Solid-state Devices And Use

[mboverload]

I have always wondered how CPUs "wear out" and if they do. Or computer parts for that matter.

I know that flash drives "wear out" but I don't know what that means.

As long as things are kept cool, power is clean, and mechnical parts are replaced, shouldn't they last forever? What am I missing?

[Aluvus]

Well, I suppose any transistor is statistically more likely to nuke itself the longer it is in operation. And transistors are the guts of most modern electronics. But many devices can continue operating for extremely long periods of time provided a suitable environment.

Flash memory is a special case, though.

Flash memory is a form of EEPROM (Electrically Erasable Programmable Read Only Memory). This means it holds its value until told otherwise, which can be accomplished with an electric signal input.

Like other EEPROMs, flash is made up of a 2-dimensional matrix of floating-gate nMOS transistors. In a nutshell, these transistors are "programmed" as "on" or "off" as desired to store data. but the process of programming them is fatiguing, and every time it is repeated the transistor(s) wear down a bit more. Over time, this causes them to eventually die and become unusable.

More here.

Most other electronics don't subject their transistors to such extremes, and so don't kill them as quickly.

There are other things to remember, of course. Capacitors, being electrochemical devices, are not good for an infinite number of charge/discharge cycles (though they can be used for quite a few). Inductors in some devices can be significantly imparied by physical damage (sometimes very slight physical damage), particularly in old-style radios. Connections can corrode over time, or weak solders can break. That sort of thing.

And some devices, naturally, can be greatly disturbed by electromagnetic interference.

Nothing is forever. But some things can keep chugging for very long periods of time.

[JDoors]

...

As long as things are kept cool, power is clean, and mechnical parts are replaced, shouldn't they last forever?  What am I missing?

<{POST_SNAPBACK}>

Cool as compared to what? Clean power? I guess if you manage to find a way to keep parts at their optimum temperature (in spite of the close proximity of parts that require different ideal operating temperatures) and provided perfect power (even if you feed the component perfect power, is the built-in power supply going to be perfect too?) it would extend the life of the components, but just how much time, money and effort would be needed?

There's a cost/benefit ratio in there somewhere and it is taken into consideration during the design and development phase. Since most electronic components are obsolete before they reach the end of their lifespan, what would be the point of extending that lifespan even further? My Atari 2600 still works but ... who cares? You could say it was over -built.

NASA does design components for longer lifespans than most consumer products are required to achieve, and they pay for it, but consumer products are subject to market pressures that NASA can (or used to) ignore.

So yes, electronics can be built to last longer if not forever, but no one can be expected to take out a second mortgage to purchase an i-Pod.

[Madbob9]

I have always wondered how CPUs "wear out" and if they do.  Or computer parts for that matter.

I know that flash drives "wear out" but I don't know what that means.

As long as things are kept cool, power is clean, and mechnical parts are replaced, shouldn't they last forever?  What am I missing?

<{POST_SNAPBACK}>

Electron migration or voltage leakage.Simply put,the transisters are slowly eroded like water will erode a rock in a stream.

[Pete_C]

I have always wondered how CPUs "wear out" and if they do.  Or computer parts for that matter.

I know that flash drives "wear out" but I don't know what that means.

As long as things are kept cool, power is clean, and mechnical parts are replaced, shouldn't they last forever?  What am I missing?

<{POST_SNAPBACK}>

The biggest problem with flash drives is wear and tear on the contact surfaces as you move them from one device (camera) to another (reader) over and over. Compact Flash are more durable than most, but SD and the like are pretty close. Some like smart media cards are very vulnerable.

Now for how do electronic devices which have no moving parts wear out?

As thing heat and cool, they expand and contract.

Different materials expand and contract at different rates.

Different materials have different flexibilities.

Thus overtime wherever two different materials meet, one , the other , or both or at least the interface will crack.

Now for a long time the greatest temperature difference was between being powered on and being powered off. So it was recommended to leave them on all the time. With modern component cooling and heat dissipation, this is less of a difference than powered on under relativley little load, and powered on under full load (gaming) so that arguement does not hold up any more.

Also, transformers, when first powered up can cause a surge . Since modern power supplies are switching, not switched (meaning that they remain powered up and just connect the appropriate feed lines when told to , using solenoids ) this too is not a real major issue.

Also, consider that heat means excited molecules and atoms. Thus you get "evaporation" and sublimation as well as increased reaction (oxidation) which alter the component structure. If the electronic component requires a certain amount of a particular metal in its metalic form to perform the function, once a certain amount of it oxidizes or migrates out of the location , it fails.

[mboverload]

Wow thanks everyone!

It is really true that all the great minds have come to BestTechie!

[uberpenguin]

Electron migration or voltage leakage.Simply put,the transisters are slowly eroded like water will erode a rock in a stream.

<{POST_SNAPBACK}>

Electron migration has only become significant fairly recently. Perhaps counter-intuitively, older MOS and even TTL circuits aren't really going to be affected by electron migration. I'm not sure what you mean by 'voltage leakage...' If you're referring to subthreshold voltage leakage I still don't see what you're getting at.

I've personally used digital equipment that was manufactured in the mid-70s that still works just fine. Generally in computers that old, the things that go first are power supply components like the big electrolytic caps. Sometimes some of the TTL components like a flip-flop will die, but even that is somewhat seldom if the computer has been kept in good conditions.

So if the subtext of your question is 'will my CPU die of natural causes within 10 years,' then the answer is probably no.

-uberpenguin

Edited August 26, 2005 by uberpenguin