Improving efficiency with COM baseboardsStory
September 10, 2013
Embedded computing in the military is trending toward wearable, handheld, and other small and lightweight systems, as well as experiencing increased downward pressure on cost. These systems require rethinking the internals in order to achieve their goals of size, weight, and cost. As a result, the traditional architectures of card cages and stackable systems, which are ideal for quick time to market but carry additional bulk and cost, are giving way to the lighter and more efficient COTS-based COM + baseboard approach. In addition to benefits in size and weight, COM + baseboard systems offer significantly longer lifetimes, fewer problems during technology refresh cycles, and equivalent ruggedness and reliability, making them ideal for long-lasting military programs.
Tasked with the overarching goals of reliability and long-term availability, military electronics designers are understandably hesitant to adopt new technologies and methods that might prove unreliable in the long term or might disappear from the market altogether in a relatively short timeframe. The historic move to COTS was a recognition, however, that in order to achieve new performance and cost goals, the military would have to find a way to embrace faster-moving commercial technologies. After 20 years of experience with COTS, engineers are now far more ready and willing to incorporate new architectures into their product designs. This flexibility has allowed the introduction of new products with reduced size, weight, and cost for mobile and even wearable applications.
Old versus new
The mainstay hardware architecture for military electronics for the past two decades has been a combination of off-the-shelf boards (or customized versions of them) with dedicated functions, connected either with a backplane or a board-to-board stacking mechanism. This approach has allowed rapid prototyping, a high degree of configuration flexibility, and fast time to market, all significant benefits to any designer.
However, these boards also add a burden to the system in terms of size, cost, and weight. The backplane and/or stacking bus connectors exist solely because the system functionality has been spread across multiple boards. These board-to-board connection schemes require circuit boards, connectors, and bus driver circuitry, all of which drive these three important metrics in the wrong direction. In particular, as systems get smaller, PCB area as well as enclosure volume are at a premium, and any and all burdens on these limited resources must be carefully considered.
To alleviate these issues, the military is increasingly turning to a Computer-On-Module (COM) + Baseboard architecture. Instead of selecting one board for each function, this architecture requires just two boards in total: One for the processor and one for everything else. Adopting this new format provides a slew of advantages to help military system designers achieve their goals.
COM versus board set architectures
In a COM-based solution, there is only one board interconnection between the two boards (for COM Express, this interconnection may consist of a pair of connectors) – see Figure 1. In most cases, the interconnection scheme has been specifically chosen for minimal impact on system size, weight, and cost, while retaining appropriate levels of reliability and ruggedness (Table 1). The result is greater efficiency in utilization of PCB area, smaller size, and significantly lower overall cost.
Figure 1: Conduction-cooled COM-based embedded system
(Click graphic to zoom by 1.9x)
Table 1: Comparison of COM baseboard versus stackable I/O architecture in terms of size and cost – System configuration: Processor, DC/DC power supply, 8 multiprotocol serial ports, dual PCIe MiniCard sockets
(Click graphic to zoom by 1.9x)
A COM + Baseboard solution architecture brings other important advantages to the designer as well:
Perhaps the most significant benefit is life-cycle extension. Military programs commonly have life expectancies of 20 years or more, whereas the typical embedded processor lifetime is only seven years. This means the typical military program can expect to encounter at least two redesign cycles during its lifetime. A common scenario with Single Board Computers (SBCs) is that when the becomes obsolete, the SBC designer has a difficult time finding an equivalent replacement. This is because no two SBCs are exactly alike, either in features or in connector type and placement. Typically the designer will engage in a series of unpleasant trade-offs among features, cost, and additional boards in order to obtain a suitable replacement to extend product life.
With a COM solution this problem simply disappears. All COMs of a given form factor offer essentially identical features. Therefore, changing to a new COM is a much simpler task. Product life can easily be extended another five or more years with minimal redesign effort.
