Why it looks like planned obsolescence not applicable at PC mobo ?

[BravoV]

When I searched stuff in my pile of junks, I happened to spot an old friend, pulled it out and its an old OC-ing mobo for early 1st gen Intel Core CPU, the Asus P5B Deluxe, and just for curiosity and giggle, I tried power it and its still working just fine when I test boot with memtest from USB stick and passed the test just fine.



And also briefly tested few of my others, especially with polymer caps for the CPU's VRM, or mostly overclocking boards, all working fine, while very-very old ones with electrolytic caps, I have some dead as they're bulging, apart from that problem, which I think is easily fixed, still many of them are still working just fine, even the one that use slot CPU.

Yes, even though my sample size is very limited, I guess they're pretty reliable, and will last for very long time.

What do you think ? Too complex and too risky to implement the planned obsolescence  ?

[greenpossum]

Easier to do it with software (bloat).

[BravoV]

Easier to do it with software (bloat).

LOL .. may be you're right. 

[Haenk]

I have a CoreDuo-based Laptop as fallback-solution (aka "just in case everything else fails"), 2GB RAM and a 120GB SSD. With Windows 10 running on it. It's slow, but considering the little RAM and slow processor, still faster than expected.
Funny enough, the installation was only possible as a 2-step upgrade path (7->8->10), for whatever reason 8 or 10 won't install directly.
Unfortunately, using older mainboards/CPUs (like from the CoreDuo-age) as a daily driver is not that great, as the power consumption is pretty high, the performance pretty slow. You'd be better off with one cheap current AMD combo (which will pay by itself soon by the power costs saved...)

[dolbeau]

What do you think ? Too complex and too risky to implement the planned obsolescence  ?

Solid-state circuits are fairly long-lived, and are pretty much the only thing on a motherboard. The one thing that ages significantly, as you noticed, are some capacitors. Of course metal can corrode, but it's an issue even short-term so there's usually a protective layer. Unless there's physical damage, motherboards/CPU/memory can last a *long* time. Just replace the batteries :-) (usually for the RTC or some non-volatile memory).

I own a still working SPARCstation 330GX from 1989, everything works fine when network booting - even though the internal 5"1/4, full-height SCSI drive died decades ago (and the NVRAM battery is dead, but it's not a fatal flaw). I have a lot of hardware from the 90s and early 00s, disks are pretty much the only recurring issue except when there was some design flaw and the occasional fried circuit (more often than not due to misuse).

[rdl]

I still have my first PC and as far as I know it still works. Last time I turned it on was in 2017 though. I had converted it over to use Compact Flash instead of the original HDD which was somewhat noisy. I was planning a power supply upgrade so it would be completely silent, but I got side-tracked.



(click for bigger)

[David Hess]

Except for aluminum electrolytic capacitors that wore out and had to be replaced, my motherboard failures have come down to:

1. Motherboards which used embedded lithium batteries for CMOS and clock - once the backup battery ran down, the CMOS became inaccessible so the motherboard would no longer boot.  Asus P/I-P55T2P4, I am looking at you.

2. I suspect the NVIDIA nForce2 chipset had a design flaw which placed a low voltage transistor on a high voltage I/O pin like the Intel C2000 Atom processors.  The result was failure after a couple years of operation.

[Siwastaja]

Planned obsolescence using electrolytic capacitors is largely a myth. It is too difficult to do properly. Usage conditions are so different. Assuming each 10degC difference doubles or halves the life, it must be designed to last at least a year or two in worst-case 1% conditions to avoid massive amount of returns. Additionally, capacitor aging isn't a carefully controlled manufacturing parameter as that would require better process control, and increase cost, just the opposite what they are doing. Using "cheap" caps requires more leeway in design, making planned obsolescence even more difficult.

Obviously, your motherboard that has sat years unused have not seen high temperatures. Put it in a poor man's non-temperature controlled "server" closet and run for 5 years and the result might be different.

With software, planned obsolescence is so much easier.

[SilverSolder]

Planned obsolescence using electrolytic capacitors is largely a myth. It is too difficult to do properly. Usage conditions are so different. Assuming each 10degC difference doubles or halves the life, it must be designed to last at least a year or two in worst-case 1% conditions to avoid massive amount of returns. Additionally, capacitor aging isn't a carefully controlled manufacturing parameter as that would require better process control, and increase cost, just the opposite what they are doing. Using "cheap" caps requires more leeway in design, making planned obsolescence even more difficult.

Obviously, your motherboard that has sat years unused have not seen high temperatures. Put it in a poor man's non-temperature controlled "server" closet and run for 5 years and the result might be different.

