20-pin PSU to 24-pin MoBo disaster?

  • Thread starter Thread starter Avoid9Pdf
  • Start date Start date
A

Avoid9Pdf

Previously I enquired here about a CPU for my SATA+IDE MoBo.
Although I failed to ID the MoBo, a shady 2nd-hand shop sold me
a CPU, which when plugged in at the shop showed life on their display.

When I got home, I removed my working PSU and tried it.
In the inconvenience/restriction, I just concentrated on lining-up
the plug-socket centerclip. Only after it failed and I removed the
connection did I realise that I had used my 20-pin plug on the MoBo's
24-pin socket !!!

I was surprised that it didn't 'smoke', and I noted that the CPU warmed
up slightly, as it did during the shady-shop test.

http://www.overclockers.com/forums/showthread.php?t=401656
writes: which seems WRONG:--
* Due to the fact that the pins are keyed, I believe there is only
one way it will go on.
* If you line it up so the clips match it will fit right in and
the spare 4 will be located correctly.

Then I bought a cheapo 24-pin PSU, but got no life.

The PSU at the shady-shop was a 'heavy' one, and the MoBo
has substantial finning and is marked "PCI express"; which
makes me think it's got a power-hungry video system, and
maybe it would work with a 'powerful' PSU, if the smart
protection circuitry prevented damage by 'center-plugging'
the 20-pin plug to the 24-pin socket.

Should I risk more expense for a Hi-wattage PSU, for this
disasterous project?

== TIA
 
Previously I enquired here about a CPU for my SATA+IDE MoBo.
Although I failed to ID the MoBo, a shady 2nd-hand shop sold me
a CPU, which when plugged in at the shop showed life on their display.

When I got home, I removed my working PSU and tried it.
In the inconvenience/restriction, I just concentrated on lining-up
the plug-socket centerclip. Only after it failed and I removed the
connection did I realise that I had used my 20-pin plug on the MoBo's
24-pin socket !!!

I was surprised that it didn't 'smoke', and I noted that the CPU warmed
up slightly, as it did during the shady-shop test.

http://www.overclockers.com/forums/showthread.php?t=401656
writes: which seems WRONG:--
* Due to the fact that the pins are keyed, I believe there is only
one way it will go on.
* If you line it up so the clips match it will fit right in and
the spare 4 will be located correctly.

Then I bought a cheapo 24-pin PSU, but got no life.

The PSU at the shady-shop was a 'heavy' one, and the MoBo
has substantial finning and is marked "PCI express"; which
makes me think it's got a power-hungry video system, and
maybe it would work with a 'powerful' PSU, if the smart
protection circuitry prevented damage by 'center-plugging'
the 20-pin plug to the 24-pin socket.

Should I risk more expense for a Hi-wattage PSU, for this
disasterous project?

== TIA

You're going to need to tell us a bit more about *exactly*
what you did.

Using this page, I can get some pictures to refer to.

http://www.playtool.com/pages/psuconnectors/connectors.html

This is a 20 pin PSU, in a 24 pin motherboard. Notice that
one end of the connectors, lines up (pin 1 with pin 1).

http://www.playtool.com/pages/psuconnectors/20in24.jpg

Now, I drew myself some pictures, and these are my observations.

1) The connector cannot be rotated 180 degrees. It will never
mate in that case, no matter how much you shift left or right.
It only mates if the latches are on the same side.

2) A linear shift of four positions in either direction, will mate.
The pin pattern repeats every four pins. If you move the connector
that far in either direction, it hangs over the end of the
24 pin connector by a lot. The latches wouldn't line up.

I drew enough examples, until it occurred to me the pattern
repeated every four pins. Then I stopped drawing them.

http://img832.imageshack.us/img832/4009/conn24.gif

There is a good chance something bad would happen, if you offset by 4.
The playtool.com page has the pinout listed, and you can see things
would get shorted together on the power supply.

The processor can only get warm, if ATX12V is connected. An exception
is Biostar motherboards, which had a habit of connecting the main connector
12V to the processor VCore circuit as well. But most of the time,
you'd need to connect ATX12V. This one powers the processor, not
the main connector. Do you have this plugged in ?

http://www.playtool.com/pages/psuconnectors/12v4pin.jpg

HTH,
Paul
 
They make 20 pin to 24 pin adapters if you are really worried about
this. However, I've put a 20 pin in a 24 pin socket without a problem.
Now this was on a low power Atom board.

I had a front panel USB cable hooked up wrong to the mobo once on a
different computer. It wouldn't boot. My point being there are other
ways to screw up building a PC regarding power, but I think a 20 into 24
isn't one of them.

I'm not sure what the OP uses this shady shop unless he/she is getting
shady deals. Most of the time I just go to New Egg or Frys.
 
They make 20 pin to 24 pin adapters if you are really worried about
this. However, I've put a 20 pin in a 24 pin socket without a problem.
Now this was on a low power Atom board.

I had a front panel USB cable hooked up wrong to the mobo once on a
different computer. It wouldn't boot. My point being there are other
ways to screw up building a PC regarding power, but I think a 20 into 24
isn't one of them.

I'm not sure what the OP uses this shady shop unless he/she is getting
shady deals. Most of the time I just go to New Egg or Frys.

