Steps to building VPX systems with an integrator

4This article looks at the steps to building a complete system. For many intelligent and critical embedded systems, the integration of the final system happens at the large prime contractors such as Lockheed Martin, Boeing, Northrop Grumman, and others. The next level of integration is done at an earlier stage in the supply chain by the board and system suppliers. The focus of this article is on that earlier level of suppliers.

The factors to consider when building High Performance Embedded Computing (HPEC) platforms for critical applications have been increasing exponentially in complexity. With VMEbus technology you could pick an enclosure with relatively few options and shop for the building block processor, memory, storage, and I/O modules to assemble a system. There were relatively few choices to be made on form factor (either 3U or 6U), with an occasional deeper or taller board in the mix. The parallel VMEbus had few options and later bus interfaces supported just about every configuration defined in the VMEbus specification.

Then along came switched serial fabrics and the number of options exploded! With VPX there are a multitude of choices, as indicated in Table 1. This is by no means a complete list of options. A rough calculation shows that VPX has hundreds of times more configurations when considering just the major options. Many more possible combinations exist.

Table 1: A multitude of possible configuration combinations exist with VPX.
(Click graphic to zoom by 1.9x)

With this jump in complexity it becomes extremely important to understand the impact of your architectural decisions. To alleviate or at least reduce the integration challenges, it is often best to start at the system level with a supplier that has extensive experience in the integration of complete VPX systems.

Application ready

VPX suppliers throughout the industry have positioned themselves to offer as much of the supply chain as possible, both to grow their business and to improve the level of service and support to their customers. What sets most suppliers apart is the amount of unique intellectual property such as third-party hardware, middleware, customized drivers, or application software that they are able to integrate into the solution.

A common strategy that many of the system integrators use is to develop families of application ready subsystems. Each application ready subsystem is a customizable configuration designed for a specific application area. The key to cost efficient customization is to use a building block design with open standards architectures like VPX. This allows systems to be put together quickly in the short market windows of many programs.

Mercury Systems’ Services and Systems Integration (SSI) team works with customers to design, develop, and deploy the ideal solution for their needs. They consider themselves a “gateway” to the newest products and latest technologies – not just from Mercury, but other suppliers as well. One of the ways they can do this is by using open standards that allow them to tap into a large reservoir of products (see Figure 1).

Figure 1: The 6-slot OpenVPX Development Chassis from Mercury Systems is a key building block in Mercury’s subsystem development infrastructure.
(Click graphic to zoom by 1.9x)

“We provide system integration services to add value to our core embedded products, including products produced by our embedded board level partners,” stated Ken Grob, Embedded Computing Products, Elma Electronics, Inc. “Elma system integration services allow us to deliver complete embedded platforms.” Across the industry, suppliers are saying that their customers want to focus on their core products and services, particularly services that provide value and return. Elma maintains a selection of Application Platform design notes that enable a fast start to common platforms.

Curtiss-Wright Defense Solutions offers Application Ready COTS and Turnkey COTS systems, plus complete system design, engineering, and integration services based on program specific requirements (see Figure 2). Their engineering team analyzes customer requirements to design a system that meets their size, weight, power, cost, technical, environmental, and programmatic requirements. Oftentimes these program specific systems require use of services.

Figure 2: Application Ready COTS systems from Curtiss-Wright are general purpose integrated subsystems that are ready for deployment.
(Click graphic to zoom by 1.9x)

What to look for in a system integrator

Do your homework when selecting a system integrator. Be sure that the value added and services they provide fill the gaps that are critical to the success of your program.

According to Ken, “At Elma we are seeing requests for services that relate to complete integrated system platforms, but not including the development of the end application.” Suppliers like Elma have the capability to provide metal fabrication, and to design and build cables, backplanes, and flex circuits. Whether it be a specific technology, time to market, or risk mitigation, each system integrator has their own strong points.

A standout in today’s environment includes the ability to consider thermal performance of the entire platform, and the design trade-offs required to deliver the solution. “GE Intelligent Platforms is most often put to task to help solve thermal management problems,” commented Mac Rothstein, Product Manager at GE Intelligent Platforms.

“Our customers require us to be able to provide electronic design and software development services,” stated Ken. “In today’s market this also includes knowledge of operating systems and FPGA code development, such as VHDL.”

Common design tools, specifically 2D and 3D CAD tool environments, which are used across the entire development phase, are important. “Through the use of common design tools, we can offer our customer a real-time, eyes-on view of the system solution that is in the development process,” Ken pointed out. The common design toolset allows designers to bring together a model that is layered and allows them to consider the system as a whole, and not just individual parts. With this approach the customer and the design team are participatory in the integrated design. This process save a lot of time, and ultimately produces a systems platform efficiently while allowing designers to check interface issues and fit issues – both mechanically, thermally, and from a cable routing point of view.

Many projects are heavy on system modeling for thermal, mechanical, and signal integrity, requiring the tools be in place to deliver on this requirement. A resurgence in proven processes and domestic manufacturing services that can be delivered cost effectively has put a renewed emphasis on local manufacturing.

