Analog Angle Article

Bitten by debug

My “old faithful” clock/radio audio started acting up the other day. The unit is pretty old and, as the saying goes, “it doesn't owe me anything”. The smart thing to do would be to toss it and get a new one, of course.

But I didn't do that, since I view a broken appliance as yet another hands-on training session for the most difficult interdisciplinary engineering skill: debug. Certainly, finding out what is wrong with a prototype or new item which has never worked–or which has a short track record–is very different than finding out what has gone wrong with something that has worked well for many years, but it's still a good mind and discipline exercise.

One thing I always keep in mind when debugging is to never assume the problem is due to a single fault. Further, the “double-fault” can take on three guises:

  1. There's the serial fault , where Problem A leads to Problem B, which in turn leads to the observed problem
  2. There is also the parallel fault , where two independent problems combine to cause the observed problem, where either one by itself would not be sufficient.
  3. And finally, there is the unrelated cause double fault , where two completely independent problems separately cause the same observed problem.

It's the last one that is the most challenging, IMO. I vividly recall once working on a LED display panel which did not work properly; the characters were jumbled and erratic. After a lot of sweat, I found that a) the DC power supply rail was not solid, and b) the outputs of the row drivers for the display did not match up with the levels of the matching receivers. These two causes were absolutely unrelated, yet produced the same class of symptom. Fixing just one of these was only a partial debug and did not make the problem go away entirely.

So what about my clock radio? My initial assumption was that one of its capacitors had dried out after all these years. But when I paid closer attention, I could hear that the audio would shift from being OK to being very distorted, which leads me to believe that, instead, there is an intermittent connection that is perhaps thermally sensitive.

Some work with a heat gun and freeze spray to induce consistent failure, plus careful visual inspection under the magnifying glass and poking with a non-conductive stick, is my next step. ♦

0 comments on “Bitten by debug

Leave a Reply

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