Also check to make sure you hooked up the other 12v (Either its a 4 wire or 8
wire connector that has only yellow and black wires)
 
Previously I enquired here about a CPU for my SATA+IDE MoBo.
Although I failed to ID the MoBo, a shady 2nd-hand shop sold
me a CPU, which when plugged in at the shop showed life on
their display. When I got home,
Only after it failed and I removed the connection did I
realise that I had used my 20-pin plug on the MoBo's 24-pin
socket !!! I was surprised that
Then I bought a cheapo 24-pin PSU, but got no life. The PSU
at the shady-shop was a 'heavy' one, and the MoBo has
substantial finning and is marked "PCI express"; which
makes me think it's got a power-hungry video system, and
maybe it would work with a 'powerful' PSU, if the smart
protection circuitry prevented damage by 'center-plugging'
the 20-pin plug to the 24-pin socket. Should I risk more
expense for a Hi-wattage PSU, for this disasterous project? == TIA

Did you do what GMAN said about the separate 4-pin or 8-pin connector
for the CPU power?

The extra 4 pins for the 24-pin connector provide one extra wire
for each voltage and ground but probably aren't needed because most
of the power for the motherboard goes through the separate 4-pin
or 8-pin connector that plugs in close to the CPU socket. I think
I've measured a maximum of 8-10 amps for any one voltage on the
20-24 pin connector, meaning it shouldn't hurt to have just the
20-pin portion of the connector plugged in because each voltage
runs on at least 3 wires, and each pin is good for at least 6-8
amps.

It actually is possible to plug the 20-24 pin connector in the
wrong way with some cheapo power supplies and motherboards, and
I managed that with an Ultra PSU made by Wintech and some really
old ECS or VisionTop motherboard -- not a tight fit. I was not
able to duplicate that with any other mobo or PSU combination.

I've never heard of a motherboard's built-in video requiring lots
of power, so any decent 350W power supply should be able to run
that video even with the fastest, most power hungry CPU (but 2
CPUs might not work). OTOH some cheapo 500W - 600W PSUs will fail
at less than 300W, as JonnyGuru.com has shown, and they all look
like they're built to put out only 200W - 300W. In the past few
months, NewEgg and TigerDirect have offered high quality 600W
PSUs for just $30-40, after rebate. I'm cheap but have never
bought a junk PSU, except when it was included with a case that
was free after rebate, because it's just not worth the risk.
 
You're going to need to tell us a bit more about *exactly*
what you did.

Using this page, I can get some pictures to refer to.
--snip--

Hi Paul, yes this is the same old garbage that I've asked about
before. No need to fetch *jpg.
Here's the schematic of the 2 types:-
x1234567890x <-20pin

123456789012 <-24pin
x is the double-pin at each end of the 24pin socket which
remained unoccupied by the 20pin plug.

That's how I mis-inseted; since the 'hook' which is between
5,6 on the 20pin, is between 6,7 on the 24pin.

As you well know, and assuming my papers on hand are
correct: if pin1 of 20pin was 'lined up' with pin1 of 24pin,
all 20pin's would be feeding correctly. And the remaining
ADJACENT 4pins are just duplicate-lines.

SO if it was plugged in thus, all 20 lines would match the
24pin requirement:-----------------

1234567890xx <-20pin

123456789012 <-24pin
-------------------

I'm beginning to think that the industry would design so that
I could NOT do what I'm claiming I did.
But I can't disable my machine now just to reconfirm.

Is it difficult to explain how pin8:PWR_OK works?

But for the way I [claim to have] plugged it:
PSU8 <-> MoBo9. So no protection.
Unless PWR_OK need a specific signal before
allowing power. I.e the default is to NOT
allow power -- hopefully?

== TIA.
 
You're going to need to tell us a bit more about *exactly*
what you did.

Using this page, I can get some pictures to refer to.
--snip--

Hi Paul, yes this is the same old garbage that I've asked about
before. No need to fetch *jpg.
Here's the schematic of the 2 types:-
x1234567890x <-20pin

123456789012 <-24pin
x is the double-pin at each end of the 24pin socket which
remained unoccupied by the 20pin plug.

That's how I mis-inseted; since the 'hook' which is between
5,6 on the 20pin, is between 6,7 on the 24pin.

As you well know, and assuming my papers on hand are
correct: if pin1 of 20pin was 'lined up' with pin1 of 24pin,
all 20pin's would be feeding correctly. And the remaining
ADJACENT 4pins are just duplicate-lines.

SO if it was plugged in thus, all 20 lines would match the
24pin requirement:-----------------

1234567890xx <-20pin

123456789012 <-24pin
-------------------

I'm beginning to think that the industry would design so that
I could NOT do what I'm claiming I did.
But I can't disable my machine now just to reconfirm.

Is it difficult to explain how pin8:PWR_OK works?

But for the way I [claim to have] plugged it:
PSU8 <-> MoBo9. So no protection.
Unless PWR_OK need a specific signal before
allowing power. I.e the default is to NOT
allow power -- hopefully?

== TIA.

The male and female plastic shells, the barrel around
each pin, has a rather unique shape. That is supposed to
prevent some of the alignment cases from mating. The
one you describe, should not be possible.

x1234567890x <-20pin

123456789012 <-24pin

The Power_OK is a signal from the power supply, to the
motherboard. It is meant to indicate when all rails
are charged pretty well towards the final voltage.
It tells the motherboard, that it can come out of
reset state and start drawing more power if it wants.

On the formfactors.org site, there are standards for
the ATX power supply. And in there, it explains how
Pwr_OK works (a waveform is shown).

Paul
 
There is a good chance something bad would happen, if you offset by 4.

