I think it depends on what hardware you look at.
Eg, early PCs were expensive and generally of good build quality. However, the very early ones had under-specced and unreliable PSUs. Replace the PSU though, and it can last a long time.
In the late 286 era/386/486 era, cheap Taiwanese clone hardware flooded the market. A lot of it was quite buggy and unreliable. But a 'good' 286/386/486 machine from a high-end brand such as IBM or Dell would probably last forever.

I guess in the Pentium era things started to stabilize more. Chipsets became more integrated, which in turn made motherboards simpler, and therefore more reliable.
But then there was the problem with poor caps. In the search of ever cheaper components, many suppliers chose cheap Chinese caps. These were built using a reverse-engineered formula for the chemical component, which was flawed. High-end boards used expensive Japanese caps, which were very reliable. The others were prone to burn out in a few years.

Once that stabilized, RoHS happened... lead was banned from solder. The replacements were not as 'elastic' in nature, causing soldered joints to crack from thermal stress (repeated warming up and cooling down, worsened by aggressive power management).

But that is under control as well now. I think we're in a reasonably reliable era currently.

I also guess that the older the hardware is, the more 'bad apples' have already been weeded out. As in: the bad hardware broke down and was thrown away already. All that has survived has to be pretty reliable by default.

Scali wrote:

I also guess that the older the hardware is, the more 'bad apples' have already been weeded out. As in: the bad hardware broke down and was thrown away already. All that has survived has to be pretty reliable by default.

That is a good observation to explain why the sentiment "back in the day, things were built to last" is uttered.

Scali wrote:

I think it depends on what hardware you look at. Eg, early PCs were expensive and generally of good build quality. However, the […]
Show full quote

I think it depends on what hardware you look at.
Eg, early PCs were expensive and generally of good build quality. However, the very early ones had under-specced and unreliable PSUs. Replace the PSU though, and it can last a long time.
In the late 286 era/386/486 era, cheap Taiwanese clone hardware flooded the market. A lot of it was quite buggy and unreliable. But a 'good' 286/386/486 machine from a high-end brand such as IBM or Dell would probably last forever.

I guess in the Pentium era things started to stabilize more. Chipsets became more integrated, which in turn made motherboards simpler, and therefore more reliable.
But then there was the problem with poor caps. In the search of ever cheaper components, many suppliers chose cheap Chinese caps. These were built using a reverse-engineered formula for the chemical component, which was flawed. High-end boards used expensive Japanese caps, which were very reliable. The others were prone to burn out in a few years.

Once that stabilized, RoHS happened... lead was banned from solder. The replacements were not as 'elastic' in nature, causing soldered joints to crack from thermal stress (repeated warming up and cooling down, worsened by aggressive power management).

But that is under control as well now. I think we're in a reasonably reliable era currently.

I also guess that the older the hardware is, the more 'bad apples' have already been weeded out. As in: the bad hardware broke down and was thrown away already. All that has survived has to be pretty reliable by default.

Thanks for this useful info. Which exact boards were the low quality Taiwanese models?

GuyTechie wrote:

Scali wrote:

I also guess that the older the hardware is, the more 'bad apples' have already been weeded out. As in: the bad hardware broke down and was thrown away already. All that has survived has to be pretty reliable by default.

That is a good observation to explain why the sentiment "back in the day, things were built to last" is uttered.

Except for the good share of still unopened retro hardware. 😉 Some bad duds are still waiting to be discovered, so watch out.

computergeek92 wrote:

Some say Pentium 1 systems can resist ESD versus later platforms, lower heat buildup increases lifespan for any computer, Athlons ran too hot, many millennial mobos had bad caps. Despite this and any other positive or negative hardware claims, the old stuff was said to be more reliable in general. What do you think was the most reliable and longest running without failures? I'm guessing 386 and 486 systems are the most reliable.

Scali wrote:

I think it depends on what hardware you look at.
Eg, early PCs were expensive and generally of good build quality. However, the very early ones had under-specced and unreliable PSUs. Replace the PSU though, and it can last a long time. [..]

I couldn't have said it better. As far as I know, very old SRAM chips were very sensible.
They were destroyed very easily by ESD. Mr. Wozniak once said something similar in an old Apple II video.

