34
$\begingroup$

The idea is for some people in a post-apocalyptic future to find these bunkers/vaults and be able to turn on the computers and charge the cellphones. How many years could the equipment last if they were stored in very good conditions (no sun light, no humidity, all covered in plastic or inside boxes)? 10 years? 100 years?

There are already some related questions (this one specific about cars), but in most of them the objects where abandoned in the environment or left behind without being properly stored.

$\endgroup$
  • 3
    $\begingroup$ Can the devices be plugged in to wall power? As Separatrix mentioned, the batteries in these devices will most definitely be the first things to go. Even with a dead battery, most modern devices can still function when provided wall power (or bypassed rather easily to function with the appropriate DC voltage being provided otherwise). $\endgroup$ – ColonelPanic Oct 24 '18 at 10:29
  • 12
    $\begingroup$ For the cellphones, if you intend to use them as phones (as opposed to use them as miniature computers, e.g. using data that is stored on them), then the networks they are compatible with will probably be the first to go. $\endgroup$ – jcaron Oct 24 '18 at 12:47
  • 3
    $\begingroup$ One thing that the answers below haven't mentioned: the screens. LED and CRT displays will last forever if protected from dust. LCD screens will not, and I suspect, neither will the various OLEDs. I've seen LCD screens grow mould if not regularly aired. $\endgroup$ – nzaman Oct 24 '18 at 13:27
  • 1
    $\begingroup$ @Narusan Actually, I'd suspect that with user's, programmer's (not that there would be much difference between the two), and service manuals - assuming, obviously, that one is able to read them - something like the Z3 might actually be easier than a tablet to use for its originally intended purpose after many years in storage. Yes, it looks rather intimidating to the uninitiated, but so does a Boeing 787 cockpit (in which tons of complexity is actually hidden, e.g. in the flight management system). $\endgroup$ – a CVn Oct 25 '18 at 16:50
  • 3
    $\begingroup$ As for "There's no-one that remembers how to use that stuff.", don't forget that there's a large, widespread interest in Retrocomputing. Sure, that might not be primarily computers from the 1940s, but you'll find plenty of people working with the real thing from around 1980. That's the better part of 40 years ago. With manuals, those systems aren't harder to use now than they were back then. $\endgroup$ – a CVn Oct 25 '18 at 16:53
39
$\begingroup$

There are three really big problem components for long term storage of computer hardware, batteries, flash memory, and electrolytic capacitors. While there are other components that can fail in long-term storage, most of them can be kept working if the storage environment is ideal. These three, however, will pretty reliably fail after a reasonably short period of time independent of the environment they are kept in. In particular:

  • Batteries: Varies, but generally not very long. Most batteries have a shelf life of at best a few years when left unused. The CMOS RTC battery in most computers won't be much of an issue barring some potentially weird configurations that rely on firmware settings. The batteries in phones or laptops however are another story. For those, you'll be lucky if they work at all after about 5 years of being unused. From a safety perspective, you should also be replacing (or at least reconditioning) any batteries in such a device before attempting to use it, failure there tends to be catastrophic and very dangerous. In some cases, you might be able to get away with having a really exotic type of battery that can safely be stored for long periods of time (Silver-Air batteries come to mind, but they're expensive, not rechargeable, and still only last at most a few years). You might also make things better by storing the device with the battery fully discharged (most rechargeable batteries have a longer shelf-life if stored discharged), but that probably also won't extend things by much.
  • Flash storage: Theoretically indefinite if there's no hard radiation, but it won't retain data past a few years without special efforts taken to do so. Flash memory is inanely durable when not being used. About the only things that can reliably make it stop working are long term exposure to hard radiation, extreme thermal stress, extreme heat, or just plain physically destroying it. However, it's actually not really all that good for long term data storage. The reason for this relates to how flash memory works. In short, flash memory stores data by trapping an electrical charge on an otherwise electrically isolated bit of conductive material. Doing this requires pushing electrons across a layer of insulating material, which degrades the insulating material over time (this is why flash memory is write-limited), causing the electrical charge to slowly leak out. For flash memory seeing active usage, this isn't much of an issue as things will get rewritten before it becomes an issue. For flash memory in archival storage however, this puts an upper limit on how long your data will last. For good quality SLC NAND flash, this limit is estimated to be somewhere around 5 years. For the cheap MLC NAND flash used in most devices these days, it's usually only 2-3 years. There's not really any practical way around this except not using flash memory, but most forms of storage media do have some type of long-term degradation they have to deal with.
  • Electrolytic capacitors: At most 15 years. These are mostly used for power handling in computers and other devices. The issue with them is that they use a gel as one of the two electrodes, and if this gel dries out or leaks out, the capacitor will stop working (and if it leaks, it may damage other components when you try to power on the device). Even if kept in otherwise perfect conditions, the sealing material will deteriorate over time, which for current designs puts a functional upper limit of about 15 years on their life expectancy. You can obviously work around this by just not using electrolytic capacitors (and some phones don't for exactly this reason), but it's non-trivial to figure out if an arbitrary system uses them or not, and most of the alternatives come with their own issues.
$\endgroup$
  • 2
    $\begingroup$ You might also make things better by storing the device with the battery fully discharged — this is not precise. There is a lowest charge that allows safe recharge. 0% on your phone means this level + safety margin, not real 0 charge left. And batteries tend to lose charge on their own when unused, so more years = more safety margin you need. More safety margin = faster deterioration, and finding the sweet spot is tricky. $\endgroup$ – Mołot Oct 25 '18 at 11:40
  • $\begingroup$ @Mołot I suspect it also depends on the exact battery design and charging circuit. Most of design's I've seen actually recommend storing the battery discharged as far as the device will let you when you plan on storing it for more than a few months without usage. $\endgroup$ – Austin Hemmelgarn Oct 25 '18 at 12:41
  • $\begingroup$ "as the device will let you" - that's what I'm talking about. Devices have their safety margins in place. But try storing one for a year or two - high chance you'll go past safety and protection circuit will simply refuse to load it. $\endgroup$ – Mołot Oct 25 '18 at 12:52
  • 1
    $\begingroup$ @lupino if the devices were deliberately built to last, the story could be quite different. If you are willing to make significant sacrifices and accept high cost, both in R&D and materials - to begin with, optimize for durability, not performance. Do not use electrolytic capacitors. Store read-only data in ROM. Do not use batteries, instead use some other kind of power source as well as instructions on how to create a power source from scratch. Etc etc.Use simple LED character display and speakers instead of graphical displays. Etc etc. $\endgroup$ – Prof. Falken Oct 25 '18 at 14:21
  • 2
    $\begingroup$ Re batteries: You might be better off removing the batteries entirely and instead providing technical specifications for new batteries on paper (or something more durable than paper). Of course, if recovering society could be post-apocalyptic or otherwise primitive, this may not work. But at that point you also have to contend with language and cultural barriers, which are substantially harder problems. $\endgroup$ – Kevin Oct 26 '18 at 3:16
25
$\begingroup$

There are two primary parts you need to consider the shelf life of for these devices

  • Magnetic storage
  • The battery

The general theme of this is that while magnetic storage shelf life can be measured in decades. LiPo battery shelf life is measured in years.

The real limiting factor here is the LiPo batteries. They don't like being overcharged, or fully discharged, or shocked, or not used, or overused. Their general reaction to all of these things is to explode.

So while visually they may be in pristine condition, the chances of being able to get any of these devices to run is minimal after anything over a decade of storage.


There's an extra factor here that needs to be considered

  • Solid state storage

Unfortunately this is currently an unknown quantity. As a fairly new and rapidly changing technology, lifespan estimates range from decay starting in only 7 days when left unpowered to lasting over 300 years with steady use.