Lets rather talk in terms of ROW numbers: like my 'diagram'.
From the docos: signals on Rows1..10 match for 20Pin & 24Pin devices.
The playtool.com page has the pinout listed, and you can see things
would get shorted together on the power supply.

Well the PSU with 20Pin is running right now.
I'm thinking of the MoBo being destroyed, rather than the PSU.
The processor can only get warm, if ATX12V is connected.

Isn't the CPU mainly a 5/3.3V device?
I would have expected fatal damage to the MoBo to cause the
CPU to not warmup. And I'm still hoping that a Hi-wattage
PSU can drive it. And I expect the system to be designed so
that a dud MoBo cannot destroy a PSU.

WDYS?

== TIA.
 
Lets rather talk in terms of ROW numbers: like my 'diagram'.
From the docos: signals on Rows1..10 match for 20Pin & 24Pin devices.


Well the PSU with 20Pin is running right now.
I'm thinking of the MoBo being destroyed, rather than the PSU.


Isn't the CPU mainly a 5/3.3V device?
I would have expected fatal damage to the MoBo to cause the
CPU to not warmup. And I'm still hoping that a Hi-wattage
PSU can drive it. And I expect the system to be designed so
that a dud MoBo cannot destroy a PSU.

WDYS?

== TIA.

There are differences in Power supplies. You may have an ATX12V 1.x
power supply and your motherboard needs possibly a EPS12V supply. See here







http://en.wikipedia.org/wiki/ATX#Interchanging_PSUs


Interchanging PSUsAlthough the ATX power supply specifications are mostly
vertically compatible in both ways (both electrically and physically), there
are potential issues with mixing old motherboards/systems with new PSUs, and
vice versa. The main issues to consider are the following:

The power distribution biases across 3.3 V, 5 V, and 12 V rails are very
different between older and newer ATX PSU designs, as well as between older
and newer PC system designs.

Older PSUs may not have connectors which are required for newer PC systems to
properly operate.

Newer systems generally have higher power requirements than older systems.
This is a practical guidance what to mix and what not to mix:

Older systems (before Pentium 4 and Athlon XP platforms) were designed to draw
most power from 5 V and 3.3 V rails.

Because of the DC-DC converters on the motherboard that convert 12 V to the
low voltages required by the Intel Pentium 4 and AMD Athlon XP (and
subsequent) processors, such systems draw most of their power from the 12 V
rail.

Original ATX PSUs have power distribution designed for pre-P4/XP PCs. They
lack the supplemental 4-pin 12-volt CPU power connector, so they most likely
cannot be used with P4/XP or newer motherboards. Adapters do exist but power
drain on the 12 V rail must be checked very carefully. There is a chance it
can work without connecting the 4-pin 12 V connector, but caution is
advised.[17][dead link]

ATX12V 1.x PSUs have power distribution designed for P4/XP PCs, but they are
also greatly suitable for older PCs, since they give plenty of power (relative
to old PCs' needs) both on 12 V and on 5 V/3.3 V. It is not recommended to use
ATX12V 1.x PSUs on ATX12V 2.x motherboards because those systems require much
more power on 12 V, and much less on 3.3 V/5 V than ATX12V 1.x PSUs provide.

ATX12V 2.x PSUs have power distribution designed for late P4/XP PCs and for
Athlon 64 and Core Duo PCs. They can be used with earlier P4/XP PCs, but the
power distribution will be significantly suboptimal, so a more powerful ATX12V
2.0 PSU should be used to compensate for that discrepancy. ATX12V 2.x PSUs can
also be used with pre-P4/XP systems, but the power distribution will be
greatly suboptimal (12 V rails will be mostly unused, while the 3.3 V/5 V
rails will be overloaded), so this is not recommended.

Systems that use an ISA bus require ATX/ATX12V 1.2 or earlier because the ISA
bus requires a -5 V power rail unless the board provides a DC to DC converter
that supplies -5 V. ATX/ATX12V 1.3 and later prohibit the -5 V rail so power
supplies built to these versions are usually unsuitable for ISA systems.

Not all computers use standard, interchangeable ATX power supplies. In
particular, some proprietary brand-name machines and high-end workstation and
server designs do not, and require an exactly matching power supply unit.
 
---
The power distribution biases across 3.3 V, 5 V, and 12 V rails are very
different between older and newer ATX PSU designs, as well as between older
and newer PC system designs.

-- Wiki info filed THANKS.
I'm looking for <the link to Pauls post>
where I'll explain my totally false report.
 
Previously I enquired here about a CPU for my SATA+IDE MoBo.
Although I failed to ID the MoBo, a shady 2nd-hand shop sold me
a CPU, which when plugged in at the shop showed life on their display.

When I got home, I removed my working PSU and tried it.
In the inconvenience/restriction, I just concentrated on lining-up
the plug-socket centerclip. Only after it failed and I removed the
connection did I realise that I had used my 20-pin plug on the MoBo's
24-pin socket !!!

I was surprised that it didn't 'smoke', and I noted that the CPU warmed
up slightly, as it did during the shady-shop test.

http://www.overclockers.com/forums/showthread.php?t=401656
writes: which seems WRONG:--
* Due to the fact that the pins are keyed, I believe there is only
one way it will go on.
* If you line it up so the clips match it will fit right in and
the spare 4 will be located correctly.

Then I bought a cheapo 24-pin PSU, but got no life.

The PSU at the shady-shop was a 'heavy' one, and the MoBo
has substantial finning and is marked "PCI express"; which
makes me think it's got a power-hungry video system, and
maybe it would work with a 'powerful' PSU, if the smart
protection circuitry prevented damage by 'center-plugging'
the 20-pin plug to the 24-pin socket.

