Thermal innovations keep high-performance electronics cool
No matter how efficient an electronic system is, it will produce unwanted heat. The design challenge is to manage heat production to balance system usability, maximize lifespan and reduce cost.
Understanding where and how heat will flow through the system while running is fundamental to thermal engineering. For example, individual cores within a multicore processor SoC will frequently become far hotter than their surroundings. Similarly, power transistors will often reach higher junction temperatures than other devices on the PCB.
Designers need to know the circumstances under which the device may reach the point of thermal shutdown and to find out how the flow of heat, whether through the board or the surrounding air, will affect other nearby devices. If they run hotter than expected, that may reduce the system’s reliability.
Thermal simulation is a common design tool to determine how excess heat production will affect different parts of the system. Managing thermal behavior calls for physical changes to optimize cooling and divert heat from the enclosure.
To avoid hotspots, designers frequently use a combination of heatsinks and forced-air cooling supplied by fans to generate enough airflow to spread hot air through and out of the system.
Fan and heat pipe design
Improvements to two aspects of fan design are helping to reduce noise and increase lifetime. One direction uses more advanced computational fluid dynamics simulations to optimize the fan blade structures for greater efficiency by allowing them to move the same amount of air at lower speeds. A second direction is in motor design. Active electronic fan control lets a microcontroller adapt fan speed to the system temperature.
Advances in heatsink design further help optimize both passive and forced-air cooling. Heatsinks have traditionally used a solid construction that takes advantage of the high thermal conductivity of metals such as copper to allow for the rapid transfer of heat to cooler surfaces.
Diamond has better heat-transfer properties. Some work has focused on using extremely thin layers of synthetic diamond to act as heat spreaders over devices like power transistors. More recent developments exploited mixtures of diamond and copper to improve the overall thermal conductivity. This approach also allows the use of finely grained diamond particles, which are easier to synthesize. Though it is difficult to maintain a reliable thermal interface between the two materials, manufacturers have found methods to improve overall performance. It has reached as high as 1000W/m.K with a diamond content of a little under 50% of the composite’s volume.
Hollow heat pipes that contain a fluid under very low pressure provide a different route to maximizing thermal conductivity. These transfer heat away from hotspots by harnessing the phase transition between liquid and vapor. Once the thermal energy carried by the vapor to the cooler end of each pipe dissipates into the surrounding atmosphere air, the gas condenses.
The liquid, through capillary action, moves back to the hot surface where it evaporates once more. The thermal conductivity offered by a heat pipe can easily be more than 1000W/m.K. However, size and system orientation may cause too many constraints. Ideally, gravity is used to assist with the vapor flow and liquid return. The condenser assembly should be placed at least several centimeters from the heat source.
Convenience of vapor
A vapor chamber can deliver the high thermal conductivity associated with heat pipes but in a form that can be more convenient. The chamber is, in effect, a planar heat pipe constructed in a similar shape to that of a solid heatsink. Thin plates with grooves etched into them form the top and bottom of the sealed vapor chamber. A porous wick structure conveys fluid to the hot surface after condensation. The chamber can be as thin as 0.2mm. Some recent experiments have shown workable designs for wick-free vapor chambers. This design overcomes the problem of the wick increasing thermal resistance in very thin vapor chambers.
Other systems, such as the accelerators used for artificial intelligence and high-performance computing, may need more cooling assistance beyond the capabilities of forced air. Both requirements are leading to a variety of innovations in thermal management.
Liquid cooling
Liquid cooling offers the best system-scale thermal-transfer properties, though it can present engineering challenges. Traditionally, liquid cooling operates by pumping liquid through pipes and chambers connected to high-temperature devices through heatsinks. In recent years, full immersion into a coolant tank has emerged as another way of implementing liquid cooling, which works well for high-power systems that also demand high spatial density, such as those used in high-performance computing (HPC). Recent experiments have shown immersion cooling could also work for cooling lithium-ion batteries in future electric vehicles and increase their recharging rates.
The immersion tanks contain a dielectric liquid, such as a mineral oil, with low chemical reactivity to ensure it does not damage the components or lead to the possibility of shorts or arcing between electrical contacts on the PCBs. As well as avoiding the use of high-velocity forced-air cooling, immersion also prevents dust buildup on PCB surfaces. The Open Compute Project has formulated guidelines for immersion cooling in server environments to help minimize the risk of damage to components by the fluid.
Peltier effect
For more localized temperature control, thermoelectric cooling provides a practical option. This takes advantage of the Peltier effect to boost the transfer of heat from one side of a heatsink structure to the other when an electric current passes through. Typically, a Peltier-effect cooler uses pillars of n-type and p-type semiconductors arranged electrically in series but arranged physically in parallel between two plates, one of which is attached to the target in need of thermal control.
The level of electrical current determines the amount of cooling up to the maximum supported by the combination of materials, which should be chosen for both high electrical and thermal conductivity. The high degree of temperature control suits Peltier-effect coolers to scientific instruments. Some recent experiments have demonstrated that it is possible to focus on hotspots within multicore processors to deliver more targeted cooling in designs where there is a risk of thermal shutdown even if only part of the device is running.
Conclusion
Though the dominant technique for cooling electronic systems remains the fan, thermal control is a sector that is innovating rapidly in response to changing system architectures. The rise of multicore and parallel processing has led to an increased focus on targeted thermal control alongside the development of technologies to deal with the high thermal densities of HPC. Simulation and analysis, in combination with advice from suppliers, will help determine which techniques will work best with the target design.
References and related links
High thermal conductive copper/diamond composites: state of the art (Journal of Materials Science),
Immersion cooling for lithium-ion batteries: a review (Journal of Power Sources),
Avnet carries thermal management solutions on its linecard, including:
Heatsinks, heatpipes, etc: Advanced Thermal Solutions, Boyd Corp
Contact a display solutions expert
