Speed is of the essence to embedded designers

5 mins read

In 1999, four authors posted a list of 95 theses on the web declaring how traditional marketing would be made obsolete by the Internet. The Cluetrain Manifesto, as this list was called, detailed how the internet would change business in general, and marketing in particular, by arguing that ‘markets were conversations’ and stating that businesses needed to take part in these conversations with customers to prosper.

The Cluetrain Manifesto has also had a profound impact on product development. Previously, the manufacturer developed products away from the public. It would decide the characteristics, functions and cost of the product with these well-defined targets in mind. When the product was eventually launched, its success was decided by the market. As this was such a well understood cycle, processes and practices evolved to increase efficiency and eliminate waste from design cycles to further increase profitability. Making the right choices at the start of the project, allowed the manufacturer to adapt the price if the market didn’t react favourably to the product launch.

It’s now 17 years since the initial publication of the Cluetrain Manifesto and its predictions have proven accurate – if anything, the authors underestimated the cataclysmic change the Internet has had on business. Several theses in particular have proven foresighted, especially when it is considered that they were made in the time before social media, and the main discussion channels were mailing lists and newsgroups. Findings such as ‘networked markets get smart fast’ and ‘the level of knowledge in a network increases at the square of the number of users, times the volume of conversation’ accurately predict the effect that sites like Facebook and Twitter have had on customers of every business.

The speed that information travels, and the increase in awareness driven by the Internet, has allowed manufacturers to make changes to the product development cycle to involve customers from an early stage. This involvement increases the chance products will be adopted by the market, and ensures they will more accurately fit customer requirements. Engaging customers directly before the final product was launched was pioneered by the Agile software development philosophy, which gave customers early access to new software and used their feedback to tailor the software to the customer’s specific needs for both function and use. Product manufacturers were also quick to see the value in early access programs and adopted the model for their closest and most trusted customers.

Another way to engage potential customers openly is seen by the recent rise of crowd-funding programs, such as Kickstarter or Indiegogo. These sites are intended to fund projects at various stages of the design cycle; from getting a project off the ground, to taking a prototype into full-scale production. Crowd-funding sites also offer another valuable service; they give the manufacturer a realistic insight into the demand for the product from paying customers. Together with social media channels, they complete the feedback loop with customers, allowing the manufacture to make changes to the design or add new features. The feedback and market insight gained from crowd-funding campaigns can even be combined by making additional features subject to reaching stretch targets.

As well as opportunities, the increased availability of information offers challenges to manufacturers. For example, it opens up customers to a much wider variety of choices. If a manufacturer cannot react quickly to customer demands, the customer has much more freedom to learn about alternative options. As mentioned earlier in this article, manufacturers used to be able to throw a veil of secrecy over a project and get to market with a unique product. The instant availability of information in modern society means that ideas propagate much more quickly, and many more manufacturers are likely to be working on similar products. Staying competitive means getting your product to market before the competition has a chance to emerge and gain a foothold. This increased pressure to get to market overturns the traditional method of new product introduction.

Nowhere is this maxim truer than in the embedded systems industry. This market has long been at the forefront of the adoption of new techniques, such as the waterfall development model for software. Since then, software has become a bigger component of embedded system, especially with the current focus on real-time control and the availability of powerful high-speed processors, such as FPGAs. These two innovations have fuelled a move towards more Agile software development. It’s not possible to exactly copy the Agile model used in the IT software industry, as the real-time demands of embedded systems require a product to have more initial functions before delivery to the customer. This has led to a slightly different implementation of Agile development, by talking to the customer and finding out the most important functions of the system and the desired interface features and developing these completely before delivery, then working on secondary functions.

The need for constant communication, gathering feedback and implementing changes has thrown out almost all of the careful cost calculations of the old design model. Previously, a company would have an accurate estimation of the overall cost of launching a new product to market. It could carefully calculate the cost of the various steps of the design process to a fully manufactured product. Because the exact specifications of the end product were known, non-recurrent engineering (NRE) costs could be calculated relatively accurately. With the current design paradigm, this assumption of cost no longer holds true. The constant implementation of new features and changes to system hardware means that the NRE costs are no longer a constant cost that can be estimated accurately. Compounding this change is the fact that there is less scope for regaining this NRE cost by selling large volumes of the product. To stop NRE costs spiralling out of control, manufacturers have had to look at off-the-shelf hardware solutions, which could be adapted easily to allow for later design specification changes.

One solution that is popular with manufacturers is off-the-shelf hardware, namely single-board computers (SBC). Although not as flexible as a custom design, SBCs offer a base module that provides a wide variety of processors, peripherals and I/O options. The base hardware of the SBC is fixed but functionality can be customised using plug-in modules. Most SBCs also offer expandability for core features like memory, ensuring that the board can be tailored for the exact application. The use of a pre-designed board ensures that the NRE cost does not spiral out of control. Often, the companies that make SBCs standardise on a processor, or a series of processors, to give customers an easy way to migrate designs if hardware requirements change, which makes porting code much easier. Digi has taken this approach with the Rabbit MCU series.

This method of using a SBC for the majority of the hardware portion of the system has a downside if there is not a large selection of SBCs to choose from. There is no one size fits all solution, so the manufacturer needs a wide and varied selection of boards to ensure it can get one that most closely fits the final function and has enough processing power to cope with future needs. For example, Digi-Key stocks nearly 600 different varieties of SBC from 15 different suppliers, which allows customers to meet the hardware requirements for almost any type of application. The manufacturer also needs a choice of boards that will ease design for specialised applications that would be difficult for standard SBCs, for instance, boards with specialised motor drivers for applications in the control and robotics industries. As well as performance features, some SBCs offer a choice of physical characteristics, such as the ability to operate over an extended temperature range. The MicroSoM series from SolidRun operates in temperatures ranging from -40°C to 105°C.

As well as having a wide choice, availability is also a huge factor for manufacturers. The Cluetrain Manifesto thesis that the customer will not wait, is as true for the supplier as it is for the manufacturer. This easy availability should not only be for a single board, or even a few units, there must be enough stock readily available for a production run or the manufacturer will go elsewhere. Some suppliers have recognised and reacted to this demand, for example Digi-Key can supply production quantities of its SBCs from stock, and furthermore has agreements in place from its suppliers to reserve stock in case a Digi-Key customer is planning a larger production run.

There has been a fundamental change in the design cycle for the vast majority of manufacturers of embedded systems. Custom designs are becoming less popular as manufacturers try to outpace the market with new designs. SBCs can offer a way to stay ahead of the curve, but that strategy will depend on forming a close relationship with a supplier that can deliver the right boards, in the right quantity, at the right time.

Author profile:
Randall Restle is vice president for applications engineering at Digi-Key Electronics.