With software, planned obsolescence is so much easier.

I have seen many PCs run 24/7 for more than a decade - they are surprisingly reliable, generally.

[T3sl4co1l]

The low operating temperature from a combination of fans blowing air everywhere, and not generally running at ratings at all times (most people aren't f@h, cryptomining, etc.), and also either intermittent use (shorter "engine hours" so to speak), or maybe reduced temp cycles from always-on operation, helps out.

AFAIK, polymers are just as prone to failure as electrolytics, they just tend not to run as hot.  Both can last quite a long time when not abused.  And the luck of the draw, and the confirmation bias where you forget about the ones you toss out, while the ones that stick around are there to actively remind you.

From personal experience, I forget what earlier mobos/PSUs died of; one mobo, the caps split open, that was pretty obvious.  Forget what PSUs were, if it was fan death leading to smoking electronics, or caps again.  Think the last one or two were at least fan related.  My current PSU seems to be pretty reliable besides the fan, which I'd already caught twice running out of lubrication and making noise and stalling; finally, earlier this year, I put in a (brand name, ball bearing) replacement which should outlast everything else pretty significantly.

Tim

[NiHaoMike]

Most PC motherboards other than the very cheapest are designed to run the highest end CPU that fits in the socket with room for overclocking. If you run something lower and don't overclock, it would be reasonable to expect it to last very long.

And the luck of the draw, and the confirmation bias where you forget about the ones you toss out, while the ones that stick around are there to actively remind you.

That's the myth that older stuff lasts longer than its modern equivalent, and while there are cases where it is indeed true, more often the ones that break down early are forgotten. Not to mention the old stuff generally was a lot more expensive, especially after factoring in inflation. What deserves criticism is that modern stuff often are not designed to be easily repaired.

[Siwastaja]

Also if you have a sample set of 20 motherboards and observe MTBF of 10 years, they are not all MTBF of 10 years, it's likely the few that failed had MTBF of just a few years and the rest are very good.

[PKTKS]

1. Motherboards which used embedded lithium batteries for CMOS and clock - once the backup battery ran down, the CMOS became inaccessible so the motherboard would no longer boot.  Asus P/I-P55T2P4, I am looking at you.

Hmm .. that recalls me a really bad experience with those imbeciles
or just retarded folks which decided to embed CMOS clock in MOBOs.

Early 90s I paid a small absurd value for somewhat early  pentium kind MOBO...
After a fairly small time the battery dried out.. and it was almost the cost
of a brand new MOBO just to replace that cretinous chip.

At that time proper SMD  solder equip. was pretty much a fortune
to get or just unavailable outside industry facilities.

I swear that I never ever ever again would buy such cretinous piece of shit...
such market  decisions are  like shooting the foot..
consumers are not imbeciles.. they walk away

The time consumed  not to just replace the MOBO but actually to
setup *everything* again  on that old hardware has no way to explain...

5 inch flops and 10M HDs are no fun.

Paul

[james_s]

I can just about count on one hand the number of computer parts I've had fail in normal operation. Almost always they were replaced due to technological obsolescence and still worked fine when they were retired. Overall I think planned obsolescence is largely a myth. Things aren't designed to fail, they're designed down to a cost and made well enough to last at least through the warranty. If a part was used that cuts the average lifespan in half, it's because the part was cheaper and would still outlast the warranty in most cases, not because they want to ensure the device fails.

[rdl]

I'm a pack rat so I still have almost every PC that was for my own personal use (one was disassembled). They were all fully functional when put into storage. All but the first two I assembled myself and each one was in in daily use one way or another for at least 10 years. The only problems over that time were 4 or 5 power supply failures, 2 HDD failures and one stick of RAM that went bad. Other than those PCs, I also have a couple of HP Microservers in use as NAS. Both are pushing 10 years of problem free usage.

[Doctorandus_P]

I haven't saved all my old PC's. Don't even remember which processors they had. My best guess:

1. 80386SX
2. 80386DX, (Later upgraded with Cyrix Coprocessor)
...
5. Some AMD with such a big slotted cartridge.
...
97. P4
98. E6600, Passmark rating 1500
99. i7 860

The E6600 I got for free, It was a leftover after a family member died of sickness, and this PC lasted for around 10 years. Bios date is from 2008, and the PC stopped working recently, and the fault turned out to be the power supply (which was already changed once). With another power supply I had lying around it chugged along where it left me.

