What is the lifetime of common parts/circuits these days ?

[CatalinaWOW]

ATmega's flash memory:
- 20 years at 85 °C
- 100 years at 25 °C

This is both good news and bad news.  Good chances for 100 years, though it is based on extrapolation of known failure modes.   Bad news is that there are undoubtedly unknown unknowns and it is still a long way from 1000 years. 

I have had relatively good luck with fifty year old stuff working, but often requires clean up of switches and pots, and sometimes electrolytic capacitors.  Luck has been much worse for gear 75 years and older.

[janoc]

Luck has been much worse for gear 75 years and older.

Given the rather primitive components of the era (paper capacitors, etc.) why is that surprising? That's not exactly indicative or even relevant to the debate how long will stuff manufactured today with today's technologies last. 

[coppice]

ATmega's flash memory:
- 20 years at 85 °C
- 100 years at 25 °C

Those are common figures, although many parts try to keep that 100 year figure at higher temperatures. What is doesn't tell you is a quirk of flash. Some small percentage, maybe 0.01%, of devices will see a bit or two drain within months, due to crystal flaws.

[radiolistener]

You forgot 1 to 2 zeros...

This may be true for equipment manufactured 10-20 years ago, but not for modern one...

[MathWizard]

The oldest stuff I own are vacuum tube radio's and an oscilloscope. So the old carbon resistor's were drifting, and of course I replaced all the electrolytics, but yeah I should measure some of the ceramic's. I don't much trust the insulation in older transformers tho, I've had one go bad.

I've seen a lot urban explorer video's, where they go to places full of old electrical equipment, some of it still powered and working.

And as someone that's played video games like the Fallout series, in a post-apocalyptic nuclear wasteland, I'd want to go around trying to restore everything.

[radiolistener]

in a post-apocalyptic nuclear wasteland, I'd want to go around trying to restore everything.

These who get high radiation dose and have proper medical treatment can live about 1-3 month.
I'm afraid if you survive there will no be medical treatment, no clean food, no clean water, no electricity, etc.

[CatalinaWOW]

Luck has been much worse for gear 75 years and older.

Given the rather primitive components of the era (paper capacitors, etc.) why is that surprising? That's not exactly indicative or even relevant to the debate how long will stuff manufactured today with today's technologies last. 

Yes we have better components today.  Sometimes.  Are you aware of the Badcaps website?  But even using the better components of today, I am not so confident of very long life.  I doubt if the designers of the twenties to fifties thought of the paper capacitors as life limiting elements.  It wasn't mentioned as I was getting started in the 1960s.  Electrolytics and vacuum tubes were the suspects, and they weren't wrong.

And remember, the original question is thousands of years.  I don't know how to maintain a protected environment that long.  A deep mine in a seismically stable zone can keep thermal excursions to a minimum, but nothing I know can keep oxygen and water out over that time frame.  I worked in aerospace and it was difficult to seal things well enough to assure non corrosive environment over twenty years, a factor of fifty two short.

Does your precious have moving parts?  Hopefully it only uses graphite as a lubricant because oils and greases evaporate.  Slowly, but surely.   

The list of difficulties is long, and since we don't usually attempt to operate over such timescales not well studied. Of course most of the work is theoretical.  The only real data we have is from simple materials and simple mechanisms preserved by the primitive methods of Egyptians and others with the limited successes coming with the aid of favorable natural environments.

[Phil1977]

You forgot 1 to 2 zeros...

This may be true for equipment manufactured 10-20 years ago, but not for modern one...

Engineering has not changed a lot over the last decades.

[Phil1977]

in a post-apocalyptic nuclear wasteland, I'd want to go around trying to restore everything.

These who get high radiation dose and have proper medical treatment can live about 1-3 month.
I'm afraid if you survive there will no be medical treatment, no clean food, no clean water, no electricity, etc.

If you want to have a realistic idea about a post-nuke society then look "Threads", not "Fallout". (Though later is much more fun.)

And to come back on-topic, yeah, people should also try that civilization survives the next 1000 years. Who else should finally answer this thread???

[janoc]

Yes we have better components today.  Sometimes.  Are you aware of the Badcaps website?  But even using the better components of today, I am not so confident of very long life.  I doubt if the designers of the twenties to fifties thought of the paper capacitors as life limiting elements.  It wasn't mentioned as I was getting started in the 1960s.  Electrolytics and vacuum tubes were the suspects, and they weren't wrong.

Yes and? How is that indicative of anything made today, from different materials, using different processes, with different constraints, operating under different conditions?

