Long-Term PC Preservation Project? 465
failcomm writes "I've been talking with my son's (middle-school) computer lab teacher about a 'time capsule' project. The school has a number of 'retirement age' PCs (5-6 years old — Dells, HPs, a couple of Compaqs), and we've been kicking around the idea of trying to preserve a working system and some media (CDs and/or DVDs), and locking them away to be preserved for some period of time (say 50 years); to be opened by students of the future. The goal would be to have instructions on how to unpack the system, plug it into the wall (we'll assume everyone is still using 110v US outlets), and get the system to boot. Also provide instructions on how to load the media and see it in action; whether it is photos or video or games or even student programs — whatever. So first, is this idea crazy? Second, how would we go about packing/preserving various components? Lastly, any suggestions on how to store it long term? (Remember, this is a school project, so we can't exactly just 'freeze it in carbonite'; practical advice would be appreciated.)"
Re:I've thought about this (Score:3, Informative)
You're assuming that in 50 years:
- the battery will still be able to hold a charge
- there will be no data loss on the magnetic media (hard drive)
- there will be no data loss on the optical media (CD rot [wikipedia.org])
- the soldered components will still work (tin whiskers [wikipedia.org])
- the display will still work (no idea about inactive LCD degradation)
Re:As for preservation (Score:3, Informative)
mineral oil... drop all of the components in, and seal it up.
Beware, mineral oil as well as some other supposedly inert liquids can act as solvents leaching certain chemicals out of plastics or other components, causing breakdown. You have to be really careful what you use, especially for long term immersion.
I think for a "time capsule" you're better off just storing it in a sealed air container. If you want to get fancy maybe go for an inert CO2 or nitrogen atmosphere.
-
Will they care? And avoid moisture. (Score:2, Informative)
1. The students won't care. They'll be concerned with whatever popular culture dominates in 2059, not with old tech. Except for the nerds.
2. If you do this, preserve other things as well. Preserve a copy of the newspaper from the Obama inauguration. (Provide instructions on how to open and read a paper newspaper.) Preserve whatever popular culture dominates in 2009. Preserve pictures of the school and letters from the students.
3. Think carefully about whether you'd really like to inflict Windows XP and Compaq hardware on a new generation of students.
4. Store it someplace dry. Moisture is your biggest enemy. Basements will flood, roofs will leak. Think mold, think corrosion.
5. Motherboard batteries will die, and may leak. Remove them and all other batteries. Forget laptops.
Flash isn't forever, either: ~10yrs (Score:2, Informative)
The information in flash memory is stored as electric charge which slowly dissipates. Last time I checked, it was recommended to refresh it periodically, or the information could be lost in as little as 10 years.
Re:Also CF Card (Score:5, Informative)
Archival DVD rated at 300 years. It might not last 300 years, but hopefully it'd last long enough for this time capsule.
http://www.smarthouse.com.au/Home_Office/Storage/U9P4F7L2 [smarthouse.com.au]
Re:CD-R DVD-R media failure (Score:5, Informative)
Definitely don't use recordable media that are dye-based or phase-change. If you can get the CDs or DVDs pressed professionally, do it - music CDs are made from durable polycarbonate with a layer of silvering applied on the top side, then covered over with lacquer or, preferably, another layer of polycarbonate.
Wrap the discs in paper, then vacuum-seal them in shrink wrap. Seal them in a padded sealed tyvec envelope. Label "Do not open until Christmas 2060" with a Sharpie.
