What is expected lifetime of IC's

[+bob1200xl]

A Seagate Cheetah drive is certainly full of new silicon and they claim a MTBF of 180 years. After 100 years, about half the drives will fail. (and, this is a majorly mechanical device!) Are they mistaken, or deceptive?

 

I want to get every one of my 10,000 write cycles out of my EEPROMs, so I'd better get busy...

 

Bob

 

 

I'd guess that the smaller the process size, the less time it'd last.

 

Our old stuff is absolutely gigantic compared to modern-day stuff, hence the 25+ years life.

Yes, and most newer stuff tends to run much hotter too (high clock speeds), there's only a handful of console/home computer ICs from the 70s and early/mid 80s that used heat spreaders, let alone dedicated fans. (totally passive ventilation in many cases at that -at least if the PSU was external)

 

We had a similar discussion on Sega-16 a while back, and a few others on AA before too.

 

 

Capacitors are the weakest link in old electronics, a few other discrete components (voltage regs, sometimes transistors, diodes, usually not resistors) tend to be more problematic than the LSI chips, but all ICs are perishable. (the silicon lasts forever, but the dopants and such that make up the logic and interconnect -I think more so the interconnect- will degrade over time)

 

Even with very old chips that get no or practically no use, they're probably not going to last hundreds of years in the best cases. (aside from painstaking cryogenic preservation or something )

The smaller the IC process, the sooner the degradation. (take modern high-quality technology and apply it back to single layer 2-5 micron manufacturing and you'd have some really long lasting chips though )

 

I'm not positive on the 70s/80s tech, but it may very well be possible that a good percentage lasts close to a century.

Hell, electrolytic capacitors in such systems (or TVs or other equipment for that matter) are generally LONG past their expected expiration date, yet a very high percentage of systems are still working fine without recap. (especially those that weren't abused -but even then quite a few) Though part of that is probably that the systems are rather tolerant to degrading (namely drying) capacitors vs newer systems which are far more sensitive. (a fair amount of the early/mid 90s Sega systems have notable capacitor problems -especially the game gear)

The sheer size of some capacitors used on the earlier models probably plays a role too. (much greater volume to surface area ratio)

 

 

 

 

In terms of hardware that gets regular use and in as far as ICs (from CPUs to ROMs to ROM to discrete logic), normal wear and tear with various risks of voltage spikes, overheating, etc will all be far more likely causes of failure than gradual degradation of the chip's circuitry alone. (like how I destroyed the chips in my light sixer ~9 years ago by using reverse polarity current -unfortunately they blew -quite literally- along with one or 2 games before the voltage regulator gave out -the motherboard is still perfect otherwise after the reg was replaced -swapped in another systems's chips- but I've still yet to find a crappy VCS to complete both systems I have with another set of TIA+RIOT+6507 -the overload blew the package right off the top of the chip for one of them -TIA or RIOT, I forget which)

[sloopy]

Believe it or not... Keeping them clean helps them to last longer! Dirt building up on chips causes over-heating = shorter life.

Same for keeping the whole system clean: less dirt and dust= better ventilation. For that matter, removing the RF shielding (where not integral) can be beneficial as well.

 

*SNIP*

 

just dont remove the RF shield from a C64, it is actually a heat sink for the video (VIC) chip...

 

and dust is a big enemy... it shields convection cooling from working... so dirt and dust, they are evil...

 

sloopy.

[Orion Pax]

Don't forget humidity! I had my dads 800xl stored in a room and forgot about it! Then the hot water tank leaked hot water in that room! I was able to save it though! Rubbing alcohol and a soft bristled toothbrush are my friends!

[AtariLeaf]

IC London, IC France, IC. . . nevermind.

[supercat]

A Seagate Cheetah drive is certainly full of new silicon and they claim a MTBF of 180 years. After 100 years, about half the drives will fail. (and, this is a majorly mechanical device!) Are they mistaken, or deceptive?

 

MTBF is a measure of reliability, not lifetime. An MTBF of 65,000 days means that if one were to run 65,000 drives continuously, you'd get about one failure per day.