Should I risk more expense for a Hi-wattage PSU, for this
disasterous project?

== TIA



Using a 20 pin PSU on a 24pin mobo should normally not hurt anything,
but it would not have been "center plugged"
To do that you would have had to have forced it and if so,
you very well might have damaged your mobo
 
I'll blame my 3rd world conditions for not being able to link-back to
Paul's post about the <geometry of 24 <-> 20 pin plugs>.

'Fortunately' I had a PC crash just-now and had the opportunity
to RECHECK the 20pin PSU in the 24pin MoBo

My analysis of WHY I falsely believed that I'd connected with the
hook-lined-up, is that, that's how you'd line-up under restricted
light. Apparently when the lined-up-hook failed, I moved the 20pin
to number-1, where the designers allowed it to go, because there
the 20pin PSU matches ELECTRICALLY to the 24pin MoBo.
At the time I didn't know that the plug & socket were not matching.

This time, the 20pin PSU's fan ran and the CPU heated up.
It's got no fan ##. It's marked core-2-duo 1.86 GHZ.
I put a drop of water on the CPU and disconnected when the CPU
became hot.

Next, I connected the VGA, and tried the new 24+4pin PSU.
This one is very 'light weight' and suspected to be of low-power.
Although with newer technology they may be able to get the
watts without the grams.

Also here the PSU fan started and the CPU heated, but the PSU
fan stopped after 40 seconds; and I suspected thermal over load.

Then I connected again, after it had cooled, but at no time
did I get a VGA signal.

Is my testing procedure correct?

## <Anecdodte>
Previously I was asking here about a CPU for this MoBo
[special since you can't buy SATA+IDE MoBos here].

Since I knew about this 2nd-hand/thief's market I decided
to try there. So this hood quotes me a good price and disapears
for 15 minutes to fetch them. And he shows me 2, without fans,
as if they were bananas. So I said that I could only pay if it
tested OK in my MoBo, which I'd brought. His associate's kiosk
has got a PSU and a monitor, so I was able to quickly [before
the CPU overheats] check that the BIOS shows on the VDU.

In the restricted time to test, I wasn't certain that the VDU's
signal came from MY MoBo. But let's assume it did.

Some weeks later, when I returned there to buy a 'FAT' PSU,
the PSU that he had for sale, was not the same one, and
the MoBo didn't work.

I believe the substantial finning on the Video chip indicates
a big power need. And I've read that PCI express is hungry.

WDYS?

== TIA.
 
I'll blame my 3rd world conditions for not being able to link-back to
Paul's post about the <geometry of 24 <-> 20 pin plugs>.

'Fortunately' I had a PC crash just-now and had the opportunity
to RECHECK the 20pin PSU in the 24pin MoBo

My analysis of WHY I falsely believed that I'd connected with the
hook-lined-up, is that, that's how you'd line-up under restricted
light. Apparently when the lined-up-hook failed, I moved the 20pin
to number-1, where the designers allowed it to go, because there
the 20pin PSU matches ELECTRICALLY to the 24pin MoBo.
At the time I didn't know that the plug & socket were not matching.

This time, the 20pin PSU's fan ran and the CPU heated up.
It's got no fan ##. It's marked core-2-duo 1.86 GHZ.
I put a drop of water on the CPU and disconnected when the CPU
became hot.

Next, I connected the VGA, and tried the new 24+4pin PSU.
This one is very 'light weight' and suspected to be of low-power.
Although with newer technology they may be able to get the
watts without the grams.

Also here the PSU fan started and the CPU heated, but the PSU
fan stopped after 40 seconds; and I suspected thermal over load.

Then I connected again, after it had cooled, but at no time
did I get a VGA signal.

Is my testing procedure correct?

## <Anecdodte>
Previously I was asking here about a CPU for this MoBo
[special since you can't buy SATA+IDE MoBos here].

Since I knew about this 2nd-hand/thief's market I decided
to try there. So this hood quotes me a good price and disapears
for 15 minutes to fetch them. And he shows me 2, without fans,
as if they were bananas. So I said that I could only pay if it
tested OK in my MoBo, which I'd brought. His associate's kiosk
has got a PSU and a monitor, so I was able to quickly [before
the CPU overheats] check that the BIOS shows on the VDU.

In the restricted time to test, I wasn't certain that the VDU's
signal came from MY MoBo. But let's assume it did.

Some weeks later, when I returned there to buy a 'FAT' PSU,
the PSU that he had for sale, was not the same one, and
the MoBo didn't work.

I believe the substantial finning on the Video chip indicates
a big power need. And I've read that PCI express is hungry.

WDYS?

== TIA.

I think the first thing I want you to do is:

1) Go outside on a sunny day. Take the side off the PC.
Now, you have light so you can see what you're doing.
The sun can be your source of light.

2) Try to connect the 20 pin connector, using the offset-by-one-pin
that you propose happened. Do the connectors mate (easily)
when pin 1 is not lined up with pin 1 ? If the alignment
feature is working, maybe there is a chance you didn't
blow up the motherboard.

*******

You can use what are known as "beep tests", to test a subset
of the hardware.

Example:

motherboard, CPU, no RAM, no video card, power supply, front
power button hooked up, computer case speaker hooked up. When
you start the computer (using the front power button), an error
beep pattern will come from the computer case speaker. Either
"beep beep silence beep beep silence" or "beep beep beep silence
beep beep beep silence". The beeping is actually a good sign,
because it means the CPU, and a major portion of circuits in
the chipset, are working.

