A WIZARD'S ELECTRONICS COMPANION

TROUBLESHOOTING

For the serious troubleshooter, I highly recommend Bob Pease's "Troubleshooting Analog Circuits", published by Butterworth-Heinemann, 1993. For my part, I will attempt to make some suggestions in this large and vital area. If you're like me, your circuits never work the first time. Perhaps we were rushed, maybe we did it to ourselves to finish our newest toy. The reality is electronics is complex and only one single thing needs to be wrong for the whole thing to act like so much scrap. It might have been useful had you spent more of your efforts getting it right the first time, but that too is a trade-off. Building it is a requirement if you want it done, but troubleshooting is nature's largesse--maybe you'll get lucky, maybe not. Here, then, are things to try:

1. First and occassionally question the need to be doing what you're doing.
2. Second is to verify that the problem exists.
3. 95% of the time it's a bad connection.
4. Most problems you will categorize after the fact as "dumb mistakes."
5. Once you've let out the magic smoke, the ensuing moments of denial while we expect it to "just start working" often just causes more damage.
6. No matter how complex, you can divide and conquer the problem. If you can conveniently do so, replace equipment, boards, or subcircuits to quickly narrow down the source of the problem. Probe for your signal partway along the signal path, and continue eliminating as much of the circuit as possible from suspicion in an efficient manner.
7. Never completely disregard a failure mode by assuming it's too unlikely.
8. Use your head. You may be in uncharted territory and quick wits now could save you a lot of unnecessary blind alleys.
9. A careful visual inspection can work wonders. Whenever cross-checking yourself, arrange a different process than you used to create it. Turn the boards upside down, place them side-by-side, focus on color, on pattern, on part numbers, on missing components, on solder bridges, on cracked connections. Check lists backwards from the way you made them the first time.
10. Reflow poor solder connections. Sometimes it's quicker and more efficient to reflow every solder connection on a board than to locate the crack, especially as it can be indicative of more than one crack from general board stress.
11. Partial short circuits can often be tracked because they act like an added fixed component resistance. If its a very short circuit, try sending excess controlled current at low voltage from an external power supply. Five amps may be low enough to keep from damaging traces but still show tiny, but measurable, voltage drops along your conductors when probed with a sensitive voltmeter. If the voltage stops changing from place to place, it indicates a path where no current is flowing. The short will show its biggest drops where the largest currents flow. I have traced shorts to within a millimeter of their location in this manner.
12. It may be possible to raise the current in a short circuit enough to open it without otherwise damaging the board in a few cases, when you get frustrated wth the above, but following the voltage gradient is preferred.
13. Use of more of your senses can help locate problems. Risking burning your finger on a hot component Using your nose to detect roughly where the "magic smoke" escaped, when you can't see the damage. Follow the sound of an electric arc or escaping electrolyte from a capacitor (try not to breath much of this.) Use your psychic ju-ju, if you've got it.
14. Extend your senses for those hot spots. A thermocouple touched to parts can be quite helpful: Take a length of thermocouple wire, strip the ends and attach to a sensitive voltmeter that can detect tens of microvolts, and strip and twist the other ends tightly together. It won't read the right temperature numbers, but you can track a heat gradient this way. There is a device which plugs into a voltmeter which converts infrared into a temperature for the price of a moderate-to-low-cost multimeter.
15. 60Hz power line noise blankets the planet. The low frequency makes it all but impossible to shield from. Check to see if your noise is at 60 or 120 Hz.
16. If it's an intermittant, this can be the worst as it won't stay put so you can locate it. Parts may be sensitive to external influences. Semiconductors are light sensitive and will generate tiny currents or show resistance changes. Some are in glass or are potted with glass or translucent plastic which may allow light leaks (even reflected). Transducers may be affected by an extraneous sources of the very things they are designed to detect (light, energy fields, sound, movement.) Even software may only react badly when certain conditions occur which you have not yet determined, thus intermittantly. So, trying to create the problem by applying what you think might be the influence can greatly help. An oxidized or bent connection, or a crack in a solder joint, may yield a noticeable effect on test equipment to a gentle tapping on the board with a screwdriver handle. Physical orientation may help you if the influence is coming from somewhere outside your circuit, but nearby. An "RF shield" composed of a metal sheet (perhaps insulated so it can be up against your circuit safely) can partially reflect some stray fields, particularly between a noisy part of your circuit and one which needs to be quiet. Check electrical noise coming down your power supply lines, too.
17. A clever (or lucky) troubleshooter can often fix things without quite knowing how things work because most will be bad connections or bad parts which are themselves the problem.
18. A bad part may not be the real problem, in which case you will probably have to know how that part of the circuit works in order to find the real culprit.