Electrolytic capacitors bad news for anything on full power for long-ish time

[Connecteur]

My brother had an Oldsmobile 88, and after a few years the door handle broke.  It was very difficult to take the door apart without proper tools, and the broken part was very expensive and impossible to repair.  It was a very flimsy piece of plastic, which obviously was made to break after a certain amount of stress being placed on it for a few years.  Several other people with the same car had the exact same failure.  It's very hard to believe this wasn't a deliberate design to drive customers to the dealerships to make money.  Door handles were built reliably a hundred years ago, so something was deliberately changed to make them fail.

[bdunham7]

My brother had an Oldsmobile 88, and after a few years the door handle broke.  It was very difficult to take the door apart without proper tools, and the broken part was very expensive and impossible to repair.  It was a very flimsy piece of plastic, which obviously was made to break after a certain amount of stress being placed on it for a few years.  Several other people with the same car had the exact same failure.  It's very hard to believe this wasn't a deliberate design to drive customers to the dealerships to make money.  Door handles were built reliably a hundred years ago, so something was deliberately changed to make them fail.

You can believe what you want but I have never had an engineer tell me that they were told to specifically design something to 'break' after a certain period--it's just too unpredictable.  The story I have heard many times is that they were told to incorporate all the latest and greatest features, often to match the competition, but to do so at the lowest possible cost, if not a bit lower than even that.  "Doing more and more with less and less". 

There are instances where things are made to stop functioning on purpose (as opposed to breaking or failing) after some period or amount of use, such as toner cartridges that indicate empty after so many pages regardless of whether they actually are empty, but that's a different story.  There are also many instances where things are made to fail in a safe manner when stressed, such as inaccessible thermal fuses, but again, that's another story.

[Siwastaja]

A proper, sustainable, environmentally friendly and also satisfactory to the customer, is a design which is designed to last maybe double the "expected" lifetime, or enough for say 99% of the customers, but no more; and then, all parts start failing approximately at the same time, so that the whole product can be replaced. This minimizes waste.

Needless to say, such engineering is very challenging, but I think Toyota for example has traditionally nailed it, making cars that have room for 200 000km in the maintenance logbook (so this is clearly the "design" lifetime), actually last for 350000-400000km with minor maintenance, then finally starts to rust through everywhere approximately at the same time. (Don't know if that's true anymore. Was true for my 1988 model which I finally got rid a year ago.)

Only few can do it.

For the rest of us, we need to calculate in larger margins. And be prepared to still have issues of premature failure. The key is not to change too much all the time, but learn from the previous product generations: just fix things that were failing early by looking at warranty repair statistics, or customer feedback if it's a throwaway product.

What many laymen and even some engineers consider "good engineering" is often a disaster consisting of some parts massively oversized to last for 1000 years, while some parts require constant maintenance. People who respect products like that like maintenance work; they think it's a good feature that "you can repair the product", "it's not a throwaway junk", "look how sturdy it is" (by looking at some overengineered visible part). What they miss is that the product would be much better if it was engineered not to require repairing at all! Being able to fix the broken piece of junk makes them happy, but majority of people will just throw it away and buy a new one if it's something say less than $1000-$2000 in value.

[bdunham7]

For many products, some failures are more acceptable than others.  Everyone will have their own view on what is and is not an acceptable sort of failure.  Also some components have characteristics that make them impractical to design for intermediate lifespans--it's either forever or you have premature failures.  Toyota, something I'm very familiar with, does not actually meet your criteria at all.  Even though your old Toyota rusted out and appeared to be all done, I can point  you to several things that are 'overengineered'  compared to competitors and have virtually a limitless lifespan under normal conditions.

I do understand what you are saying and it is one reason that it is not necessarily a good idea to ask your mechanic what the best cars are.  My disagreement is over the issue of how large and predictable your desired 'intermediate zone' --between premature failure and 'forever' life at greater expense--actually is.  A power supply with a 5-year lifespan is probably designed right at the ragged edge of the component datasheets, if not well beyond--if the components even have datasheets.

[Siwastaja]

I said it's "very challenging" for a good reason. I presented an ideal, which is obviously unreachable, even by Toyota. Still, it should be the general idea and target. Especially damaging is to have very early failures on key components. Or having 99.9% of the components last for 100 000 hours, yet have one small part define the 10 000 hour lifetime. Large imbalances are the problem.

Many parts are, as you say, either "die soon" or "last almost indefinitely". In these cases, you can't do anything else but to choose the latter and be done with it.

[coppice]

I said it's "very challenging" for a good reason. I presented an ideal, which is obviously unreachable, even by Toyota. Still, it should be the general idea and target. Especially damaging is to have very early failures on key components. Or having 99.9% of the components last for 100 000 hours, yet have one small part define the 10 000 hour lifetime. Large imbalances are the problem.

