OT: Motherboard and CPU interaction...

[Ken Springer]

This is actually a hardware question, but hoping someone here has an
answer, idea, or opinion...

My nephew and I were discussing computer failures, and we both came up
wondering about the following:

If some portion of the motherboard fails, can it take out the CPU also?
As well as the opposite, if the CPU fails, can it take the motherboard
out?

Not that part 1 of either question will definitely cause part 2, but can it?

And if the answer is yes, what's the best way, if any, of determining
how much of the system has failed?



--
Ken

Mac OS X 10.8.3
Firefox 20.0
Thunderbird 17.0.5
LibreOffice 4.0.1.2

 

[Drew]

This is actually a hardware question, but hoping someone here has an
answer, idea, or opinion...

My nephew and I were discussing computer failures, and we both came up
wondering about the following:

If some portion of the motherboard fails, can it take out the CPU also?
As well as the opposite, if the CPU fails, can it take the motherboard
out?

Not that part 1 of either question will definitely cause part 2, but can
it?

And if the answer is yes, what's the best way, if any, of determining
how much of the system has failed?

A simple overheat can destroy both.. Been there done that. Keep it clean
and keep it cool is what I live by. As for determining damage it is a
case of, Is it worth the time and expense? If you are loaded then what
the heck but if you are not then upgrade or replace with the same.

 

[Ken Springer]

A simple overheat can destroy both.. Been there done that. Keep it clean
and keep it cool is what I live by. As for determining damage it is a
case of, Is it worth the time and expense? If you are loaded then what
the heck but if you are not then upgrade or replace with the same.

We were looking at it as a simple component failure, not due to
something like excessive heat due to dirt, power supply going out, etc.
Just one day works normally, the next, nothing works right! LOL

--
Ken

Mac OS X 10.8.3
Firefox 20.0
Thunderbird 17.0.5
LibreOffice 4.0.1.2

 

[Drew]

We were looking at it as a simple component failure, not due to
something like excessive heat due to dirt, power supply going out, etc.
Just one day works normally, the next, nothing works right! LOL

Geez Ken!.. A lot of time on your hands?. Now you got me thinking and
that in itself is a pain on a Friday night LOL!!!!

 

[Wolf K]

Depends on what caused the mobo failure. NB that a motherboard is a
collection of many components. If thge mobo fries because of bad
cooling, the CPU will likely fail too. IMO, the CPU would fail before
other components, because it runs at a higher temp than teh rest of the
system.

Not likely. Eg, if the CPU fails because of internal overheating caused
by overclocking, the mobo will not be affected.

If there's no visual or olfactory evidence, pull the suspected
component, replace it with a known good one, and see what happens. Or
take it to a tech and pay him to do the testing with his fancy gadgets.

We were looking at it as a simple component failure, not due to
something like excessive heat due to dirt, power supply going out, etc.
Just one day works normally, the next, nothing works right! LOL

You've actually listed the most common causes of "simple component
failure". ;-)

 

[charlie]

Without knowing the age of the motherboard - -
A past common failure had to do with capacitors failing. The bad ones
usually show signs of venting.
Works one day and not the next can be something as simple as a poor
connection. Bad solder joint, poor contact at connectors, crimped
connector pins, and so forth.

It's difficult for the average person to replace components on a
multilayer motherboard without damaging the motherboard.

 

[Paul]

This is actually a hardware question, but hoping someone here has an
answer, idea, or opinion...

My nephew and I were discussing computer failures, and we both came up
wondering about the following:

If some portion of the motherboard fails, can it take out the CPU also?
As well as the opposite, if the CPU fails, can it take the motherboard
out?

Not that part 1 of either question will definitely cause part 2, but can
it?

And if the answer is yes, what's the best way, if any, of determining
how much of the system has failed?

Yes and yes.

Generally, CPUs are pretty reliable. Considering the complexity involved,
it's really surprising what small percentage of system failures, are from
the silicon. It's more power and analog components (failed MOSFETs,
capacitors, coils) than anything else.

In the motherboard groups, in the past, there have been "domino failures"
while debugging. Meaning you move hardware from the suspect system,
to a test system, and the test system gets damaged, and in turn the
test system damages something else. That can mean, if you're not
thinking very carefully about what just happened, you can end up
with a pile of ruined hardware. That's a worst case scenario.

So while on the one hand, you can get lucky when testing, and figure
out immediately what broke. The worst case, you'd lose enough financially,
to make it better to go to the computer shop, and ruin their test system :-(
Not every DIY repair project is a "success" from the perspective
of avoiding unnecessary computer repair shop fees.

I would say on balance, it's usually financially in your favor to
attempt the testing yourself (if you already own the gear). For
example, I have two LGA775 motherboards, and I could do some swapping
if one of the systems shows problems. But I also have to consider
I could end up taking a fair hit financially, if the testing transfers
a fault from system to system, component to component.

