That's right on the money. Something else not mentioned specifically is that many electronic device manufacturers pull in the best(most of the time, lowest) bids for outside companies to produce the circuit board components, lcds, electronic ribbon tapes, etc. Unfortunately, as the bid price goes down(and the electronic device manufacturers profit goes up) so may the quality go down. The failure rate of the components being sold may also substantially be increased as the bid price goes lower. And there are different standards in different countries as to what "ingredients" are used in building the components.

 

Hard to say on this one. Transformers and caps are actually considered to be in the passive component category.

Transformer failures include dielectric breakdown between windings or overheating. Both can be avoided if the designer didn't push the poor thing assuming there wasn't an inherent flaw in the materials or construction of the device. Caps typically blow out or die from overheating, dielectric breakdown, loss of electrolyte, internal shorts and opens from all sorts of reasons as well. I've seen caps literally burn up in seemingly benign designs where they are exposed to excessive switching currents up in the MHz range. Ultimately they failed due to high ESR in those specific frequencies. Again, proper design and component selection and construction can avoid such things.

Example: caps supposedly measure open-circuit or in the very high megaohms with a ohmmeter (DC). Stick a 200V electrolytic across 120VAC and watch it blow up... You'd think nothing would happen because it's open circuit and it's got the voltage rating. However, at 60Hz, it looks like a low ohmage resistor that simply heats up.

Try that with one of those audio amp stiffening caps and have some more fun!

Resistors can die open-circuit from overheating as well.

Wiring technically doesn't have fixed values per se. As with everything else, they have resistive, capacitive, and inductive properties that change across frequency and temperature amongst other variables. Prime example: a piece of romex measures about 0ohms at DC. Try passing 6GHz, and you'll see a huge loss in signal at the other end.

BTW, this is where some of the marketing comes into play to make people believe their products are superior to others and therefore charge them extra $$. There are lots of "audiophile" cable companies that claim their audio patch cables have significantly lower loss than cheapy standard RCA wires that come with whatever you buy from the store. They claim some fantastic dB measurements that are lower. Lower than what? What they also fail to state is what frequency range they are comparing and show pretty graphs which are seemingly convincing. Their price: $100-200bux for a 3ft cable with a ton of exotic materials that make it look like something that goes into a Space Shuttle. An ugly $2 radio shack cable will perform equally well down in the 20KHz region; it might not perform at all out at 10GHz--but who cares? Nobody can hear anything out there anyways, assuming the product even has that bandwidth to work at 10GHz.


Bottom line is--given an unlimited cost budget, a smart design engineer designs a circuit with all these things in mind and selects the proper components so that it works robustly. The product life is then a function of mortality rates inherent in the components used (which is typically very low) rather than failure due to a crappy design and crappy materials.

 

True! The huge challenge in consumer-grade commodity products is building something while maintaining a profit. It's brutal. No joke. The profit margins on commodity goods (like hard drives and pc motherboards) are very low despite what some think. They survive on huge volumes.

We as consumers naturally don't want to pay, yet want more for less. There's ultimately a price floor in how low the raw materials/components cost to build something. Then factor in labor costs... hence the move offshore to China etc to build stuff. BTW, China is getting expensive now, so I see movement to other countries like India and Vietnam now. Then those offshore mfrs are pushed to squeeze even more to reduce the cost, so they resort to using cheaper grade materials and manufacturing methods. It gets real bad. That old adage, "You pay for what you get" is all so true :-(

Fortunately, some of the high end, more expensive stuff remains here and is still built here the da U.S.A.

 

Seein that this is basically been my life for 20 years, I quess I gotta chime in. Each type of component has its own top failure modes, but some are common:
- physical damage, could be before or after you receive the gear. Ceramic caps are notorious for being 'brittle', the larger they are, the more prone they are.
- solder quality. includes many possible issues from contaminants on the PCB or in the solder itself. 'Cold' solder joints, or those that develop cracks. Also, all solder joints are in permanent stress due to their solidifying at a higher temp than they 'exist' in afterwards - the solder does actually 'move' from this stress.
- heat. as someone mentioned the rule of thumb is 10C cuts the MTBF (mean-time-between-failure) in half.
- electrical overstress (EOS). Unfortunately this is usually the lazy repsonse to something more involved - its the darling of many of some component mfg who aren't really interested in figuring out why something failed (although EOS can and certainly has claimed it share of failures).
- PCBs. Although it is a 'passive' component, PCB's can easily be a driver for reliability, typically due to thermal cycling that basically rips the bd apart causing opens, maybe shorts, or other 'defects' that can cause things like moisture to get into bds and eventually cause issues.
- Maybe its was posted here, but typical electronic failure rates follow a 'bathtub' curve - higher failure rates in the first xx hours/days/weeks or service, settles down to a lower value, and then increases again as end-of-life approaches. So when you go buy that new stereo or TV, its likely that if the electronics last the first 10 or so hours it will last the expected life (physical stuff like switches, potentiometers, etc have mechanic wear so the failure curve is different).

One thing of note, the switch to lead-free solder. This is being driven mainly by the EU (effective July '06) and has and is being adopted by more and more countries. Lead solder has its quality issues but they are very well understood, being used in the electronic industry for many decades. Lead-free paste on the other hand has little history on a large scale. There are issues, such as 'tin whiskers' (tiny 'hairs' of tin will actually sprout from the solder joint and grow which can lead to shorts) where the root cause, effective screening and mitigation is not known - typical pols putting the cart before the horse. 2 concerns here - although the industry 'thinks' they'll be little impact on reliability, knowbody really knows (and the increased melting point stresses everything that much more, many components are marginal here); #2 MTBE - need I say more (Hey, that stuff is real bad, why not try (i.e. we'll legislate) this new-fangled stuff and see what happens...).

