Overcoming design challenges for ultra-compact, rugged embedded computing
Unmanned vehicles must be supported by a new generation of rugged small form factors computers.
It's not easy to squeeze a lot of computing power into a small space, but many applications call for this capability. An inside look at a new small form factor and the reasons behind developing it shows how design teams are approaching system problems in the rugged computing space.
Military planners are seeking to achieve persistent surveillance over areas of interest. Accomplishing this goal will require the deployment of various intelligence-gathering sensors, from simple video to sophisticated radar, in expanded numbers of small, unmanned vehicles. To be truly effective platforms for sophisticated sensors, these unmanned vehicles must be supported by a new generation of powerful, rugged, and ultra-compact signal processing computers.
Identifying design requirements
In an effort to meet the demand for this advanced capability, Mercury Computer Systems engineers designed a new type of rugged small form factor computer from the ground up. The design team determined that the unit needed to fulfill several challenging design requirements, including:
- Weight of less than 10 lbs (4.5 kg)
- Volume of less than one-half cubic foot
- The ability to function in extremely harsh environments
- Maximum processing power no fewer than 100 GFLOPS
- Based on a high-speed interprocessor communication infrastructure
The design team recognized that common rugged computing form factors such as 6U or 3U would not meet the weight or volume requirements. Also, to address processing-intensive applications, the platform needed to be configurable to support multiple processing engines with diverse capabilities.
Ultimately, the designers created a 6-slot ultra-compact chassis specification that at 4" x 5" x 6" (10 cm x 13 cm x 15 cm) would adequately address the space constraints of small platforms. Several processor modules, each supporting I/O daughtercards, were designed to fit into this form factor.
Any system that delivers 100 GFLOPS of processing power will generate significant amounts of heat. To be able to function in extremely harsh environments, the Ensemble Series 1000 system (shown in Figure 1) had to incorporate a tightly enclosed system chassis and the most effective means possible to dissipate heat in its confined space.
Conduction cooling and spray cooling were two potential options for removing heat from the individual modules. Because the overall system design was already constrained to a small size, conduction cooling provided an efficient means to draw heat outward from the boards to the chassis without the spatial requirements or mechanical complexities introduced by nozzles or pumps.
An aluminum heat spreader (heat sink) is an integral part of each module, moving heat from electronic components to the chassis. This heat spreader (diagrammed in Figure 2) is fixed to the chassis walls using a wedge-locking mechanism that ensures maximum heat conductivity with the positive side effect of increasing reliability across a range of environmental conditions, including shock, vibration, and temperature changes. Further ruggedization includes high-reliability connectors and o-ring sealing for pressure, humidity, and EMI isolation.
The final step in dealing with heat dissipation was to remove the heat generated by multiple modules in the chassis. The engineers created one design that uses flow-through liquid sidewalls in the chassis, exploiting the fact that the thermal capacity of a liquid is much greater than that of air. In addition, unlike air, the cooling capacity of a liquid is unaffected by altitude.
This approach puts the final cooling burden on the platform to supply the liquid flow to take away the heat, but many platforms already support some form of liquid-cooled electronics. Almost any liquid can be used, provided the liquid temperature and flow rate are sufficient. Platform cooling strategies can also be very creative, such as platforms that use their own fuel, moving from storage tank to engine to cool electronics.
The other design option for dissipating heat from the chassis is simple conduction to the outside environment. In this technique, a ribbed chassis maximizes heat transfer to surrounding air, while the metal base conducts heat directly to any supporting platform. Although this approach is much simpler to implement than liquid flow-through, heat removal efficiency is highly dependent on the surrounding temperature and altitude during operation.
To deliver maximum computing power to multiple sensor types, the system needed to be configurable with a selection of processing elements and had to provide an efficient way to move an input data stream between these elements for multistage processing.
The system meets these requirements with six slots interconnected by a high-bandwidth PCI Express switch fabric backplane. This backplane provides four lanes of simultaneous point-to-point full-duplex communication to each module with an aggregate bandwidth of 10 Gbps of raw data. Processing options include the P.A. Semi PA6T-1682M, Freescale PowerQUICC III and Intel processors, Graphics Processing Units (GPUs), and Xilinx Virtex-4 and Virtex-5 FPGAs.
Surpassing the goals
The final result is a system that exceeds the aforementioned design goals. A fully configured Ensemble Series 1000 weighs just under 7 lbs (3.2 kg) and can be held comfortably in one hand. Configured with six PA6T-1682M modules, it delivers up to 172 GFLOPS of processing power. ➤