HOW STUFF BREAKS

 

 

Last month’s column wound up with the engineer’s lament:  no matter where you are, something’s always trying to break on you!  Like most sad stories, that one’s completely true.  The good news is that there are predictable patterns to components’ malfeasance.  This month, we catalog their misdeeds:

 

We all know that transistors fuse faster than the fuses that are there to protect them.  But fuses get tired, too, particularly the slow-blow variety—from metal fatigue in the expansion during each initial surge, supposedly—and finally blow when they shouldn’t.  Thermal problems cause intermittent faults in rectifier bridges, too.

 

Dust particles are attracted to anything at a high voltage potential.  They’ll stick and eventually provide a conducting path, which will then carbonize and probably blow to pieces.  This is partially counteracted because all physical connections, meantime, are busy expanding and contracting as they’re cycled, trying to work themselves loose.  Typically they’ll get hot and burn before they open up completely, though.

 

Electrolytic capacitors are always trying to either leak or dry out.  If they dry out, they’ll intermittently go open.  If they leak, the corrosive electrolyte will proceed to wreck any printed circuit board in the vicinity.  Printed circuit board material, meanwhile, will gradually carbonize under the influence of heat.

 

The heat generated by power carbon composition resistors actually causes the carbon granules inside the resistor to regranulate over the years, causing the resistance to drop over time.  Of course, in most circuits, this results in more current, and more heat, etc., etc., until the inevitable short.  Usually after that they’ll present a very high resistance—kind of like they’re trying to reform for their previous current-hogging ways…

 

Transmitting mica capacitors develop a series resistance that increases over time, often variable with ambient temperature.  The resistive component makes them get hotter and hotter, until the inevitable fire.  The old style, with the white ceramic coating, will sometimes contain themselves—the newer black plastic ones will usually spray a flaming rubbery plastic all over the place until the transmitter overloads and gives up.  They smell bad, too.

 

Metal oxide varistors will always fail by suddenly providing a dead short.  Of course, generally they’ll blow up when this happens, blowing off the leads and noisily restoring a nice open circuit that means no harm to anyone.  But then the ex-varistor is no longer providing any protection, either.

 

Mylar capacitors, or plate blockers, are prone to pinholes hidden under the plate bypass element.  If they were easy to see, there wouldn’t be any challenge!

 

Hollow doorknob capacitors heat up until the solder connections to their terminals melt.  The solid red doorknobs get pinholes or carbon traces, sometimes internal, usually very hard to see.  Typically they’re very intermittent, too.  O joy!

 

Oil-filled capacitors generally leak, or short out and explode.  Tantalum capacitors prefer the dead short.  Disc ceramics go “leaky,” developing a shunt resistance, or just absorb water vapour and start drifting in value.

 

High-tension wire will break down from heat and ozone, eventually carbonizing, typically near one end.  Carbon traces can travel several inches at the end, however.  Neoprene insulation will rot from heat and ozone, and flake off, exposing the copper beneath.

 

An old trick of the high-power inductors in the plate supply is to maintain their inductance, but short a point in the windings to the case.  You can sometimes solve this by slipping last year’s phone book under the case, insulating it from ground.

 

Ceramic tower guy line insulators (“eggs”) will develop carbon traces, then start arcing, creating RF interference.  The fiberglass ones will break down in the sun, absorb water, and arc until they’re fully carbonized.  Then they explode.

 

Old circuit breakers, like old engineers, just get “tired” and start tripping over nothing.