SWaP and ease of integration keep CompactPCI deployed in unmanned vehicles

5The growing number of autonomous and semi-autonomous platforms require a diverse range of technology subsystems, prompting engineers to select 3U CompactPCI (CPCI) for unmanned systems deployments because of its low power and ease of use.

This year’s announcement that the number of active-duty Army personnel would be reduced to pre-World War II levels has positioned unmanned systems technology for a much larger role within the Department of Defense (DoD). Throughout the wars in Iraq and Afghanistan, unmanned systems were leveraged heavily for Intelligence, Surveillance, Target Acquisition, and Reconnaissance (ISTAR) missions, and advancing sensor technology only affirms their place in the Pentagon’s tactical situational awareness strategy.

However, while the sophisticated sensors of an RQ-4 Global Hawk or MQ-9 Reaper require the best processing performance and bandwidth capacity available, the fact is those larger platforms represent only a small fraction of the DoD’s Unmanned Aerial Systems (UASs) inventory (Table 1). For smaller unmanned vehicles the Size, Weight, and Power (SWaP) of systems like VME and VPX are simply too great, and represent an overkill for less compute-intensive applications in vehicle management and Payload Interface Units (PIUs). As a result, engineers continue to leverage CompactPCI (CPCI) technology for its low power and expansion capabilities, among other benefits.

21
Table 1: The Department of Defense (DoD) employs five Groups of Unmanned Aerial Systems (UASs), each with different tactical roles and technology requirements. Source: Unmanned Systems Integrated Roadmap FY2013-2038.

“In the ever slow-changing defense market there’s still plenty of demand out there for lower power, management-type vehicle applications where CPCI – because it doesn’t consume nearly as much power as a VPX system does – is still looked to as advantageous,” says Mac Rothstein, Systems Product Manager at GE Intelligent Platforms in Huntsville, AL (defense.ge-ip.com). “There is still a good demand for CPCI out there, and we really see it more in non-processing intensive types of applications where a more low-power processor is sufficient to communicate and process the data as it comes across.”

“CPCI has served unmanned vehicle applications well when backplane data rates are reasonable, outperforming VME in this regard,” says Rodger Hosking, Vice President and Cofounder of Pentek, Inc. in Upper Saddle River, NJ (www.pentek.com). “When high-speed data transfers between boards are necessary, however, VPX and MicroTCA (mTCA) both offer major advantages because dedicated gigabit serial links replace a common, shared data bus.

“Nevertheless, XMC modules with new technology still can be installed on CPCI carriers or CPU boards,” Hosking continues. “By using the ubiquitous PCI bus, CPCI can accommodate thousands of different custom and standard I/O modules in both PMC and XMC format (Figure 1).”

21
Figure 1: The Pentek 71610 is an XMC digital I/O module for control and data acquisition applications based on the Xilinx Virtex-6 FPGA.

Mike Horan, CEO of Dynatem, Inc., a Eurotech subsidiary headquartered in Mission Viejo, CA (www.dynatem.com), concurs that support for PMC, XMC, and FMC mezzanine expansion on CPCI enables the use of specialized memory cards from a variety of vendors, such as the PMCR-SATA2 RAID controller designed for unmanned vehicle applications (Figure 2, page 12).

22
Figure 2: Mezzanine modules such as the Dynatem PMCR-SATA2 RAID controller can be installed to insert new technology onto CompactPCI (CPCI) carriers or CPU boards.

PCI eases custom I/O integration in unmanned

Beyond mezzanine expansion, the PCI bus provides development and integration benefits when using CPCI technology because its universality often simplifies customization. In addition to eliminating complexities of PCI Express (PCIe) that result from functions such as Peer-to-Peer (P2P) communications, this can result in a lower Total Cost of Ownership (TCO) for CPCI-based subsystems, Rothstein says.

“We have a standard set of CPCI boards and they won’t always meet our customers’ specifications. A lot of times it comes down to making modifications to the backplane or the I/O interface with the 38999 connector to be able to route our customer-specific I/O out through the system,” says Rothstein.

“With the VPX world, normally you have to spin both the backplane and the I/O interface to accomplish that,” he continues. “A lot of times with CPCI we’re able to reuse the backplane due to the parallel interface and only modify the I/O interface. From that perspective, it’s a lower cost of ownership to make those types of modifications if they’re needed, and therefore really gives you a more proven solution in that there are a fewer number of parts that need to be modified to meet those demands.”

“CPCI system architectures are far simpler to integrate than VPX, with its vast number of different backplane topologies, data lane widths, and protocols,” says Hosking. “Because it uses a synchronous backplane bus, CPCI may also be easier to use than the asynchronous VME backplane. So, CPCI probably wins the ease-of-integration race.”

“Pricing is lower than VPX with a number of vendors offering specialized boards compatible with the backplane. This gives CPCI many advantages for unmanned systems,” says Horan. “PCI is universal and does not require special software drivers. Board-to-board communication [in CPCI] is usually simple CPU-to-I/O, and usually not more complex multiprocessing as seen in VMEbus or VPX. This is one of CPCI’s key benefits.”

“Based on the industry-standard PCI bus, CPCI boards can communicate across the backplane using standard drivers supported in virtually all Operating Systems (OSs),” says Hosking. “New technology insertions usually mean adding a new driver, but often much of the application software can be preserved for similar functions.”

CompactPCI alleviates UAS SWaP sensitivities

Unlike VME, CPCI is common in a 3U form factor, which is critical for SWaP-sensitive unmanned systems with strict weight and thermal limitations. Along with lightweight Conduction-Cooled Aluminum (CCA) housing options, this has enabled CPCI to carve out a niche in some lower category UASs, says Doug Patterson, Vice President, Military and Aerospace Business Sector at Aitech in Chatsworth, CA (www.rugged.com).

“Generally, 3U CPCI has fared well in some of the smaller UASs compared to the larger 6U formats due to smaller SwaP and sufficient backplane pins for the 3U size to bring I/O to the outside world interfaces,” Patterson says. “From Aitech’s point of reference, 3U CPCI tends to do well in the platform management and control segments of the platform (Figure 3).”

23
Figure 3: The Aitech C925 is a Freescale PowerQICC III-based 3U CompactPCI (CPCI) Single Board Computer (SBC), that features high radiation tolerances for high-altitude Unmanned Aerial Vehicles (UAVs) and uses the CAN bus for control applications.

“We have some programs that we sell 6U CPCI subsystems into, but by and large – mostly driven by SWaP and that you can get the system more than half as small – 3U environments are where we see most of the demand and where our product offerings are based around,” Rothstein says. “From the power standpoint, in most of our CPCI Single Board Computers (SBCs) you might see 20 W or so per board whereas VPX you’re looking at more 30-40 W just because we realize that in VPX systems it’s more processing intensive, and from that standpoint it does call for more power to be provided to a system.

“Thermal is nearly always the biggest challenge, even if it’s a low-power system,” he says. “That’s one of the reasons our mechanical engineers love CPCI because normally they’re dealing with a subsystem that’s less than 100 W, and it’s a lot easier to dissipate that heat than say a 200-300 W VPX system. You have to get rid of that heat somehow. Really it’s just making sure that you’ve got really good thermal contact between the sidewalls of the chassis and the conduction-cooled board to be able to move the heat off (Figure 4).

24
Figure 4: The CRS-C2I-3CC1 Rugged COTS System from GE Intelligent Platforms is a conduction-cooled, two-slot control computer based on 3U CompactPCI (CPCI) that is well-suited for military Unmanned Aerial Vehicles (UAVs).

“In CPCI systems, if we can stay in the 50-60 W, and sometimes even 70 W range, we can get by with just natural convection, which is just a thin chassis and if there’s enough air movement within the vehicle the heat can be moved away from the system sufficiently enough,” Rothstein continues. “Certainly that’s where our customers prefer to be because all they have to do is mount it and use it, and don’t have to worry about having a fan blowing right behind it or maintaining a cold plate.”

CompactPCI also uses a rugged pin and socket connector, and provides options for extended ruggedization with soldered memory, ECC memory, and thermally screened components, which are essential for harsh unmanned environments, adds Horan.

Prospects for CompactPCI in unmanned systems

Though it is unlikely that CPCI subsystems will be leveraged in many new deployments, long military lifecycles and the shortage of new defense contracts should extend the technology’s service. For current design ins, CPCI will continue to benefit from low cost and wide vendor support (see sidebar “Can Serial maintain the CompactPCI legacy of service?”).

21
Sidebar 1: Can Serial maintain the CompactPCI legacy of service?

“Let’s not bury CPCI just yet; CPCI still has plenty of horsepower for smaller, machine control applications when power and throughput are measured in multiples of milliseconds, not picoseconds where ISR lives today,” Patterson says. “Like 3U VME and some 6U CPCI, [3U CPCI] will remain in military service for years to come and either be enhanced with technology insertions or be replaced with some newer, new-fangled something or other technology du jour. It all comes down to mission definition and profiles, service personnel to be put into or taken out of harm’s way, collateral damage assessments, cost, and program lifecycles.

“Over the fullness of time, CPCI-based hardware platforms will become displaced by PCIe-based and distributed parallel processing solutions as the chip interconnect architectures move from the older parallel bus chip interconnects to multiple high-speed serial architectures – and as the Real-Time Operating System (RTOS) providers adapt to multi-processor, multi-threaded applications processes,” he adds.

“I was looking through a market study I received at the beginning of the year with respect to both system- and board-level components broken down by architecture,” says Rothstein. “The results there were similar to what my feelings are: I think we’ll see a steady decline in CPCI. It’s never going to be like you have a lot of it today and tomorrow you have none, just because of the market we live in. Once CPCI is designed into a platform, the likelihood of that being replaced within the next 10-15 years is pretty small.”