Re:As for preservation (Score:3, Informative)
As far as making sure it's ready for the future... (Score:5, Informative)
For long time storage, I would suggest taking apart the entire system and giving it a good cleaning to remove any dust, Also inspect all electrolytic capacitors for any leakage or damage, you don't want an out-of-box experience to have to include replacing all the capacitors (although it may end up needing it anyway) This will obviously include voiding the warranty on the power supply to clean it out properly (be careful of the capacitors inside as they could hold a deadly charge, even after 15 minutes if the internal resistors don't work correctly) and inspect it. You should remove things like the CMOS battery, usually a button lithium cr2032, which would leak and destroy circuits on the motherboard, or at least go dead, and you should also pack some spare parts and components with it (at least a spare motherboard, ram, cpu, power supply, optical drive, spare fans, expansion cards, etc) , along with the documentation for them, which might not be available then. Pack at least 2 hard drives, pre-loaded with all the software you want them to see, including iso's of the discs that you will include, as you don't know how long the cdrom/dvd media will actually last.. you might want to include a fully bootable flash drive or two with the software and os as well. Include a complete listing of the bios settings for when they do have to put a battery in... if you can, make a written writing with all the electrolytic capacitors values and voltages, as that might come in handy for later. Include as many operating systems as is possible, to give a flavor of what pc's used to be like and what used to run on them, make sure all the licensing information is both in paper and digital form for any piece of commercial software, as they may need it to run the software, even if the companies who made it are long out of business by then. if the pc uses a standard db15 for vga, you should leave a crt and a lcd if possible, and if it uses a dvi connection you should also leave a DVI-DB15 adapter. Make as many video output options as you can available in case things have changed....
Include a nice strong keyboard (like an old IBM Model M) along with a couple of the other keyboards you have (use different models and brands if possible), as the rubber membrane keyboards will likely not age very well. Include a ball and a optical mouse for snickers, and possibly a document on how each works...
Of my years of collecting old pc's, that's what I've always wished was done for me! =)
Re:50 yrs is not that long (Score:3, Informative)
My uncle still fires up his Apple LISA every few months to do his accounting on it.
Which is why he can still get it to fire up. 50 years with no power will probably condemn the capacitors to oblivion. Without proper attention to moisture the disk drive (hard and floppy) heads will have probably oxidized beyond usability. I have a circa 1992 386/20 that is still running but a shortwave radio of about the same time period is inoperative due to capacitors that crapped out since it sat idle for that same 15 years.
If you have the inclination then go for it but I'd suspect you'll have more of a static display (and there is some value to that I suppose) than a functioning PC after 50 years.
Ten years may be doable and that's a bunch of PC generations.
Re:boot sequence? Re:Way too many unknowns (Score:1, Informative)
2^7 is 49. That's 7 generations of change as per Moore's law [for CPU usage] in less than 50 years. There will be significant REVOLUTIONARY changes.
2^7 = 128. 7^2 = 49. 50 years is about 2^5.644 (lg(50)), so it's about 5 and a half generations.
Re:As for preservation (Score:5, Informative)
No, definitely no vacuum. The pressure difference will cause damage. If you think you need to provide more than a stable, not too humid climate, use an inert gas.
Most components will last 50 years without problems, but the BIOS battery won't. Modern hard disks with fluid dynamic bearings may be a problem. Software should be stored on low density magnetic and optical media: Tapes are still the longest lasting archival format that is directly readable by a computer. CDs are more likely to last 50 years than DVDs.
The best way to keep a system in working order is to use it every once in a while.
Capacitors are your enemy (Score:5, Informative)
The electrolytic capacitors are going to dry out in 50 years and will cease to function. There's a chance they will damage other components when the power supply is powered up again. I've seen it happen with equipment that is less than 25 years old. I don't think there is any known solution to this problem.
I'm currently restoring a 50 year old stereo receiver (Harmon Kardon TA230) and the electrolytics are almost completely gone. Everything else is in excellent shape; the resistors, coils, tubes, even the lamps test good but the caps are all shot. This receiver has a old style transformer power supply, so I can bring the voltage up slowly using a Variac for testing. Your computers are going to have switching power supplies which will not like having a lower voltages applied to them so that's not an option.
I honestly have my doubts that much from this era will survive 50 years. It's all made as quickly and as cheaply as possible with the expectation that it will be replaced in 3 or 4 years.
I currently have an Apple ][ that no longer can read its boot disks, a PC XT that doesn't always recognize one of its ST-506 drives and a few months ago I went through my Amiga disks and found that most of them were no longer readable. All of these are far less than 50 years old and have been stored carefully and well cared for.
However, my AIM-65 made in 1977 is still able to read data from my ASR 33's paper tape reader, which is 45 years old and still working fine.
Yeah, my wife hates me for keeping all this junk.