The same reasoning applies when it comes time for a performance upgrade. The existing COM can be swapped out easily for a higher-performance one when additional processing power is needed to run new software applications, which is a common occurrence in long-life products.
Time to market
Another advantage of using COMs is that most vendors offer the newest processors in COM formats long before they offer the same processors in small form factor SBCs. Therefore, a COM can potentially offer a longer life cycle than an SBC with the same processor even before the technology refresh cycle kicks in. Case in point: At Embedded World 2013 in Nuremberg, Germany, the Intel booth had a display of no less than eight vendors showing their implementations of Intel’s fourth generation “Haswell” Core i7 processor, all in COM Express Basic format. No standard form factor SBCs with this processor were on display. The processor had not yet even been released to the public at this time. This means that customers selecting Haswell in a COM Express form factor can enjoy the full seven years of expected lifetime, whereas waiting for an SBC might cut one to two years off that life expectancy.
Physical system design
Physical system design is another important consideration. Backplane systems typically severely restrict the overall size and shape of the system. In fact, most backplane architecture boards include front-panel slices with built-in I/O connectors, forcing the designer to use what the vendor has chosen, as well as impacting ruggedness and ingress protection. (The vast majority of such boards use commercial connectors, such as Dsub, rather than the rugged ones preferred for military applications, such as 38999.) With stackable architectures, the designer has much more freedom in designing the enclosure and selecting I/O connectors. However, cabling from the boards to the I/O panel can be a challenge, with board connections potentially on all four edges of the boards and no ready method to secure the cables internally. With a COM baseboard, the designer has total flexibility to choose what type of connectors to use and where to put them. In many cases, cabling can be eliminated by mounting user-accessible I/O connectors directly on the PCB. Benefits of this approach include reduced weight (eliminated cables), reduced assembly time, and increased ruggedness.
Addressing COM concerns
A common criticism of COM-based solutions is the expense of designing a custom baseboard. There are two answers to this complaint: First, the industry is starting to see off-the-shelf COM baseboards in the same size as the COMs themselves. These baseboards offer fast time to market as well as a convenient form factor for creating the total solution.
Secondly, although the cost of the baseboard project might be significant, amortizing it over the life of the project will usually result in a lower total unit cost than paying a vendor for an off-the-shelf SBC. For example, a $50,000 design project amortized over a 500 unit project lifetime is only $100 per unit, which is probably much less than the gross margin a vendor is charging over the production cost of the SBC alternative. Even with the cost of the baseboard rolled in, the overall cost of a COM-based solution is much lower than that of an off-the-shelf SBC + I/O module.
Another frequent question is the ruggedness of COM-based solutions, compared to the proven ruggedness of stackable architectures such as PC/104 (Figure 2), SUMIT, and EMX. Perhaps the most attention focuses on the fine-pitch gold fingers found on Qseven modules. Diamond has conducted shock and vibration tests according to MIL-STD-202G test criteria on the Qseven connector and found it to be sufficiently reliable to use in military vehicle applications. Part of the reason that Qseven survives this level of treatment is that it offers four mounting holes to help minimize relative movement between the module and the carrier board. COM Express vendors are starting to introduce substantially thicker PCBs (.093" versus the typical .062") that offer better resistance to flexing from vibration. In a nutshell, a COM-based solution can offer equivalent ruggedness to a stackable SBC.
Figure 2: PC/104 stackable embedded system
(Click graphic to zoom)
All of these factors combine to provide a compelling argument for using COMs plus custom baseboards in military applications, especially where long life, size, weight, and cost are concerned. As discussed, designers can have full confidence that COM-based solutions will exhibit equivalent ruggedness to traditional SBC or backplane systems, and their extended life cycles plus easier technology refresh efforts provide additional appeal.
Jonathan Miller is Founder and President of Diamond Systems Corporation. He can be contacted at [email protected]
Diamond Systems Corporation 650-810-2500 www.diamondsystems.com