Now, remove all power, add one stick of RAM. Start the
computer again. The beep pattern should change. If it does,
it might be interpreted to mean the RAM module was "accepted"
and it is OK. Since the video card is missing, that's why the
BIOS code still beeps the speaker.

Those two tests, help prove the core of the motherboard
is working.

An even simpler test is:

motherboard, front power button hooked up, power supply,
no cpu, no RAM, no video card. If the front panel power
button, properly controls the power supply, it means
the PS_ON# wire and path in the logic on the motherboard
are working. And the board isn't completely "blown". The
power supply should respond, when you push the front power
button.

So the idea in these tests is, you don't install all
the components. And in the test, you look for a particular
expected result (presence of a beep pattern coming from
the computer case speaker).

This is an example of a computer case speaker, complete
with the 1x4 pin header connector labeled "SPKR" on the end.
This is what "beeps" when the computer starts. In order for
the beep to come out, the processor must be working. And
that can only happen, if the ATX12V is hooked up.

http://ecx.images-amazon.com/images/I/417+mPQHUDL.jpg

On this motherboard, the SPKR connector would go to the
four orange-colored pins inside the ellipse. The white
letters on the motherboard, just below the PANEL header,
show "+ speak -" to show where the speaker goes. And that
speaker is good for debugging.

http://img208.imageshack.us/img208/1041/fpanel.jpg

That speaker is separate from the external amplified
speakers that a user connects to their computer so
they can listen to music. The tiny speaker inside
the case, is mainly there to create a "beep" tone.

If your computer case doesn't have a speaker (the
high end cases don't), you can buy one of these
instead. This is a piezo speaker. And can be connected
any time you want to listen for the startup "beeps".
My computer store doesn't carry these. And all my cases,
have the standard transistor radio style of speaker.

http://benchmarkreviews.com/images/reviews/cases/NZXT_ALPHA/Speaker-Battery.jpg

Paul
 
You really need to think twice about using less than reputable dealers.
Years ago I would mess with the search engines to get every freakin'
penny off the items. I bought a CPU from a Russian in Texas that arrived
as a drop shipment from California. The sender turned out to be robbing
CPUs and eventually things went sour and they killed a guard during a
robbery. The guy is a lifer at San Quentin, but that is little
consolation for me.

One place where I used to work used Prestige Computers:

My conclusion was dealing with such scum was generating bad karma, not
to mention aiding and abetting.
 
Paul <[email protected]> said:
(e-mail address removed) wrote:
Previously I enquired here about a CPU for my SATA+IDE MoBo. Although
I failed to ID the MoBo, a shady 2nd-hand shop sold me a CPU, which
when plugged in at the shop showed life on their display.

When I got home, I removed my working PSU and tried it. In the
inconvenience/restriction, I just concentrated on lining-up the
plug-socket centerclip. Only after it failed and I removed the
connection did I realise that I had used my 20-pin plug on the MoBo's
24-pin socket !!!


You're going to need to tell us a bit more about *exactly* what you
did.

Using this page, I can get some pictures to refer to.
--snip--

Hi Paul, yes this is the same old garbage that I've asked about before.
No need to fetch *jpg.
Here's the schematic of the 2 types:- x1234567890x <-20pin

123456789012 <-24pin
x is the double-pin at each end of the 24pin socket which remained
unoccupied by the 20pin plug.

That's how I mis-inseted; since the 'hook' which is between 5,6 on the
20pin, is between 6,7 on the 24pin.

As you well know, and assuming my papers on hand are correct: if pin1
of 20pin was 'lined up' with pin1 of 24pin, all 20pin's would be
feeding correctly. And the remaining ADJACENT 4pins are just
duplicate-lines.

SO if it was plugged in thus, all 20 lines would match the 24pin
requirement:-----------------

1234567890xx <-20pin

123456789012 <-24pin
-------------------

I'm beginning to think that the industry would design so that I could
NOT do what I'm claiming I did. But I can't disable my machine now just
to reconfirm.

Is it difficult to explain how pin8:PWR_OK works?

But for the way I [claim to have] plugged it: PSU8 <-> MoBo9. So no
protection.
Unless PWR_OK need a specific signal before allowing power. I.e the
default is to NOT allow power -- hopefully?

== TIA.
The male and female plastic shells, the barrel around each pin, has a
rather unique shape. That is supposed to prevent some of the alignment
cases from mating. The one you describe, should not be possible.
Yes, I've described my mistaken opinion's basis.
x1234567890x <-20pin

123456789012 <-24pin

The Power_OK is a signal from the power supply, to the motherboard. It
is meant to indicate when all rails are charged pretty well towards the
final voltage. It tells the motherboard, that it can come out of reset
state and start drawing more power if it wants.

On the formfactors.org site, there are standards for the ATX power
supply. And in there, it explains how Pwr_OK works (a waveform is
shown).

Paul
= This thread is getting increasingly chaotic; I thought I'd posted this.
already; but some additions: bell-test is good, because theoretically the.
VDU test could fail AFTER the bell-test. But I'm thinking it should test
like it did at the thieves-market. Besides, it's been years since I've
had a civilised/sealed PC. When I move between weekday <-> weekend
locations, I just unplug my IDEs and take them. That's why I desperately
needed a IDE + SATA MoBo: for backups..

Initially [which would probably become permanently] I'd just lay the
MoBo on a table a plug the PSU, disks, keybrd, mouse, VDU in.

