Battling the heat in military CPUs

Story

December 02, 2015

Tim Fleury

Mercury Systems

Demand for higher functionalities in defense electronics has led to conflicting demands for more heat management, more sensitive signals, shorter design cycles, and higher test coverage, all within ever-tighter budgets. Heat is the primary enemy of reliability in central processing units (CPUs).

Military reconnaissance and surveillance platforms rely on CPUs in intensive signal processing systems to handle real-time radar, video, and signals intelligence data. Hardware for these applications often must meet size, weight, and power (SWaP) constraints for use in aircraft, unmanned aerial vehicles (UAVs), ships, and other platforms.

With advancing technological developments, the power levels for CPU modules and mezzanine buses used in military electronics has increased dramatically. Devices such as microprocessors and field-programmable gate arrays (FPGAs) have been running ever faster while their size has been constantly shrinking, which has in turn increased heat densities and threatened product reliability.

This demand for higher and higher functionalities in defense electronics has led to conflicting demands for more heat management, more sensitive signals, shorter design cycles, and higher test coverage, all within ever-tighter defense budgets. Moreover, these products have to be highly reliable with years of operational run time in a wide range of harsh environments. These demands make designing new printed circuit boards (PCBs) and enclosures challenging for test engineers, signal-integrity engineers, and mechanical engineers. Many of today’s high-powered modules cannot be cooled using legacy cooling approaches.

Three different types of CPU modules include air-cooled (A/C), conduction-cooled (C/C), and what can be called air flow-by (AFB) modules. Each type or technique has a different cooling efficiency.

Challenge: Cooling computing systems in a rugged environment

Air-cooling provides easy access to module debug connectors, front panel I/O, and mezzanine modules. This combination simplifies system development and configurability while the system is in its greatest state of flux and all requirements are not yet identified. A major drawback is that air-cooled modules are not typically designed to be deployed in rugged environments. Conduction cooling has been the preferred method of cooling for deployed systems for many years.

The modules are designed to handle the rugged shock and vibration levels, while the systems seal the modules away from harmful elements. A major challenge with conduction cooling, however, is that it is heavier than air-cooled and thermally challenged with higher power modules. Air flow-by delivers the best of both worlds. It provides the efficient point-source cooling of an air-cooled module with the rugged deployment capabilities of conduction-cooling.

Analysis on the thermal design of one of Mercury’s air-cooled products (Figure 1) was performed, using a standards-based approach to bring heat from the mezzanine modules to the carrier module’s heatsink.

Figure 1: Cooling system developed after analyzing with CFD, using thermal bridge hooks.

(Click graphic to zoom by 1.9x)

Designers discovered in the computational fluid dynamics (CFD) simulation (performed with the Mentor Graphics FloTHERM 3D thermal simulation software) that this was possible by adding “hooks” for a thermal bridge between the carrier module heatsink and the mezzanine module heatsink. The net effect was a thermal solution that complied with standards and allowed for a wide range of mezzanine modules to be placed on a host, while limiting any potential changes to a single component. This method was also found to lower the cooling by half, to a 5 °C processor thermal reduction. It also had a significant effect on mean time between failures (MTBF).

These new thermal-management solutions are capable of dissipating tremendous amounts of thermal energy, while still meeting the same or smaller size, weight, and power requirements for the overall solution. By understanding the thermal profile for each specific component that makes up a system using CFD, it is possible to innovate methods for the mass transfer of thermal energy that work at the individual component, module, and subsystem level.

Tim Fleury manages the Thermal Analysis and Environmental Test department at Mercury Systems. Before joining Mercury, he was director of engineering services at Harvard Thermal and a principal engineer at Raytheon with 20 years as a thermal analyst. He holds a master’s degree in thermofluids from Northeastern University and a bachelor’s degree in mechanical engineering from the University of Rhode Island.

Mercury Systems www.mrcy.com