Controlling the Power and Heat of Rugged, Embedded Electronics

   < Back to Articles

Systems designers struggle with ever-increasing computing power and the intense levels of heat today's high-performance computing can cause, and look for innovative ways to cool systems without compromising performance.

By Skyler Frink

Embedded systems must perform many different tasks to support war- fighters, and as technology advances the power consumption increases, leading to greater excess heat and a higher demand for systems that can control their power use.

Vehicles today, for example, are packed with electronics, and it's important that those electronics make the most of their power while staying cool enough to function in the harsh environments they inhabit without damage.

While increasing chip densities and performance have proven to be relatively easy, keeping up with the increased heat produced by high-performance systems has shown itself to be a difficult task. Fortunately, the defense industry has risen to the challenge with new techniques, materials, and projects emerging to solve the problems facing today's electronics.

Difficult Environments

A command vehicle can have dozens of systems, and in the best-case scenario, a vehicle with cooling, that environment is going to be hot. To keep those systems up and running, the electronics need to avoid using power for unneeded components and have thermal management systems that allow for easy heat dissipation. In addition to requiring power control and thermal management, these systems need to be rugged, which means some traditional cooling methods, such as fan cooling, are not an option.

For an unmanned vehicle, this problem is made even greater by the environment not needing to support human life. An unmanned aerial vehicle can fly at altitudes where conduction cooling isn't an option because there simply isn't enough air. An unmanned ground vehicle can be driven into areas that are far too hot for an actual warfighter to enter. "If there's a human in the environment, it's usually more benign," explains Martin Sweeney, lead project engineer at Parvus in Salt Lake City. "When it's more automated by default, the environment can get much more demanding."

Preventing Power Waste

Commercial off-the-shelf (COTS) devices are common now, as they offer cost-saving without sacrificing performance and can fit into a wide variety of systems. However, COTS systems are not always power efficient due to their need to be versatile. Recent advances have led to systems that can be capable of doing more, but can also adapt to save power based on whatever mission they are selected for.

"Sometimes we have a device that can perform 20 functions, but we're only using 10 of them," says Parvus's Sweeney. "In general, there's a lot more functionality being put into single devices. A lot of times they're more than you need. You buy it with all the bells and whistles, but you're only using half of them. You can shut the other half off if they're not being used by the device."

Firmware makes COTS products more power efficient and also cost-effective. By making COTS devices that can perform many different functions but also have the power efficiency of a custom-designed device, the COTS devices can go in systems that require high efficiency or strict thermal requirements.

"I think the whole root of it is the movement towards system-on-chip at a component manufacturer level. The people who make FPGAs [field-programmable gate arrays], they're no longer just FPGAs; they have a processor on there. They're trying to be all things to all people, but when you buy them you find out you'll never use 25 percent of it, but they give you the option to power those parts down," explains Sweeney. "What we gain from this is long-term average power reliability. If you can reduce the number of watts your box uses for the long-term average, it means your box is less hot over its lifetime."

Thermal Management

Without good thermal management, systems can sustain damage, and will have short life spans. New thermal management techniques are being used, along with standard techniques and improved materials. "The trends are the drive for more power and more performance," says Gregg Baldassarre. "There's a lot more wasted heat, there's smaller spaces because everyone wants to package things smaller, and there's demand for low mass because you want to make things more affordable."

Currently, thermal management is one of the major limiting factors of advanced computing systems for the military. Since the most effective cooling methods are difficult to ruggedize, expensive to implement, or both, the search for cost-effective thermal management systems with the right size and cooling capabilities has been a major focus of the Defense Advanced Research Projects Agency (DARPA) and the defense industry as a whole.

A k-Core® cooling uses encapsulated graphite to create a lightweight cooling plate that is often used in weight-critical applications, such as aircraft or satellites.


The DARPA Intrachip/Interchip Enhanced Cooling (ICECool) Applications program seeks to place cooling directly into the chip by constructing the chip with cooling channels that will pump coolant into hot areas. The program uses a non-conductive, dye-electric fluid that flows into the chip, but the program is still relatively new and is still dealing with many problems. "As you're pumping fluids through the chip, you're adding vibrations from the pumps and all the circulating fluids," explains Thermacore's Baldassarre. "How do chips respond to fluid flowing through it when before it was just static?"

