Perhaps it’s just me, but I get the feeling that components just aren’t as good as they used to be. I’ve had a run of defective wall-wart power supplies, an IC that suddenly failed in-circuit with a puff of smoke, and potentiometers for an audio preamp project that just don’t meet specifications for resistance. I suppose it’s a reflection of cost-cutting measures in the electronics industry.
So-called early life or infant mortality problems are nothing new in the electronics industry, and there are numerous standards established by the military and industries to indicate the likelihood of component failure over time. My experience with component and board failures suggests that if a circuit or component functions adequately for 24 hours, it will last indefinitely if operated within design limits. For example, all bets are off if you subject a board to electrostatic discharge (ESD) or your power grid is hit by lightning.
Component and board failures can be soft (a change in value or performance) or hard (the component fails completely). Hard failures in high voltage components are often spectacular. I’ve had several electrolytic capacitors explode and flame out because of dry electrolyte. Low voltage components, in contrast, tend to simply smolder.
Soft failures are especially problematic in analog circuits. For example, resistor and capacitor values may change significantly over time, and op-amp performance may suddenly diminish. Such failures can be difficult to diagnose or correct in a complex analog circuit. I’ve only had one soft failure over the past few years on a digital circuit board — it was the linear voltage regulator chip, and not the digital electronics.
My most recent failure was that of an analog-to-digital converter (ADC) circuit. I built six identical boards for a project. As luck would have it, I had only five IC sockets for the key analog chip, and the chip that was soldered directly to the board failed. After about two hours, the chip made a snapping sound, followed by the telltale smell of a burned component. The chip’s failure could have been due to the thermal stress of soldering or simply a defective chip. Whatever the cause, removing the chip and preparing the board for the replacement chip was time consuming.
My experience highlights two aspects of component failure. The first is you should be proactive in your designs. This includes using sockets for expensive or difficult to replace components, and ordering a few spare components when you first place your order. You don’t want to be in the position of having to order a single component for 32 cents and paying $11 for shipping.
The second take-away is that you should make a habit of burning-in high value circuits. By this, I mean circuits that have to work — a garage door opener, controller for your R/C helicopter, or a device you’re selling or providing as a gift. I have a burn-in area on my workbench. It’s a 12 x 12 inch square of flame and heat resistant Formica laminate. I place the circuit in the center of the platform and connect a regulated, current-limited power supply to it, then let it cook for as long as I’m in the room.
Moreover, if the circuitry is rated at, say 9-12 volts, I stress it with 12.5 volts. My goal is to have the circuit fail on my workbench — before I spend time mounting the circuit in an expensive enclosure.
This brings me to something else that I’ve noticed lately — cost cutting in test equipment. In particular, if your new bench supply is rated at say, five amps, don’t assume that the leads can handle that level of current indefinitely. After one burn-in session, I found that a component had a hard failure, resulting in a short. The power supply was unharmed, but the insulation on the wimpy leads had partially melted, exposing the copper wire. I’ve since replaced the power supply leads with 12 gauge, heat-resistant wire, solid banana plugs, and beefy alligator clips.
If you’d like more information on component reliability, check out military standard MIL-HDBK-217, commonly referred to as MIL-217. NV