$\endgroup$
  • 11
    $\begingroup$ +1, the batteries, they will degrade within a decade no matter how well stored. I'd also add mobile phones won't call anywhere without cellular towers and computers that have electricity, and in a post-apoc world, I doubt you have the full electrical grid up and powering everything. If you do, get those manufacturing plants running to manufacture new batteries! $\endgroup$ – Amadeus-Reinstate-Monica Oct 24 '18 at 10:51
  • $\begingroup$ The storage on these devices is solid state, not magnetic. Still good answer because the Li batteries are the big issue. $\endgroup$ – pojo-guy Oct 24 '18 at 11:02
  • $\begingroup$ @pojo-guy, the trouble is that solid state drives are too new for there to be real data. I've seen estimated values from over 300years of solid use to 7 days when unpowered. $\endgroup$ – Separatrix Oct 24 '18 at 11:25
  • 6
    $\begingroup$ 'The secret to the SSD is that it used electrical arcing to melt metal to create and break physical connections' well, I'm pretty sure that the mechansim is a bit different than that, eg. electrons trapped, (or not), between insulated gates in a transistor. If the SSD data could be refreshed at intervals to mitigate occasional electrons tunneling their way to freedom, the lifetime could be very long. especially if the drive has a much larger capacity than the data and multiple copies can be stored/checked, $\endgroup$ – Martin James Oct 24 '18 at 12:22
  • 2
    $\begingroup$ @SSight3 I believe that question has come up on Retrocomputing a few times, in different forms. Certainly there is a question about What precautions to take when powering on old computer for the first time in years. $\endgroup$ – a CVn Oct 24 '18 at 15:11
8
$\begingroup$

One more cause of failure to note are “Tin Whiskers”.

https://en.wikipedia.org/wiki/Whisker_(metallurgy)

At present, any lead-free solder used will form tin/zinc/silver whiskers at some unpredictable time in the future. These hairs are a few micrometers across and once started can grow as fast as 1mm per year. Eventually the whiskers will bridge pins and traces on the circuit board and destroy the device. There has been a large amount of research into using conformal coatings to prevent this growth, but the massive pressure these growths generate in tiny pinpoint locations has been able to eventually penetrate or lift all coatings that have been tested.

This brings up an interesting side of your question. Electronics built before the mid-2000s may not have this type of failure for centuries, while electronics built after the lead soldier ban are unlikely to last much beyond 20 years. However, safety critical and space flight circuits are currently excepted from the ban. This means computers in cars, airplanes, satellites, and power plants are allowed to use lead soldier and will not suffer from this problem.

$\endgroup$
  • $\begingroup$ Allowed to, but sometimes do not use lead. (Source - I worked at an industrial and train equipments manufacturer.) $\endgroup$ – Prof. Falken Oct 25 '18 at 21:59
  • $\begingroup$ With whiskers that fine, a brief wash with a weak acid may be sufficient to 'revive' them. $\endgroup$ – Sherwood Botsford Oct 31 '18 at 13:47
5
$\begingroup$

Another issue to consider is the overall system aspect of cell phones. They don't work without a cell tower network, including functioning computers to keep track of everything. So cell phones simply won't work at all once the external tech has died.

$\endgroup$
  • 1
    $\begingroup$ They would work as portable computers. I read a trilogy (can't remember the titles or author at the moment) where one key plot point was a character who, when on an alternate timeline mixed-up Earth of sorts (it gets complicated...) has essentially a very-smart-phone with a huge database of information, so that even without any cell phone network at all (effectively the only phone on Earth) it is still extremely useful as a computer. $\endgroup$ – manassehkatz Oct 25 '18 at 2:31
  • $\begingroup$ manassehkatz is correct, you don't need any kind of network link to access the data stored on most computers and smart-phones, you only need it to access the network. $\endgroup$ – Austin Hemmelgarn Oct 25 '18 at 12:42
  • 1
    $\begingroup$ @AustinHemmelgarn But most smart-phones really don't have that much data on them (yay "cloud"). A database like manassehkatz referenced would have had to be deliberately and explicitly installed. $\endgroup$ – T.J.L. Oct 25 '18 at 12:47
  • 1
    $\begingroup$ @T.J.L.It really depends on what data you care about. Many people have quite a few photos and a reasonable amount of music on their smartphones. $\endgroup$ – Austin Hemmelgarn Oct 25 '18 at 13:13
3
$\begingroup$

I would think that any computer equipment (PC, notebook, phone) would last pretty well, as long as:

  • Any battery (including e.g. a coin-cell motherboard battery) was removed
  • It didn't rust (temperature and humidity were low)
  • Doesn't rely on EPROM holding state (it will degrade in a decade or so)
  • The plastics didn't degrade too much
  • Any capacitors didn't leak

So if the computers were intentionally stored for long-term survival, and there were instructions on how to put them back together, build new batteries, etc. then they could last for.. let's say multiple decades but probably not much longer than 100-150 years.

