It’s easy to assume that implementing a small design change will be trivial when the actual impact is often more extensive and more painful than unanticipated.
At its core, engineering design is finding the right balance among tradeoffs, driven by product objectives while, at the same time, being bounded by constraints. If it weren’t for limitations due to noise, power, cost, size, weight, or other factors (pick one or more), the design would be easy, right?
Sorry, it wouldn’t be easy: it would still be hard. It doesn’t just end with the design, either. Often, getting the product through a pilot-run phase and into production with the needed fixtures, tolerances, tests, and more is as challenging as the prototype design phase.
That’s why it’s essential to be very cautious when someone – especially someone not involved in the planning phase or the design review – suggests or insists on a seemingly minor change in a product design, especially late in the cycle (Figure 1). Their key phrase is usually “that should be no big deal.”
But is it really “no big deal?” Sure…but only for those not involved in actually implementing and dealing with its likely impact cascade. There are a few phrases that capture this situation clearly: “the law of unintended/unforeseen consequences,” “the ripple effect”, or perhaps more colorful, “the butterfly effect.”
What’s the butterfly effect? It’s a dramatic example of “chaos theory” and “deterministic chaos” – that should be warning enough. It is a term credited to meteorologist Edward Lorenz who discovered in the 1960s that tiny, butterfly-scale changes to the starting point of his weather models resulted in anything from sunny skies to violent storms, and with no way to predict in advance what the outcome might be. In colorful terms, it’s when a butterfly flaps its wings in China, and the result is a tornado a few days later in Kansas (Figure 2). In engineering terms, small perturbations early in a process or inaccuracies in modeling can have enormous implications as the process or model unfolds.
I suspect you can cite some examples you have encountered of this ripple effect on a large or small scale. For example:
- You’re asked to add yet another useful feature to a product, but now you need just a bit of memory, and, whoops, you’ve maxed out what you have left unused. That necessitates going to a higher-capacity processor or memory IC, but that comes in a bigger package. Now the entire board needs a new layout, which brings all sorts of signal routing, noise, and even additional IR-drop issues.
- Or someone decides the product needs another external I/O port, and fortunately, there’s room for the interface IC and connector on the PC board – that’s the “no big deal” part. But now you have to redo the bill of materials (BOM) for the new I/O interface IC, connector, and any protection devices it requires, and redo the enclosure’s design for access to that connector. Then, checking on the power budget, you find you’ve just exceeded what the internal DC regulator can supply, so you need to go to a larger-capacity regulator with a larger package and passives and a different layout.
- You change the gain-setting resistor values on an op amp – ostensibly no big deal at all – but this increases the circuit noise. The degradation in signal/noise ratio (SNR) and impact on low-end resolution requires the inclusion of a new analog filter which, in turn, bumps the design up to a larger circuit board.
There are countless cases of this ripple effect ranging from small-scale perturbations which soon damp themselves out to major projects which require a near-total redo to accommodate an apparently small change mandate. So the next time someone casually says, “that should be no big deal”, ask them – and yourself – “have we rigorously gone through all the implications before committing to this?” You and the person making the request may be surprised at the answer.
Related EE World Content
Chaos Reigns Even In Simple Electronics
Can chaos theory help predict heart attacks?
Reference
Paul Halpern, Forbes, “Chaos Theory: The Butterfly Effect, And The Computer Glitch That Started It All”
Leave a Reply
You must be logged in to post a comment.