So how long will electronic circuit last? Far too many variable to give any 'global' answer, but I'd quess many of us have 'electronics' that have lasted 20, 30 or more years. Using 'calculations' (Mil-STD, Bellcore) can 'estimate' this (not well in my experience); intense Accelerated Life Testing gives actual data but is 'accelerated' and its not trivial to project this to a device under normal circumstances; actual Life testing is the best way to estimate actual in service use - typically done with many units (could be 10 to 1000's depending on the expected failure rate). The more units, typically the faster you start to put boundaries on that bathtub curve. As you'd think its also typically the most expensive (due to the # of units you start with) and time consuming (the acceleration factors are typically less). Most mfgs will use actual field data for this, which is arguably the best indication as its as the customer receives, installs, and uses the product, but it also has gaps from misuse, unknown time from shipment to actual use, and (usually the biggest issue with consumer stuff) only a fraction of the failures are reported - makes the #s look a whole lot better though...

 

Wow - this thread is going ballistic!
I'm going to vote it a star.

Construction technique has a lot to do with it also. Solder flux left on a board can cause a corrosion track between printed runs, I have seen that one many times over.

Funny to fix a high-end avionic device with nothing but an acid brush and isopropyl alky, but I have done that many times.

Heat can also induce brittleness in multi-stranded conductors. The only thing you then need is vibration, and the strands part...

I cite specifically the A-8 card(s) in APX/72 transponder heads. Single stranded wire could save a lot of those.

 

This thread is headed down the wayward path of no return... I'll try to bring it back. Electronics die when the kid pours water into the back of a TV.

 

Actually when an electronic device dies, it dies in the same manner as a mechanical device - the most expensive part is the one that will fail....unless of course there is an inexpensive part that is almost impossible to get to, whereup it will fail, and the most expensive part won't fail until the inexpensive one has been replaced. Murphy said so.

 

I have a stereo that is about 40 to 50 years old I use it all the time. I have had to replace a few caps tubes and a few resisters. But every part in this thing is rated 3 to 4 times or more than voltage or wattage that is going throu those parts. Like it has 2 caps rated at 2500 volts but only 375 goes throu it. I think if some thing is built correctly it will last a life time. Today every thing is how cheap it can be made so they can sell it for less.

 

The failure rate on semiconductors (ICs, transistors, diodes, etc.) follow a bathtub type of failure rate curve (failure rate vs time). There is a high failure rate early in the life of the device (called Infant Mortality), and then
the failure rate increases again at "end of life", due to wearout mechanisms.
Most of these reliability failures are due to random photolithography process defects, that can't be tested for. MIL specs require burnin, or HTOL (High Temperature Operating Life Test) that operated the device at elevated voltage, and temperature, to try and identify the devices that would fall in the Infant Mortality portion of the failure rate curve.

Just some examples of common process defects that will cause reliability failures:
1) Gate oxide defects: These are defects that causes the gate oxide to break down, causing a gate short. The gate oxide can also have "trap regions" that hot carriers fall into, shifting the Vt of the transistor.

2) Metalization & via defects: These type of defects cause a high current density in the wiring on the chip that results in electromigration, which means the wire opens up.

Somebody mentioned .65nm technologies. This dimension is the minimum polysilicon design width, which defines the channel length of the PFETs &
NFETs in a CMOS technology. Probably a more important dimension is the
gate oxide thickness. This thickness is in the 5-10 angstrom range with the
latest CMOS technologies. With the gate oxide this thin, it is no longer an
insulator. There will be "tunnel currents" that will flow in oxide this thin.

Beside defects, EOS (Electrical Over Stress) is another cause of semiconductor failures. EOS can be in the form of ESD (Electro Static Discharge) or high current latchup with CMOS technologies. These normally happen in the card build & test processes. The EOS in the card processes
can cause a "Walking Wounded" scenario with semiconductor devices.

When semiconductor processes are developed, reliability is a major concern.
When these processes are qualified, extensive life test work is done to monitor hard failures, and also monitor any change in electrical parameters.
My first patent back in the late 70's (US patent 4113512) was for a novel bipolar transistor structure that eliminated a degradation problem.

Ok, enough for the semiconductor failures, now the card failures. The biggest
failure rate item in cards are the solder joints. Problems with the solder joints can also be a result of process defects, causing bad solder joints. Current and thermal cycling (going from cold to hot, and hot to cold) can create opens in these defective solder joints. Automotive applications have this thermal cycling!!!

 

You left out one on the card failures ---- copper migration....(he he)

 

You do bring up a good point!!!! One of the fail modes for the new "lead free"
technologies is conductive whisker growth, causing shorts.

 

This should be 65nM, or 0.065uM

 

Or 0.000002441 inches, (if you really want to split hairs). Of course, a hair is pretty huge at about 0.004 inches. How could a little stray static affect that?

 

Nano, nano, nano.

Some of my electronic devices die a sudden death by means of a clenched fist, or the 5th century B.C. age old sport of the discus throw.

Usually I just recycle the larger dead components, though.

Todays components/merchandise is pretty much limited use, throw away crap.

 

Actually I collect "dead electronics" from other family members & friends for
"recycling"....open 'em up and see what goodie parts I can find. When your "experimenting" tastes run amok, any thing helps!