[lgleason]

Repairability would help as well. Many times the only viable option to fix something is to swap a board, or replace the entire item, instead of replacing the one failed component that caused the board to fail, or reflowing the board etc.. Many components also do not offer batteries that can be replaced, such as the magic mouse, so you end up needing to replace the entire item.

It's interesting how as certain things age, such as cars, cottage industries pop up to do just that when new replacement boards and parts are not available.

The other issue is cost cutting. Many components are made cheaply and fail pre-maturely. Great examples of this are mains voltage LED bulbs where the rectifier circuits that power the LED's fail, but the only real option is to replace the entire thing, creating a lot of e-waste in the process.

[lefra]

Changing the PCB for a known-good one: $10 + maybe half an hour of low-skill work.

Changing the failing component: maybe a few minutes, probably a few hours of an electronics engineer that's familiar with the design (plus his expensive tools). He's probably bad at soldering, so you'll need someone else to do that. Then you need to revalidate the board.

It almost never make economical sense to try to repair the board.

[bigyabai]

If you have a surplus of donor components, board-level repair can be very feasible and often even profitable depending on the board.

[AngryData]

If we were provided board and part diagrams it might be worth it because then you don't need an actual engineer or super highly knowledgeable person to waste a few hours of time just to diagnose most problems. But because we lack such diagrams whoever is diagnosing it also has to reverse engineer how it works in their head.

[duped]

The fact that we tolerate creating waste because it's "economical" is frankly disgusting. The disposal fees for e waste should make it uneconomical to dispose of boards.

Also training techs to repair SMD parts is really easy and cheap, you're grossly overestimating the costs. The real waste comes from boards with designs that can't be repaired so we tolerate a certain yield. For many small devices the yields are shockingly low.

The other thing is that yields are low because of bad designs. If it became uneconomical for you to throw half your boards out then designers would fix their crappy boards with tombstoned jellybean parts because they used shitty footprint libraries. This is a solvable engineering problem and it's gross that it's cheaper to throw shit into a landfill instead of fixing it.

[oulipo2]

Exactly! That's what motivated us to design a repairable e-bike battery at https://infinite-battery.com

[kube-system]

I'm all for repairability, but as labor costs go up and manufacturing costs go down, the window for which there is incentive to repair narrows.

e.g. there's no amount of repairability design that you could apply to a $3 light bulb which would encourage people to pay someone western wages to repair. I think we're better off lobbying for better standards to communicate the quality of a bulb's design. The whole reason we have crappy LED bulbs to begin with is because the $3 overdriven bulb with crap components jammed into a tiny enclosure looks like a better deal on the shelf than a bulky $20 bulb with a large heat sink and lower output.

And the labor required to do component level repair is wildly expensive and limited (YouTubers who do it on principle notwithstanding), even further narrowing that window.

[AngryData]

If you could disassemble and diagnose a failing $3 bulb in 60 seconds, you wouldn't need to hire someone at western wages to fix it. But because it is glued together to not be taken apart, and there are no diagrams for how anything in it works or is put together, it isn't worth the time even if you have a station and equipment all ready setup and replacement component on hand. 95% of the time fixing electronics is just figuring out how they were put together in the first place so you can diagnostically trace along the circuit.

Not that I think lightbulbs are probably worth saving, but expand it to any other device which gets exponentially more complex and it is easy to see why they don't get diagnosed, not to mention repaired. With a board diagram I can point at a spot on the board and say "I should see 15 volts here", without a board diagram i gotta draw out and figure out how the power supply even works so I know what it is suppose to be outputting and then trace that all the way to the test point to make sure there isn't other crap inline before then that might change what I see.

[Findecanor]

Human-scale engineering is underrated. It is very satisfying when you can repair something yourself using your hands, without having to need specialist equipment.

For example when you have a circuit board that can be serviced with a soldering iron, without having to use a microscope or reflow-oven.

[kube-system]

It is underrated in terms of personal satisfaction. It is overrated in terms of potential impact to municipal waste management.

[kerblang]

Try buying an LED flashlight.. when the LED circuitry/bulb goes out, the whole thing's a brick, battery, assembly, everything. You have to throw it all out. The bulb assembly is usually fused to the frame so that it's hard just to recycle that frame.

[AlexandrB]

Seeing LED bulb reliability rapidly degrade as the technology matured was like seeing the Phoebus Cartel[1] play out in real time.

[1] https://en.wikipedia.org/wiki/Phoebus_cartel

[MostlyStable]

There was a cartel, but this is one of those "more complicated than it appears" situations. In incandescent bulbs, there is a real tradeoff between durability of the bulb, efficiency (lumens/watt), and brightness/quality of the light, for physics reasons you _can't_ improve one without degrading the other.

Since "quality of light" is a very difficult thing to market, there was an incentive to push "lifetime of the bulb" in marketing and just make the light quality increasingly worse. The cartel attempted to halt that by making everyone agree on a lifetime/quality to hit and not participate in a race to the bottom (and yes, there was also the obvious benefit to the cartel members of increased sales and profits, which they explicitly talked about in internal documents).

I want to be very clear that I'm anti cartels and I'm not trying to say "so this was all hunky dory", just that this was not (and these things very rarely are) a simple case of "they made the product objectively worse for the sole sake of more money". Instead, they chose a different point on the pareto-frontier of brightness/efficiency/lifespan that also had the benefit of making them more money.

But yes, LED bulbs are currently mostly garbage and have terrible heat/power management electronics which means that in practice you almost never get anywhere close to the theoretical life span increases (because the electronics die from overheating far before the actual LEDs themselves would go out), and finding out information on how well a given bulb brand does on heat/power management is essentially impossible.

[coryrc]

I wish we had a Phoebus Cartel to enforce an expected 50k-hour lifetime.

[schmidtleonard]

TechnologyConnections debunked the Phoebus Cartel a while ago.

tl;dw incandescent bulbs can be made more efficient and brighter by running them hotter, but this reduces the lifetime. The obvious Nash Equilibrium involves increasingly hot/bright/efficient bulbs and as much lying about lifetime as a typical consumer would accept, which is a lot. The idea behind the Phoebus Cartel was to force honesty on the dimension where it was most likely to disappear. You are free to disapprove of this and reject bulb lifetime policing, but if so you support the "everybody lies" alternative. Pick your poison.