The i7 I also got for free. A friend gave it to me, and he got it from a colleague at work. It's got a passmark of around 2900, is twice as fast as the E6600, also has faster video and 8GiB of RAM instead of 4GiB. It probably also cost more 10 years ago... When I got it a few weeks ago (around 2020-08-20), the power supply looked like it had already been replaced (different brand, which is also sold in local computer shops) and it had a 2.5" HDD, which I replaced with an SSD I had lying around, and it's now my daily workhorse.
It is a slow PC for today's standards. The slowest / cheapest which can be considered a "normal" pc (not those atoms with onboard processor) is about an AMD 3000G, and this EUR55 processor has a Passmark rating nearing 5000. I probably would have bought something like that if I had not gotten the i7 Freebie.

[LootMaster]

https://www.eevblog.com/forum/repair/pre-emptive-capacitor-change-on-beloved-laptop/msg3284742/#msg3284742

was searching for "obsolescence"...

Does anybody have confirmations on good recent gen laptops with pristine design intent?

[golden_labels]

BravoV:
Because “planned obsolence” is usually⁽¹⁾ a myth: a mix of conflation of normal optimization with malicious intent, ignorance about operation of the devices, being oblivious of forces that shape them and a bunch of fallacies. As such it can’t apply to anything.

PC motherboards are not produced near the point at which it becomes unprofitable, they have plently of room and proper thermal design, no extreme miniaturization, there are only 2 or 3 complex chips that could easily fail in the first place, they deal with mostly digital signals with little transients. If there is no failure at manufacturing, there is little that may fail later.

Get a 740cm² ATX mobo with CPU, RAM and chipset on it, scale it down onto a 1cm² ceramic substrate with no cooling and then ask why the former is failing less often than the latter.
____
⁽¹⁾ There certainly are companies, who try to cheat like that, but nearly everything attributed to “planned obsolence” is easily explained otherwise. And there really are better methods to scam consumers.

[tooki]

BravoV:
Because “planned obsolence” is usually⁽¹⁾ a myth: a mix of conflation of normal optimization with malicious intent, ignorance about operation of the devices, being oblivious of forces that shape them and a bunch of fallacies. As such it can’t apply to anything.

PC motherboards are not produced near the point at which it becomes unprofitable, they have plently of room and proper thermal design, no extreme miniaturization, there are only 2 or 3 complex chips that could easily fail in the first place, they deal with mostly digital signals with little transients. If there is no failure at manufacturing, there is little that may fail later.

Get a 740cm² ATX mobo with CPU, RAM and chipset on it, scale it down onto a 1cm² ceramic substrate with no cooling and then ask why the former is failing less often than the latter.
____
⁽¹⁾ There certainly are companies, who try to cheat like that, but nearly everything attributed to “planned obsolence” is easily explained otherwise. And there really are better methods to scam consumers.

Well said. Unfortunately most people prefer to wallow in their conspiracies, insisting there is malicious intent where none exists. They want quality that costs $2000 to make, but are only willing to spend $200. And yet they don’t understand that this requires cost optimization...

[SilverSolder]

BravoV:
Because “planned obsolence” is usually⁽¹⁾ a myth: a mix of conflation of normal optimization with malicious intent, ignorance about operation of the devices, being oblivious of forces that shape them and a bunch of fallacies. As such it can’t apply to anything.

PC motherboards are not produced near the point at which it becomes unprofitable, they have plently of room and proper thermal design, no extreme miniaturization, there are only 2 or 3 complex chips that could easily fail in the first place, they deal with mostly digital signals with little transients. If there is no failure at manufacturing, there is little that may fail later.

Get a 740cm² ATX mobo with CPU, RAM and chipset on it, scale it down onto a 1cm² ceramic substrate with no cooling and then ask why the former is failing less often than the latter.
____
⁽¹⁾ There certainly are companies, who try to cheat like that, but nearly everything attributed to “planned obsolence” is easily explained otherwise. And there really are better methods to scam consumers.

Well said. Unfortunately most people prefer to wallow in their conspiracies, insisting there is malicious intent where none exists. They want quality that costs $2000 to make, but are only willing to spend $200. And yet they don’t understand that this requires cost optimization...

Against that, most things do have a design life...   car components are made to a specification of 100K miles, 10 years - in the worst case environment.  So, in a friendly environment, they last longer, sometimes much longer.

Another way of looking at "design life" is that obsolescence / scrapping has indeed been baked in from the very beginning.

You can fight back by taking good care of your stuff, though.   That's the one remaining way of beating the system!   

[tooki]

Except that design life (“needs to last at least X amount”) isn’t the same thing as planned obsolescence (“it needs to fail after X amount”). As you said, companies don’t want design life to be too short or they risk too many units failing under warranty. So the design life needs to be chosen such that it intersects  only the very tail end of the early-failure side of the bell curve. Consequently, most units will last longer than needed.