Preservation (Score:4, Informative)
Northeast Document Conservation Center [nedcc.org] is another good resource with guidance pertaining to specific types of materials.
NIST's PDF guide Care and Handling of CDs and DVDs [nist.gov] contains best-practices for optical media storage/handling.
Re:The primary problem with your idea (Score:5, Informative)
boron magically electromigrates or keeps on diffusing? Um, I have been doing ultra high reliability electronics for over 25 years and this is total news to me. Don't think so. I know of electronics in geostationary satellites still humming along over 20 years and still going strong. There probably is even older out there, but I wasn't involved, so I cant say.
Suggest:
I would put any and all data storage in multiple formats and multiple copies. Cover your bases.
Find a way to seal the system against moisture. (Airtight containers and Silica Gel packets inside.)
Provide lots of text based paper documenting the system, and all its hardware and software. (information printed on high quality paper)
Multiple disk drives that can be booted from would also be good.
As a side note, the 6800 microprocessor still gets made and used today and thats about 30 years old right now. I have both an Atwater Kent and a Crosley radio from 1928 and 1932 and both of them still run just fine. Not too shabby for 70 year old electronics!
Re:As for preservation (Score:3, Informative)
It'll take some reliability engineering... (Score:5, Informative)
You will have to replace every single electrolytic capacitor (in everything - mainboard, PSU, every drive, monitor, mouse, keyboard and speaker amp) with solid-state versions. Electrolytics dry out and it's very unlikely that anyone other than a computer historian would think of this before powering the computer up. Altair 8800s and Imsai 8080s from the late 1970s are now to the point that their power supplies and electrolytics must be replaced for them to work reliably - don't expect your machine to fare any better.
It's also a safe assumption that the lubrication in any rotating media drive will be gone by 2060 - not sure how to deal with that other than providing lube in a hermetically sealed package along with instructions to disassemble the CD drive and apply it.
How are you going to have your data last? Tapes and hard drives will demagnetize by 2060. Flash may have a prayer; Your best bet is to get some extremely long-lasting batteries and interface a microcontroller with a plugged-in thumb drive. Store the data along with error-correction codes on the drive. Have the system wake up every ten or twenty years and "scrub" the drive, reading every block and writing it back. Do the same with the system's bios EEPROM - the system will be useless if that gets killed by a cosmic ray. You should also pay to have data CDs gold-mastered - redundancy is the only way to go here.
The display is another problem. The only technology I'd really trust to just work without needing any repair is an LED display; LEDs can run continuously for decades. After the LEDs, a CRT is probably the best bet (despite a decent one having hundreds of precision electrolytics that'll need replacing) - After all, we've got examples of working CRTs from the 50s and 60s. Newer technologies haven't been around long enough to prove themselves yet.
Get a corrosion resistant, hermetically sealed package for the whole kit and kaboodle and flood it with a dense inert gas like SF6 to keep anything from growing. Thoroughly sterilize every square millimeter with a hard UV light just to be safe. Put the HDD in its own sealed bag full of nitrogen if you include one.
For power, your best bet is probably a primary battery (Mg-Cu) with seperately-stored electrolyte feeding an inverter - The shelf-life is "forever until mixed," at which point the machine will probably have a few hours of power depending on how much you include.
Assume that the people who recover the device will still speak your local language and have libraries where they can look up terms such as volt/byte/etc. If they can't, I doubt there will be enough of civilization left to care about some artifact from before The Fall. I think that it will take far more time and money than you're prepared to casually expend if you want to entomb a computer and have any reasonable probability of it turning on and actually working after 5 decades alone, rather than just popping a PSU capacitor or being a dead relic.
Re:The primary problem with your idea (Score:3, Informative)
During manufacturing Boron is diffused into selected regions of purified silicon (doping) to create the transistors....Most CMOS process technologies have anticipated lifespans of 5-10yrs.
Uhh.. I've had plenty of machines that are more than 10 years old that work just fine. I've got one machine that's been running continuously for 6 1/2 years (a few reboots, but no crashes). The motherboard is approaching 10 years old.
