I hope I can get some guidance here. My desktop computer that I built
several years ago just died on me - as I was watching, it appeared to
be going into sleep mode, which it does after some hours of
inactivity. However, it could not be awakened, it was dead. I
immediately thought the power supply had gone bad, and in fact, I
tested it with a multimeter and it was, indeed dead - no voltage
detectable on any pin. So, I bought a new PS of similar specs. Now, the
motherboard (ASUS Prime Z270-K) lights up, but there is no other
activity, no fans going, no display, no disc drives spinning. I can
only wonder if when the PS blew it took out the mobo or some other
component.

So, my question is, 'what can/should I do now to confirm that the mobo
is truly, irretrievably gone?' If I can't, I will proceed toward a new
build or buy, but if I can revive the existing one, VERY much the
better. After all, it's tuned to a fair-thee-well and I don't look
forward to going through months of tweaking to get it to that point
again.

Aside... A year or so ago I replaced the mechanical boot drive with an
SSD drive. If THAT had gone bad, would the mobo not still boot with at
least a display and error message?

On 3/20/2024 9:10 PM, Nil wrote:
> I hope I can get some guidance here. My desktop computer that I built
> several years ago just died on me - as I was watching, it appeared to
> be going into sleep mode, which it does after some hours of
> inactivity. However, it could not be awakened, it was dead. I
> immediately thought the power supply had gone bad, and in fact, I
> tested it with a multimeter and it was, indeed dead - no voltage
> detectable on any pin. So, I bought a new PS of similar specs. Now, the
> motherboard (ASUS Prime Z270-K) lights up, but there is no other
> activity, no fans going, no display, no disc drives spinning. I can
> only wonder if when the PS blew it took out the mobo or some other
> component.
>
> So, my question is, 'what can/should I do now to confirm that the mobo
> is truly, irretrievably gone?' If I can't, I will proceed toward a new
> build or buy, but if I can revive the existing one, VERY much the
> better. After all, it's tuned to a fair-thee-well and I don't look
> forward to going through months of tweaking to get it to that point
> again.
>
> Aside... A year or so ago I replaced the mechanical boot drive with an
> SSD drive. If THAT had gone bad, would the mobo not still boot with at
> least a display and error message?
>

The indicator of PSU death, might be something like the state of Pin 9.
If the PSU is plugged in, the PSU makes +5VSB which functions as a
"supervisor voltage" and it's on Pin 9.

Anything on the motherboard needing supervision (while asleep, soft off,
hibernating) then some +5VSB would be handy for that.

If PS_ON# is driven to Ground (logic 0), that's a Wired-Or capable
signal that you can ground to test that a PSU works (the main portion,
and its internal fan). If you ground PS_ON#, the fan should spin (on a
PSU with a "normal" kind of fan curve -- some PSU don't spin the fan
until the unit hits 50C inside which is unfortunate).

https://en.wikipedia.org/wiki/ATX#Power_supply

Purple +5VSB Pin 9

When the PSU is doing the job of making +5VSB (used by USB ports),
it does not need to use the fan for that. Some PSUs get kinda warm
doing that (a Bestek in front of me, blows a cloud of warm air out
at startup, as proof the +5VSB thing was running). More modern supplies
use an SMPS for +5VSB and are more efficient and have less waste heat.

The integrated circuit that drives PS_ON# to ground, it needs
power and that power source is +5VSB. You can drive PS_ON# to
ground yourself, with a wire to a black ground, or a wire plus a
switch (if you want a convenience switch for turning it on and off).

A motherboard that cannot drive PS_ON# to ground, then that motherboard
will never get to spin the fans or make any other happy noises.

And PS_ON# signal failures are fairly frequent, for unknown reasons.

At one time, the driver for that on some motherboards, was a Fairchild 74F with
64mA drive capability. Which gave me a good chuckle, because it
said "this here motherboard is not taking NO for an answer".

Later designs only had 8mA drive and were more easily pistol-whipped
into failure. The circuit really should never need more than 2mA
of sinking capability. But I think the individual who was using the
64mA drive, that guy deserves a medal :-) You want an IC that will
"catch fire" rather than take NO for an answer.

*******

You CAN do ohms tests between rails. With the 24-pin connector
open, you could ohm from +5V to GND, +12V to GND and so on.
But be aware there could be 2000uF of bypass caps on such a rail,
and this makes it hard to distinguish what you're measuring
(the dead short of a charging capacitor).

The PSU has a Power_Good signal. It indicates the rails have
charged to min-voltages. Maybe Power_Good is logic 1, any time
+12V is above about +11V or so.

The motherboard also has an internal Power_Good of its own.
All the tiny power converters on the motherboard, that make
1.5V, 1.8V, 2.5V, and so on, those can have status indicators
that say they are working nominally.

The motherboard can only come out of RESET, when all the
Power_Good are logic 1. The Reset generator might have a
half-second pulse on it, but the Power_Good extends the
board-level RESET for as long as it takes for Power_Good
to go to logic 1.

After about 35 milliseconds or so, the PSU starts to check
for overcurrent. The 35 milliseconds gives time for the PSU
to charge the motherboard bypass capacitors. The 35 millisecond
period where it does not check for overcurrent, causes computer
case fans to "twitch" and then you know PS_ON# was asserted,
the PSU did try to start, but it tripped off on overcurrent.
And normally, you have to (slowly( toggle the switch on the
back OFF. Then wait 60 seconds and turn it ON again, to attempt
another start. Seeing a fan twitch, is a "symptom" to record
in your lab report :-)

Mobo VCore is a rather large SMPS, and capable of a couple hundred
watts of output. On the machine I'm typing on, VCore varies
between 0.35V and maybe 1.2V or so, at up to 200 amps. It should
have a Power_Good indicator too, and the CPU cannot be started,
unless VCore is nominally working. When the CPU socket is empty,
the VID signals float to logic 1, and the VCore all-1s code
is "shut off". The motherboard will not arm VCore, if the socket
is empty. This also means the motherboard would not enable all
Power_Good, if the CPU was missing. Not that it would matter
in that case, of course.

