How would we design electronics to last 100 years?

[GMB]

Last time I studied this it looked to me that what the manufacturers mean by its "life" is being in specification. So after the stated number of hours it doesn't necessarily go bang - it may just mean it has drifted in value so much that it isn't what it written on the side. But many components in circuits don't need an accurate value so this is often not important.

[Maarten]

Originally Posted by emeritus

1000 hours is just under 6 weeks, and I had been leaving the Icecrypt box switched on all the time, only turning off the TV, but will now be turning both off when not in use.

In modern electrolytics, the lifetime is defined at 105 degrees ambient (which has already been pointed out, I see now) AND maximum allowed ripple current. For every 10 degrees less, the lifetime doubles. Reducing the ripple current will also result in a significantly longer life. Low ESR electrolytics allow for higher ripple currents at 10kHz to 100kHz than traditional types so are especially suitable for such applications.

[duncanlowe]

Some other things that might help, depending on the particular application of said electronics.

Avoid straining any PCB during manufacture as this can lead to component failures. MLCCs on a PCB are particularly prone to this. As an example, it's quite common to snap PCBs out of a panel. It's much less likely to cause a problem if they are machined out.

Ensure the finished PCBs are clean. Flux or other residues can absorb atmospheric moisture and become conductive causing dendritic growth leading to shorts or partial shorts. Conformal coating can help with this, but also make it worse. I have personally seen extremely early life failures caused by flux residue that ultimately caused a MOSFET drive to oscillate, and the MOSFET overheat and cease to exist.

[The Philpott]

Duncan touches on 'breakout' PCB's. It made me cringe last week when i had to separate 5 charging modules from each other by force. They did of course have a V slot on either side, but even so..

dave

[Catkins]

Originally Posted by G6Tanuki

There's also the issue that the world as you designed your device for moves on over that 100 years.

I'm reminded of the UK's introduction of radio; initially a broadcast-receiver only really needed to receive one station - because that was all there was.

So simple was elegant/cheap. Then the BBC came along and Mr. Eckersley implemented his 'Regional plan' which meant that loads of people then found their early, simple/cheap receivers had to have selectivity to separate-out the regional and national stations. A single signal-frequency tuned-circuit was no longer good-enough.

With so little on medium wave nowadays, a pre-regional single tuned circuit radio is now again sufficiently selective

FYI This is a video of my "1924 Sterling Anodion" single tuned circuit radio receiving Radio Wales on 882 kHz from the Washford transmitter.

https://youtu.be/3ZxiaJvIb1g

This radio is almost 100 years old and still functional.

[Radio Wrangler]

The need for electronics is likely to have changed significantly if the last hundred years is any guide, so current electronic stuff is likely to have become museum exhibits. Still, it would be nice if they could be demonstrated working, just to show what we had to put up with back in our era.

David

[emeritus]

Thanks very much for the info on the effect of temperarure on electrolytic capacitor life. My Icecrypt STB is mounted on end on the side of its TV, reminiscent of how our Band III convertor was mounted to get ITV, so should have good thermal convection via the ventilation slots in its sides.

[bluepilot]

Not exactly electronics, but close:
Oxford bell

[jjl]

A characteristic of traditional aluminium electrolytics that I was unaware of until recently is that ESR increases with decreasing temperature. ESR more or less doubles at -10 degrees C when compared to that at 20 degrees C.
This may mean that, at work, we'll need to replace at least some standard electrolytics with polymer types in a system that is to be used at low temperatures.

John

[Radio Wrangler]

Electrolytic capacitor life reduces dramatically with altitude. More pressure difference across seals leading to faster loss of water vapour as the reducing pressure inside promotes faster evaporation.

Solid dielectric electrolytics are better, and then there's big ceramics.

David

[G0HZU_JMR]

Originally Posted by jjl

A characteristic of traditional aluminium electrolytics that I was unaware of until recently is that ESR increases with decreasing temperature. ESR more or less doubles at -10 degrees C when compared to that at 20 degrees C.
This may mean that, at work, we'll need to replace at least some standard electrolytics with polymer types in a system that is to be used at low temperatures.

John

The electrolyte in these caps changes at sub-zero temperatures and it becomes quite viscous by -30 degrees C. In the past I have put an electrolytic cap in a small environmental chamber and measured it with a network analyser as the temperature was reduced. At about -20 degrees C the capacitance started to reduce noticeably and the ESR started to increase. By -30 degrees C the capacitance had collapsed and the ESR had increased a lot. The datasheet stated that the cap was rated to -30 degrees C but this clearly doesn't mean full performance is maintained at low temperatures.

[kalee20]

Yes. If the electrolyte freezes it becomes non-conducting, so all bets are off!

At low temperatures, it becomes more viscous (even plain water is about 3x more viscous at 20 deg C than at 80 deg C). So resistance changes by a corresponding amount.

[G6ONEDave]

Use good quality ceramic capacitors and not the ones as fitted to the Samsung radio in another thread. Shame there was no actual makers name on them.

Also make sure that you have an independant signal source suitable for the equipment being designed, so that the correct signal can also be created in 100 years time. This would mean even more design consideration though.

Dave

[Maarten]

Just use amplitude modulation on a medium wave frequency. It has lasted 100 years already, so might be a good candidate to last another 100 years. Come to think of it, a receiver which only contains a tuned circuit and a diode, can probably also last 100 years so only the transmitter would need some careful engineering.