What killed my mobo ?

[ZTK]

A week ago, I posted some questions re installation of Win98 on an MSI
RS480M2 mobo. No problem installing Win XP and I could also install
98SE by disabling the L2 cache. But the mobo just died a couple of
days later. It happened this way :

After installing Win XP, I put in a Pinnacle TV tuner card. I'd
misplaced the new driver CD and tried to install an older version. I
don't know if it's because of that, but it reported a conflict towards
the end of the self-test process (I don't remember exactly what it
said). I restarted with the intention of changing the IRQ in BIOS but
decided against it. I did change the Reset Configuration Data from the
default 'Disabled' to 'Enabled'. The computer refused to start up
after that, not even a POST screen. CPU and chassis fans turn, power
LED lights up.

Removing the tuner card and clearing the CMOS makes no difference.
Neither does disconnecting all drives. Mobo's new and battery voltage
is OK. The PSU is OK on another computer, and substituting it with a
known good one doesn't help. RAM is OK too.

In any case, I replaced the mobo with a new one of the same model and
everything's running fine. But neither mobo gives out any beep from
the onboard piezo buzzer, so that's no help. I haven't yet tried
installing the tuner again.

I took all the usual precautions against ESD, and though I'm not an
expert, I've never had a piece of hardware die in my hands except an
EDO RAM stick a long time ago. I called my supplier and they've
already despatched a replacement, but I'd love to know if I did
anything to kill the mobo. Thoughts please ?

 

[Timbertea]

A week ago, I posted some questions re installation of Win98 on an MSI
RS480M2 mobo. No problem installing Win XP and I could also install
98SE by disabling the L2 cache. But the mobo just died a couple of
days later. It happened this way :

After installing Win XP, I put in a Pinnacle TV tuner card. I'd
misplaced the new driver CD and tried to install an older version. I
don't know if it's because of that, but it reported a conflict towards
the end of the self-test process (I don't remember exactly what it
said). I restarted with the intention of changing the IRQ in BIOS but
decided against it. I did change the Reset Configuration Data from the
default 'Disabled' to 'Enabled'. The computer refused to start up
after that, not even a POST screen. CPU and chassis fans turn, power
LED lights up.

Removing the tuner card and clearing the CMOS makes no difference.
Neither does disconnecting all drives. Mobo's new and battery voltage
is OK. The PSU is OK on another computer, and substituting it with a
known good one doesn't help. RAM is OK too.

In any case, I replaced the mobo with a new one of the same model and
everything's running fine. But neither mobo gives out any beep from
the onboard piezo buzzer, so that's no help. I haven't yet tried
installing the tuner again.

I took all the usual precautions against ESD, and though I'm not an
expert, I've never had a piece of hardware die in my hands except an
EDO RAM stick a long time ago. I called my supplier and they've
already despatched a replacement, but I'd love to know if I did
anything to kill the mobo. Thoughts please ?

Probably just a victim of the bathtub curve...

------ ------ High Failure Rate
\ /
\ /
\ /
------------- Low Failure Rate

1-6 month ...years... end of life

Generally speaking, if a device is going to fail it's going to do it in
it's early part of service, the first couple months. If it survives that
it will probably go on to last for years until it hits the wear out point.

From the sound of it, it was probably defective anyway. Does the new
one force you to disable L2 cache to boot into windows 98 as well?

 

[kony]

A week ago, I posted some questions re installation of Win98 on an MSI
RS480M2 mobo. No problem installing Win XP and I could also install
98SE by disabling the L2 cache. But the mobo just died a couple of
days later. It happened this way :

After installing Win XP, I put in a Pinnacle TV tuner card. I'd
misplaced the new driver CD and tried to install an older version. I
don't know if it's because of that, but it reported a conflict towards
the end of the self-test process (I don't remember exactly what it
said). I restarted with the intention of changing the IRQ in BIOS but
decided against it. I did change the Reset Configuration Data from the
default 'Disabled' to 'Enabled'. The computer refused to start up
after that, not even a POST screen. CPU and chassis fans turn, power
LED lights up.

It would seem the board (or other, power maybe?) was on it's
last leg and your most recent configurations merely happened
to coincide.

Removing the tuner card and clearing the CMOS makes no difference.
Neither does disconnecting all drives. Mobo's new and battery voltage
is OK. The PSU is OK on another computer, and substituting it with a
known good one doesn't help. RAM is OK too.