I still think in-terms-of the old/pre-feedback-controlled-poweron
PSUs, and I'm confused that the marking for the 'spkr' seems 4-pins?
Besides I haven't got a spare spkr+lead-plug here.

I'm still thinking this MoBo needs more power that the cheapo PSU
that I bought.
==================== Previous post starts here===============

I'll blame my 3rd world conditions for not being able to link-back to
Paul's post about the <geometry of 24 <-> 20 pin plugs>.

'Fortunately' I had a PC crash just-now and had the opportunity
to RECHECK the 20pin PSU in the 24pin MoBo

My analysis of WHY I falsely believed that I'd connected with the
hook-lined-up, is that, that's how you'd line-up under restricted
light. Apparently when the lined-up-hook failed, I moved the 20pin
to number-1, where the designers allowed it to go, because there
the 20pin PSU matches ELECTRICALLY to the 24pin MoBo.
At the time I didn't know that the plug & socket were not matching.

This time, the 20pin PSU's fan ran and the CPU heated up.
It's got no fan ##. It's marked core-2-duo 1.86 GHZ.
I put a drop of water on the CPU and disconnected when the CPU
became hot.

Next, I connected the VGA, and tried the new 24+4pin PSU.
This one is very 'light weight' and suspected to be of low-power.
Although with newer technology they may be able to get the
watts without the grams.

Also here the PSU fan started and the CPU heated, but the PSU
fan stopped after 40 seconds; and I suspected thermal over load.

Then I connected again, after it had cooled, but at no time
did I get a VGA signal.

Is my testing procedure correct?

## <Anecdodte>
Previously I was asking here about a CPU for this MoBo
[special since you can't buy SATA+IDE MoBos here].

Since I knew about this 2nd-hand/thief's market I decided
to try there. So this hood quotes me a good price and disapears
for 15 minutes to fetch them. And he shows me 2, without fans,
as if they were bananas. So I said that I could only pay if it
tested OK in my MoBo, which I'd brought. His associate's kiosk
has got a PSU and a monitor, so I was able to quickly [before
the CPU overheats] check that the BIOS shows on the VDU.

In the restricted time to test, I wasn't certain that the VDU's
signal came from MY MoBo. But let's assume it did.

Some weeks later, when I returned there to buy a 'FAT' PSU,
the PSU that he had for sale, was not the same one, and.
the MoBo didn't work..

I believe the substantial finning on the Video chip indicates
a big power need. And I've read that PCI express is hungry.

WDYS?

== TIA.
 
I think the first thing I want you to do is:

1) Go outside on a sunny day. Take the side off the PC.
Now, you have light so you can see what you're doing. The sun can be
your source of light.

2) Try to connect the 20 pin connector, using the offset-by-one-pin
that you propose happened. Do the connectors mate (easily) when pin
1 is not lined up with pin 1 ? If the alignment feature is working,
maybe there is a chance you didn't blow up the motherboard.

*******
We know now that I COULD NOT have connected with the 'hook lined up', as I
previously claimed, so the PSU voltages DID match-up with the MoBo's
requirements,and the way the CPU is heating up looks good too.
You can use what are known as "beep tests", to test a subset of the
hardware.

Example:

motherboard, CPU, no RAM, no video card, power supply, front power
button hooked up, computer case speaker hooked up. When you start the
computer (using the front power button), an error beep pattern will come
from the computer case speaker. Either "beep beep silence beep beep
silence" or "beep beep beep silence beep beep beep silence". The beeping
is actually a good sign, because it means the CPU, and a major portion
of circuits in the chipset, are working.

Now, remove all power, add one stick of RAM. Start the computer again.
The beep pattern should change. If it does, it might be interpreted to
mean the RAM module was "accepted" and it is OK. Since the video card is
missing, that's why the BIOS code still beeps the speaker.

Those two tests, help prove the core of the motherboard is working.
I was thinking, what's the point of removing RAM in my case ?
But the reduced power load could be relevant so I did: check for VDU-live.
An even simpler test is:

motherboard, front power button hooked up, power supply, no cpu, no RAM,
no video card. If the front panel power button, properly controls the
power supply, it means the PS_ON# wire and path in the logic on the
motherboard are working. And the board isn't completely "blown". The
power supply should respond, when you push the front power button.
Well that looks bad. Originally it needed to hoods to short the
reset pins. Now the PSU fan comes on immediately and the CPU heats,
without any wires connected to the reset.
BTW, the VDU connector is on the MoBo and I believe the video-chip is
the one with BIG finning.
So the idea in these tests is, you don't install all the components. And
in the test, you look for a particular expected result (presence of a
beep pattern coming from the computer case speaker).

This is an example of a computer case speaker, complete with the 1x4 pin
header connector labeled "SPKR" on the end. This is what "beeps" when
the computer starts. In order for the beep to come out, the processor
must be working. And that can only happen, if the ATX12V is hooked up.
I'm confused by the 4 pins labeled SPK, since the speaker should be a 2
terminal, non-polarised device, as they were on old MoBos ?!

I can't take-down my working PC to see now, but I believe it's the old
type.
http://ecx.images-amazon.com/images/I/417+mPQHUDL.jpg

On this motherboard, the SPKR connector would go to the four
orange-colored pins inside the ellipse. The white letters on the
motherboard, just below the PANEL header, show "+ speak -" to show where
the speaker goes. And that speaker is good for debugging.

http://img208.imageshack.us/img208/1041/fpanel.jpg

That speaker is separate from the external amplified speakers that a
user connects to their computer so they can listen to music. The tiny
speaker inside the case, is mainly there to create a "beep" tone.