As for any data, the storage medium would need to be carefully considered.

$\endgroup$
  • 2
    $\begingroup$ With respect to "any battery" - in addition to the primary batteries, which are easily removable on most laptops & phones, but most definitely not on all of them, there are often smaller batteries (e.g., CR2032 or similar for clock) that (a) often require disassembly to get to them for removal and (b) are sometimes not designed to be easily removable even once you find them. Those smaller batteries can degrade - e.g., leak - over time too, which is definitely a consideration for any > 10 years situation. $\endgroup$ – manassehkatz Oct 24 '18 at 14:18
  • 3
    $\begingroup$ Um, no UV radiation does not corrupt EEPROMS, it erases EPROMS (which are different and largely unused anymore), but they have to be have it directly shining through the quartz window on the packaging of the chip, which would be hidden by the case of whatever the chip is in. $\endgroup$ – Austin Hemmelgarn Oct 24 '18 at 14:38
  • 2
    $\begingroup$ Also, flash memory (including most solid state media) is good for only a couple of years without prophylactic rewrites, magnetic tape can last longer than you probably think if kept in ideal conditions, and optical media actually does degrade over the course of a few decades. $\endgroup$ – Austin Hemmelgarn Oct 24 '18 at 14:39
  • 1
    $\begingroup$ @manassehkatz: Yes indeed, any battery (which is why I mentioned the motherboard battery) $\endgroup$ – K. Morgan Oct 24 '18 at 15:02
  • 1
    $\begingroup$ Oops. I somehow missed that you did already mention the motherboard battery. But my main point is those batteries are often very hard to remove/replace on a laptop. Desktop (unless soldered) - takes < 1 minute (mostly to remove the case screws). $\endgroup$ – manassehkatz Oct 24 '18 at 15:22
2
$\begingroup$

If the devices were deliberately built to last, the story could be quite different.

If you are willing to make significant sacrifices and accept high cost, both in R&D and materials - to begin with, optimize for durability, not performance.

  • Do not use electrolytic capacitors or rechargeable batteries.
  • Store read-only data in old school ROM. (Not EPROM even, ROM.)
  • Make all circuits essentially space grade, so not even accumulated background radiation can flip something.
  • Do not use batteries, instead use some other kind of power source as well as instructions on how to create a power source from scratch.
  • Provide dry batteries (non-rechargeable) for boot-strapping in the first phase.
  • Wet batteries but with chemicals stored separately in large vats with instructions on how to activate them?
  • Use simple LED character displays and small speakers for output instead of relying on normal, complex bitmapped displays.
  • Provide extensive documentation on all systems, including physical paper (or plastic) on everything
  • Build the same system with different technologies in several versions, from discrete transistors up to ICs with everything in between, so a society can bootstrap and make replicas in ever increasing sophistication.

and so on...

a bit on a tangent but I thought it was a cool idea. Like a Rosetta stone for computing.

$\endgroup$
1
$\begingroup$

Atmel ATMega series microcontrollers give 100 year data retention guarantee. This is at regular shelf around 25C. Under better conditions they can last longer and even better performing devices can be built for the purpose. I can see a computer system that will be able to work after several 100 years. LEDs do last for a very long time. Even while working non-stop, I have seen LEDs working for 20+ years. Thus a simple (say 200x100) LED screen could easily last several centuries. Just protect everything from rusting and temperature swings.

Unfortunately, batteries will not last that long. There are lithium batteries that can last for 40 years, that is the best I could find. So it would be an issue to power these devices. The best bet could be a mechanical device that can be cranked to supply power to a low power device. Of course this mechanical device should be stored in an airtight storage.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.