Analog Angle Article

Sometimes it’s smarter not to be too clever

We all admire clever engineers and their designs, and with good reason. (Maybe, sometimes we're the clever one!) Cleverness is often the only way to get around seemingly intractable design requirements, requirements which are often in conflict with each other.

But there is such a thing as being too clever. Years ago, when I worked doing circuit and system design, one of the members of the project tram was a brilliant circuit designer. His initial schematics and designs were often near-perfect; the mistakes were usually due to errors in vendor data sheets, not his doing. For one design, he used a basic digital gate available from at least a dozen vendors, but insisted that the bill on materials (BOM) only allow the version from a particular vendor. Since it was a small and straightforward part a much larger circuit, no one asked why, we had other things to worry about.

That was the situation until the final design review, which was attended not only by the design team, but also marketers, field service, and the purchasing/production people. One of them asked the obvious: why that specific vendor only for that gate IC? They were concerned about availability and pricing flexibility.

Long story short: the very clever designer had noticed that the internal design of the gate output from that vendor was such that he could use it for another function as well. I don't remember the details, but it was some sort of improvised “OR-ing” function. The good news was that this implementation eliminated another IC.

But the negative implications of this approach far outweighed its admitted cleverness. The IC vendor did not guarantee this internal structure, and might change it at any time without notice, while still meeting all published data sheet specs. We were tied down to that one vendor in both up-front purchasing and field-replacement stocking, yet there were plenty of alternate good sources for this commodity part. And from an engineering perspective, you could not tell from looking at the schematic what was going on internal to the IC, so in a year or two no one would be able to figure it out. If the design-background documentation was lost over time (hey, it happens), or that designer ever left, or his memory faded, we'd really be in trouble.

In the end, the legitimate concerns of the various team members prevailed, and the circuit was redesigned to be less clever, more straightforward, even at the cost of another IC. That was the right decision, although it was perhaps less pleasing from a clever-designer's esthetic perspective! ♦

0 comments on “Sometimes it’s smarter not to be too clever

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.