Many parts are, as you say, either "die soon" or "last almost indefinitely". In these cases, you can't do anything else than to choose the latter and be done with it.

A trouble free design needs parts to be engineered to be reliable wherever the car is sold, and that can make their lives almost infinite in some markets. If the design is stripped back there is a danger of it not holding up to extreme stresses. like deep potholes. A car has to sell globally. If you are lucky enough to live in a country where roads are properly maintained, you might well be OK with a significantly stripped back design.

[bdunham7]

If we're talking about a television, for example, I'm in complete agreement with  you.  It doesn't matter what breaks, it usually goes in the trash.  And nobody wants a 20 year old TV anyway.

 However, there are some components of some products, like a manual transmission clutch in a Toyota, where a failure is 'acceptable' because we all understand the limitations of clutch technology. If you object that a clutch is a 'wear item', then just use a wheel bearing as the example instead.  However, if your door handle broke from anything other than abuse or freezing up, you would consider that substandard.  Or at least I would.  The expectations are different for different parts. 

Another example--refrigerators.  If any part of the sealed system or compressor failed on my refrigerator, I would be upset and consider the product poorly made even if it lasted 10 years or more.  However, if the icemaker jams up and breaks, well after all it's a plastic mechanism working in cold temperatures--so that happens, right? 

[Siwastaja]

Oh, wheel bearings started to sound a bit weird at 380 000 km and were acceptable up to 420 000 km, which was the time most of the body started to finally rust through everywhere. Clutch started to show signs of starting to slip soon at similar time but would have been fine for maybe 50000km more.

I have seen cars where wheel bearings are end-of-life after 50000km, while the body lasts for 6-7 times as long.

I thing reducing number of "wear items" is great design. Some "wear items" cannot be avoided; I think brake pads and tyres are such examples.

I think a refrigerator should last for at least 20 years or so. The reason is, if done properly, this can be achieved with normal level of good engineering. Another reason is, I don't see any reason to "update" to a newer refrigerator (efficiency claims are mostly bullshit).

But a smartphone does not need to last 20 years. People want to update for good reasons. Also, they are stressed to external forces and wear.

[coppice]

Another example--refrigerators.  If any part of the sealed system or compressor failed on my refrigerator, I would be upset and consider the product poorly made even if it lasted 10 years or more.  However, if the icemaker jams up and breaks, well after all it's a plastic mechanism working in cold temperatures--so that happens, right?

I think you've skirted around something, but missed the target. If it were technically difficult to make the cooling system in a fridge run continuously without trouble for 10 years, we would have to get used to that, and accept a shorter life. However, decades of mass manufacture have taught us to expect at least 10 years, and anything less is unacceptable. That is our baseline for solid practical reasons. On the other hand, its easy to make the icemaker strong enough to work well for 10 years, and we know that. We can see how flimsy it is, and how the additional cost of making it stronger wouldn't be huge. However, for some reason people seem more and more accepting of things like fridges gradually losing bits, like ice makers, and shelf supports, and unreasonably thin glass shelves, and door handles (the door handle of our 2 year old Samsung fridge has fallen apart just this week. Its still under warranty, we should be OK). People even blame themselves for failures. "I suppose we shouldn't have put so much on those thin glass shelves". Damnit, those shelves are supposed to be strong enough for you not to worry, or the product fails the fitness for purpose test.

People need to less tolerant.

[bdunham7]

I think you've skirted around something, but missed the target. If it were technically difficult to make the cooling system in a fridge run continuously without trouble for 10 years, we would have to get used to that, and accept a shorter life. However, decades of mass manufacture have taught us to expect at least 10 years, and anything less is unacceptable. That is our baseline for solid practical reasons.

That's more or less my exact point.  I didn't say it explicitly, but my expectation about that system is based on experience and a general knowledge of the basic systems.

On the other hand, its easy to make the icemaker strong enough to work well for 10 years, and we know that. We can see how flimsy it is, and how the additional cost of making it stronger wouldn't be huge. However, for some reason people seem more and more accepting of things like fridges gradually losing bits, like ice makers, and shelf supports, and unreasonably thin glass shelves, and door handles (the door handle of our 2 year old Samsung fridge has fallen apart just this week. Its still under warranty, we should be OK). People even blame themselves for failures. "I suppose we shouldn't have put so much on those thin glass shelves". Damnit, those shelves are supposed to be strong enough for you not to worry, or the product fails the fitness for purpose test.

People need to less tolerant.

I disagree about icemakers (although mine is 25 years old and works fine) simply because they are exposed, mechanical systems.  I understand they can break and I don't think you can easily design that possibility out of them.  And, of course, they are less important.  As for the rest of your stuff breaking, see my comments elsewhere about 'doing more and more with less and less'.  Not only should people be less tolerant, they should be more willing to fork out a premium for 'invisible' quality like stronger, more robust parts rather than an infinite variety of doodads with marginal utility. 