My record for broken systems here, is having two computers down,
and requiring the services of a third computer, to look for advice
on the web. So in terms of "how many computers should I own",
based on first hand experience, the answer is "three"

Paul

 

[Ed Cryer]

Yes and yes.

Generally, CPUs are pretty reliable. Considering the complexity involved,
it's really surprising what small percentage of system failures, are from
the silicon. It's more power and analog components (failed MOSFETs,
capacitors, coils) than anything else.

In the motherboard groups, in the past, there have been "domino failures"
while debugging. Meaning you move hardware from the suspect system,
to a test system, and the test system gets damaged, and in turn the
test system damages something else. That can mean, if you're not
thinking very carefully about what just happened, you can end up
with a pile of ruined hardware. That's a worst case scenario.

So while on the one hand, you can get lucky when testing, and figure
out immediately what broke. The worst case, you'd lose enough financially,
to make it better to go to the computer shop, and ruin their test system
:-(
Not every DIY repair project is a "success" from the perspective
of avoiding unnecessary computer repair shop fees.

I would say on balance, it's usually financially in your favor to
attempt the testing yourself (if you already own the gear). For
example, I have two LGA775 motherboards, and I could do some swapping
if one of the systems shows problems. But I also have to consider
I could end up taking a fair hit financially, if the testing transfers
a fault from system to system, component to component.

My record for broken systems here, is having two computers down,
and requiring the services of a third computer, to look for advice
on the web. So in terms of "how many computers should I own",
based on first hand experience, the answer is "three"

Paul

What makes things worse is the way manufacturers repair broken units.
You take one back where you bought it because it "don't work". They send
it to Acer or Dell or whatever. Back it comes. You pick it up, and
they've "replaced the motherboard".
So what went wrong?
Don't know sir, but if you have any more trouble, bring it back again.

What was it? A blown fuse, burnt out component? Multiple organ failure?

You're left not knowing.

Ed

 

[Paul]

What makes things worse is the way manufacturers repair broken units.
You take one back where you bought it because it "don't work". They send
it to Acer or Dell or whatever. Back it comes. You pick it up, and
they've "replaced the motherboard".
So what went wrong?
Don't know sir, but if you have any more trouble, bring it back again.

What was it? A blown fuse, burnt out component? Multiple organ failure?

You're left not knowing.

Ed

The "front end" of the organization, isolates to the nearest
sub-assembly, and swaps it out. That allows less skilled
labor to be used. They swap stuff in, until it works.

Your motherboard eventually works its way back to the factory,
and the staff there are more of your "TV repair shop" style of
staff. They actually know how it works, and if there is a
burnt out component (or bad capacitors), they'll find it
pretty quickly. And they have access to automated test
equipment, that can find some kinds of faults. But a large
part of it, still requires "old fashioned TV repair guy"
skills. At my little factory, there might have been around
forty people doing that work, each with their own bench
to work at. At the Asus factory, there would be thousands.
The stuff done at my factory, was termed "low volume", whereas
right now, I think Asus is down to around 3 million motherboards
per month, of production. And you'd need a ton of staff, to test
and clean up the mess from that level of production. Reworking
busted boards, would be in one corner of the factory. And have
access to any equipment that might be needed.

Even with all of that, there can be thousands of motherboards
thrown in a garbage bin, out back. A certain percentage of
boards, just can't be fixed in a reasonable period of time.
If a PCB was cracked, they'd probably just toss it.

Paul

 

[charlie]

The "front end" of the organization, isolates to the nearest
sub-assembly, and swaps it out. That allows less skilled
labor to be used. They swap stuff in, until it works.

Your motherboard eventually works its way back to the factory,
and the staff there are more of your "TV repair shop" style of
staff. They actually know how it works, and if there is a
burnt out component (or bad capacitors), they'll find it
pretty quickly. And they have access to automated test
equipment, that can find some kinds of faults. But a large
part of it, still requires "old fashioned TV repair guy"
skills. At my little factory, there might have been around
forty people doing that work, each with their own bench
to work at. At the Asus factory, there would be thousands.
The stuff done at my factory, was termed "low volume", whereas
right now, I think Asus is down to around 3 million motherboards
per month, of production. And you'd need a ton of staff, to test
and clean up the mess from that level of production. Reworking
busted boards, would be in one corner of the factory. And have
access to any equipment that might be needed.

Even with all of that, there can be thousands of motherboards
thrown in a garbage bin, out back. A certain percentage of
boards, just can't be fixed in a reasonable period of time.
If a PCB was cracked, they'd probably just toss it.