The goal of ICECool is to make thermal management an important aspect of chip design, which will help speed the evolution of advanced chip integration. This will enable systems to overcome the size, weight, and power (SWaP) bottleneck that faces advanced electronics.

The DARPA Thermal Ground Plane (TGP) program seeks to improve thermal management by developing a practical, high-performance thermal substrate to improve the performance of defense electronics. "The goal is to get cooling closer to where the heat is generated," says Nelson Gernert, vice president of engineering and technology at Thermacore. "You're patching spreaders and vapor chambers to the chips and using material matching."

In areas where electronics are densely packed, such as vehicles, there has been a trend to include thermal management systems that electronics can use. "Usually the cooling system is part of the vehicle or end system, unless the [base] system is very large," says Parvus's Sweeney. "It's more of a recent trend, probably in the last five years, forced by larger power densities." By including thermal management technology in the end systems, electronics that are placed in that system can be designed to work with whatever thermal management tools are given to them rather than relying on conduction cooling.

Thermal management varies based on the end system. For amphibious vehicles and ships, liquid can be pumped in from the outside to provide cooling for electronics, while other systems can make use of forced air or other cooling methods.

The Future of Thermal Management

Many DARPA projects are in the works to improve thermal management. Programs range from ICECool to the TGP program, as well as the Microtechnologies for Air Cooled Exchangers (MACE) program, which aims to replace some of the expensive and difficult to maintain cooling systems with simple, inexpensive air-cooled exchangers, and the Active Cooling Module (ACM) program, which is meant to cool electronics using novel materials and structures.

With so many programs in progress, it's unclear how thermal management will continue to evolve. Aside from DARPA, the defense industry is finding new ways to cool electronics to provide the best equipment at the lowest cost.

The industry has been using new thermal interface materials and techniques, such as three-dimensional cooling, to help break through the bottleneck created by current thermal management practices. While there is no one true way to cool electronics, with so much new technology in the works it will depend on which projects work well and which fall to the wayside.

Smart Motion Control

Motion control has become more and more important as unmanned vehicles become more popular and more systems become automated. "We're more and more getting into smart devices," says Amir Shafy, applications engineering manager at North Atlantic Industries in Bohemia, N.Y. "We're seeing a lot more coming out in the UAVs and robotics, and even standard machinery and controls. They're scaling back on personnel and are having a lot more automation. There are a lot more I/O sensors that need to be measured of controlled, and along with that are a lot of positioning devices."

Motion control typically has been done with analog technology that requires additional systems to process the information and make it useful. "A resolver is basically a volter, tied with an AC voltage, and attached to the motor shaft," explains Shafy. "Basically, 3 transformers in a Y configuration. Depending on where the motor shaft is it will give you a different proportional voltage relationship. It's a very simple, easy, and rugged way to tell you absolute position, but the challenge is these are AC voltages and there's a lot of analog-type, front-end stuff that needs to be measured, managed properly, and then give you a digital version of that so the computer can take and use that information." That's where smart motion control comes in.

"If the sensors get smarter, they provide just an Ethernet or serial communications string, or CAN bus, stream of data," Shafy continues. "Then a lot of that analog part of it that gets put into whatever subsystems box it's measuring, it doesn't need that. All you're getting is communications messages that are already digitized."

By having sensors skip the analog process, or simply process the analog information themselves, systems can become smaller by not requiring subsystems that turn the analog data into digital data.

"As the processing power gets better, we're able to provide basically 10 functions on the modules now," Shafy says. "Instead of just giving you a raw A to D voltage we can process that and give you an average over time. What that does is we're able to offload a lot of the pre-processing of the data before it gets to the mission computer or the controller, so that computer can worry about its own processing. It doesn't have to worry about raw data input processing."

The benefits of using sensors that can communicate to the mission computer not only results in freeing up processing power for the mission computer, but it also results in reduced size, weight, and power (SWaP).

"There's not some brand new widget that now does everything," Shafy says. "But as technology evolves and we're better able to use it, we're able to offer more features, more channel density, and that all reduces SWaP requirements and allows us to keep up with the size of systems. For the feedback side of it, there may be new sensors that come out. When it comes right down to it, it's voltage changes, current changes, perhaps the sensors themselves get a little smarter so instead of analog and digital voltage changes, maybe it's a serial communications type string, it's a little hard to say."


Request More Information


   < Back to Articles