You can see that all of the power conversion machinery must
be in top-rate form, for the CPU to start running firmware code.
Nothing short of perfection, will work, when it comes to power.

But if the +12V comes on, you should be hearing case fans spin.

*******

I've had two motherboard failures now (P4C800-E ICH, P5E X48/ICH10),
and the failures seemed to be Southbridge/PCH and something to do
with one of the five converters that feed the Southbridge. But
once the SB is cooked, you'd have a load of fun trying to fix that
(hot air station, on motherboards not intended to be repaired).

*******

The PSU can fail if the slow-blow fuse inside it, blows open circuit.
That hardly ever happens. If the 11kV outside your house, fell on
the 230V wire, then that would be sufficient to blow the fuse.
The fuse of course, is not rated for 11kV, so there would be
mondo damage. In any case, it would take an extreme event to
blow the fuse. Minimum would be a close lightning hit. That
might do it. Plus fry a lot of other shit.

Paul

Lotsa interesting info here, thanks. I've been poking at this thing all
day with no joy. I can verify that the old PS is dead - no voltage on
any pins, no matter what. I can verify that the new PS is good - it's
one of the ones you mention that the fan doesn't come on unless it
heats up to a certain temperature, but the expected voltages are on all
pins, including the CPU connector. But that's as far as it goes. The
motherboard has some lights on it that glow when the PS is connected
and turns on, but nothing else happens, no fans, no drives spinning up,
no display, not a peep or twitch. I can only conclude that the mobo has
failed. Probably the PSU killed it while itself dying. Murder-suicide.

I wonder if it's possible that the PSU might test good with the
multimeter but can't power up mobo + peripherals when all loaded up?

This is a bummer - I've always enjoyed building PCs for myself, but
I've always known the current machine was ready to be replaced and I
had time to leisurely deal with it. This is a catastrophic failure and
I'm in the middle of some projects and don't have time to research and
gather components for a new build. I'm inclined to go buy a Dell or
something off-the-shelf.

On 20 Mar 2024, Paul <nospam@needed.invalid> wrote in
alt.comp.hardware.pc-homebuilt:

