Virtual thermal sensors (VTS) are an optimal software solution to complement or replace physical sensors in high-power density applications for real-time control.

They enable improved power-derating strategies, enhancing system performance, safety, and cost efficiency.

Alternative technologies, such as 1D thermal networks, often involve greater trade-offs in terms of accuracy and computational power.

These limitations compromise the reliability of the solutions, requiring engineering teams to apply larger safety margins.

VTS provides superior accuracy with lower computational demands, ensuring a more reliable and optimized solution.

A virtual thermal sensor is a specialized multiphysics model embedded in the microcontroller of a component. It operates in real-time by providing input directly to closed-loop control strategies, enabling immediate actions, predictions and adjustments.

To achieve real-time, high-resolution simulations without compromising accuracy, two types of data are typically needed:

Simulation Data: Includes geometry, material properties, power loss models, and computational fluid dynamics (CFD) data (if available).

Test Bench Data: A robust dataset derived from test bench experiments, representing the entire operational range of the system (varied operating points, boundary conditions, etc.).

This data is used to calibrate the simulation model to ensure accuracy.

This kind of data is usually already available within the existing product development roadmap.

Our methodology and tools are designed to be highly flexible and compatible with a wide range of simulation tools, including ANSYS, Siemens, and others.

This ensures seamless integration into your existing workflow.

The VTS is delivered as a state-space model, ensuring compatibility with various systems. It can be exported in multiple formats, including .CSV, .MAT, Simulink, and C code, making integration into the final firmware straightforward and efficient.