Even the electrolytic caps don't have the same failure modes - the 50+ year old ones if they are bad (and many are just fine!) are mostly just dried out. In many cases they work fine after reforming, because they were under minimal stress and non-criticial in those circuits anyway. When modern ones fail it is in 90% of cases bulging/leaking from high ESR caused heating because they operate in switchmode supplies with high ripple. Something pretty much unheard of 50+ years ago.

So you are still extrapolating from data that is possibly completely irrelevant to the construction and use of your components today.  That's a bit like people who go and destroy even otherwise perfectly good hardware by "recapping it" because that's the only thing they know - it is always the capacitors, right? Never mind that the caps they are replacing are e.g. polymer or tantalum capacitors that don't have at all the same failure modes and are not even relevant to the fault they are seeing ...

That's not to say that the modern components or devices will fare better over the span of 50,100 or even 1000 years. Certainly not, most of them, esp. consumer gear, are designed to barely last through the warranty, which was not the case with the massively built equipment from the 50s and 60s. Where often the tubes were the thing that usually failed first due to the limited lifetime of them but the hardware itself was designed to last decades - like furniture (which it often was - e.g. old radios or TV sets).

So the correct answer is - we literally don't know and we will only discover how well we did once we arrive there.

Stuff like dried out lubricants are an easy fix compared to failed semiconductors that can't be sourced anymore.

[Phil1977]

Sometimes it´s even worse.

I´ve heard the rumor (can't proof it with citations) that the tanks of the Saturn-rockets for the Apollo-missions have been made of a certain high tensile stress steel foil that is not in production any more. And even if you would find a company that "cooks" the steel after the old recipe and manages to deliver an in-spec steel foil, then there would be no single company any more that knows how to handle this type of steel in the right manner.

What does it mean: Each era of human history has it´s technologies that they master at this time. Preserving an "active" technology for future use will always mean you have to spend resources on that and you have to preserve the knowledge of the workers by letting them continue what they are trained on. Else you just loose this technology.

We can't build the pyramids as the old Egyptians have done today, (maybe) we can't just build a new Saturn 5 today but we have falcon 9, 3nm-nodes and flatscreens everywhere. It´s just evolution.

[SteveThackery]

Most of modern electronics is designed to failure within 1 year. If it works longer, then just feel lucky.

I think this is just nonsense. It's extremely difficult to design something to fail after a certain time. Obviously you can design something to fail - that bit is easy, just under-specify capacitor voltage ratings, resistor power ratings, or over-volt your chips. But to design something that works for around a year and then fails?

Well, how would you do it? Electronic components aren't aware of the passage of time. You can't tell a resistor "work for 12 months, then burn out". Obviously you could install a software bomb if you are using a microcontroller, but that would be better described as sabotage rather than failure.

So you tell me - how do you design an electronic widget that will fail after a year? And how would it differ from one you've designed to last, say, three years? How will your widget know when to fail?

Honestly, it's easier just to design something that works, and whose life is limited by the intrinsic failure rate of the components. Yes, you can deliberately push the safe operating areas and expect to lower the MTBF, but how far would you push them to get a failure after 12 months? That's a proper question I'm asking you to answer.

The lack of ageing mechanisms in modern components makes such a task a major challenge.

[vk6zgo]

I was watching a review of a bad/good movie 'Battlefield Earth', and in it they find some +1,000 year old fighter jets that still work perfect after sitting for that long.

Take for example, some BJT RC oscillator I soldered up, with cheap metal film resistors, cheap 50V  ceramic caps, and some TO-92 BJT's.

So how long might something like that last, if it was only powered up to test it, and then it just sits in climate controlled roomtemp box. Maybe it should be Lead-lined too.

Maybe some of the Lead/Tin solder would be the 1st thing to change, with tin whiskers. But what about common unpowered parts, like above ? What about common 8-pin DIP op-amps ?

IDK how long it takes for plastics or cermaics to change/breakdown. And molecules from the air can make their way into the materials too, and vise-versa. And what would happen to the metal film over the centuries ??

I would like to go 1,000yrs into the future and see what becomes of my little proto-board circuits.


I know there's old salt mines that have been converted to storage vaults for various things like documents, movie films, etc. But does anyone know if there's any amount of tech stored in these places ? I know there's tech/computer museum's, but yeah I wonder how much is in some secure vault, meant to last as long as possible.

I dont think that you would ever get to worry about the electronics.
1000 year old jet fuel would have deteriorated to unusability.