By "known good" do you mean "good" or do you merely mean,
"known working"? Difference being that any
two/three/four/etc systems may have different power needs
per rail. It's quite common that a marginal power supply
will work on one system but not another. Even so, I suspect
the board as you shouldn't have had to disable the L2 cache
in the first place.

In any case, I replaced the mobo with a new one of the same model and
everything's running fine. But neither mobo gives out any beep from
the onboard piezo buzzer, so that's no help. I haven't yet tried
installing the tuner again.

See if it's configured to output from the onboard sound in
the bios? Perhaps a jumper onboard needs changed- check the
manual as this was just a random thought, I've no idea if
this particular board needs that.

I took all the usual precautions against ESD, and though I'm not an
expert, I've never had a piece of hardware die in my hands except an
EDO RAM stick a long time ago. I called my supplier and they've
already despatched a replacement, but I'd love to know if I did
anything to kill the mobo. Thoughts please ?

It'd tend to be thinks you didn't mention, IF it were
anything you did at all. Adding cards (while system is
unplugged of course) or resetting ECD or the CMOS shouldn't
have killed the board. Boards with component problems will
tend to fail sooner rather than later, it would be more odd
for the system to run fine for another month and THEN fail
(unless you had a severe overheating or power problem but no
mention of that).

You didn't mention the power supply on this post so I'll
presume you deem it sufficient- all that should be done at
this point is to try the replacement board, though sometimes
it's best to try (if possible) to use a completely different
board. If the first had manufacturing defects or bios bugs,
etc, then a replacement board from same vendor's stock could
be a board produced on same line, same day, similar
potential for bugs of a physical or bios nature.

 

[thefriend]

Not that I could see, things like that happen, 17 years as a technician I
have seen strager things.
The board was just bad.

 

[ZTK]

It would seem the board (or other, power maybe?) was on it's
last leg and your most recent configurations merely happened
to coincide.

I'm inclined to think that was it. Pls see below re the PSU.

By "known good" do you mean "good" or do you merely mean,
"known working"? Difference being that any
two/three/four/etc systems may have different power needs
per rail. It's quite common that a marginal power supply
will work on one system but not another. Even so, I suspect
the board as you shouldn't have had to disable the L2 cache
in the first place.

By "known good" I mean one that's being used in a system that has at
least the same power requirement. The same PSU that was used when the
mobo died is now being used to power the system with the new mobo. No
problem.

See if it's configured to output from the onboard sound in
the bios? Perhaps a jumper onboard needs changed- check the
manual as this was just a random thought, I've no idea if
this particular board needs that.

Will check. I don't have access to it right now because I handed it
over to the owner. I do know that the mobo has no header pins for
connecting a separate speaker.

It'd tend to be thinks you didn't mention, IF it were
anything you did at all. Adding cards (while system is
unplugged of course) or resetting ECD or the CMOS shouldn't
have killed the board. Boards with component problems will
tend to fail sooner rather than later, it would be more odd
for the system to run fine for another month and THEN fail
(unless you had a severe overheating or power problem but no
mention of that).

You didn't mention the power supply on this post so I'll
presume you deem it sufficient-

Yes. I have a bit of experience in this regard. With the limited
choice we have in this area, I often tweak a readymade PSU to get a
little more juice and/or safety margin out of it. It's a 300W unit.

all that should be done at
this point is to try the replacement board, though sometimes
it's best to try (if possible) to use a completely different
board. If the first had manufacturing defects or bios bugs,
etc, then a replacement board from same vendor's stock could
be a board produced on same line, same day, similar
potential for bugs of a physical or bios nature.

True. Thing is, this friend needs a reasonable gaming performance but
cannot afford a separate gfx card right now, and AFAIK the RS480M2-IL
is the only A64 board currently available with decent onboard
graphics. He could add a PCI-E card later, perhaps next year. He won't
be able to buy a new computer for the next few years, so I didn't want
him to be stuck with a 32-bit system either.

 

[kony]

By "known good" I mean one that's being used in a system that has at
least the same power requirement. The same PSU that was used when the
mobo died is now being used to power the system with the new mobo. No
problem.

OK, but that's not evidence that the power supply is
sufficient. A power supply with marginal capacity will
cause great ripple to the board. That progressively wears
out the board. Board ends up dead but power supply
continues working ok on a not-so-worn-out board. This is
not only theory but has been observed to happen- but I can't
remotely diagnose this, it's only a potential scenario.

