Engineering Failures - Report Ideas

[wraper]

Found that they lacked PI layer.
Nvidia wiithout that layer:

Some chip with it:

[jeremy]

Wow that's really interesting, I would have never guessed that. Thanks wraper! So it's still a thermal expansion mismatch, but the problem was clearly worse at the die rather than at the PCB.

[Circlotron]

HiSense TV sets.   

[Tris20]

Can't believe I didn't think of this before, but the XBOX 360 Red Ring of Death.

Basically caused by bad head dissipation which lead to the motherboard warping and some of the chips subsequently become detached. Some known fixes (albeit temporary) were to wrap a towel around the 360, force it to over heat to the point it turns itself off and then you've effectively reached a temperature that melts the solder and re-attaches the chips to the board! 

[scientist]

The Mars Beagle lander that converted miles directly into kilometers. It deployed the parachutes... minutes after impacting the ground. 

[wraper]

Can't believe I didn't think of this before, but the XBOX 360 Red Ring of Death.

Basically caused by bad head dissipation which lead to the motherboard warping and some of the chips subsequently become detached. Some known fixes (albeit temporary) were to wrap a towel around the 360, force it to over heat to the point it turns itself off and then you've effectively reached a temperature that melts the solder and re-attaches the chips to the board!

Do you seriosly believe in what you have wrote? Melt the solder? 217^oC?

[miguelvp]

Can't believe I didn't think of this before, but the XBOX 360 Red Ring of Death.

Basically caused by bad head dissipation which lead to the motherboard warping and some of the chips subsequently become detached. Some known fixes (albeit temporary) were to wrap a towel around the 360, force it to over heat to the point it turns itself off and then you've effectively reached a temperature that melts the solder and re-attaches the chips to the board!

Do you seriosly believe in what you have wrote? Melt the solder? 217^oC?

From wiki, so your miles may vary on how true this is:

German computer magazine c't blamed the problem primarily on the use of the wrong type of lead-free solder, a type that when exposed to elevated temperatures for extended periods of time becomes brittle and can develop hair-line cracks that are almost irreparable.

But there is no proof and that article referenced by wiki is gone as far as I can tell. (Edit: or maybe just hard to find and it was originally in German)

[wraper]

The real problem is similar to Nvidia I wrote about. That fix is temporary. Just connection between substrate and die reestablishes for a short time.

[miguelvp]

The real problem is similar to Nvidia I wrote about. That fix is temporary. Just connection between substrate and die reestablishes for a short time.

Looks that way, but it was 1 in 3 failure rate and on older models using a kinect but who knows what is true or urban legend

[GiskardReventlov]

That fix is temporary. Just connection between substrate and die reestablishes for a short time.

This  sounds to me just like the PS3 problem. I was hoping Dave would tear down that PS3 and we'd find out.

[TerminalJack505]

The Mars Pathfinder priority inversion issue would be an interesting engineering failure to write about. 

It isn't one of the more well-known failures.  This is most likely due to the fact that they were able to recover from the problem so it wasn't a complete and total failure.  Still, a very teachable sort of failure.

There are a several interesting engineering-related aspects of the Three Mile Island incident as well. 

I always thought the fact that the alarms couldn't be disabled was a huge engineering blunder.  Nothing like having a siren blare while you are trying to figure out why a nuclear reactor is melting down. 

Or, that there wasn't CCTVs inside of the reactor building so that they could actually see what was going on--they only had their instruments and best-guesses to go by. 

Or, the fact that it was just one failed part that caused the whole accident (a valve that failed to close.)

Another interesting one is the Hubble space telescope mirror.

[Richard Crowley]

News reports about our local nuclear power plant (Trojan, in Oregon*) indicated that the system was "scrammed" (emergency-shutdown) when a technician attempted to make a VOLTAGE measurement with a DMM, but had the leads in the CURRENT socket (shorting out the control system supply rail).

* Reputedly, that nuclear plant was the inspiration for the setting of "The Simpsons".

[scientist]

Outside of electrical flaws, there's this: http://eastwickpress.com/news/2011/07/a-mishap-at-the-beacon-power-frequency-flywheel-plant/

A giant carbon fiber flywheel exploded underground, blasting asbestos-like carbon fiber debris all over the town, while the plant operators claimed it was harmless carbon dust.