The other thing that goes into the equation is the expected useful life of the product. Sure, one could design a cellphone that lasts 20 years without needing so much as a battery swap. But it’d cost a lot more, weigh more, and be bulkier. And to what end, since people replace their phones long before they fail.

[golden_labels]

Another way of looking at "design life" is that obsolescence / scrapping has indeed been baked in from the very beginning.

It’s baked in into the universe from the very begining. It not only older than humans, but has created them. It’s called entropy.

[SilverSolder]

Except that design life (“needs to last at least X amount”) isn’t the same thing as planned obsolescence (“it needs to fail after X amount”). As you said, companies don’t want design life to be too short or they risk too many units failing under warranty. So the design life needs to be chosen such that it intersects  only the very tail end of the early-failure side of the bell curve. Consequently, most units will last longer than needed.

The other thing that goes into the equation is the expected useful life of the product. Sure, one could design a cellphone that lasts 20 years without needing so much as a battery swap. But it’d cost a lot more, weigh more, and be bulkier. And to what end, since people replace their phones long before they fail.

I'm cynical enough to think that "they would if they could".   Cloud products are the perfect example of how a product can be made to "fail" instantly if you don't pay up (or the company goes bust, or your nation has bad relations with the USA, or whatever!). 

That said, if nothing was ever scrapped, we would still be driving Model T's today...  obviously there has to be a limit to the lifetime of many (most?) products to allow evolution to take place...

[golden_labels]

I'm cynical enough to think that "they would if they could".   Cloud products are the perfect example of how a product can be made to "fail" instantly if you don't pay up (or the company goes bust, or your nation has bad relations with the USA, or whatever!). 

Yet another pain of the concept of “planned obsolescence”: pulling anything that has even the most remote connection with expiring, in any possible meaning of that word, into that term.

I’m not sure what you are referring to with “cloud products”. If cloud VPS solutions: it’s no different than non-cloud VPS in the scope discussed here. The difference is in the implementation. If you ask someone else to handle something for you, it’s not planned obsolescence if they refuse to provide service after you stop paying. You pay for the specified period of time, they do their job, period. If IoT devices: those are usually not products, but services with gear for which you pay once. Not different from getting a modem from your ISP to access internet or a credit card from your bank. While I’m myself not happy with products being replaced with services, it’s not planned obsolescence that service ceases if you do not pay or company goes out of business.

[SilverSolder]

I'm cynical enough to think that "they would if they could".   Cloud products are the perfect example of how a product can be made to "fail" instantly if you don't pay up (or the company goes bust, or your nation has bad relations with the USA, or whatever!). 

Yet another pain of the concept of “planned obsolescence”: pulling anything that has even the most remote connection with expiring, in any possible meaning of that word, into that term.

I’m not sure what you are referring to with “cloud products”. If cloud VPS solutions: it’s no different than non-cloud VPS in the scope discussed here. The difference is in the implementation. If you ask someone else to handle something for you, it’s not planned obsolescence if they refuse to provide service after you stop paying. You pay for the specified period of time, they do their job, period. If IoT devices: those are usually not products, but services with gear for which you pay once. Not different from getting a modem from your ISP to access internet or a credit card from your bank. While I’m myself not happy with products being replaced with services, it’s not planned obsolescence that service ceases if you do not pay or company goes out of business.

Fair point, there isn't much use in collecting all our gripes in one bucket.

Having now been forced to actually think about planned obsolescence, I'm back to thinking that there isn't much difference between the concept of a design life (e.g. a car's 10 years / 100K miles) and planned obsolescence.   Obsolescence isn't the same thing as "self destruction" - it just means it is no longer expected to work and is no longer supported.

If there is a design life of 10 years, then the product is by definition obsolete after 10 years - in the case of cars,  the dealers will no longer carry spare parts for it, for example - you will have to buy components from third parties who make their money from extending the life of vehicles that are beyond their design life.

I have worked on projects where the product had a design life of as little as two years.  Nobody called it planned obsolescence - but that's what it was, because the company also planned the introduction of the next product to take over when the old ones reached end of life.

The key word here is "planning".

You could argue that if a product doesn't have planned obsolescence baked in, you will be dealing with something even worse -  Unplanned obsolescence!   ...  which makes it harder to design and make a product, because nobody will know how "good" it is supposed to be, and different people and suppliers could make different decisions.

Entropy is always with us, on different levels - we either allow for it and plan for it in our products, or...   what?