[MEtalGuy66]

The smaller the IC process, the sooner the degradation. (take modern high-quality technology and apply it back to single layer 2-5 micron manufacturing and you'd have some really long lasting chips though )

 

 

Yeah.. We could make a core2duo that's the size of a city block, runs at a clockspeed of 10mhz and needs a huge industrial heat exchanger facility to cool it..

[Chilly Willy]

A Seagate Cheetah drive is certainly full of new silicon and they claim a MTBF of 180 years. After 100 years, about half the drives will fail. (and, this is a majorly mechanical device!) Are they mistaken, or deceptive?

 

MTBF is a measure of reliability, not lifetime. An MTBF of 65,000 days means that if one were to run 65,000 drives continuously, you'd get about one failure per day.

 

Exactly, and a FIT of 100,000 years does NOT mean the chip will last 100,000 years. If that were the case, there would yet to be ANY failures, period.

 

And I do NOT believe there are any Atari computers anywhere that have been running a BBS 24/7 without fail for 20+ years. Sorry, not possible. I'm an engineer, not a magician. I deal with reality, not fiction.

[MEtalGuy66]

I know of one C=64 that ran a BBS 24/7 from 1987 through 2007.. and the only "failures" during that time were when the hardisk crapped out.. He had to replace it several times..

 

Seems to me you are the one who needs to separate fiction from reality.. What some professor told you (based on theory, probability, and his knowledge/experience) and what DOES happen in the real world in SPECIFIC INSTANCES are two EXTREMELY different things..

 

I saw in the paper one time, that they finally changed a lightbulb in a farmhouse that had been burning since the 1930s.. There was no switch in the circuit.. it was hard wired, and had been burning constantly.. Industry experts confirmed that the lightbulb was actually from the 30s.. They said it defied all known characteristics of the technology.. Probability was extremely low that the bulb would last that long.. But it did..

 

So to say that it DOES NOT HAPPEN... is a little rediculous.. To say that its very unlikely is a much more REALISTIC statement..

[svenski]

Wow, that light bulb story is incredible.

 

I think a lot of our classic Atari stuff will continue to outlive its more modern equivalents. I was reading somewhere today (might have been AA) that we can expect a lot more problems with our electronics equipment since the switch to lead free solder.

[Chilly Willy]

I know of one C=64 that ran a BBS 24/7 from 1987 through 2007.. and the only "failures" during that time were when the hardisk crapped out.. He had to replace it several times..

 

Seems to me you are the one who needs to separate fiction from reality.. What some professor told you (based on theory, probability, and his knowledge/experience) and what DOES happen in the real world in SPECIFIC INSTANCES are two EXTREMELY different things..

 

I saw in the paper one time, that they finally changed a lightbulb in a farmhouse that had been burning since the 1930s.. There was no switch in the circuit.. it was hard wired, and had been burning constantly.. Industry experts confirmed that the lightbulb was actually from the 30s.. They said it defied all known characteristics of the technology.. Probability was extremely low that the bulb would last that long.. But it did..

 

So to say that it DOES NOT HAPPEN... is a little rediculous.. To say that its very unlikely is a much more REALISTIC statement..

 

Sorry, I've been an engineer for going on 25 years, and it is MUCH more realistic to say it NEVER happens, and you have to supply extraordinary proof for any exceptions. You say there is a C64 that ran 24/7 for 20 years? I say PROVE IT! It's probably been swapped out a few times (probably when the "harddrive" failed - his euphemism for replacing the computer) and the story told is pure Grade-A bull.

 

I don't buy beach front property in Arizona, and I sure don't buy 20+ year-old computers running 24/7 without failures. I was born at night, but it sure wasn't last night. If you buy that C64 story, I've got a Bigfoot fur I want to sell you.

[Marius]

Well fox-1 130xe computer is running bbs thunderdome since 1995. Not 20 years yet but no issues so far I know.

 

Perhaps your teacher meant that those ic's had to be 20 years in full activity? Well a computer that is switched on has moments of activity and idle time. I guess in idle time they last longer in that theory... In fact most of the time a switched on computer is rather idle ...

