Analog Angle Article

Recheck your assumptions, for better design and debug

We design and debug using many assumptions: how much to bypass those power supplies, what constitutes an adequate ground, where to separate analog and digital grounds are just a few examples. Some of these assumptions are apparent, but many of them are built into our thinking and not called out explicitly. Experts, tutorials, experience, and learned lessons have embedded them in our engineering psyche.

Working with these assumptions is usually a good thing, because without them, little would get done in any reasonable time, and we'd make mistakes that could have been avoided. We'd be learning the hard way what others already have found out.

At the same time, these spoken and unspoken assumptions can blind us to when things are, to put it bluntly, different. There are occasions when the standard guidelines (also called “rules of thumb”) don't apply. Sometimes you have to ground the other end of that cable's shield, or ground both ends, even though it's usually better to ground just one. Or sometimes, things just don't add up right and some real out-of-the-box diagnostic thinking is needed, to figure out why the bugs in the system won't succumb to any logical troubleshooting analysis.

It's not just our niche that has assumptions which need reassessment. A brief January 2007 article in Popular Mechanics entitled “Your Dad Was Wrong” showed that a lot of traditional automotive wisdom no longer holds up. For example, “let the engine idle to warm up on cold days” is not true any more, given today's sensored and computer-controlled cars and fuel injection.

There are two reasons that your assumptions need to be reexamined. First, the conditions under which they are valid may not be present in a specific situation. Secondly, and no surprise: things change. When you have technology that moves as quickly as it does in our field, there are eternal truths but also constant re-visiting of the implications of those truths.

Perhaps we should remember there is nothing certain except death and taxes (attributed to various sources, see To these I would add Maxwell's equations, plus
F = ma, E = mc2 , and perhaps a few others!

-x x x-

0 comments on “Recheck your assumptions, for better design and debug

Leave a Reply

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