Who killed that component?
Q: Why do unstressed components sometimes fail for no obvious reason?
A: Sometimes they die of old age, sometimes the stress is there, but, as you say, it is not obvious.
“Old age” in a component is cumulative degradation due to physical or chemical changes. It is well-known that electrolytic capacitors and some types of film capacitor eventually die as a result of chemical reactions in their dielectric caused by a combination of trace impurities (oxygen, among others) and electrical stress. As integrated circuit structures follow Moore’s Law and become smaller, the risk of dopant migration at normal operating temperatures causing failure within decades rather than centuries does increase; and mechanical fatigue in inductors due to fatigue arising from magnetostriction is a well-known effect. Some types of resistance material oxidize slowly in air, and more quickly as the air becomes more humid1. Nobody expects batteries to last forever, either.
So when choosing components it is well to understand their structures and the possible age-related failure mechanisms which may operate even when the device is used under ideal conditions. This column is not the place to discuss such mechanisms in detail, but most reputable manufacturers will understand the aging of their products and are usually prepared to discuss service life and potential failure mechanisms – and many system manufacturers have publications on the safe service life of their products and the mechanisms limiting it2.
However most electronic components can be expected to last for many decades, or even longer, under the correct operating conditions – but some of them still die. The reason is often unconsidered stresses.
As we continuously remind the readers of this RAQ column, one of the more useful formulations of Murphy’s Law is “The Laws of Physics don’t stop working just because you’re not paying attention”. Many stress mechanisms are easily overlooked.
Everyone who designs electronics for use in a marine environment considers the effects of salt spray and humidity – and well they should, they’re horrible! But many electronic devices may encounter lesser, but still potentially damaging, chemical challenges. Human (and animal) breath is humid, and slightly acidic. Kitchen and other domestic environments contain mildly corrosive fumes of various types (bleach, disinfectant, cooking fumes of various sorts, and oils and spirits) – none of which is very damaging but we should not assume that our circuits are going to spend their entire lives in perfectly protected safety. Designers should always consider the environmental challenges their circuits will encounter and, where economically possible, design to minimize any potential damage.
Electrostatic damage (ESD) is one stress mechanism which we are continually warned about, but still regularly overlook. After being built in a factory where every care is taken to eliminate ESD during manufacture, many PCBs are used in systems without adequate protection from the ESD induced by normal handling. Adequate protection is not hard, but it may add a few pennies to the cost and so is omitted. That can be poor economy. Part of every design should be an assessment of what ESD protection is necessary for the system electronics in the most extreme conditions of normal use, and its implementation.
Overvoltage is another factor. Few people expect semiconductors or capacitors, to survive gross overvoltage, but it is common to see high value resistors subjected to voltages massively larger than the absolute maximum on their data sheet. The problem is that if their resistance is high enough they don’t get warm – but they may suffer from microscopic internal arcing and slowly drift out of specification, and eventually short-circuit. Large wire-ended resistors usually have breakdown voltages of many hundreds of volts so the problem was uncommon in the past, but today’s tiny surface-mount resistors may have breakdown voltages below 30 V and be quite vulnerable to overvoltage.
High currents also cause problems. Everyone is familiar with the common fuse – a piece of wire which heats and melts if too large a current flows in it, thus protecting power supplies from short-circuits and similar problems. But where there is a very high current density in very small conductors they may not get very hot – but may still, eventually, fail. The cause is electromigration3 (sometimes called ion migration). This is the transport of material caused by the gradual movement of the ions in a conductor due to the momentum transfer between conducting electrons and diffusing metal atoms. This causes thin conductors carrying high DC currents to become thinner with time and eventually to fail.
But some components fail like fuses – wires, or the conducting tracks on semiconductor chips, melt. A common cause of the high currents which make this happen is high capacitor charge current. Consider a 1 μ F capacitor with an ESR of 1 Ω: if it is connected across 110 V, 60 Hz, mains, an ac current of about 41 mA rms will flow in it. But if it is connected to the mains at the exact time the voltage is at a maximum (110 √ 2 = 155.6 V) the only current limit is the ESR and a peak current of 155.6 A will flow, albeit for less than a microsecond. But this is long enough to destroy many small-signal semiconductor devices, and repeated current surges may damage the capacitor itself, especially if it is an electrolytic one. This is a particularly common failure mechanism in the cheap low-voltage switching supplies (“wall-warts”) used for charging small electronic devices – if plugged in at the wrong part of an ac cycle the rectifiers and capacitors carry a very large surge, which may eventually, when it has happened many times, destroy them. A small resistor in series with the rectifier limits this surge current and minimizes the problem4.
If we are lucky ESD or overvoltage/overcurrent will destroy components instantly – so it is clear that there is a problem. More commonly, though, such stress may cause damage, which results in premature death long after the stress which initiated the failure has gone. Diagnosing the cause of such failure is very difficult and may be impossible.
When designing any circuit it is well to consider the lifetimes and failure mechanisms of the components used, and whether there are any potential problems and possible sources of stress damage to the components under the most extreme of the allowable conditions of use. Any such problems should be considered, and where possible minimized, in the final design.
1 See the Vishay Application Note:- “Predictable Components: Stability of Thin Film Resistors“
2 Some useful papers from the Emerson Corp are:- SL-24617 The Effect of Regular, Skilled Preventive Maintenance and Remote Monitoring on Critical Power System Reliability / SL-24628 Longevity of Key Components / SL-24630 Replacing-Capacitors.
3 See Electromigration.
4 For low power supplies a suitable resistor is unlikely to dissipate significant power (for instance a 33 Ω resistor in a 5 W/110 V supply will keep the surge below 5 A and will dissipate <70 mW) but with larger supplies of this type a thermistor may be necessary.
James Bryant was an applications manager at Analog Devices from 1982 until his retirement in 2009 and still writes and consults for the company.
This article first appeared on EE Times’ Planet Analog website.
Related links and articles: