides nearly 82% improvement in hotspot temperature prediction accuracy, enabling precise thermal monitoring and reducing overheating margins, compatible with various processor types and architectures.
Implementation Method 1
measuring a temperature of the semiconductor chip at each predetermined location using a corresponding one of the plurality of thermal sensing elements
Implementation Method 2
combines sensor readings and temperature gradients to accurately predict hotspot temperatures
Implementation Method 3
temperature gradients to accurately predict hotspot temperatures
Data Source
AI summary
A method includes forming a plurality of thermal sensing elements at predetermined locations on a semiconductor chip proximate to a target location, measuring a temperature of the semiconductor chip at each predetermined location using a corresponding one of the plurality of thermal sensing elements, and determining a temperature at the target location using the temperatures measured at each of the predetermined locations.