> On 3/20/2024 9:10 PM, Nil wrote:
>> I hope I can get some guidance here. My desktop computer that I
>> built several years ago just died on me - as I was watching, it
>> appeared to be going into sleep mode, which it does after some
>> hours of inactivity. However, it could not be awakened, it was
>> dead. I immediately thought the power supply had gone bad, and in
>> fact, I tested it with a multimeter and it was, indeed dead - no
>> voltage detectable on any pin. So, I bought a new PS of similar
>> specs. Now, the motherboard (ASUS Prime Z270-K) lights up, but
>> there is no other activity, no fans going, no display, no disc
>> drives spinning. I can only wonder if when the PS blew it took
>> out the mobo or some other component.
>>
>> So, my question is, 'what can/should I do now to confirm that the
>> mobo is truly, irretrievably gone?' If I can't, I will proceed
>> toward a new build or buy, but if I can revive the existing one,
>> VERY much the better. After all, it's tuned to a fair-thee-well
>> and I don't look forward to going through months of tweaking to
>> get it to that point again.
>>
>> Aside... A year or so ago I replaced the mechanical boot drive
>> with an SSD drive. If THAT had gone bad, would the mobo not still
>> boot with at least a display and error message?
>>
>
> The indicator of PSU death, might be something like the state of
> Pin 9. If the PSU is plugged in, the PSU makes +5VSB which
> functions as a "supervisor voltage" and it's on Pin 9.
>
> Anything on the motherboard needing supervision (while asleep,
> soft off, hibernating) then some +5VSB would be handy for that.
>
> If PS_ON# is driven to Ground (logic 0), that's a Wired-Or capable
> signal that you can ground to test that a PSU works (the main
> portion, and its internal fan). If you ground PS_ON#, the fan
> should spin (on a PSU with a "normal" kind of fan curve -- some
> PSU don't spin the fan until the unit hits 50C inside which is
> unfortunate).
>
> https://en.wikipedia.org/wiki/ATX#Power_supply
>
> Purple +5VSB Pin 9
>
> When the PSU is doing the job of making +5VSB (used by USB ports),
> it does not need to use the fan for that. Some PSUs get kinda warm
> doing that (a Bestek in front of me, blows a cloud of warm air out
> at startup, as proof the +5VSB thing was running). More modern
> supplies use an SMPS for +5VSB and are more efficient and have
> less waste heat.
>
> The integrated circuit that drives PS_ON# to ground, it needs
> power and that power source is +5VSB. You can drive PS_ON# to
> ground yourself, with a wire to a black ground, or a wire plus a
> switch (if you want a convenience switch for turning it on and
> off).
>
> A motherboard that cannot drive PS_ON# to ground, then that
> motherboard will never get to spin the fans or make any other
> happy noises.
>
> And PS_ON# signal failures are fairly frequent, for unknown
> reasons.
>
> At one time, the driver for that on some motherboards, was a
> Fairchild 74F with 64mA drive capability. Which gave me a good
> chuckle, because it said "this here motherboard is not taking NO
> for an answer".
>
> Later designs only had 8mA drive and were more easily
> pistol-whipped into failure. The circuit really should never need
> more than 2mA of sinking capability. But I think the individual
> who was using the 64mA drive, that guy deserves a medal :-) You
> want an IC that will "catch fire" rather than take NO for an
> answer.
>
> *******
>
> You CAN do ohms tests between rails. With the 24-pin connector
> open, you could ohm from +5V to GND, +12V to GND and so on.
> But be aware there could be 2000uF of bypass caps on such a rail,
> and this makes it hard to distinguish what you're measuring
> (the dead short of a charging capacitor).
>
> The PSU has a Power_Good signal. It indicates the rails have
> charged to min-voltages. Maybe Power_Good is logic 1, any time
> +12V is above about +11V or so.
>
> The motherboard also has an internal Power_Good of its own.
> All the tiny power converters on the motherboard, that make
> 1.5V, 1.8V, 2.5V, and so on, those can have status indicators
> that say they are working nominally.
>
> The motherboard can only come out of RESET, when all the
> Power_Good are logic 1. The Reset generator might have a
> half-second pulse on it, but the Power_Good extends the
> board-level RESET for as long as it takes for Power_Good
> to go to logic 1.
>
> After about 35 milliseconds or so, the PSU starts to check
> for overcurrent. The 35 milliseconds gives time for the PSU
> to charge the motherboard bypass capacitors. The 35 millisecond
> period where it does not check for overcurrent, causes computer
> case fans to "twitch" and then you know PS_ON# was asserted,
> the PSU did try to start, but it tripped off on overcurrent.
> And normally, you have to (slowly( toggle the switch on the
> back OFF. Then wait 60 seconds and turn it ON again, to attempt
> another start. Seeing a fan twitch, is a "symptom" to record
> in your lab report :-)
>
> Mobo VCore is a rather large SMPS, and capable of a couple hundred
> watts of output. On the machine I'm typing on, VCore varies
> between 0.35V and maybe 1.2V or so, at up to 200 amps. It should
> have a Power_Good indicator too, and the CPU cannot be started,
> unless VCore is nominally working. When the CPU socket is empty,
> the VID signals float to logic 1, and the VCore all-1s code
> is "shut off". The motherboard will not arm VCore, if the socket
> is empty. This also means the motherboard would not enable all
> Power_Good, if the CPU was missing. Not that it would matter
> in that case, of course.
>
> You can see that all of the power conversion machinery must
> be in top-rate form, for the CPU to start running firmware code.
> Nothing short of perfection, will work, when it comes to power.
>
> But if the +12V comes on, you should be hearing case fans spin.
>
> *******
>
> I've had two motherboard failures now (P4C800-E ICH, P5E
> X48/ICH10), and the failures seemed to be Southbridge/PCH and
> something to do with one of the five converters that feed the
> Southbridge. But once the SB is cooked, you'd have a load of fun
> trying to fix that (hot air station, on motherboards not intended
> to be repaired).
>
> *******
>
> The PSU can fail if the slow-blow fuse inside it, blows open
> circuit. That hardly ever happens. If the 11kV outside your house,
> fell on the 230V wire, then that would be sufficient to blow the
> fuse. The fuse of course, is not rated for 11kV, so there would be
> mondo damage. In any case, it would take an extreme event to
> blow the fuse. Minimum would be a close lightning hit. That
> might do it. Plus fry a lot of other shit.
>
> Paul
>
>

On 3/21/2024 6:44 PM, Nil wrote:
> Lotsa interesting info here, thanks. I've been poking at this thing all
> day with no joy. I can verify that the old PS is dead - no voltage on
> any pins, no matter what. I can verify that the new PS is good - it's
> one of the ones you mention that the fan doesn't come on unless it
> heats up to a certain temperature, but the expected voltages are on all
> pins, including the CPU connector. But that's as far as it goes. The
> motherboard has some lights on it that glow when the PS is connected
> and turns on, but nothing else happens, no fans, no drives spinning up,
> no display, not a peep or twitch. I can only conclude that the mobo has
> failed. Probably the PSU killed it while itself dying. Murder-suicide.
>
> I wonder if it's possible that the PSU might test good with the
> multimeter but can't power up mobo + peripherals when all loaded up?
>
> This is a bummer - I've always enjoyed building PCs for myself, but
> I've always known the current machine was ready to be replaced and I
> had time to leisurely deal with it. This is a catastrophic failure and
> I'm in the middle of some projects and don't have time to research and
> gather components for a new build. I'm inclined to go buy a Dell or
> something off-the-shelf.

In the old days, I would have taken a clamp-on DC ammeter and
checked what kinds of currents were being drawn. It it was
doing that (drawing a lot of current), there would likely be
some other symptoms. But you would probably need an IR camera
to locate where the power was going, without a DC ammeter.
The nice thing about clamp meters, is there are no fuses to blow
on the instrument :-)

If it wont start, you could have a grounded RESET wire to the
front panel. Or, a Power_Good signal from one of the onboard
regulators, is indicating it is Not Ready, and the CPU stays
in the reset state if that is the case.

With modern kit and all those black wires, it's pretty hard
to tell what is what (when positioning the clamp meter, and having
a wrap around the wires that needs to be cut.

Even when I bought a 24-wire extension cable,
the fools used colored wire, but did not assign standard pins
for the colors, so the coloration was of no use.