Step 1: Start the engagement

Begin the process by finding an integrator that aligns with your needs. A good place to start is by taking a look at application ready platforms that many of the integrators have to offer. Ken explains that when considering an integrated system, a rule of thumb would be the percent of native content that the supplier can provide when considering the packaging, and subsequently, how much of the hardware payload they can provide themselves from their partner base. “If we reach 60 to 70 percent and it is an application where Elma is familiar with the required deployment standards, then we are usually a pretty good fit,” stated Ken. He also warns in the case of application specific platforms to watch for application skills that are in alignment with requirements. Not all suppliers have the requisite application knowledge to get the job done.

Since many application specific systems are not readily available off-the-shelf, be sure to understand the development cost model. Is there Non-Recurring Engineering (NRE) or is the cost bundled into the product price? NRE is used to pay for the custom work that is needed. Bundling is often necessary when customers eschew upfront fees, believing that these costs are not in sync with their product revenues. Suppliers that are willing to bundle NRE are, by default, accepting some risk in return for increased product revenues.

Step 2: Define the system

The biggest challenge is often to define the system to be built. You may know what you want but then begin the negotiations to match up requirements with capabilities. Do not skimp on this effort – take the time necessary to develop a clear statement of work that you are comfortable using for the project.

“Our system specialists work with our sales team to address customer issues,” stated Mac. “Supplier system architects are assigned when it is time to develop a proposal or Statement Of Work (SOW). They meet with customers to fully understand the application and the platform configuration required to address the problems the platform is designed to solve. They guide customers through the proposal and lead in the development effort.” GE Intelligent Platforms primarily does hardware integration for customers that are going to add the application software and any additional hardware integration.

Figure 3: GE Intelligent Platforms’ MAGIC1 Rugged Display Computer combines CPU technology with the latest Graphics Processing Units (GPUs) for high performance in rugged applications.
(Click graphic to zoom by 1.9x)

An integration project can start from a part, or a set of parts, or it can start from the top down. Some customers already know that they want a system level solution to be produced. Other customers are exploring the make vs. buy premise, and are not sure what makes sense for their opportunity. “To address the process, we use a team approach. Our customer facing team includes our sales, field application engineers, and embedded computing architects,” mentions Ken.

At this step, everyone gets on board with the project. Integrators bring in their product managers and engineering teams. Projects may start with a set of specifications that are complete or a goal to work collaboratively to develop the SOW. A typical scenario would leverage standards, such as VPX, to help support the definition of the system platform. The customer would then drive the design with a SOW, quality specifications, performance and interconnect requirements, and a test procedure.

Step 3: Add value

With a completed and signed SOW, the responsibility of added value can be assigned. The integrator will determine what is available off-the-shelf, what must be custom designed, and identify sources for missing components.

Software, middleware, and FPGA IP are big drivers for added value. Adding value with higher levels of software services, such as device driver integration and test, and now FPGA code development is a growing part of an integrator’s business model, says Ken. Various Linux distributions and Real-Time Operating System (RTOS) vendors can be involved as determined by the project.

Some level of customization will most likely be required. This is where a building block approach with architectures based on open standards has great benefit. Every integrator has their own strengths. For instance, the main areas of customization for GE Intelligent Platforms are with thermal technologies for cooling and addressing different levels of ruggedization (see Figure 3).

Step 4: Shake and bake

Testing is easy to overlook or minimize but it is the most important step. Be sure to include detailed testing parameters in the SOW and then verify and monitor at this step. Take advantage of the expertise from the integrators as well as your own to be sure the best test and acceptance procedure possible is developed.

Integrators work with customers to develop test plans, procedures, and often purpose-built test suites necessary to exercise and verify the platforms they are building. Typical testing at the platform level includes continuity testing, functional testing, and qualification testing. Higher level categories include environmental and regulatory testing. Experienced integrators have extensive testing capability that should be leveraged. Application specific testing should be considered when selecting an integrator.

Step 5: Deliver on time

This is what it’s all about, isn’t it? Reducing risk, saving cost on resources, filling skill gaps, focusing on expertise. If you can’t get the product delivered on schedule then is it all worth it? Use diligent program management to track progress. Meet frequently enough to ensure that issues are being addressed appropriately. Stay on top of the process!


According to recent findings released by VDC Research, Avnet and Curtiss-Wright have increased their footprint in the systems integration market through their respective 2013 acquisitions of MSC Group and Parvus. ELMA also increased its presence through a recent partnership with ADLINK. Meanwhile, Kontron undertook significant efforts in 2013 to reshape its company regarding vertical markets and, at the same time, streamline its operations.

While these steps are not intended to be all-inclusive, it does provide a starting point for consideration. In my last days at Motorola, we were using a product lifecycle process that had over 350 separate steps. When shopping for a systems supplier or integrator, look for expertise in the technology that is essential to the success of your project. The right supplier can shorten time to market while reducing program risk.

Check the VITA membership roster for possible integration sources: