The infamous smoke test: for engineers the culmination of hours or weeks or years of design and simulation and analysis; for managers the first tangible indicator of possible success or failure. It’s also the ubiquitous engineering term, channeling a swashbuckling approach that we all wish was our style. The term can even bring a little fear to the uninitiated bean counters and marketing mavens, and it’s the rare engineer who has not over-dramatized the literalness of an upcoming smoke test a tad.
For those who are afraid to ask what a smoke test is, it is the first gross functional check of a new product. For electronics, its the moment when power is applied. For plumbing, the analog is a wet test, when fluids are unleashed. For a boat, trying to float it. The name perhaps comes from the early days of electronics where firing up a new circuit could indeed lead to smoke. Even today, this is not as far-fetched as it may sound–a single integrated circuit with a power or ground pin reversed; a polarized component like a diode or a capacitor stuffed backward; these can all lead to that puff of smoke.
I’ve seen it happen, on my own bench back, in the day. And as we were building 500 V circuits we sometimes got something better than smoke–plasma arcing! On occasion you could see the aluminum wiring turn molten… But I digress.
Engineers, and their safety departments, all have different approaches to the smoke test. Some will take the direct approach–hook it up, throw the switch and have a fire extinguisher handy. Others will test as many parameters as they can short of actually hitting the “on” button. And of course many shades of gray in the middle. Beyond the courage factor, of course, real business and technical factors can and do come into play. A million dollar prototype is a different beast than a $10 circuit board. And flippancy aside, products that have potential for real safety issues should never be subjected to avoidable safety compromises. And if said $10 circuit board is the only one on the continent, and Christmas is coming fast, then caution may still be required. On the other hand, a smoke test also is one of the fastest ways to learn a lot about just how good or bad the design is, to get to the core issues.
We’ve smoke tested a couple of electronic products in the past month. The first one was a relatively straightforward redesign to fix a noise problem. Nevertheless we checked the polarity of the tantalums and did a thoroughly visual check before flipping the switch. It worked darn near perfectly, and no smoke.
The second smoke test was two hours ago. The product is not that complex but we currently have the only prototype in existence and, yes, Christmas is coming (for those unfamiliar with consumer electronics, Christmas manufacturing is happening right now–by December its way too late). So we did go through a fair amount of checking before we threw the switch. No smoke, but the supply current was 4X what we expected (which could lead to damaging sensitive components), so the kill switch got flipped really fast. We spent a couple of hours digging into the issue, trying the part several more times, each time disengaging quickly before anything could overheat. The culprit was fairly rapidly found–an oddity in the boot code that had some power circuitry unexpectedly on at boot. No harm, no foul, so smoke. And we can now proceed to the real engineering challenges ahead.
Smoke ’em if you got ’em. Just have that fire extinguisher handy.