[David Hess]

failure in intel chipset that affect sata port,don't remember which chip.

I forgot about that one.

Even better would be their RAMBUS i820 chipset problems or the famous Pentium floating point division bug.  More obscure would be the Pentium 4 clock wall.

[Tris20]

Can't believe I didn't think of this before, but the XBOX 360 Red Ring of Death.

Basically caused by bad head dissipation which lead to the motherboard warping and some of the chips subsequently become detached. Some known fixes (albeit temporary) were to wrap a towel around the 360, force it to over heat to the point it turns itself off and then you've effectively reached a temperature that melts the solder and re-attaches the chips to the board!

Do you seriosly believe in what you have wrote? Melt the solder? 217^oC?

To be fair, I did this about three or four years ago and didn't pay too much attention to the details at the time. I just remember it being something along those lines. Overheating the xbox had a positive effect for a while anyway

The real problem is similar to Nvidia I wrote about. That fix is temporary. Just connection between substrate and die reestablishes for a short time

See the large bold statement above.


Apology accepted...

[Richard Crowley]

There are a several interesting engineering-related aspects of the Three Mile Island incident as well. 

And from the Fukushima Daiichi reactors which got hit by the tsunami and earthquake.
I watched an NHK documentary on YouTube and was struck by the same kind of oversights
as we saw at Three Mile Island (and doubtless at Chernobyl, as well).
For example, at Fukushima Daiichi

1) Gas-pressure-actuated relief valves which just assumed that the "ambient" pressure inside the containment vessel would "never" go higher than the actuation-gas pressure.  Of course this was a big FAIL when the "ambient" pressure went far above the "expected" levels.
2) While the primary emergency relief valves were properly rigged to withstand earthquake, the plumbing GOING TO the relief valves was installed like ordinary infrastructure with no higher priority of protection. The earthquake probably caused the plumbing to fail in one or more places, rendering the valves useless in the emergency.
3) Locating the emergency generation gear (giant deisel generators) BELOW SEA LEVEL when they were that close to the ocean just seems like a dumb idea. At least from my armchair with hindsight.

[David Hess]

I always thought the fact that the alarms couldn't be disabled was a huge engineering blunder.  Nothing like having a siren blare while you are trying to figure out why a nuclear reactor is melting down. 

It is more of a regulatory issue than an engineering issue but I have seen discussions about this in connection with fire alarms.  The loud sirens hinder communication and the strobes end up causing significant disorientation by themselves.

A similar problem exists with the requirements for fire retardant materials.  The same materials which help prevent a fire from starting or spreading are much more toxic when burned.

[wraper]

Can't believe I didn't think of this before, but the XBOX 360 Red Ring of Death.

Basically caused by bad head dissipation which lead to the motherboard warping and some of the chips subsequently become detached. Some known fixes (albeit temporary) were to wrap a towel around the 360, force it to over heat to the point it turns itself off and then you've effectively reached a temperature that melts the solder and re-attaches the chips to the board!

Do you seriosly believe in what you have wrote? Melt the solder? 217^oC?

To be fair, I did this about three or four years ago and didn't pay too much attention to the details at the time. I just remember it being something along those lines. Overheating the xbox had a positive effect for a while anyway

The real problem is similar to Nvidia I wrote about. That fix is temporary. Just connection between substrate and die reestablishes for a short time

See the large bold statement above.


Apology accepted...

Apology for what? I just noted that you should thought that it is impossible to melt solder by chip heating itself. And then why it is not permanent. Chip would fry itself at least 100^oC before solder even starts to melt. Albeit it cannot dissipate such amount of heat to even closely reach this temperature with heatsink attached on top of it. But even if it could, all balls would flatten out and become short because of pressure applied on top by heatsink. Connection reestablishes because of thermal expansion, not melting solder.

[wraper]

Sorry for harsh tone but it really pisses me off how much bullshit is written everywhere about this problem. DIY worthless repairs, endless reballs, etc.

[Tris20]

Sorry for harsh tone but it really pisses me off how much bullshit is written everywhere about this problem. DIY worthless repairs, endless reballs, etc.

I agree. But that fix did work for a while for a lot of people.

[wraper]

Problem is that in very many cases when consoles/notebooks arrived to the real repairman after DIY repairs, kitchen oven or stupid reballers (they are not so stupid in many cases, just take your money), board was already fried or something else broken.