Designing for longevity questions and request for long-lived extremophiles!

[newastrocity]

Hi All,

I'm working on some research for school and could use some answers to specific questions, as well as some background and reference materials I might not have come across yet.  I *also* want to pidgeonhole some questions about a physical project I would like to build.

I'm against the throw-away culture, I understand the economies of scale and how it is more economically feasible to buy a new $150 TV than fix it for most people.  However, I would prefer to buy something once and then never have to buy a replacement as much as possible (I love Toyotas!).  I am trying to research how electronics can be designed to last as long as possible (ignoring obsolescence).  Unfortunately there is not a lot of money to be made by designing something to last decades rather than for a few years when we all know most consumers will move on before then.

Specific questions about longevity:
1)  Are there more good examples of electronic systems still being used as workhorses today?  I know about the C64 in the Polish tire shop, Voyager probes, and the Amiga that was powering that school district HVAC system for a long time...industry and government examples would be great.  They don't necessarily have to be still working if I can find documentation of them running for an extreme amount of time in real-world situations.

2) Does potting compound help with extending life?  I understand that many devices would die prematurely without potting due to moisture and vibration, but if two identical components were sitting on a desk in a temperature controlled environment, would the potting help otherwise?  How does thermal interact with potting?

3)  What is an example of a device DESIGNED to last a very long time, decades+?

4)  Other than heat, moisture etc, all the standard common sense stuff, what design considerations can be taken to extend component life if money was not as big an object?  Even little stuff that doesn't immediately matter but would add up over time would be something really interesting and useful.  Obviously, you need to stay away from Electrolytics as much as possible.

5) Any good reference materials for very long lasting electronics?  It seems real good material on this subject is scant except for batteries and storing data for long periods of time, but people aren't terribly concerned with the hardware itself working.

Part 2:  My project

Continuing on the theme above.  I'd like to make a device that can last as long as possible to do something as simple as possible and to be the most robust.  I was thinking about a timer/counter that was ultra-low power, and only displayed the "count" since it was turned on for a brief second when a switch was pressed. 

If I wanted to make a counter that would keep the count for 40+ years, what would be the best device to keep the count?  I see some ultra low powered ICs in the microamps, but would I fare better longevity wise with a larger more robust DIP package?  I'm imagining sealing something with potting compound and an Edison Battery.  I'd also be all ears for something that could be fully electronic and could be passed down as a cool keepsake.

Thanks for the help, any anecdotes are welcome, would *REALLY* love to hear some design considerations.

[Teun]

First of all. Most engineers are unlike you branded on getting new things once in a while. Newer versions of hard/software will always be apealing because of the improved and enhanced features they will bring. This is something that is beneficially on an economic level. Keeps most of us engineers here in buisiness  .

1. Most test-equipment such as oscilloscopes, Network Analyzers etc. have a very long lifetime. In many company's these devices will be older models because they are very expensive. I use a HP network analyser from the early 90's for example. Device works great, but the floppy disks for the data are getting a problem. Had to use a very old computer and look very hard for a conversion program to use the file format

2. Potting can help. As you said mostly for moisture etc. But it depends on the compound used. When it doesn't dissipate heat well it can cause a lot of trouble as well.

3. Can't really think of a good example right away. Factory macines maybe? CNC milling, turning etc.

4. If money doesn't matter every EE can probably give you hundreds of different examples. EMC and EMI related issues for example. Conducting materials, design space this list can go on and on.

5. All depends on the environment products are designed for. If you would design something for use in the open desert you wouldn't expect it to last long in these conditions(hot/cold, sand an dust). The hardware for a far less intensive used device could last for decades.

Basically all these things are to be taken in consideration and that is what it means to be an engineer.


Your project:

I think something as simple as a counter could last for maybe 100+ years. I think my dad still has an alarmclock from the 70's he still uses every day to get up. That's +/-50 years and counting!

[dmills]

Industrial controls can have surprisingly long lives, I know of one set of elevator control gear that is still in service and is relay logic built on a slate bed for example, it just works so nobody touches it (Which is probably why it stays working!).

There are still more then a few sets of 1970s phase angle controlled lighting dimmers in active service in theatres up and down the land, and plenty of similar kit in dark corners of older plants that just sit there doing the same thing it did for the last 30 years, long forgotten about in some cases and only touched when something breaks.

The trick to this stuff is build it big (so you don't need forced cooling (moving parts)), and build it simple, stuff that does not have custom programmed microprocessors is generally easy to keep working, don't add features just because you can. Simple linear power supplies, with caps that are not stressed remotely heavily, with good airflow and sane derating help. Electrolytics are not the automatic problem you might expect, we have good models for their behaviour... Old Tants are a bigger issue, as are very heavily doped junctions (Think tunnel diodes).

Of course sometimes you just get unlucky in unforseen ways (Old RIFA class X and Y caps for example, nobody at the time saw that coming).

Then we get to test gear, there is a lot of the old stuff in peoples labs, my 4192A, 8753C and the like are all over 30 years old, and still do what they do rather well, but they were both the price of a nice house when new (And having military tools budgets when setting the BOM target cost helps at the design stage).

Generally you will also find that infrastructure stuff is built to have a very long life, look to railway signalling, power network stuff, things like that.

Regards, Dan.

[coppice]

What do you consider a long life? There are 2 aspects to this - design for serviceability to achieve a long life, and design for an extremely long typical time to failure. If you want to put something somewhere inaccessible, like the sea bed or in space, you need to design for an extremely long time to failure. For things like industrial equipment, you might just need to ensure the system is serviceable for many years, with a few failures being accepted along the way. If you are looking for lifetimes of thousands of years, which might be needed for space applications at some point, I don't think we have a clear idea how to achieve that. You will, however, find some specialist systems trying to achieve lifetimes up to 100 years.