On the other had, other TTL ICs of that same era were one of the most durable and approved ICs of all times.
Like the NE555 timer chip, the LM386 amplifier or 780x/790x voltage regulator series.

There are also other things to consider. Like fabrication process and integration density/structural size.
Ancient processors like the 8088/286 or Z80 were initally fabricated using NMOS technology.
This technology contained resistors and was quite hotheaded (in comparison to CMOS).
Yet still, it was very reliable and well understood. Same for TTL. It is ancient, but still needed.
The NE558 for example, has also a CMOS variant. So is the TTL version still needed ?
Yes, because the older model can still operate at higher frequencies.

It wasn't a joke that the Space Shuttle had 8086 processors.
The larger structural size and higher operation voltages made these chips more resistend
against cosmic radiation. In case of a solar flare, these things and electron tubes have a higher chance
to withstand than our modern, yet overbred technology.

Every generation of hardware had its failings, if it wasn't something like shitty power supplies or rotting caps it was always something else like the barrels of cancer during the 386/486 era that leaked eating away at everything. Modern stuff fails more often because too many corners are cut quality wise while being so much more complex. Doesn't help that the skill level of builders these days is lower than past eras as they let things like vrms get far too hot for far too long so they end up losing $500-800+ dollar graphics cards or doing stupid things that kill their SSDs ect. Yea those cheap planar TLC nand SSDs are cheap for a reason....

Personally I've settled into playing with older hardware and keeping things to a minimum when I need to deal with something modern like a new build ect. I don't like modern laptops much and despise the shitty ultra thins!

As far as most reliable computers , anything with a fully dipped pcb like military grade computers for fighter jets,missiles , ect will essentially be 100% reliable.

On the comsumer level , automotive computers from the late 80s to early 90s will be some of the oldest computers still in regular use for 25 to 30 years without failing very much.

as far as comsumer desktops anything with no moving parts from the 80s will be more reliable than more complicated ones. Single board computers like the commodore and zx spectrum. Ect...

@nforce4max Yup, but what do we expect ? Nowerdays, electronics are cheaper than food.
Back in the day, someone had to sell his car for a computer.

Quality products didn't wanish. We just don't see them anymore.
Because we look for cheap things and don't want to spend the same money we were once used to.

Jo22 wrote:

The NE558 for example, has also a CMOS variant. So is the TTL version still needed ?
Yes, because the older model can still operate at higher frequencies.

Actually the CMOS 558 ICs are much faster than the TTL versions as well as being lower power. They can't output as much current as the TTL versions though, but that's easily worked around if it happens to be a problem.

Jo22 wrote:

@nforce4max Yup, but what do we expect ? Nowerdays, electronics are cheaper than food. Back in the day, someone had to sell his […]
Show full quote

@nforce4max Yup, but what do we expect ? Nowerdays, electronics are cheaper than food.
Back in the day, someone had to sell his car for a computer.

Quality products didn't wanish. We just don't see them anymore.
Because we look for cheap things and don't want to spend the same money we were once used to.

This... As much as it hurts, is true.

Although there's a small big fail in this reasoning. Many products being sell at hq-professional price come with shitty electronics. Just check the last "pro" notebook lines from Dell/HP/Lenovo or the HQ dell displays with these fucking Lelon caps in the power supply, or dry solder joints, etc... Like any Acer crap. That's not fucking fair, because they lie about quality, charge a premium price, and you can't tell the scam until you open the damn thing, killing the warranty along with the seal sticker.

And isn't only about computer hardware... There are zillions of posts in EE forums about how shitty are the electronics of the last Keysight (Agilent) stuff, etc...

lissajous wrote:

Jo22 wrote:

The NE558 for example, has also a CMOS variant. So is the TTL version still needed ?
Yes, because the older model can still operate at higher frequencies.

Actually the CMOS 558 ICs are much faster than the TTL versions as well as being lower power. They can't output as much current as the TTL versions though, but that's easily worked around if it happens to be a problem.

My bad. That's what you get if you're using +20 years old catalogues. 😅

hyoenmadan wrote:

Many products being sell at hq-professional price come with shitty electronics. Just check the last "pro" notebook lines from Dell/HP/Lenovo or the HQ dell displays with these fucking Lelon caps in the power supply, or dry solder joints, etc... Like any Acer crap. That's not fucking fair, because they lie about quality, charge a premium price, and you can't tell the scam until you open the damn thing, killing the warranty along with the seal sticker.