[DamageX]

I was reading somewhere today (might have been AA) that we can expect a lot more problems with our electronics equipment since the switch to lead free solder.

That has already happened. Video chips were desoldering themselves left and right on video cards, laptops, xbox 360, etc.

[MEtalGuy66]

I know of one C=64 that ran a BBS 24/7 from 1987 through 2007.. and the only "failures" during that time were when the hardisk crapped out.. He had to replace it several times..

 

Seems to me you are the one who needs to separate fiction from reality.. What some professor told you (based on theory, probability, and his knowledge/experience) and what DOES happen in the real world in SPECIFIC INSTANCES are two EXTREMELY different things..

 

I saw in the paper one time, that they finally changed a lightbulb in a farmhouse that had been burning since the 1930s.. There was no switch in the circuit.. it was hard wired, and had been burning constantly.. Industry experts confirmed that the lightbulb was actually from the 30s.. They said it defied all known characteristics of the technology.. Probability was extremely low that the bulb would last that long.. But it did..

 

So to say that it DOES NOT HAPPEN... is a little rediculous.. To say that its very unlikely is a much more REALISTIC statement..

 

Sorry, I've been an engineer for going on 25 years, and it is MUCH more realistic to say it NEVER happens, and you have to supply extraordinary proof for any exceptions. You say there is a C64 that ran 24/7 for 20 years? I say PROVE IT! It's probably been swapped out a few times (probably when the "harddrive" failed - his euphemism for replacing the computer) and the story told is pure Grade-A bull.

 

I don't buy beach front property in Arizona, and I sure don't buy 20+ year-old computers running 24/7 without failures. I was born at night, but it sure wasn't last night. If you buy that C64 story, I've got a Bigfoot fur I want to sell you.

 

Ya know.. The problem with a WHOLE LOT of engineers (of which you are an excellent example) is that your head is so crammed full of theory, that when you get where the "rubber meets the road" you're unable to think outside of what you think you know as fact. This is why all of your designs have to go through numerous testing phases.. If what you say you KNOW was 100% applicable in real-world conditions, all of your designs would just work, exactly as you envisioned them and never need any testing, or revision..

 

I know the guy with the C=64 personally. I believe what he says.. Hes not an unreliable idiot. I'm not gonna ask him to prove anything to you. I've seen the machine, back then, and now. I also believe what Fox-1 says about his experience with ATARIs running BBSes.

 

I say, prove it failed. You have no proof. SO without proof, having never seen the object in question, you should not definitively state that it did fail.

 

And I'll tell you something else. If you do not realize that the WILL OF GOD is the most powerful force in the universe, then you are a REALLY piss poor engineer, and you don't qualify to even make coffee for the best/brightest scientific minds in the world..

 

DO you think it's impossible that GOD WILLED that particular C=64 to last as long as it did? If you admit this is possible, then do you have any reason to believe that he didn't?

 

If you think i'm actually making a religeous statement, your are wrong.. (My name is not Atariksi and I don't fix computers with philosophy) My point is that inexplicable shit happens everywhere, every day.. And just because it seems unlikely or improbable does not mean that it doesn't happen. There are INFINITELY MORE factors influencing anything and everything in the world/universe that surrounds us that we CANNOT explain than those we can. If you are unable to figure this basic premise/principle into your mode of thinking, then you're gonna be at a loss as far as dealing with all kinds of circumstances where life in general is concerned.. Nevermind Engineering.. Asking someone for proof of anything you deem improbable is gonna make you a really skeptical, if not neurotic individual.

Edited February 14, 2011 by MEtalGuy66

[kool kitty89]

Believe it or not... Keeping them clean helps them to last longer! Dirt building up on chips causes over-heating = shorter life.

Same for keeping the whole system clean: less dirt and dust= better ventilation. For that matter, removing the RF shielding (where not integral) can be beneficial as well.

 

*SNIP*

 

just dont remove the RF shield from a C64, it is actually a heat sink for the video (VIC) chip...

 

and dust is a big enemy... it shields convection cooling from working... so dirt and dust, they are evil...

 

sloopy.