Will check. I don't have access to it right now because I handed it
over to the owner. I do know that the mobo has no header pins for
connecting a separate speaker.

When a board outputs from the onboard sound, it goes to the
"main" speakers for the system, the rear (or front panel out
if case is so equipped) output 3.5mm stereo jack.
Regardless, the beeping is of little consequence if it's
working OK.

Yes. I have a bit of experience in this regard. With the limited
choice we have in this area, I often tweak a readymade PSU to get a
little more juice and/or safety margin out of it. It's a 300W unit.

Tweak for more juice?
How, EXACTLY did you do that?
Did you replace the transformer? That is manditory to
actually get any significant benefit, "more juice".
If all you did was mod it (or turn a POT) to raise the
output voltage you may have then made things worse, because
when the voltage is low due to insufficiency, trying to
force a higher peak voltage just introduces more ripple. A
given transfomer can only susstain a certain # of watts
within the tolerance of the input, such that raising the
output voltage lowers the amperage it can deliver (more
ripple).

I only mentioned it this much (random speculations) because
it seems quite possible the power supply simply wasn't
sufficient. The typical 300W would be marginal and a
generic 300W certainly not sufficient for that system unless
it had been completely rebuild from the ground up which
isn't really cost or time effective.

True. Thing is, this friend needs a reasonable gaming performance but
cannot afford a separate gfx card right now, and AFAIK the RS480M2-IL
is the only A64 board currently available with decent onboard
graphics.

The gaming performance will be incredibly poor.
It would be better to use an older, less expensive platform
and a fairly good gaming-oriented video card.

He could add a PCI-E card later, perhaps next year. He won't
be able to buy a new computer for the next few years, so I didn't want
him to be stuck with a 32-bit system either.

64 bit isn't all that relevent unless he has a real need for
applications using over 2GB of memory (and his system has
over 2GB of memory in it). There are other uses but for a
minimal system without even a video card, 64 bit is quite
far down on the list of useful features.

 

[ZTK]

....

Not that I could see, things like that happen, 17 years as a technician I
have seen strager things.
The board was just bad.

Thanks to you and others who replied. You've put my mind at ease.

 

[ZTK]

When a board outputs from the onboard sound, it goes to the
"main" speakers for the system, the rear (or front panel out
if case is so equipped) output 3.5mm stereo jack.
Regardless, the beeping is of little consequence if it's
working OK.

By "separate speaker" I mean the small speaker that often comes with
the computer case to serve as the POST diagnostic beeper.

Tweak for more juice?
How, EXACTLY did you do that?
Did you replace the transformer? That is manditory to
actually get any significant benefit, "more juice".
If all you did was mod it (or turn a POT) to raise the
output voltage you may have then made things worse, because
when the voltage is low due to insufficiency, trying to
force a higher peak voltage just introduces more ripple. A
given transfomer can only susstain a certain # of watts
within the tolerance of the input, such that raising the
output voltage lowers the amperage it can deliver (more
ripple).

I only mentioned it this much (random speculations) because
it seems quite possible the power supply simply wasn't
sufficient. The typical 300W would be marginal and a
generic 300W certainly not sufficient for that system unless
it had been completely rebuild from the ground up which
isn't really cost or time effective.

Nothing quite as crude as raising the voltages. I do have some 30
years of experience in various fields of applied electronics. The
extent of modification I do varies. It may be as simple as making sure
all power devices are firmly fastened to their heatsinks - the
uninitiated will be surprised at the sloppy handiwork of many
manufacturers. The next step may involve replacing certain key
components - e.g., some reputable manufacturers use a 15-ampere
Schottky rectifier to supply a 5-volt rail rated for 30+ amps.
Paralleling the wires to reduce copper loss, making sure the crimps
and soldering are perfect, etc. The next level, which I don't do
often, is actual modification of the design - the PWM controller
circuit, the transformer...... Heck, I know it's not cost-effective
from a business POV, but I own my (small) company, have a few helpers
who do the routine work, and treat most of my work as a hobby, so I
can indulge in such things.

 

[ZTK]

When a board outputs from the onboard sound, it goes to the
"main" speakers for the system, the rear (or front panel out
if case is so equipped) output 3.5mm stereo jack.
Regardless, the beeping is of little consequence if it's
working OK.

