Understanding the risks associated with component selection during the design and product definition phases requires a deep understanding of the timeline for long-term system development, and of when components are introduced by the semiconductor company. It is essential to consider potential misalignments in the market, architecture, and board. Component selection during development is a deciding factor in potential premature product redesigns and requalification.
Market misalignment
At times, the easiest or most efficient component selection can be the wrong choice, due to market misalignment. This situation may be feasible if there is a planned and budgeted last-time purchase within a few years of selecting the components; however, this is rarely the case. For example, graphics driver products have a very short lifetime compared with military or commercial avionics displays. Opting for a PC-oriented component in a market where those components are intended solely for product development will lead to obsolescence even before the first production units are shipped for long-term systems.
Architecture misalignment
Commercial avionics has long relied on the PowerPC processor multi-core architecture, due to its control of multi-core operation and speculative execution across multiple processor cores. The existing multi core PowerPC products have been certified for commercial avionics and software development is proven.
However, the end of the PowerPC architecture is on the horizon. It is only a matter of time before the commercial avionics market adopts ARM or RISC-V as their architectures.
Board design misalignment
There is always the temptation to pack DRAM as tightly as possible. Many systems use variable amounts of DRAM, either to enhance theirmproduct or to provide tiers within a product family. Packing that DRAM into as small a space as possible can be advantageous. However, the challenge with long-term systems, designed to last 15-20 years, is that DRAM technology will evolve significantly in that time. It is therefore crucial to anticipate this change by strategically designing board layouts upfront, thereby minimising the need for future modifications.
Key questions to consider:
- What is the component’s lifecycle status across the application’s lifetime? Not only must the end-product’s lifetime be considered in component selection, but also when the component’s lifecycle began, as well as the start and end dates of the product’s lifecycle.
2. Are the key components of the design comprehensively documented? Software is around 10x the cost of change compared with hardware. Any component directly controlled by software will be the most valuable in keeping a long-term system shipping.
3. Can the true design files be archived at the design phase to offer a chance of rebuild if the unexpected occurs?
4. Does the design contain proprietary intellectual property? If so, the ability to ‘port’ such designs when components are made obsolete may be compromised, or subject to relicensing and royalties.
Embedded IP blocks, particularly within FPGAs and ASICs, are commonplace. However, these IP blocks can also make portability and sustainability almost impossible if steps are not taken to ensure there is a plan for these products. This plan may involve a fully funded last-time buy, or upfront IP licenses that allow porting from one technology to another. Many companies undertake little or none of this planning for long term systems during the design phases, where the impact would be maximised. From component selection through to IP block selection, there are many ways a long term systems company can mitigate risk and plan for system longevity.
Ultimately, partnering with an authorised distributor and licensed manufacturer such as Rochester Electronics is the best way to ensure long-term system availability.
By Dan Deisz, Rochester Electronics’ Vice President, Design Technology
This article originally appeared in the embedded world North America supplement of Electronic Specifier Design – see ES’s Magazine Archives for more featured publications.