*******

You have some DDR4 in hand. You could look for a different solution,
and try to reuse the RAM at least.

A lot of the boards now, no longer have a PCI slot. They may still have
a kbd/mouse PS/2 hole, for at least one of the two appliances to go. Some
even have a serial RS232 port. Both my machines have RS232, even though
it was hardly advertised as being present. On one machine, I just noticed
it one day while examining USB2 headers, and I noticed "one header has a
strange pinout". And that was an RS232 when I looked it up.

The PCI Express slots may not be as useful as before. Before, we had
x8/x8 on two slots, and bifurcation logic. Today, the mainstream
boards are still sporting x16 on one slot (no external logic for it),
and x4 (off the PCH) for the second x16 slot. The bandwidth is not shared any more,
to the same extent it once was. Any arch changes, the extra lanes are
used for NVMe devices, rather than as card slots.

You have to watch modern machines, for how they're doing VCore. The
heatsinks are too small. The processor draws over 200W under turbo,
and this stresses the VCore a bit. I prefer to add cooling so
it is less stressed, but, they don't make it easy. To do it properly,
I'd have to change cases to make room for a fan in the right position.
On the Test machine, I did a kind of "blower", mounted off the
end of the drive bay, to get cooling onto VCore on that one. I burned
my finger on VCore while testing on the kitchen table on the Test Machine,
and that's when I determined I had not made a clever product choice.

If you get a Dell or an HP, you know the VCore is going to be
"even thinner". And at least in the case of Dell, the BIOS is
likely to be some sort of unacceptable piece of work.

I guess I still prefer to build, but for slightly different
reasons than before. Getting a good BIOS is part of the puzzle,
one with settings that work.

A relatively cheap way to build a system, would be a good AM4 board,
an AMD Zen3 5600G, and your DDR4 RAM reused. But I suspect the 5600G
are out of production, because they're too cheap and don't make
AMD enough money. You could go upscale a bit, and get the one with
additional L3 on it.

[Picture] 5600G

https://i.postimg.cc/BZYR3B7x/cpuz.gif

Another thing to watch, is the CPU cooler. The 5600G system in that
picture, it uses the stock aluminum cooler. the adaptive cooling
system in the BIOS, doesn't even modulate the fan at all, because
there isn't enough stress for it to track what is going on.

When you put a 200W CPU in there, the instructions say "AMD recommends
a water cooler". I thought this was some kind of joke, but if you
want the Turbo to get close to the "rating on the tin", it really
appears you do need water.

For a cooler, I decided to skip the Noctua and tried a AK620 Zero Dark
(deepcool.com) and it was a bit cheaper than a Noctua would have been.
And it actually handles the heat OK. If the cooler is under-sized on
a high power processor, you might see weird "thermal spikes" in a
temperature trace. With the more capable cooler fitted, the control loop
settles right down. With water, I would guess I'd get more than 4.55GHz,
but not enough for me to want water inside the poor thing. I guess I
just like air cooling.

[Picture] 5950x at idle and full power (62C) on Deepcool AK620

https://i.postimg.cc/XJbVph8f/power-control-recent-CPU.gif

You don't get to use the cores all that often, and I got the
setup just for 7ZIP.

[Picture]

https://i.postimg.cc/RVJz4kZC/cpu-flat-out.gif

It still takes all day, to compress a backup file. And at
224W wall power measured, you only want to be doing that
sort of compression in winter, when the air is cooler in here.

Maybe an Alienware would give you some of the properties of
an enthusiast machine, but then you're paying $1000 more for
the name (like "Cadillac").

With your experience, I'm sure it won't take you that
long to pick something.

Paul

I've let the panic fade away and after checking out the cost of off-
the-shelf systems, my persecutive today is that I can do better for
less if I build myself one. I should be able to recycle the case, RAM,
cooler, and drives. I already have this new PSU a home. So I'll need a
new motherboard and CPU, and I guess a new Windows license and I can be
back in business. I'll have a better system for no more than the flimsy
Dell-types.

One of the main duties for this computer has been music recording. I
had been clinging to my beloved ancient PCi audio adapter, but it was a
problem even in this computer and I've been casually searching for a
USB replacement. This situation will force the issue. But at least I
can permanently forget the PCi requirement.

I actually have another desktop (a Dell!) that I can make my main
computer until the new build comes together. My first task is to get
some data off the SSD in the dead computer. I have a SSD adapter on its
way - after that I can relax and take my time to make the right
choices.

Thanks for your good ideas. I needed someone to bounce this off of and
nobody around hear understands or cares about this kind of computer
nerd stuff. To most people I know, a desktop computer is an antique
curiosity, like a rotary telephone.

On 21 Mar 2024, Paul <nospam@needed.invalid> wrote in
alt.comp.hardware.pc-homebuilt:

> You have some DDR4 in hand. You could look for a different
> solution, and try to reuse the RAM at least.
>
> A lot of the boards now, no longer have a PCI slot. They may still
> have a kbd/mouse PS/2 hole, for at least one of the two appliances
> to go. Some even have a serial RS232 port. Both my machines have
> RS232, even though it was hardly advertised as being present. On
> one machine, I just noticed it one day while examining USB2
> headers, and I noticed "one header has a strange pinout". And that
> was an RS232 when I looked it up.
>
> The PCI Express slots may not be as useful as before. Before, we
> had x8/x8 on two slots, and bifurcation logic. Today, the
> mainstream boards are still sporting x16 on one slot (no external
> logic for it), and x4 (off the PCH) for the second x16 slot. The
> bandwidth is not shared any more, to the same extent it once was.
> Any arch changes, the extra lanes are used for NVMe devices,
> rather than as card slots.
>
> You have to watch modern machines, for how they're doing VCore.
> The heatsinks are too small. The processor draws over 200W under
> turbo, and this stresses the VCore a bit. I prefer to add cooling
> so it is less stressed, but, they don't make it easy. To do it
> properly, I'd have to change cases to make room for a fan in the
> right position. On the Test machine, I did a kind of "blower",
> mounted off the end of the drive bay, to get cooling onto VCore on
> that one. I burned my finger on VCore while testing on the kitchen
> table on the Test Machine, and that's when I determined I had not
> made a clever product choice.
>
> If you get a Dell or an HP, you know the VCore is going to be
> "even thinner". And at least in the case of Dell, the BIOS is
> likely to be some sort of unacceptable piece of work.
>
> I guess I still prefer to build, but for slightly different
> reasons than before. Getting a good BIOS is part of the puzzle,
> one with settings that work.
>
> A relatively cheap way to build a system, would be a good AM4
> board, an AMD Zen3 5600G, and your DDR4 RAM reused. But I suspect
> the 5600G are out of production, because they're too cheap and
> don't make AMD enough money. You could go upscale a bit, and get
> the one with additional L3 on it.
>
> [Picture] 5600G
>
> https://i.postimg.cc/BZYR3B7x/cpuz.gif
>
> Another thing to watch, is the CPU cooler. The 5600G system in
> that picture, it uses the stock aluminum cooler. the adaptive
> cooling system in the BIOS, doesn't even modulate the fan at all,
> because there isn't enough stress for it to track what is going
> on.
>
> When you put a 200W CPU in there, the instructions say "AMD
> recommends a water cooler". I thought this was some kind of joke,
> but if you want the Turbo to get close to the "rating on the tin",
> it really appears you do need water.
>
> For a cooler, I decided to skip the Noctua and tried a AK620 Zero
> Dark (deepcool.com) and it was a bit cheaper than a Noctua would
> have been. And it actually handles the heat OK. If the cooler is
> under-sized on a high power processor, you might see weird
> "thermal spikes" in a temperature trace. With the more capable
> cooler fitted, the control loop settles right down. With water, I
> would guess I'd get more than 4.55GHz, but not enough for me to
> want water inside the poor thing. I guess I just like air cooling.
>
> [Picture] 5950x at idle and full power (62C) on Deepcool AK620
>
> https://i.postimg.cc/XJbVph8f/power-control-recent-CPU.gif
>
> You don't get to use the cores all that often, and I got the
> setup just for 7ZIP.
>
> [Picture]
>
> https://i.postimg.cc/RVJz4kZC/cpu-flat-out.gif
>
> It still takes all day, to compress a backup file. And at
> 224W wall power measured, you only want to be doing that
> sort of compression in winter, when the air is cooler in here.
>
> Maybe an Alienware would give you some of the properties of
> an enthusiast machine, but then you're paying $1000 more for
> the name (like "Cadillac").
>
> With your experience, I'm sure it won't take you that
> long to pick something.
>
> Paul
>

It's turning out different than I anticipated. After my computer died,
I figured that that PSU was bad and I replaced it. When it still
wouldn't boot up I figured the motherboard was bad, too. Time to make a
new computer sez I, so today I bought a new motherboard and CPU, and,
because I still had a sneaking suspicion, yet another PSU.

When I got home, to check my suspicion, I put PSU #3 in the old
computer... and it booted up! So, it turns out that new PSU #2, the
first replacement, was defective. The guy at the checkout at Micro
Center had actually warned me that he had recently seen a lot of those
Corsair PSUs come back as DOA, but I had already paid and was in a
hurry and (wrongly) figured the odds were in my favor.

So, everything would be hunky-dory now except that I discovered that
when the original PSU died, it killed my D hard drive where some of my
most precious data was stored. My most recent backup is not all that
recent. Nothing I can do about that now.

Now I have a motherboard (ASUS Prime B550-PLUS) and CPU (AMD Ryzen 7
5700G) that I can either return to the store or go ahead and build a
new computer. I need to sleep on it.

On 22 Mar 2024, Nil <rednoise9@rednoise9.invalid> wrote in
alt.comp.hardware.pc-homebuilt:

> I've let the panic fade away and after checking out the cost of
> off- the-shelf systems, my persecutive today is that I can do
> better for less if I build myself one. I should be able to recycle
> the case, RAM, cooler, and drives. I already have this new PSU a
> home. So I'll need a new motherboard and CPU, and I guess a new
> Windows license and I can be back in business. I'll have a better
> system for no more than the flimsy Dell-types.
>
> One of the main duties for this computer has been music recording.
> I had been clinging to my beloved ancient PCi audio adapter, but
> it was a problem even in this computer and I've been casually
> searching for a USB replacement. This situation will force the
> issue. But at least I can permanently forget the PCi requirement.
>
> I actually have another desktop (a Dell!) that I can make my main
> computer until the new build comes together. My first task is to
> get some data off the SSD in the dead computer. I have a SSD
> adapter on its way - after that I can relax and take my time to
> make the right choices.
>
> Thanks for your good ideas. I needed someone to bounce this off of
> and nobody around hear understands or cares about this kind of
> computer nerd stuff. To most people I know, a desktop computer is
> an antique curiosity, like a rotary telephone.
>
>
> On 21 Mar 2024, Paul <nospam@needed.invalid> wrote in
> alt.comp.hardware.pc-homebuilt:
>
>> You have some DDR4 in hand. You could look for a different
>> solution, and try to reuse the RAM at least.
>>
>> A lot of the boards now, no longer have a PCI slot. They may
>> still have a kbd/mouse PS/2 hole, for at least one of the two
>> appliances to go. Some even have a serial RS232 port. Both my
>> machines have RS232, even though it was hardly advertised as
>> being present. On one machine, I just noticed it one day while
>> examining USB2 headers, and I noticed "one header has a strange
>> pinout". And that was an RS232 when I looked it up.
>>
>> The PCI Express slots may not be as useful as before. Before, we
>> had x8/x8 on two slots, and bifurcation logic. Today, the
>> mainstream boards are still sporting x16 on one slot (no external
>> logic for it), and x4 (off the PCH) for the second x16 slot. The
>> bandwidth is not shared any more, to the same extent it once was.
>> Any arch changes, the extra lanes are used for NVMe devices,
>> rather than as card slots.
>>
>> You have to watch modern machines, for how they're doing VCore.
>> The heatsinks are too small. The processor draws over 200W under
>> turbo, and this stresses the VCore a bit. I prefer to add cooling
>> so it is less stressed, but, they don't make it easy. To do it
>> properly, I'd have to change cases to make room for a fan in the
>> right position. On the Test machine, I did a kind of "blower",
>> mounted off the end of the drive bay, to get cooling onto VCore
>> on that one. I burned my finger on VCore while testing on the
>> kitchen table on the Test Machine, and that's when I determined I
>> had not made a clever product choice.
>>
>> If you get a Dell or an HP, you know the VCore is going to be
>> "even thinner". And at least in the case of Dell, the BIOS is
>> likely to be some sort of unacceptable piece of work.
>>
>> I guess I still prefer to build, but for slightly different
>> reasons than before. Getting a good BIOS is part of the puzzle,
>> one with settings that work.
>>
>> A relatively cheap way to build a system, would be a good AM4
>> board, an AMD Zen3 5600G, and your DDR4 RAM reused. But I suspect
>> the 5600G are out of production, because they're too cheap and
>> don't make AMD enough money. You could go upscale a bit, and get
>> the one with additional L3 on it.
>>
>> [Picture] 5600G
>>
>> https://i.postimg.cc/BZYR3B7x/cpuz.gif
>>
>> Another thing to watch, is the CPU cooler. The 5600G system in
>> that picture, it uses the stock aluminum cooler. the adaptive
>> cooling system in the BIOS, doesn't even modulate the fan at all,
>> because there isn't enough stress for it to track what is going
>> on.
>>
>> When you put a 200W CPU in there, the instructions say "AMD
>> recommends a water cooler". I thought this was some kind of joke,
>> but if you want the Turbo to get close to the "rating on the
>> tin", it really appears you do need water.
>>
>> For a cooler, I decided to skip the Noctua and tried a AK620 Zero
>> Dark (deepcool.com) and it was a bit cheaper than a Noctua would
>> have been. And it actually handles the heat OK. If the cooler is
>> under-sized on a high power processor, you might see weird
>> "thermal spikes" in a temperature trace. With the more capable
>> cooler fitted, the control loop settles right down. With water, I
>> would guess I'd get more than 4.55GHz, but not enough for me to
>> want water inside the poor thing. I guess I just like air
>> cooling.
>>
>> [Picture] 5950x at idle and full power (62C) on Deepcool AK620
>>
>> https://i.postimg.cc/XJbVph8f/power-control-recent-CPU.gif
>>
>> You don't get to use the cores all that often, and I got the
>> setup just for 7ZIP.
>>
>> [Picture]
>>
>> https://i.postimg.cc/RVJz4kZC/cpu-flat-out.gif
>>
>> It still takes all day, to compress a backup file. And at
>> 224W wall power measured, you only want to be doing that
>> sort of compression in winter, when the air is cooler in here.
>>
>> Maybe an Alienware would give you some of the properties of
>> an enthusiast machine, but then you're paying $1000 more for
>> the name (like "Cadillac").
>>
>> With your experience, I'm sure it won't take you that
>> long to pick something.
>>
>> Paul

On 3/24/2024 12:06 AM, Nil wrote:
> It's turning out different than I anticipated. After my computer died,
> I figured that that PSU was bad and I replaced it. When it still
> wouldn't boot up I figured the motherboard was bad, too. Time to make a
> new computer sez I, so today I bought a new motherboard and CPU, and,
> because I still had a sneaking suspicion, yet another PSU.
>
> When I got home, to check my suspicion, I put PSU #3 in the old
> computer... and it booted up! So, it turns out that new PSU #2, the
> first replacement, was defective. The guy at the checkout at Micro
> Center had actually warned me that he had recently seen a lot of those
> Corsair PSUs come back as DOA, but I had already paid and was in a
> hurry and (wrongly) figured the odds were in my favor.
>
> So, everything would be hunky-dory now except that I discovered that
> when the original PSU died, it killed my D hard drive where some of my
> most precious data was stored. My most recent backup is not all that
> recent. Nothing I can do about that now.
>
> Now I have a motherboard (ASUS Prime B550-PLUS) and CPU (AMD Ryzen 7
> 5700G) that I can either return to the store or go ahead and build a
> new computer. I need to sleep on it.

I wonder how these DOA power supplies manage to ship ?

Corsair has them contract manufactured, but they're supposed
to at least turn them on, after assembly.

*******

Your software licenses align with the old system,
so that has some value.