Several years ago I had an AMD 386-40, circa 1992. I ran it as my mail server until maybe 2005 when I decided to upgrade to a faster machine and run spamassassin. It ran continuously for several years, and with the exception of HD and power supply failures ran without a hitch.
The ONLY component failures beyond the normal PS/HD failures I've seen are blown capacitors. This happened after only a few years, and was part of the larger problem people have been experiencing with bad caps.
So.. I'd say your theory about boron diffusion destroying computers in 5-10 years doesn't stand up to the evidence. There's a LOT of people with rather old everyday machines that run just fine after 10 years.
Re:Will they care? And avoid moisture. (Score:2, Informative)
Re:I've thought about this (Score:4, Informative)
Yeah, because in the last 50 years, there have been massive changes to the voltage and frequency of electrical utilities. And why protocols? IP is well documented, and any changes would be as well. There haven't been massive library burnings in quite a while, so we should be safe. There have been very few changes in the last 50 years. There is some "quality" change, but the basics are the same. The microwave oven has taken over, and TVs are flatter. Gadgets are everywhere, but no ones that cause problems. Anyone that learned on a 50 year old car could drive a current one with no more than 30 seconds of training, if that.
Re:The primary problem with your idea (Score:3, Informative)
You apparently have no knowledge of deep-submicron VLSI. Regardless I did not say that the device magically fails at year 10, I said that it might need to be clocked more slowly. If you are conservative with your operating frequency to begin with, you are going to see many more years of life. Consumer electronics are not conservative.
There is nothing magical about "diffusion", an concentration gradient will result in diffusion, only a question of how fast.
Last, joining "electromigration" with "boron" is just nonsense. Electromigration effects the metalization, not the dopant atoms.
Re:Way too many unknowns (Score:2, Informative)
Re:As for preservation (Score:2, Informative)
Its really hard to change power standards once they're in place. I guess there just isn't enough incentive in saving a bit of copper to go through the mess of throwing out everything that uses power in the country.
Re:As for preservation (Score:3, Informative)
It's actually 230V in the EU, not sure about anywhere else.
To get the same power, a higher voltage gives a lower current (P=IV). The lower current means wires don't have to be as thick (alternatively you can draw more power for the same thickness of wire).
A normal UK appliance is allowed up to 13A at 230V = 3kW. Stuff like vacuum cleaners and electric heaters might draw this much power.
Re:As for preservation (Score:4, Informative)
um 10 to 20 ha, try 200.
Also just so you know Every house wired in the USA has had 240 volt since the 70's. The average receptacle might be 110, but the mains coming in, every stove, and most electric heaters are all 240 volt. We get the best of both worlds. efficiency for high power draw items, and cord ends smaller than bricks. Also we have more than 3 receptacles in a room.
Re:The primary problem with your idea (Score:3, Informative)
When one of today's computers is booted up 50 years from now, a few caps will probably need to be replaced, many of the plastics will be brittle, the silicon will probably be like new; but the firmware will be gone, and who will have a copy of "Dell GX270 A08" in 2059?
Re:As for preservation (Score:3, Informative)
The main advantage of 220/240v is that everyone else in Europe uses it. The main advantage of 100/110v is that everyone else in the US/Japan uses it.
Germany standardised on 220v and the US standardised on 110v. Everyone else pretty much followed one or the other. I don't think there was much thought in those days about what the optimum voltage might be.
Re:As for preservation (Score:5, Informative)
Have a read at the problems faced in preserving the BBC Domesday project, and that's only about 20 years old.
Re:As for preservation (Score:5, Informative)
Re:As for preservation (Score:3, Informative)
Re:As for preservation (Score:5, Informative)
There was a lot of thought.
The US standardized at ~110V because that was what basically as high of a voltage as you could put through a ligthbulb at that time (higer voltage light bulbs have more delicate filiments, and therefor arn't as robust). Germany on the other hand standardized several years later, when better lightbulbs were available, so they standardized on a higher voltage which made more efficient use of wire (higher voltage lets you put more power through a given cross-section of wire).
Both made the best choice given the state of technology at the time.
Re:As for preservation (Score:2, Informative)
In Europa we have 230V for "small items", high power items like stoves etc usually uses 400V.