Paul

One of the problems Mfrs face is the complexity of fault isolation.
Most of the MBDs are tested with some sort of automated process.
If a board is an older board, the automated process uses fixtures that
are likely history.

At least the processors, I/O connections, memory sockets, and so forth
are standardized. Thus a great deal of testing can be done with software
on say a "generic" MBD tester that is more or less the same for a P/C.

On the other hand, think of military electronics for a moment.
Some of the critical systems have been out of production for more than a
decade. A fair number of parts, including integrated circuits, were
"purpose" designed and built, and produced in finite quantities.
When the production line shut down, that was it!

When enough failed assemblies exist, either they must be fixed, a new
assembly designed and built, or the entire system replaced.

Manual testing - -
When a MBD is as cheap as they are, does it make sense to tie up
expensive test equipment and skilled techs? Even if you pay the techs
$1.98 an hour? Not unless there is some sort of common failure that can
be easily identified and repaired!

 

[Paul]

One of the problems Mfrs face is the complexity of fault isolation.
Most of the MBDs are tested with some sort of automated process.
If a board is an older board, the automated process uses fixtures that
are likely history.

At least the processors, I/O connections, memory sockets, and so forth
are standardized. Thus a great deal of testing can be done with software
on say a "generic" MBD tester that is more or less the same for a P/C.

On the other hand, think of military electronics for a moment.
Some of the critical systems have been out of production for more than a
decade. A fair number of parts, including integrated circuits, were
"purpose" designed and built, and produced in finite quantities.
When the production line shut down, that was it!

When enough failed assemblies exist, either they must be fixed, a new
assembly designed and built, or the entire system replaced.

Manual testing - -
When a MBD is as cheap as they are, does it make sense to tie up
expensive test equipment and skilled techs? Even if you pay the techs
$1.98 an hour? Not unless there is some sort of common failure that can
be easily identified and repaired!

The two kinds of testing are called structural test and functional test.

At the factory, they wouldn't throw away the adapter boards made for
structural test. It's possible the bed-of-nails tester could get upgraded,
resulting in the adapters being a dead loss. And making adapters is expensive
enough, you need a reason to have one (i.e. current production). I
never got involved with the making of those adapters, for the products
I designed. Staff at the factory took care of that.

The structural test, for the most part ignores the function of the
product, and just evaluates the wiring and scan chains. Both military
and civilian systems can benefit from it.

For final test, you can do a functional test. At that point, you're
fairly sure all onboard regulators are pumping out the correct
voltages, clock signals are present, I/O is present and accounted for
and so on. So then, it's worthwhile to involve humans ($$$) to verify
the thing actually performs the desired function. In a video one of
the motherboard enthusiast sites had, it takes about two minutes for
a functional test on a production motherboard. And they use standard
plugin assemblies, which go into the slots and sockets of the motherboard,
as part of that test. Two minutes, times three million motherboards
a month, is a lot of man-hours.

Paul

 

[philo]

This is actually a hardware question, but hoping someone here has an
answer, idea, or opinion...

My nephew and I were discussing computer failures, and we both came up
wondering about the following:

If some portion of the motherboard fails, can it take out the CPU also?
As well as the opposite, if the CPU fails, can it take the motherboard
out?

Not that part 1 of either question will definitely cause part 2, but can
it?

And if the answer is yes, what's the best way, if any, of determining
how much of the system has failed?

In the 12 years or so that I've been doing computer repairs I've seen
*many* bad motherboards. In all cases, the cpu was still good when used
elsewhere.

 

[Ken Springer]

In the 12 years or so that I've been doing computer repairs I've seen
*many* bad motherboards. In all cases, the cpu was still good when used
elsewhere.

This is something we were really curious about, as they get much more
expensive than some motherboards.


--
Ken

Mac OS X 10.8.3
Firefox 20.0
Thunderbird 17.0.5
LibreOffice 4.0.1.2

 

[philo]

This is something we were really curious about, as they get much more
expensive than some motherboards.

Every time I scrap out a motherboard I save the CPU if it's a p-IV or
better.

Then, I re-use the CPU when a motherboard comes along.

Since I've recently retired and have more spare time, I finally went
through my box of CPU's and tested them and they were all good.


So in general I can say that whatever takes out a motherboard is not
likely to kill the cpu in the process.

 

[charlie]

Every time I scrap out a motherboard I save the CPU if it's a p-IV or
better.

Then, I re-use the CPU when a motherboard comes along.

Since I've recently retired and have more spare time, I finally went
through my box of CPU's and tested them and they were all good.


So in general I can say that whatever takes out a motherboard is not
likely to kill the cpu in the process.

In more years that I like to think about, I've only had
a couple of MBDs that did in the CPU. The first was a 386 CPU,
and the last one of the older AMD 32 bit processors.

MBD regulators were the likely culprit.