Thermal Management Tips and Considerations for Embedded Devices

Long life and high reliability under demanding conditions and continuous industrial use: this is a basic requirement of embedded systems.

Present technological demands are driving computing power to increase while components and systems become smaller and smaller. This makes modern system-on-modules (SOMs) and single-board computers (SBCs) ideal for use in self-sufficient mobile systems, smart medical and household appliances, controllers for solar energy or charging stations for electromobility, and numerous other areas.

The higher the performance and smaller the system, the more relevant heat generation and cooling become.

Typically, embedded systems and their components are already designed for an extended temperature range and can be scaled down to the expected operating temperatures in their respective use cases, down to the component level. At PHYTEC, we use capacitors of the quality level X7R at critical points, for example, which show the lowest possible degeneration over the life cycle thus maintaining their technical properties even at higher temperatures throughout the entire life of the product.

The purpose of thermal management is to operate with the lowest possible temperature and optimal heat dissipation.

Embedded systems achieve their maximum service life and minimum failure rate when they remain as cool as possible during operation. Overheating can cause serious damage to the entire system.

Efficient thermal management prevents the need to protect against overheating with reduced cycle rate and computing power by ensuring that electronics can maintain their performance even over a long period of time, at higher ambient temperatures.

If the resulting heat of electronic components cannot be completely and sufficiently released via the components themselves, measures for cooling are necessary.

A distinction is made between active and passive cooling:

Active cooling, with fans, for example, is very effective and works even with large amounts of heat. However, due to its mechanical components and the required energy for embedded systems it is less suitable.

Passive cooling systems, which do not require additional moving parts, are preferred. They do not need energy, do not generate noise, and do not wear out or require maintenance.

For the passive cooling of the hardware in embedded systems, several methods have been established that can be used individually or in combination:

• Housings can be designed to ensure an optimal natural air flow – either in the housing itself or on the outside of the housing, so that heat transmitted to the housing is discharged.
• Heat sinks are glued to heat-generating components such as controllers, power management ICs (PMICs) and RAM devices. They often have a large surface area to efficiently dissipate heat to the environment. Such heat sinks can also serve as a buffer to absorb and distribute short-term heat generation.
• Heat spreaders also dissipate heat from the source and release it, for example, to a heat sink. Or if the heat in the housing itself can no longer be dissipated, a heat spreader could allow for better distribution of the heat to the housing of a system or to its surroundings.
• Thermal guide pads or thermal paste are used in conjunction with almost all the aforementioned cooling methods. They enable the best possible heat transfer between the component and the heat sink while avoiding air gaps.
• Regulating the environmental ambient conditions of the embedded system

If embedded processors reach certain temperature values, the clocking and voltage of the cores are reduced. This method called “Dynamic Voltage and Frequency Scaling” for throttling the processor power is defined in the Linux BSP.

Due to this dynamic scaling, one must consider the available computing power may be reduced when testing and selecting a cooling solution. Control of a fan for active cooling is integrated in the BSP and can be used to supplement passive cooling methods to prevent or limit throttling of the processor.

Critical temperature values can lead to an emergency shutdown of the processor to prevent total failure and defect. This behavior and the consequences – for example, how a system can then be reset and restarted – should be considered in the design.

Heat generation, cooling requirements, and the selection and dimensioning of critical components requires extensive expertise and experience.

The environmental conditions for the use of electronics must be known or defined. The thermal specification of the processor and other components as well as power consumption and heat development under various load scenarios provide information about the cooling requirement. These values can be simulated and validated with a prototype in the laboratory.

For the selection of the appropriate heat sinks, their thermal resistance and possible power consumption are compared with the data of simulation or laboratory tests. In conjunction with other requirements and conditions, such as the available space and the nature of the housing, this will provide the ideal cooling system for the application.