By "separate speaker" I mean the small speaker that often comes with
the computer case to serve as the POST diagnostic beeper.

Tweak for more juice?
How, EXACTLY did you do that?
Did you replace the transformer? That is manditory to
actually get any significant benefit, "more juice".
If all you did was mod it (or turn a POT) to raise the
output voltage you may have then made things worse, because
when the voltage is low due to insufficiency, trying to
force a higher peak voltage just introduces more ripple. A
given transfomer can only susstain a certain # of watts
within the tolerance of the input, such that raising the
output voltage lowers the amperage it can deliver (more
ripple).

I only mentioned it this much (random speculations) because
it seems quite possible the power supply simply wasn't
sufficient. The typical 300W would be marginal and a
generic 300W certainly not sufficient for that system unless
it had been completely rebuild from the ground up which
isn't really cost or time effective.

Nothing quite as crude as raising the voltages. I do have some 30
years of experience in various fields of applied electronics. The
extent of modification I do varies. It may be as simple as making sure
all power devices are firmly fastened to their heatsinks - the
uninitiated will be surprised at the sloppy handiwork of many
manufacturers. The next step may involve replacing certain key
components - e.g., some reputable manufacturers use a 15-ampere
Schottky rectifier to supply a 5-volt rail rated for 30+ amps.
Paralleling the wires to reduce copper loss, making sure the crimps
and soldering are perfect, etc. The next level, which I don't do
often, is actual modification of the design - the PWM controller
circuit, the transformer...... Heck, I know it's not cost-effective
from a business POV, but I own my (small) company, have a few helpers
who do the routine work, and treat most of my work as a hobby, so I
can indulge in such things.

 

[kony]

Nothing quite as crude as raising the voltages. I do have some 30
years of experience in various fields of applied electronics. The
extent of modification I do varies. It may be as simple as making sure
all power devices are firmly fastened to their heatsinks - the
uninitiated will be surprised at the sloppy handiwork of many
manufacturers. The next step may involve replacing certain key
components - e.g., some reputable manufacturers use a 15-ampere
Schottky rectifier to supply a 5-volt rail rated for 30+ amps.

I"m not sure if I'd call them reputable then?
Often when I see such things it didn't matter much as the
rail wasn't actually capable of delivering that 30 amps.

Paralleling the wires to reduce copper loss, making sure the crimps
and soldering are perfect, etc.

Those are good things to do if needed, but I have to wonder
about the quality of the supply if you found those things
necessary? Beyond manufacturing defects it's rarely needed.
Starting out with a poor supply it could be necessary to
spend quite a bit of time and cost on parts... even then
possibly having to redrill the PCB just to get parts to fit,
implementing risers just to relocate some parts in some
cases. Far too much work for the benefit, IMO.

The next level, which I don't do
often, is actual modification of the design - the PWM controller
circuit, the transformer...... Heck, I know it's not cost-effective
from a business POV, but I own my (small) company, have a few helpers
who do the routine work, and treat most of my work as a hobby, so I
can indulge in such things.

If you enjoy it that's all that matters I suppose, but IMO
you'd be far better off just buying a suitable PSU instead.
There are plenty at the under $50 range that can deliver
enough power without any modifications at all. FWIW, I find
most often the questionable supplies have their output
filter caps pop, which may in itself be partially due to the
insufficiently sized transformer but they're also typically,
notibly smaller capacity and unknown/generic quality &
specs.

 

[ZTK]

If you enjoy it that's all that matters I suppose, but IMO
you'd be far better off just buying a suitable PSU instead.
There are plenty at the under $50 range that can deliver
enough power without any modifications at all. FWIW, I find
most often the questionable supplies have their output
filter caps pop, which may in itself be partially due to the
insufficiently sized transformer but they're also typically,
notibly smaller capacity and unknown/generic quality &
specs.

I do such things partly because I enjoy it, and partly because, as I
indicated earlier, I live in such a remote place that we simply do not
have the kind of options you have. We don't have any online stores to
speak of in my country, and even in the nearest metro 500km away by
air, where I order most my stuff from, many well-known products are
simply not available - e.g., Viewsonic monitors, Arctic Silver......
hell, I had a hard time getting a spare Socket A HSF. I even had to
explain to my ISP's chief tech that Usenet is NOT the same as
usa.net.......

And most of my clients cannot afford to spend $50 on a PSU anyway. Oh
well, I'm one of those people who will never get rich