[tszaboo]

I think this is just nonsense. It's extremely difficult to design something to fail after a certain time.

While the 1 year is a hyperbole, I kind-of agree with the general statement. Why do you think Volkswagen has tens of thousands of engineers? It's not to be able to make cars that don't break down, we figured that out in the 1980s. It's so your turbocharger breaks one day out of warranty.

Now things might seem to change with subscriptions, to become worse. Your things will break down the moment it cannot find the mothership or you stop paying.

[Phil1977]

I think this is just nonsense. It's extremely difficult to design something to fail after a certain time.

While the 1 year is a hyperbole, I kind-of agree with the general statement. Why do you think Volkswagen has tens of thousands of engineers? It's not to be able to make cars that don't break down, we figured that out in the 1980s. It's so your turbocharger breaks one day out of warranty.

That´s just nonsense.

True is that thousands of engineers try to make production of the car cheaper. For that reason they may reduce the quality of parts that are too good for their requirements - no one pays the supplier for exceeding what's wanted.

Anyhow the lifetime targets are usually in the range of 10years. And there are good economic reasons to do this, because if you set the lifetime target of all car components to 10years, then you anyhow risk having some things fail during warranty or while the leasing contracts are running.

Of course, you could set the quality standards in a way that the car lasts 25 years and a million km. But nearly no single customer would like to pay this.

[MrAl]

Back in the 1980's MTBF was considered to be longer with less connections.  In other words, something with more connections would be assumed to fail sooner than something with less connections.  That's external connections though, external to the IC chip or transistor.  So a chip with 16 pins would be assumed to fail sooner than a chip with 8 pins.

The thing that worries me most though is not the longevity as much as the designs themselves.  I have seen a lot of poor designs over the past 20 years.  They just suck really bad, and the assumed workings do not play out.
Microwave ovens for example.  I have not had a single 'inverter' microwave oven work the way it should, or the way you think it should work.  For example, both my Panasonic and my Toshiba do not use the "inverter technology" to cook below power level 3 (with power levels 1 through 10).  That means that power levels 1 and 2 do the PULSE on and off thing to get to a lower power, exactly what a non inverter oven does for all power levels.  That's a big problem even though it may sound not so bad.  The problem is that while it is pulsing it STILL burns the crap out of the food.  Heating a slice of pizza for example the Toshiba on power level 1 (lowest power) will apply 700 watts of cooking power to one little slice of pizza on and off.  It's 'on' for 20 seconds and 'off' for 40 seconds.  The problem is that during that 20 seconds it still hits the slice with 700 watts of cooking power which still burns the pizza.  It has to be a more gentle cooking power like 350 watts maybe even lower.  There's no way in hell to get that oven to power down to anything under 700 watts pulsed.  The Panasonic did a little better with a power of 500 min, and pulsing for less time such as 10 seconds on and 20 seconds off.
The power settings are very bad and the time settings are a nightmare as well.  You only get to set the time in the increments that they provide you cannot choose any time.  We are talking here about $250 to $300 price point ovens too not the run of the mill $100 unit.  But even the $30 to $100 microwave ovens will pulse for less time than the Toshiba, which is nuts, and give you full control of the time (such as if you want to set it for 2:01 instead of 2:00 minutes).
I think there is an Oster microwave oven that does go all the way down to level 1 with the inverter but that's the only one I've seen so far.

Who the hell is designing these ovens that's what I'd like to know.  They obviously do not know what it takes to cook and heat items in a microwave oven.  I suppose they got their college degrees at ShopRite.

[tszaboo]

I think this is just nonsense. It's extremely difficult to design something to fail after a certain time.

While the 1 year is a hyperbole, I kind-of agree with the general statement. Why do you think Volkswagen has tens of thousands of engineers? It's not to be able to make cars that don't break down, we figured that out in the 1980s. It's so your turbocharger breaks one day out of warranty.

That´s just nonsense.

True is that thousands of engineers try to make production of the car cheaper. For that reason they may reduce the quality of parts that are too good for their requirements - no one pays the supplier for exceeding what's wanted.

Anyhow the lifetime targets are usually in the range of 10years. And there are good economic reasons to do this, because if you set the lifetime target of all car components to 10years, then you anyhow risk having some things fail during warranty or while the leasing contracts are running.

Of course, you could set the quality standards in a way that the car lasts 25 years and a million km. But nearly no single customer would like to pay this.