And isn't only about computer hardware... There are zillions of posts in EE forums about how shitty are the electronics of the last Keysight (Agilent) stuff, etc...

Agreed. Maybe small companies (family businesses ?) do still care about quality and for their customers.
At least in the Hi-Fi market they should be still existing.

If you need a good latop, perhaps you can get a second-hand "Thoughbook".
They are said to be very rugged and long-living.

The one without Super Socket 7, capacitor plague, ISA sound cards and leaky batteries.

computergeek92 wrote:

GuyTechie wrote:

Scali wrote:

I also guess that the older the hardware is, the more 'bad apples' have already been weeded out. As in: the bad hardware broke down and was thrown away already. All that has survived has to be pretty reliable by default.

That is a good observation to explain why the sentiment "back in the day, things were built to last" is uttered.

Except for the good share of still unopened retro hardware. 😉 Some bad duds are still waiting to be discovered, so watch out.

LGR opened up a "new in box" CD-ROM drive or multimedia kit (if I recall) when he was building his retro PC and it ended up not working, 🤣. So yes, dud. It's possible the belt dried out while still sealed in the box.

computergeek92 wrote:

Some say Pentium 1 systems can resist ESD versus later platforms, lower heat buildup increases lifespan for any computer, Athlons ran too hot, many millennial mobos had bad caps. Despite this and any other positive or negative hardware claims, the old stuff was said to be more reliable in general. What do you think was the most reliable and longest running without failures? I'm guessing 386 and 486 systems are the most reliable.

Early slot A and socket A non-XP Athlons did run hot - but they ran OK even at 60C. Nobody seems to mention 478 machines tough. They always ran hot, were noisy and required expensive power supplies - at least anything over 2.6 ghz did.

Reliability wise I'd have to say xt/286/386/486 era gear. These things are seemingly indestructible. Some socket 7 boards as well - but as mainboards and video cards became more complicated and power hungry, they became less reliable and more fragile.

The most expensive boards of the era (high end workstation , embedded industrial, and server boards) were the most reliable. They also had the best chipsets, BIOS ,and voltage regulation.

Around the time of the Pentium MMX (dual voltage CPU) and the later SS7 boards you had quite a few venders making sometimes good and sometimes crap boards along with chipset and BIOS that were hit or miss. Around this time you also started seeing motherboard companies go under. It didn't help that you had IDT, AMD, Intel, Winchip , IBM/Cyrix chips all being used on the same boards. Competition made prices drop like a rock and so did design, fudge factor, and reliability. You don't expect a $500 computer to be as reliable as a $2000 computer.

I think board design software got good enough that engineers could reliably design something that barely worked over the warranty period and then the bean counters started taking stuff away to save a penny. I never seen a 286/386/486 era board have spots for capacitors being empty like you did in the early Athlon age.

You may well say so! Do you remember the times when PC chassis were made of steel, not just aluminium ? 😉

I tend to trust old PSUs from 15 years ago more than modern ones. I know this sounds crazy, but I've had more luck and reliability using power supplies salvaged from old Pentium 4 Dells, for example. I've tried brand new modern PSUs, but I’ve had problems like unstable rails on a new Seasonic, and ESD from the cables of a black painted Corsair PSU. (PSUs are supposed to be unpainted to ensure proper grounding) Old power supplies are all over the place at PC recycling centers. Salvage up the old 10+ year units while you can!

Sure the black paint wasn't conductive ?
I mean, someone should think that whoever has the ability to produce such complicated PSUs should
also have the sources for something simple as conductive paint..

Jo22 wrote:

Sure the black paint wasn't conductive ?
I mean, someone should think that whoever has the ability to produce such complicated PSUs should
also have the sources for something simple as conductive paint..

I know right? Guess companies don't really care much these days. More about making money. I talked to my friend and he also agrees that good power supplies should be unpainted. It even says so in static precaution references to touch the unpainted metal of your chassis, especially the power supply, to ground yourself. Many computer cases today are painted all over. Now combine that with a painted PSU.