[coppice]

I disagree about icemakers (although mine is 25 years old and works fine) simply because they are exposed, mechanical systems.  I understand they can break and I don't think you can easily design that possibility out of them.  And, of course, they are less important.  As for the rest of your stuff breaking, see my comments elsewhere about 'doing more and more with less and less'.  Not only should people be less tolerant, they should be more willing to fork out a premium for 'invisible' quality like stronger, more robust parts rather than an infinite variety of doodads with marginal utility.

The customer no longer has a choice. When we bought our current fridge 2 years ago I was very concerned about how fragile things like the shelves seemed in every fridge I looked at. It didn't matter if it was a low end or a high end model. It didn't matter which brand I looked at. I couldn't find anything with the robust feel of even a cheap fridge from the 80s or earlier.

[Zero999]

Thanks, ...I mean, if you search digikey for  “Aluminium electrolytic capacitor, 450v, 47uF”…..there are virtually none above 105degc.

There are loads rated to 105^oC and I found this one rated to 125^oC on the first page of results.
https://www.digikey.co.uk/product-detail/en/tdk-electronics-inc/B43866C5476M/495-6100-ND/3500156

This one is 100µF, 250V 130^oC. Use two in series, with suitable voltage sharing resistors.
https://www.digikey.co.uk/products/en/capacitors/aluminum-electrolytic-capacitors/58?k=&pkeyword=&sv=0&pv252=156406&sf=1&FV=-8%7C58%2Cmu100%C2%B5F%7C2049%2Cmu250V%7C2079%2Cmu280V%7C2079%2Cmu300V%7C2079%2Cmu315V%7C2079%2Cmu330V%7C2079%2Cmu350V%7C2079&quantity=&ColumnSort=0&page=1&pageSize=500

[SilverSolder]

A proper, sustainable, environmentally friendly and also satisfactory to the customer, is a design which is designed to last maybe double the "expected" lifetime, or enough for say 99% of the customers, but no more; and then, all parts start failing approximately at the same time, so that the whole product can be replaced. This minimizes waste.

Needless to say, such engineering is very challenging, but I think Toyota for example has traditionally nailed it, making cars that have room for 200 000km in the maintenance logbook (so this is clearly the "design" lifetime), actually last for 350000-400000km with minor maintenance, then finally starts to rust through everywhere approximately at the same time. (Don't know if that's true anymore. Was true for my 1988 model which I finally got rid a year ago.)

Only few can do it.

For the rest of us, we need to calculate in larger margins. And be prepared to still have issues of premature failure. The key is not to change too much all the time, but learn from the previous product generations: just fix things that were failing early by looking at warranty repair statistics, or customer feedback if it's a throwaway product.

What many laymen and even some engineers consider "good engineering" is often a disaster consisting of some parts massively oversized to last for 1000 years, while some parts require constant maintenance. People who respect products like that like maintenance work; they think it's a good feature that "you can repair the product", "it's not a throwaway junk", "look how sturdy it is" (by looking at some overengineered visible part). What they miss is that the product would be much better if it was engineered not to require repairing at all! Being able to fix the broken piece of junk makes them happy, but majority of people will just throw it away and buy a new one if it's something say less than $1000-$2000 in value.

I worked in the automotive industry for many years, for my sins.  Including in QA.  Basically, a car (and every component that goes into it) is rated and tested for 100,000 miles and 10 years, in specific "worst case" environmental conditions.  This is what all the component makers have to contractually fulfill - and woe betide a supplier that doesn't meet the quality criteria (which includes corrosion resistance).  (GM would have had words (and money) from the supplier of the Oldsmobile door handles mentioned above, if the failure rate was above contracted levels.)

So your observation is 100% correct - cars ARE designed to have every component fail at the same time -  but in the worst case conditions of heavy use and corrosive environment!  This requirement means that most cars can last 200K miles or even more, if taken care of (kept clean, oil changed, etc.).

It also means that if you buy a used car with more than 100K miles or 10 years on it, you are buying a vehicle that is past its designed end of life...   not that you would know, based on used car prices...

[TimFox]

With respect to all subsystems failing simultaneously, the classic account is an 1858 poem by Oliver Wendell Holmes, Sr (1809 -1894), American physician and the father of Oliver Wendell Holmes, Jr who served on the US Supreme Court.
http://holyjoe.org/poetry/holmes1.htm
The "One Hoss Shay" is a carriage built by a deacon who knew that a machine failed at its weakest point, and therefore avoided having any weak points in the construction.  Economists use the term to denote a depreciation calculation where the depreciated asset has no scrap value.