If your computer case doesn't have a speaker (the high end cases don't),
you can buy one of these instead. This is a piezo speaker. And can be
connected any time you want to listen for the startup "beeps". My
computer store doesn't carry these. And all my cases, have the standard
transistor radio style of speaker.

http://benchmarkreviews.com/images/reviews/cases/NZXT_ALPHA/Speaker- Battery.jpg

Paul

I'm still hoping that the video circuitry will work if it gets sufficient
wattage. Are the 4 extra [yellow] wires for particular circuitry?

Thanks.
 
Unknown said:
We know now that I COULD NOT have connected with the 'hook lined up', as I
previously claimed, so the PSU voltages DID match-up with the MoBo's
requirements,and the way the CPU is heating up looks good too.

OK, we'll assume something else is wrong, and no bad hookup was done.
I was thinking, what's the point of removing RAM in my case ?
But the reduced power load could be relevant so I did: check for VDU-live.

When you remove RAM from a working computer, the BIOS detects that
no RAM is present, and "beeps" the speaker. But we do this test
with another motive. To make the beeps, the CPU must be running.
If we hear the beeps, then we know the CPU works, the chipset is
also proven a bit. So we're not really "testing for RAM", we're
checking to see if the CPU will run enough to beep the speaker.
Well that looks bad. Originally it needed to hoods to short the
reset pins. Now the PSU fan comes on immediately and the CPU heats,
without any wires connected to the reset.
BTW, the VDU connector is on the MoBo and I believe the video-chip is
the one with BIG finning.

PS_ON# could be jammed in the low state. That causes the power
supply to come on immediately.

I've had that happen here once. If you have an IDE ribbon cable,
and it is only partially seated, the electrical loading that
causes, upsets the logic on the Southbridge, and causes PS_ON#
to be asserted. Basically, the Southbridge is a multi-rail
chip, and applying a stress (overload) to logic signals,
can cause the logic in the +5VSB well to be upset. On my
computer, as soon as I re-seated the IDE cable, it worked
properly again.

If the chipset is damaged, burned, PS_ON# can also be jammed
on in that case. Look for a burn mark on the top of the chip,
if the Southbridge does not have a heatsink. The Northbridge
doesn't generally play a part in it. The Southbridge and SuperI/O
chips contribute logic terms to the operation of PS_ON#. So something
in that area of the design, is not working right.

....
I'm confused by the 4 pins labeled SPK, since the speaker should be a 2
terminal, non-polarised device, as they were on old MoBos ?!

I can't take-down my working PC to see now, but I believe it's the old
type.

Speaker pin spacing, can be 1x2 or 1x4 style. On newer systems,
like maybe a Dell or HP, the PANEL header might be a 2x4 or 2x5, and
the speaker could be a 1x2. On older motherboards, or retail motherboards,
the PANEL header can be longer (maybe a 2x10) with a lot more pins,
and there is room for a 1x4 SPKR connector. So both variations exist.

If you own a 1x4, and the connector shell has the small tabs on it,
you can actually back out the pin from the shell, and move the pin
to the 1x2 spacing. This works, as long as the body of the connector
does not bump into anything. So you don't have to cut off the connector
and install a new one, to change formats.

http://www.frontx.com/head_con.html
http://ecx.images-amazon.com/images/I/417+mPQHUDL.jpg

On this motherboard, the SPKR connector would go to the four
orange-colored pins inside the ellipse. The white letters on the
motherboard, just below the PANEL header, show "+ speak -" to show where
the speaker goes. And that speaker is good for debugging.

http://img208.imageshack.us/img208/1041/fpanel.jpg

That speaker is separate from the external amplified speakers that a
user connects to their computer so they can listen to music. The tiny
speaker inside the case, is mainly there to create a "beep" tone.

If your computer case doesn't have a speaker (the high end cases don't),
you can buy one of these instead. This is a piezo speaker. And can be
connected any time you want to listen for the startup "beeps". My
computer store doesn't carry these. And all my cases, have the standard
transistor radio style of speaker.

http://benchmarkreviews.com/images/reviews/cases/NZXT_ALPHA/Speaker- Battery.jpg
Paul

I'm still hoping that the video circuitry will work if it gets sufficient
wattage. Are the 4 extra [yellow] wires for particular circuitry?

Thanks.

The ATX12V, with two yellow and two black wires, provides +12V to
the Vcore regulator around the CPU socket. That's all it powers.
It is just for the CPU.

Everything else on the motherboard is powered by the main connector.
The main connector also carries +12V, which can go to things like the
PCI Express slots.

The difference between the 20 pin and 24 pin, is some redundant pins
that could carry more current if it was required. A 24 pin connector
is only absolutely required, if you're running two Geforce 6600 video
cards in SLI (about 8 amps loading on 12V yellow wire of a 20 pin
connector). Using the 24 pin connector in that case, the two yellow
wires on the 24 pin, help power two cards in SLI. Many more modern
video cards, don't draw as much slot power as the 6600 did, so it
is less of a concern. So in many cases, using more modern cards,
two of those in SLI will draw around 4 amps, and a 20 pin connector
on a 24 pin motherboard is then fine. Only a few video card combinations
*require* using the 24 pin connector with a 24 pin motherboard.

You're unlikely to be using two video cards at the moment,
so the 20 pin connector is fine by itself. As long as
the power supply is not extremely weak (~250W or 200W or less),
the PC should really be able to start.