Most military systems are built for at least 25 years operational life.They usually have stringent project requirements for complete service information being supplied by the developers, and all parts being available for at least 25 years - either through continuous production, or by stockpiling when the system is new. Stockpiling is a non-trivial task, as you need to ensure the parts are kept pristine. If you don't, then after a few years they may be in no better condition than the parts being replaced. For industrial users, companies like Rochester Electronics exist to provide a long term availability service. Military users usually deal with this themselves.

Telephone network, and other infrastructure equipment was once built for a 25 year service life, but most of that has been reduced to 15 years or less. The equipment was becoming obsolete too quickly for a 25 year life to matter much. Even undersea cables now have a reduced life time, because expanding capacity needs keep making the older cables worthless.

As for consumer equipment, lots of people seem to think things like TVs are only designed to last for the warranty period. Having worked with a few TV makers, their internal design goals are usually 8 to 10 years of trouble free life. Things like the electronics in ovens and fridges may be designed for twice that.

[atmfjstc]

I think the Voyager probes are a good example of complex man-made mechanisms that have survived waaay past their designed life, under extremely inhospitable conditions and virtually no maintenance to speak of.

[coppice]

I think the Voyager probes are a good example of complex man-made mechanisms that have survived waaay past their designed life, under extremely inhospitable conditions and virtually no maintenance to speak of.

Its amazing how long many things last when there's nobody around to tinker and break things. 

[atmfjstc]

I just thought of something else. If you want something to last, design it without electrochemicals like batteries and electrolytic capacitors. Chemical stuff is much more vulnerable to temperature changes and degrades much faster than anything else, even when the device is completely powered off.

And also make sure the chip making process is legit. I recall a story about some early MOS technology chips failing en masse after a few years due to a fatal flaw in the manufacturing process that caused them to corrode themselves from the inside.

[grifftech]

Some parts of the power grid are quite old

[coppice]

Some parts of the power grid are quite old

So far, we have been much better able to make to make electrical and electromechanical things last than electronics. This is, perhaps, not surprising with simple, no moving parts, electrical equipment, like transformers. Its a bit sad that we still struggle to make no moving parts electronics last better than electromechanical stuff with lots of opportunity for wear and tear.

[helius]

I just thought of something else. If you want something to last, design it without electrochemicals like batteries and electrolytic capacitors. Chemical stuff is much more vulnerable to temperature changes and degrades much faster than anything else, even when the device is completely powered off.

https://en.wikipedia.org/wiki/Oxford_Electric_Bell

Seriously, not all electrochemical systems are the same and some will operate longer than others. Even if you carefully select the longest-lived components, the complexity of your device plays a major role in failure. The space probes were engineered with reliability in mind and include redundant systems because failures were anticipated.

[MadScientist]

Given the budget , most electronics can be designed to operate well past the end of their engineers life or to survive conditions that would quickly extinguish human life.

The reality is that cost constraints drive everything , why make a telly that can survive for 50 years , there no customers for that device

Today most half decent electronics lasts quite a long time , I can’t remember the last time I dumped a non functioning TV. I have MacBooks that are 10 years old , etc etc

[Tomorokoshi]

Old machines that just keeps working like the HP equipment referenced above have likely been repaired at one point or another. What keeps them around is the relatively easy access to clear schematics, part lists, and spare parts. One strategy is to use widely available parts that have been manufactured for a long time.

The schematic is only a mnemonic of the circuit. It doesn't completely represent the realization of the implementation. (To use Blaauw's terminology). The schematic can miss influences such as:

Environmental:
- Temperature
- Humidity
- Dust contamination
- Aging

Chemical:
- Corrosion
- Tarnishing

Mechanical:
- Flex
- Vibration
- Spacings

Electrical:
- Out of range inputs
- Out of range outputs
- EMC in all its various forms
- Frequency of operation

User:
- Foolishness
- Idiocy
- Ineptitude

Some of these can be mitigated using well-known and understood design practices.

For instance:
- Instead of potting the entire assembly, conformal coating will provide a lot of protection without making serviceability impossible.
- Protect all inputs and outputs with filters and devices to limit the input and output voltages, currents, and frequencies to acceptable levels.
- Use an enclosure that provides appropriate protective features.
- Operate parts well inside their ratings. Note that some parts may drift more when operated well below their ratings.
- All the above influences can change the operating value of the part in various ways.
- Add redundant circuits in critical areas.

That said, it isn't too difficult to find equipment that is quite old that never needed servicing. However, it's easier to find examples of newer equipment that failed for preventable reasons.

[chickenHeadKnob]

If you havn't seen this before:

[Audioguru]

I thought you were asking how will I last a long time.
It is because I feel and act much younger than my age (73), my wife keeps me young and I have routine maintenance from my doctor and medications.

[Nusa]

1)  Are there more good examples of electronic systems still being used as workhorses today?
Tube-based amplifiers are still valued and used by many musicians and audiophiles. Including antiques.

2) Does potting compound help with extending life?
Maybe. Depends on nature of the circuit and the environment it needs to operate in. On the flip side, a broken device that is NOT potted is more likely to be economically repairable if it fails.

3)  What is an example of a device DESIGNED to last a very long time, decades+?
Virtually all commercial avionics. The design life of most aircraft are 30+ years.
A lot of higher-end electronics test equipment. There's a healthy market in hobbyist arena for still-functioning test equipment from decades ago, as the new stuff is out of their price range.

4)  Other than heat, moisture etc, all the standard common sense stuff, what design considerations can be taken to extend component life if money was not as big an object?
5) Any good reference materials for very long lasting electronics?
NASA has a lot of public documentation on design standards for space equipment.