Except the models with cardboard shielding that actually insulates the chips and worsens overheating. (for other cases, you could always take a small strip of aluminum -or copper or steel for that matter- and stick it to the VIC-II with a bit of thermal paste -and that's assuming the added convection cooling from no shielding isn't better than the poorly attached steel shield fudged as a *heat sink* -apolloboy's C64 has no shielding at all, but his was cardboard to start with anyway iirc)

[Marius]

The ONLY thing I wanted to know was I need more than 3 68b21P, R6551AP etc. chips as spare parts, to be safe for the next let's say 20 years.

 

Then I got a very nice mail from Fox-1 himself.

 

I asked him: how long do you think expected lifetime of an IC is.

 

His fabulous answer was:

 

Somewhere between 1.5 day and 5 centuries.

 

Yeah!

 

Greetz

M.

[+poobah]

 

Ya know.. The problem with a WHOLE LOT of engineers (of which you are an excellent example) is that your head is so crammed full of theory, that when you get where the "rubber meets the road" you're unable to think outside of what you think you know as fact. This is why all of your designs have to go through numerous testing phases.. If what you say you KNOW was 100% applicable in real-world conditions, all of your designs would just work, exactly as you envisioned them and never need any testing, or revision..

 

I work with "MATLAB engineers". *Everything* works in MATLAB.

It's that darn real-world that screws stuff up

[ClausB]

The ONLY thing I wanted to know was I need more than 3 68b21P, R6551AP etc. chips as spare parts, to be safe for the next let's say 20 years.

If you buy spare chips, then your originals will never fail. Some other part of the circuit will fail, however.

 

Am I a pessimist? No, just an optimist with experience.

[+bob1200xl]

Yes, what you both say is true - nobody thinks ICs will last 100,000 years, or, perhaps even 180 years. MTBF, MTTF, AFR are all reliability measurements. I don't have any 'lifetime' numbers - do you? They test these things by jacking up the temperature, voltage and current to see what fails in a short period of time. Real 'lifetime' numbers will take real time measurements. Are there any? Is anyone tracking the failure rates on LM741s over time? TI may have estimated 'lifetime' numbers for the 7400, but what is it turning out to be in reality? Saying that there is a bathtub curve (there is) doesn't determine our position on it.

 

Marius1976 wanted some idea as to the life expectancy of his Atari ICs. To put it in the timeframe of 2-20 years is not accurate. I know from personal experience that there are rooms (huge rooms) full of computers that run 24/7 for many years. I can put my hand on one tape controller that contains thousands of ICs and has been running for 30 years or more. There's a DEC system nearby that is at least 20 years old - same story.

 

So, in as much as I can walk into a room with at least a million ICs that have been running continuously for 10 to 30 years, and there are hundreds of such facilities in the US, my answer to Marius1976 is 'a good, long time' - not to worry.

 

Bob

 

one other thing... an Atari that has been running for 20+ years can fail and still be a valid example. The failure does not show EOL on the ICs unless the MTBF is ramping up significantly. As you said, failures are a measure of reliability not lifetime.

 

 

A Seagate Cheetah drive is certainly full of new silicon and they claim a MTBF of 180 years. After 100 years, about half the drives will fail. (and, this is a majorly mechanical device!) Are they mistaken, or deceptive?

 

MTBF is a measure of reliability, not lifetime. An MTBF of 65,000 days means that if one were to run 65,000 drives continuously, you'd get about one failure per day.

 

Exactly, and a FIT of 100,000 years does NOT mean the chip will last 100,000 years. If that were the case, there would yet to be ANY failures, period.

 

And I do NOT believe there are any Atari computers anywhere that have been running a BBS 24/7 without fail for 20+ years. Sorry, not possible. I'm an engineer, not a magician. I deal with reality, not fiction.

[atx4us]

Hi

 

Yes I know... I am a but neurotic person ... I am really addicted to all my Atari stuff and related items.

 

I'm collecting all kind of spare parts, to be sure in future of a long and happy Atari life.

 

Some components are harder to get, or a bit too expensive to buy giant amounts of them.