*******

The hard drive is supposed to have two protection devices.
One sits across the 12V rail. If the rail rises to 15V,
it starts to conduct. The purpose of a transient suppressor,
is for "hot disconnect" which is a feature of SATA. If the
power connector were to be pulled away, the DC current wants
to continue to flow, and the voltage may have an inductive
spike. The transient device is intended for very short spikes,
and helps protect the components.

Both the 5V rail and the 12V rail have one.

If an incident happens, where an elevated voltage stays
there for a duration of a second or two, that can fry those
components and perhaps burn the label on them.

The "concern points" would be, whether a head amplifier
got damaged, or a motor winding got damaged. The controller
board could be replaced (but you have to move a ROM from
the old controller board to the new controller board,
and you don't know whether the ROM is still good). The
drives now tend to have FDE support, and it's just a guess
on my part, that the ROM they tell you to move, has the
FDE key in it. You can't get the data off, if the key
information is not correct. Whether every drive has
a unique key, or whether unencrypted drives all use the
same key, again, I don't know. Since the HGST website
that "explained hard drives" is gone, that was my only
source of information.

Paul

On 24 Mar 2024, Paul <nospam@needed.invalid> wrote in
alt.comp.hardware.pc-homebuilt:

> The "concern points" would be, whether a head amplifier
> got damaged, or a motor winding got damaged. The controller
> board could be replaced (but you have to move a ROM from
> the old controller board to the new controller board,
> and you don't know whether the ROM is still good). The
> drives now tend to have FDE support, and it's just a guess
> on my part, that the ROM they tell you to move, has the
> FDE key in it. You can't get the data off, if the key
> information is not correct. Whether every drive has
> a unique key, or whether unencrypted drives all use the
> same key, again, I don't know. Since the HGST website
> that "explained hard drives" is gone, that was my only
> source of information.

I watched a few videos about hard disk recovery. I doubt the
mechanicals in my drive are damaged, just the electronics, so I was
wondering if I could just replace the controller board myself - I even
found a site where you could buy them for various brands. The recovery
videos I saw showed the guy discovering and replacing various fried
components using a microscope and surgical tools. I can solder, but not
like that! I didn't understand the role of the ROMs then, but you clued
me in. This is almost certainly not a job I can do. Commercial data
recovery services are VERY VERY expensive. I can recover much of what I
lost from my spotty backups, and there was a lot of junk I had
collected. There were a few irreplaceable things. I may just have to
learn to live without them.

On 3/24/2024 2:00 PM, Nil wrote:
> On 24 Mar 2024, Paul <nospam@needed.invalid> wrote in
> alt.comp.hardware.pc-homebuilt:
>
>> The "concern points" would be, whether a head amplifier
>> got damaged, or a motor winding got damaged. The controller
>> board could be replaced (but you have to move a ROM from
>> the old controller board to the new controller board,
>> and you don't know whether the ROM is still good). The
>> drives now tend to have FDE support, and it's just a guess
>> on my part, that the ROM they tell you to move, has the
>> FDE key in it. You can't get the data off, if the key
>> information is not correct. Whether every drive has
>> a unique key, or whether unencrypted drives all use the
>> same key, again, I don't know. Since the HGST website
>> that "explained hard drives" is gone, that was my only
>> source of information.
>
> I watched a few videos about hard disk recovery. I doubt the
> mechanicals in my drive are damaged, just the electronics, so I was
> wondering if I could just replace the controller board myself - I even
> found a site where you could buy them for various brands. The recovery
> videos I saw showed the guy discovering and replacing various fried
> components using a microscope and surgical tools. I can solder, but not
> like that! I didn't understand the role of the ROMs then, but you clued
> me in. This is almost certainly not a job I can do. Commercial data
> recovery services are VERY VERY expensive. I can recover much of what I
> lost from my spotty backups, and there was a lot of junk I had
> collected. There were a few irreplaceable things. I may just have to
> learn to live without them.
>

I just wanted to get the message across, whereas in the past
you could just do a board swap and "try your hand at it",
on modern drives this is not going to work, unless you
move the ROM from the old board to the new. There are
actually several ROM devices, but one of them appears
to be a "custom", programmed at the factory just for that
particular serial number of drive. Once information is written,
with respect to that ROM, the ROM stays with the HDA. That's my
guess as to what is going on. Old drives did not need
this care and attention.

You may be able to find an article somewhere, demonstrating
the swapping of the thing.

Paul

Just curious, will it go into BIOS Mode? If not, I would clear the BIOS
and try again, and if nothing, I would give up on the MB. But I am a
"hardware idiot" compared to Paul. Maybe you can just replace the MB,
to get your hardware running again, at a relatively inexpensive price?
Good luck! I understand how frustrating this stuff can be--I was there
2 weeks ago with my router that became finicky (and I quickly
replaced/upgraded).

On 3/20/2024 9:10 PM, Nil wrote:
> I hope I can get some guidance here. My desktop computer that I built
> several years ago just died on me - as I was watching, it appeared to
> be going into sleep mode, which it does after some hours of
> inactivity. However, it could not be awakened, it was dead. I
> immediately thought the power supply had gone bad, and in fact, I
> tested it with a multimeter and it was, indeed dead - no voltage
> detectable on any pin. So, I bought a new PS of similar specs. Now, the
> motherboard (ASUS Prime Z270-K) lights up, but there is no other
> activity, no fans going, no display, no disc drives spinning. I can
> only wonder if when the PS blew it took out the mobo or some other
> component.
>
> So, my question is, 'what can/should I do now to confirm that the mobo
> is truly, irretrievably gone?' If I can't, I will proceed toward a new
> build or buy, but if I can revive the existing one, VERY much the
> better. After all, it's tuned to a fair-thee-well and I don't look
> forward to going through months of tweaking to get it to that point
> again.
>
> Aside... A year or so ago I replaced the mechanical boot drive with an
> SSD drive. If THAT had gone bad, would the mobo not still boot with at
> least a display and error message?