Here is some statistics for a VW Jetta:
https://caredge.com/volkswagen/golf-gti/maintenance#more-info

The sales price (in the US, I don't like European prices as there is soo much tax on them) is 22000 USD.
Consider that VW will make somewhere around 10% or 2000 USD on the sales. I've added together the maintenance costs for 12 years, that came up to 8000 USD. Profit on repairs is generally considered 50% or more, so VW makes twice as  much on repairs than the sale of the car. Over 12 years.
We are not talking about companies from England, or Pegueot, that doesn't know how to build reliable cars. But if you really want to see something, check out the BMWs on that website. And ask yourself, how can Lexus give 10 year warranty.

[Phil1977]

Here is some statistics for a VW Jetta:
https://caredge.com/volkswagen/golf-gti/maintenance#more-info

The sales price (in the US, I don't like European prices as there is soo much tax on them) is 22000 USD.
Consider that VW will make somewhere around 10% or 2000 USD on the sales. I've added together the maintenance costs for 12 years, that came up to 8000 USD. Profit on repairs is generally considered 50% or more, so VW makes twice as  much on repairs than the sale of the car. Over 12 years.
We are not talking about companies from England, or Pegueot, that doesn't know how to build reliable cars. But if you really want to see something, check out the BMWs on that website. And ask yourself, how can Lexus give 10 year warranty.

This site states that 4% of the TCO of a golf GTI are maintenance. And that´s exactly the reason why no customer would like to pay 20% more for a more reliable car that probably is also heavier and less powerful.

[SteveThackery]

Anyhow the lifetime targets are usually in the range of 10years. And there are good economic reasons to do this, because if you set the lifetime target of all car components to 10years, then you anyhow risk having some things fail during warranty or while the leasing contracts are running.

Indeed. The thing is, cars are a poor example in terms of the OP's question, which was about the lifespan of electronic components and devices.  Cars are different because they have loads of mechanical parts, and those have loads of well-understood ageing mechanisms.

I had an acquaintance who worked for Siemens in Canada. They make lots and lots of components and modules for cars. He explained to me that the generally expected service life of a car is around 4000 hours, which sounds scarily short (less than 6 months). But in reality it means 120,000 to 160,000 miles when the vehicles overall average speed is 30 to 40 miles per hour.  Most of the components are specced to meet that lifespan. He gave me an interesting example of a BMW component he was working on - a radiator fan (I can't remember the model of car). The specified service life that BMW wanted was 1000 hours. Now this does not mean the car will need four radiator fans in its life; rather, it means that BMW expected the fan to run about 25% of the time the car is in use. At the same time he was developing a fan for Ford, who wanted 1500 hours. He said the Ford most likely has a lower bonnet profile and shallower grill, requiring the fan to run more often.

Unlike electronic components, mechanical ones are very predictable in terms of ageing. Friction, wear and tribology are well understood. The lifespan of the brushes used in a DC motor can be predicted with great accuracy. Ball and roller bearings have charts published by the manufacturer showing how the bearings ages with load and usage pattern. The piston rings, camshafts and followers use exceptionally well-controlled materials and manufacturing processes in order to meet the designed durability. Toothed cambelts are so familiar by now that their life can be predicted with confidence, although they are slightly different as they are considered to be a consumable. Plain bearings, like rolling element bearings, are extremely well understood and easy to design for a particular lifespan.

The only really serious cock-ups that have occurred in recent decades arose from the innovative use of new materials. Everything from cam chain tensioners and thermostat housings to complete, variable-length inlet manifolds were re-engineered in plastic, and it is true to say that many unanticipated failures occurred. Jaguar got burned in the late 1990s when they introduced plastic cam chain guides and tensioners. At the next revision they went back to using aluminium.

It is important to point out that the 4000 hour lifespan of a car does not mean it's scrap after that time. Rather, it means that the overall failure rate will go up as the various components come the the end of their design life. It is often feasible and cost effective to keep the car going for significantly longer.

So yes: designing mechanical devices like cars, aircraft, even spaceships for a particular service life is possible, even routine. But electronic components have so few ageing mechanisms (but not none) that designing them to fail after a given amount of service is extremely difficult, and for most purposes impractical.

[CatalinaWOW]

I in no way meant to imply that performance of 75 year old gear was useful in predicting life of current gear.  More intended to show our lack of knowledge.  Both ways.  I doubt anyone 75 years ago seriously thought the gear would still be working.  But it often is.  And the weaknesses that were known then didn't include all the failure modes that have turned up. 

