When R&D is designing a product, it is usually the result of top management's agreement that the new product will eclipse other similar products already in the marketplace only if it has a superior function, acceptable cost, and a significant market appeal.
The superior function of an electronics or electrical product is itself a function of the feature set and better electrical or performance specifications than what is currently available. So, with superior electrical specifications in mind, better components or a more sophisticated, leading-edge technology usually comes into play. The more difficult it becomes to meet the specification requirement, the more involved is the design effort. The design engineers must design-in operating margins or windows of acceptable specifications, or the production floor test yields may become so poor that the product in volume will not be economically viable.
Engineers do not want to "just barely" meet a specification, but they want to design with 100 percent yield likelihood in mind. Individual off-the-shelf components have operating tolerances that are specified by the manufacturer. A resistor may be sold as a 10 percent tolerance part, or it may be sold as a 0.1 percent part. This means that when purchasing a 100 Ohm resistor at 10 percent, you might be buying a 110 Ohm part or a 90 Ohm part, or +/- 10 percent of 100. But when buying a tighter-tolerance part, you are more likely to hit your value right on the money.
Tight-tolerance parts cost more than loose-tolerance parts. Semiconductors have specifications like "Minimum, Typical, and Maximum." So, generally speaking, you don't want to design something where it is running at its maximum rating, or you will be putting the part under extensive stress and thereby shortening its useful life. So, engineers have to design their products with the maximum volume production yield in mind.
One way to do this is to design with electrical, mechanical, and environmental specifications that exceed factory requirements for acceptability. So if I want to make sure that my factory can absolutely guarantee, say, a 10 percent tolerance window to the end user, my engineering team may design with 2 percent tolerances in mind. That provides a window of yield that will definitely fall within the 10 percent factory test parametric.
As I stated earlier, tighter tolerances cost more money, so the engineering group will have to make the critical cost decisions to determine where the cost vs. specification allowance is best served. Maybe, after testing with the 2 percent tolerance metric, there was a 100 percent yield, so the engineer may slip the specification to 5 percent and still realize a 100 percent yield on the factory floor. The 5 percent part cost less than a 2 percent part, and so the engineering team has successfully designed with cost optimization in mind.
Considering all mechanical, electrical, and environmental specifications optimized for cost and yield, the product now has the best chance to satisfy the end user at the lowest possible production cost. That is just your basic formula for success.
Think in terms of a single-lane mountain road. If you drive too close to either edge, you are putting yourself in danger. If you drive right down the middle, you are increasing your chances of getting to your destination in good shape. Proper specifications, or as we call it in the industry, "good specmanship," will drive your product right down the middle of your production floor highway, right on through your inspection and shipping department, and out the backdoor without incident.