The computer was dead and lifeless as a doornail. I correctly guessed
that the power supply had died, and I bought and installed a
replacement. What I didn't anticipate was that the REPLACEMENT PS was
also defective and I ran around in circles for a few days thinking the
motherboard was bad, instead or also. I finally installed a second
replacement PSU and I'm back in the saddle. Unfortunately, the dying
PSU killed one of my hard disk drives and it will take me a while to
recover from that.

Frustrating, yes. It has probably happened, but I don't recall a PS
just winking out like that. Usually they get flaky and let me know they
are on their way out. I used to work in a computer lab where we had
lots of spare parts to test things, but now, at home, it's a lot of
guesswork and blind buying of parts.

On 24 Mar 2024, Bill <nonegiven@att.net> wrote in
alt.comp.hardware.pc-homebuilt:

> Just curious, will it go into BIOS Mode? If not, I would clear
> the BIOS and try again, and if nothing, I would give up on the MB.
> But I am a "hardware idiot" compared to Paul. Maybe you can just
> replace the MB, to get your hardware running again, at a
> relatively inexpensive price? Good luck! I understand how
> frustrating this stuff can be--I was there 2 weeks ago with my
> router that became finicky (and I quickly replaced/upgraded).
>
>
> On 3/20/2024 9:10 PM, Nil wrote:
>> I hope I can get some guidance here. My desktop computer that I
>> built several years ago just died on me - as I was watching, it
>> appeared to be going into sleep mode, which it does after some
>> hours of inactivity. However, it could not be awakened, it was
>> dead. I immediately thought the power supply had gone bad, and in
>> fact, I tested it with a multimeter and it was, indeed dead - no
>> voltage detectable on any pin. So, I bought a new PS of similar
>> specs. Now, the motherboard (ASUS Prime Z270-K) lights up, but
>> there is no other activity, no fans going, no display, no disc
>> drives spinning. I can only wonder if when the PS blew it took
>> out the mobo or some other component.
>>
>> So, my question is, 'what can/should I do now to confirm that the
>> mobo is truly, irretrievably gone?' If I can't, I will proceed
>> toward a new build or buy, but if I can revive the existing one,
>> VERY much the better. After all, it's tuned to a fair-thee-well
>> and I don't look forward to going through months of tweaking to
>> get it to that point again.
>>
>> Aside... A year or so ago I replaced the mechanical boot drive
>> with an SSD drive. If THAT had gone bad, would the mobo not still
>> boot with at least a display and error message?

On 3/27/2024 1:59 AM, Nil wrote:
> The computer was dead and lifeless as a doornail. I correctly guessed
> that the power supply had died, and I bought and installed a
> replacement. What I didn't anticipate was that the REPLACEMENT PS was
> also defective and I ran around in circles for a few days thinking the
> motherboard was bad, instead or also. I finally installed a second
> replacement PSU and I'm back in the saddle.

* Congratulations! *

<end of post>

Unfortunately, the dying
> PSU killed one of my hard disk drives and it will take me a while to
> recover from that.
>
> Frustrating, yes. It has probably happened, but I don't recall a PS
> just winking out like that. Usually they get flaky and let me know they
> are on their way out. I used to work in a computer lab where we had
> lots of spare parts to test things, but now, at home, it's a lot of
> guesswork and blind buying of parts.
>
>
> On 24 Mar 2024, Bill <nonegiven@att.net> wrote in
> alt.comp.hardware.pc-homebuilt:
>
>> Just curious, will it go into BIOS Mode? If not, I would clear
>> the BIOS and try again, and if nothing, I would give up on the MB.
>> But I am a "hardware idiot" compared to Paul. Maybe you can just
>> replace the MB, to get your hardware running again, at a
>> relatively inexpensive price? Good luck! I understand how
>> frustrating this stuff can be--I was there 2 weeks ago with my
>> router that became finicky (and I quickly replaced/upgraded).
>>
>>
>> On 3/20/2024 9:10 PM, Nil wrote:
>>> I hope I can get some guidance here. My desktop computer that I
>>> built several years ago just died on me - as I was watching, it
>>> appeared to be going into sleep mode, which it does after some
>>> hours of inactivity. However, it could not be awakened, it was
>>> dead. I immediately thought the power supply had gone bad, and in
>>> fact, I tested it with a multimeter and it was, indeed dead - no
>>> voltage detectable on any pin. So, I bought a new PS of similar
>>> specs. Now, the motherboard (ASUS Prime Z270-K) lights up, but
>>> there is no other activity, no fans going, no display, no disc
>>> drives spinning. I can only wonder if when the PS blew it took
>>> out the mobo or some other component.
>>>
>>> So, my question is, 'what can/should I do now to confirm that the
>>> mobo is truly, irretrievably gone?' If I can't, I will proceed
>>> toward a new build or buy, but if I can revive the existing one,
>>> VERY much the better. After all, it's tuned to a fair-thee-well
>>> and I don't look forward to going through months of tweaking to
>>> get it to that point again.
>>>
>>> Aside... A year or so ago I replaced the mechanical boot drive
>>> with an SSD drive. If THAT had gone bad, would the mobo not still
>>> boot with at least a display and error message?