The broad coverage offered by wireless communications systems such as cellular networks and Wi-Fi has proved to be a boon for embedded developers. Piggybacking on technologies that are essentially consumer-oriented, embedded systems can achieve fast data rates, low network connection costs and reliable operation almost anywhere in the world.
While the wireless technology is shared with the consumer world, the development process for typical embedded systems – with production volumes of a few thousand units rather than a few million – is somewhat different. To design wireless connectivity into a low- to medium-volume embedded device, development teams will typically adopt one of two approaches:
-
Modem Solution: Drop a PTCRB and network-approved modem on to the board. In this case, the only additional design work required is interfacing via a serial bus or USB. Multi-Tech Systems is the best known supplier of such modems for embedded applications.
-
Module Solution: Design an appropriate RF module into the system, adding components to provide required functions such as control, power supply, bus interface, antenna, circuit protection and so on. Widely used suppliers of such modules include Telit , Sierra Wireless, ublox, or Gemalto
This presents the development team with a 'make or buy' decision. By designing a wireless module into the system, the development team may achieve a Bill of Materials (BOM) cost reduction of approximately $10-30 for a typical cellular network-enabled product compared to the cost of a modem solution.
However, this lower BOM cost needs to be balanced against the cost of developing a wireless module in- house and gaining the necessary regulatory and carrier approvals. In the case of cellular systems, the approval process can be lengthy, expensive, and uncertain. Alternatively, a modem solution from a reputable supplier will already have network provider approvals for multiple carriers.
For the module solution, it is important to note the design and approvals costs are incurred up-front, BEFORE the product even reaches the market. It is a simple matter to calculate the break-even point – the number of units at which the higher initial development costs for the module is offset by the lower BOM cost of the module. These development costs for the module include not only the circuit design and prototyping costs, but also the need to gain network provider approvals and regulatory certifications. It shows a typical scenario, in which the break-even point occurs at around 6,000 – 7,000 units.
The make or buy decision is primarily a business, not a technical judgment. Operating in a standards-rich environment, there should be little difference in performance between a design-your-own module-based solution and a drop in modem solution. But engineering professionals can help business units make informed decisions take into account the design, competitive, marketing and supply-chain risks that can validate or invalidate the simple break-even calculation shown in Figure 1.
It is tempting for businesses to base decisions on the 'known known’s'; the BOM and development costs, which can be precisely forecast with great accuracy. By contrast, the risks listed above, and the potential costs that arise from them, are more difficult to calculate precisely.
Nevertheless, development teams can estimate with some accuracy a range in which the costs would fall based on assumptions that they can assign a probability to. This provides a far more robust method for calculating costs and benefits than the practice of ignoring the hard-to-estimate costs entirely. When this more robust method is used, development teams can find that the actual break-even point for the modem solution rises far in excess of 7,000 units.
The Risks of In-House Wireless Modem Development
The first potential cost to consider is the risk that dare not speak its name: design risk. Embedded design engineers might not want to acknowledge it, but there is always the risk that an in-house team will fail to produce a right-first-time design. RF design is particularly prone to glitches, because interference from elsewhere in the system can impair the performance of the RF module or antenna in ways that are hard to predict or diagnose. Experienced, capable RF design teams are unlikely ever to be tripped up. But in embedded projects that are adding connectivity for the first time, or that are replacing wired with wireless communication, the difficulty of implementing cellular or local-area wireless should not be underestimated.
The second important risk to take into account is competitive risk. Business units should ask the question, how much of the profit in any given market segment will be claimed by the first entrant? What is the risk of a competitor entering the market before us?
Multi-Tech's estimate, based on customer input, of the potential impact of choosing to make rather than buy: an additional nine months before an end product reaches the market. Every manufacturer will make its own forecast for development time, but in cellular applications in particular, the effect on project duration of the approvals process is defined – and largely outside the manufacturer's control.
The impact of another risk is also affected by the actions of customers: this is the risk that the manufacturer will need to develop product variants or upgraded versions of the original product. For instance, should the original product be successful, there might be marketing-led requirement to develop variants that use the original as a platform. These could be cost-reduced versions, 'lite' versions with less functionality or premium versions with extra features.
These variants might also require different wireless functionality, and a different modem. If the original used a module, any new variant might lead the project team to jettison the original design and the tooling used to manufacture it, as any replacement wireless chipset could require different peripheral circuitry.
The fourth risk for module-based designs is related to this previous one: what if a redesign is forced on a manufacturer by the obsolescence of a module’s wireless chipset? Cellular and Wi-Fi chipset manufacturers, which are driven by the dynamics of the consumer market, regularly make their devices obsolete and replace them with new, improved versions. Obsolescence can also result from the mergers and acquisitions that wireless chip manufacturers are subject to. Embedded product manufacturers, with their much longer product life cycles, risk redesign and retooling costs that are generated solely by their supply chain.
Finally, any manufacturer with ambitions to sell products globally will encounter the confusing and time- consuming world of approvals. Certifications for RF products are devised and regulated locally or regionally, not globally. Some countries require yearly certification renewal. If a certification expires, the product has to be withdrawn from sale until it is renewed. Regulations can also change during the life of any given model, and the corresponding certifications cannot always be grandfathered on to existing products. Embedded modem manufacturers have specialist teams that continuously monitor certification regimes throughout the world and maintain all their products’ approvals. A manufacturer requires a large and management and technical effort on the part of a manufacturer using a module.
How to Predict and Manage the Impact
The risks described above are by their nature difficult to quantify and cost. But it is possible to make a reasonable estimate of their cost that reflects individual circumstances. Each customer will have more or less skilled RF designers, quicker or slower competition, a wider or narrower range of end user requirements, and longer or shorter product life cycles. This allows each customer to estimate the likelihood of each risk occurring, and to estimate the cost. So, if there is a 50% chance of a module design being faulty, and the cost of a faulty design is $100,000, $50,000 might be added to the cost estimate for making a module solution, on top of the costs already accounted for.
A modem solution dramatically reduces exposure to the certification and approval risks. It is straightforward to drop on to a board, and the end product development time is considerably shortened. Multi-Tech SocketModem products (see Figure 3) also mitigate marketing and supply-chain risks. Each of the SocketModem products offer a combination of communications technologies, including 2G and 3G, GSM and CDMA, Bluetooth and GPS tracking and they all share the same Universal Socket pin-out and (where appropriate) Multi-Tech’s 'Universal IP® IP stack for persistent/automatic connectivity.
This means that the designer can swap different technologies in and out of the board at will, with no hardware redesign or retooling required, no requirement to write a new IP protocol and minimal changes at the application software level. Multi-Tech SocketModem’s are approved by carriers and regulatory agencies and support long embedded life cycles to reduce redesign cost and time.
The ability of SocketModem or similar modems to eliminate the normal risks inherent in in-house design makes the task of estimating and quantifying the risks worthwhile. As the analysis above shows, the 'make or buy' break-even point is driven by more than just BOM and product design costs.