 

So I was wondering:

 

When there are no 'crazy' things like too much current, or short-circuit and that kind of problems. What is an expected lifetime for such IC's in regular use?

 

Is that 5 year? 10 year? 2 year?

 

I know my Atari's do last a long time... but most of them haven't been running for ages.

 

Thanks for any feedback!

 

Marius

 

p.s. the IC's I described are components of my blackbox.

 

 

This is an interesting question and is certainly one that I had pondered upon many times myself. My Atari 800 is still working - Got it back in 1984. I don't know how much longer it will last which is your question. I was VERY glad to find out about emulation. Even though it's not perfect, Atari800Win PLus 4.0 has been able to run just about all of my favorite programs that I had converted from floppy disks (which are not very reliable!) by using the excellent APE setup for Windows. So, I'm relieved to have a Plan B should my Atari 800 ever fail.

 

I'm also glad to see how the interest in an old 8-bit computer has lasted for so long and is still going strong! Before I found AtariAge, I thought that I was the only one left in the whole world who is holding on to the good 'ol Atari.

 

Now to my question, which of the ICs in the Atari 800 may be the most likely to fail first? (i.e. the DRAM ICs?)

[flashjazzcat]

Heh... nothing is impossible. At school I had a friend whose brand new XEGS failed about ten minutes after he took it out of the box. That's probably reasonably unlikely, but certainly not impossible. My first 65XE ran for a year then died (probably brick PSU related, I now realize), and the second one ran about an average of six hours a day, every day for a decade. Surely the power cycling on that thing (at least 3,652 powerups) would put more stress on the components than leaving it switched on all the time. As the revered local comic Bobby Thompson once said: "I'm nee engineer", but I don't find the story of an 8-bit computer with no moving parts running for twenty years too outlandish. My second 65XE probably ran for 30,000 hours, and I still have the motherboard here, still working.

 

 

 

[+Stephen]

I know of one C=64 that ran a BBS 24/7 from 1987 through 2007.. and the only "failures" during that time were when the hardisk crapped out.. He had to replace it several times..

 

Seems to me you are the one who needs to separate fiction from reality.. What some professor told you (based on theory, probability, and his knowledge/experience) and what DOES happen in the real world in SPECIFIC INSTANCES are two EXTREMELY different things..

 

I saw in the paper one time, that they finally changed a lightbulb in a farmhouse that had been burning since the 1930s.. There was no switch in the circuit.. it was hard wired, and had been burning constantly.. Industry experts confirmed that the lightbulb was actually from the 30s.. They said it defied all known characteristics of the technology.. Probability was extremely low that the bulb would last that long.. But it did..

 

So to say that it DOES NOT HAPPEN... is a little rediculous.. To say that its very unlikely is a much more REALISTIC statement..

 

Sorry, I've been an engineer for going on 25 years, and it is MUCH more realistic to say it NEVER happens, and you have to supply extraordinary proof for any exceptions. You say there is a C64 that ran 24/7 for 20 years? I say PROVE IT! It's probably been swapped out a few times (probably when the "harddrive" failed - his euphemism for replacing the computer) and the story told is pure Grade-A bull.

 

I don't buy beach front property in Arizona, and I sure don't buy 20+ year-old computers running 24/7 without failures. I was born at night, but it sure wasn't last night. If you buy that C64 story, I've got a Bigfoot fur I want to sell you.

Your loss if you don't think these old machines are capable of surviving more than 20 years. I don't know what kind of proof you want us to offer but I won't waste my time arguing on this forum any longer. You had one person tell you what you want to believe even though hundreds of examples of the contrary exist, suit yourself. I'll continue using my 30 year old equipment and marvel at how it is possible.

[Lynxman]

I think the ICs will make much more hours of work if they run nonstop compared with ICs with many Power on/off.

 

Of cause ICs are no light bulbs but do you remember in which moment the light bulbs die? Yes, most of the time if you switch on.

 

Every time someone told me that any computer etc was "suddenly" defect i asked "what have you done?" the answer is always "yesterday everything was fine and today i switch on and..."