I do think it is naive to say that current components have no aging mechanisms.  No known aging mechanisms, maybe.  Or at least none in a sufficiently benign environment.  Anyone who has ever operated equipment on as equatorial oceanic Beach knows there are aging mechanisms.  Which just returns us to the problem of maintaining a sufficiently benign environment over 1000 years.  And the problems we know about their are very, very difficult and there is much we don't know

[coppice]

Anyhow the lifetime targets are usually in the range of 10years. And there are good economic reasons to do this, because if you set the lifetime target of all car components to 10years, then you anyhow risk having some things fail during warranty or while the leasing contracts are running.

Indeed. The thing is, cars are a poor example in terms of the OP's question, which was about the lifespan of electronic components and devices.  Cars are different because they have loads of mechanical parts, and those have loads of well-understood ageing mechanisms.

I think cars are a pretty good example. The infotainment in modern cars is so central to key operations, like heating and cooling, that you can't really use the car without it. They generally won't repair them, and they cost thousands of dollars to replace. Our neighbour just had an electronics failure in a 6 year old VW Polo. The quote to fix was 2500 pounds. They looked around at third parties, who said they couldn't get parts to fix it, and would also need to charge 2500 for a complete replacement. That's quite a hit in a car that is only worth about 4 times that with a working infotainment system. If it failed a bit further down the road it would have been a write off issue. This would not be a big deal if these failures were rare. However, if you look at the annual statistics some of the third party warranty companies publish, these things are not just expensive, they fail quite a lot. With things like emergency braking becoming mandated, you won't even be legally able to drive a car with some of the electronics not functioning properly. I think this is just the kind of development the car makers must love.

[SteveThackery]

I think cars are a pretty good example. The infotainment in modern cars is so central to key operations, like heating and cooling, that you can't really use the car without it. .

I believe you have misunderstood my point. Of course the electronics in cars behave like electronics everywhere. That is to say, the ageing mechanisms are few and far between. Very few. Electronics can still fail, despite having hardly any ageing mechanisms. Perhaps you are confusing ageing with intrinsic failure rates. 

But a car is much more than electronics. It is a complex machine with a lot of mechanical parts. And the reason a car is a bad analogy is because all these mechanical parts do have numerous ageing mechanisms, all of which are well understood. It is easy to design in a specific service life to these mechanical components, and it is easy to predict when they will wear out.  Neither of these apply to electronic components.

So yes, you can compare car electronics to any other electronics and say they are analogous. But you cannot usefully compare the mechanical elements of a car with electronics in general, because they have almost nothing analogous.

[SteveThackery]

I do think it is naive to say that current components have no aging mechanisms. 

Trust me, it isn't naive. People like me spend a whole career studying reliability and failure modes at all levels in a system, from individual semiconductor junctions to vastly complex systems like a spacecraft or an international telecommunications network.

No known aging mechanisms, maybe.  Or at least none in a sufficiently benign environment.  Anyone who has ever operated equipment on as equatorial oceanic Beach knows there are aging mechanisms.  Which just returns us to the problem of maintaining a sufficiently benign environment over 1000 years.  And the problems we know about their are very, very difficult and there is much we don't know

Well, I guess I could live with that. But maybe you didn’t read my earlier post fully. I did explain in detail that nowadays the operating environment totally swamps any intrinsic ageing effects. Environmental stress is not the same as ageing. For a great example of environmental stress, drop your electronic device into a bowl of salty water. It will almost certainly fail, and quickly. But this is damage, not ageing. Same as putting it in an oven (which is what an equatorial beach might resemble) - it is very likely that temperatures beyond those expected by the designers will push components out of their safe operating areas. But again, this is damage, not ageing. Be assured that it is possible to design electronics to operate at very high (or low) temperatures for a long time.

The OP was asking about how long electronic components and systems will last. I responded by saying that many of the ageing mechanisms common in older devices are understood and largely eliminated in modern devices. I then explained that electronic devices will still fail - they all have an intrinsic failure rate - and environmental stress can greatly increase that failure rate. Indeed, the operating environment is by far the most important factor. But the time-related increases in failure rates - that is to say, ageing - are rare. Not non-existent (LEDs age, for example), but rare. Very rare. I hope I've expressed that more clearly now. 

[Messtechniker]

I expect electronics including my own to work without problems for 10 years or so.
After that and within the next 10 years I expect some problems
which can be repaired. Beyond 20 years I don't expect anything.

[SteveThackery]

Oh, and MTBFs and MTTFs don't address the lifespan of a device. They address the intrinsic failure rate.

Suppose a device has a 10 year MTBF. What do you think is the probability of it lasting 10 years before it fails? If you don’t know the answer, you will be surprised!