 

Watch to the power supplies of computers, don´t care if old or new, if you switch on, there is always a very short time periode with overvoltage. Of cause, not seconds or ms, i think it is less than 1 ns but it is not good for the components.

Edited February 14, 2011 by Lynxman

[Chilly Willy]

Ya know.. The problem with a WHOLE LOT of engineers (of which you are an excellent example) is that your head is so crammed full of theory, that when you get where the "rubber meets the road" you're unable to think outside of what you think you know as fact. This is why all of your designs have to go through numerous testing phases.. If what you say you KNOW was 100% applicable in real-world conditions, all of your designs would just work, exactly as you envisioned them and never need any testing, or revision..

 

What I said is based on EXPERIENCE, not theory. I've seen enough "never been opened in 20 years" boxes that have parts with 2 year-old manufacturer's dates to take any claims of 20 years 24/7 usage with a grain of salt.

 

 

I know the guy with the C=64 personally. I believe what he says.. Hes not an unreliable idiot. I'm not gonna ask him to prove anything to you. I've seen the machine, back then, and now. I also believe what Fox-1 says about his experience with ATARIs running BBSes.

 

I say, prove it failed. You have no proof. SO without proof, having never seen the object in question, you should not definitively state that it did fail.

 

You don't have to be an unreliable idiot to tell your friends a Whopper. And you should know it's impossible to prove a negative, but easy to prove a positive - just take a photo of the mobo so that the manufacturer's date is clear.

 

 

And I'll tell you something else. If you do not realize that the WILL OF GOD is the most powerful force in the universe, then you are a REALLY piss poor engineer, and you don't qualify to even make coffee for the best/brightest scientific minds in the world..

 

DO you think it's impossible that GOD WILLED that particular C=64 to last as long as it did? If you admit this is possible, then do you have any reason to believe that he didn't?

 

Well, yes, if it's the Will of God that a computer lasts 20+ years running 24/7, it's certainly possible, but I'll still want photos showing those chips really are 20 years old and not a 2 year-old mobo someone stuck in the shell when no one was looking. In God we trust - all others pay in advance.

[AtariLeaf]

[+bob1200xl]

...and, just where would someone get a two year-old 'mobo'?

 

Do you have any instances where IC products are at EOL due to chip degradation in anything, anywhere? Mass failures of 2114s?

 

What?

 

Bob

 

 

 

Ya know.. The problem with a WHOLE LOT of engineers (of which you are an excellent example) is that your head is so crammed full of theory, that when you get where the "rubber meets the road" you're unable to think outside of what you think you know as fact. This is why all of your designs have to go through numerous testing phases.. If what you say you KNOW was 100% applicable in real-world conditions, all of your designs would just work, exactly as you envisioned them and never need any testing, or revision..

 

What I said is based on EXPERIENCE, not theory. I've seen enough "never been opened in 20 years" boxes that have parts with 2 year-old manufacturer's dates to take any claims of 20 years 24/7 usage with a grain of salt.

 

 

I know the guy with the C=64 personally. I believe what he says.. Hes not an unreliable idiot. I'm not gonna ask him to prove anything to you. I've seen the machine, back then, and now. I also believe what Fox-1 says about his experience with ATARIs running BBSes.

 

I say, prove it failed. You have no proof. SO without proof, having never seen the object in question, you should not definitively state that it did fail.

 

You don't have to be an unreliable idiot to tell your friends a Whopper. And you should know it's impossible to prove a negative, but easy to prove a positive - just take a photo of the mobo so that the manufacturer's date is clear.

 

 

And I'll tell you something else. If you do not realize that the WILL OF GOD is the most powerful force in the universe, then you are a REALLY piss poor engineer, and you don't qualify to even make coffee for the best/brightest scientific minds in the world..

 

DO you think it's impossible that GOD WILLED that particular C=64 to last as long as it did? If you admit this is possible, then do you have any reason to believe that he didn't?

 

Well, yes, if it's the Will of God that a computer lasts 20+ years running 24/7, it's certainly possible, but I'll still want photos showing those chips really are 20 years old and not a 2 year-old mobo someone stuck in the shell when no one was looking. In God we trust - all others pay in advance.