While a multimeter can help you make a few electrical measurements,
there are really limits as to how much good equipment like that can do.
I would not recommend the purchase of such equipment now, because
it still might not answer the question why your motherboard
does not work.

The "beep" test, is about the cheapest test you can do,
which will help pinpoint a CPU or motherboard problem. We
can't tell which is at fault, but with the video card and
RAM removed, they're not to blame if it won't beep.

The fact your PC starts immediately when the power supply
switch at the back is turned on, that is *not* a good sign.
While a beep test might still be successful in such a situation,
I'm betting nothing is going to work on that motherboard, until the
reason for PS_ON# being jammed on, is known. If you're a user
like me, you try to do as much visual examination as you can,
for things that are physically broken, but at some point all
you can do is try to get a warranty repair from the motherboard
manufacturer.

If the motherboard is too old for any warranty to be left, then
your board is junk. (The Asus motherboards I buy here,
have a three year warranty, measured from the day it leaves the
Asus factory. The last Asus motherboard I bought, was already
1.5 years old, so only 1.5 years of warranty remained. Just so
you understand, the warranty is not very generous, unless the
product is purchased "factory fresh".)

Paul
 
[QUOTE="Paul said:
I believe the substantial finning on the Video chip indicates
a big power need. And I've read that PCI express is hungry.
[/QUOTE]

So my guess was/is that the MoBo needs 'more' wattage.
I think the first thing I want you to do is:

1) Go outside on a sunny day. Take the side off the PC.
Now, you have light so you can see what you're doing.
The sun can be your source of light.

2) Try to connect the 20 pin connector, using the offset-by-one-pin
that you propose happened. Do the connectors mate (easily)
when pin 1 is not lined up with pin 1 ? If the alignment
feature is working, maybe there is a chance you didn't
blow up the motherboard.

You can use what are known as "beep tests", to test a subset
of the hardware.
As previously explained the new PSU that I bought [before I
realised that wattage requirement vary greatly on new PCs]
'warmed' the cpu [which has so far got no cooling facilities],
and 'starts' the PSU fan, without <CPU reset>.

Which we both suspect could indicate serious problems.
The measured the 5V, 12V were perfect, underload, so I was confused;
but I'd forgotten about the <3.xV> which didn't exist in my days.

In fact the start without cpu-reset is not the MoBo's problem,
with a bettter PSU. So I gave the new/flimsy PSU away.

After the MoBo seller failed to sell me a 400W PSU as promised, I
went on to my alternate location to find a beeper. [strange that a
non-polarised 2-terminal devices is designed to use a 4-pin header].
And on the PC that I use there, I noticed the 4-wire-socket hanging
from the PSU, which I'd never had connect to any [old] PC, and used
to assume was for some new-kiddy-external-device.

Although that is a 20 pins PSU, as you know the extra 4 pins seem
only for spreading the PCB-track-current, and I had brought the
MoBo, to test with the intended, but failed, new PSU purchase;
so I tried the MoBo with this PSU.

It beeped and worked and reset-power-down correctly!

So I made a flat-bottomed aluminum water cup, for CPU-cooling
and managed to boot a spare IDE. I shows <Nvidia..> in 'graphic
mode' without loading a graphic driver from disk.

When I got back to my normal location, the tests <powered down>
before completion, although I had increased the cpu-cooling.
Perhaps because the ambient had increased. It's mid summer in
S. hemisphere. My mate, who I gave the flimsy PSU to, thinks the
cpu is shutting-off the PSU due to over temperature, but I think
the PSU is shutting off due to overload.

The MoBo's got a chip with substantial finning, which gets hot,
which I guess is an advanced video processor, using much more
wattage that the old video daughter-boards.

So far, I've not been able to confirm if the MoBo can r/w
IDE <-> SATA, which was the whole reason for buying it.

I suspect that this 2nd-hand BoMo has some functionality broken.
Like I expected it to automatically 'use' a USB-keyboard.
Perhaps that needs to be set in the BIOS?

Before I spend more to buy a 'bigger' PSU, how would I test
if the present power-down after 3 minutes is caused by the PSU?

== TIA.
 
[QUOTE="Paul said:
I believe the substantial finning on the Video chip indicates
a big power need. And I've read that PCI express is hungry.

So my guess was/is that the MoBo needs 'more' wattage.
I think the first thing I want you to do is:

1) Go outside on a sunny day. Take the side off the PC.
Now, you have light so you can see what you're doing.
The sun can be your source of light.

2) Try to connect the 20 pin connector, using the offset-by-one-pin
that you propose happened. Do the connectors mate (easily)
when pin 1 is not lined up with pin 1 ? If the alignment
feature is working, maybe there is a chance you didn't
blow up the motherboard.

You can use what are known as "beep tests", to test a subset
of the hardware.
As previously explained the new PSU that I bought [before I
realised that wattage requirement vary greatly on new PCs]
'warmed' the cpu [which has so far got no cooling facilities],
and 'starts' the PSU fan, without <CPU reset>.

Which we both suspect could indicate serious problems.
The measured the 5V, 12V were perfect, underload, so I was confused;
but I'd forgotten about the <3.xV> which didn't exist in my days.

In fact the start without cpu-reset is not the MoBo's problem,
with a bettter PSU. So I gave the new/flimsy PSU away.

After the MoBo seller failed to sell me a 400W PSU as promised, I
went on to my alternate location to find a beeper. [strange that a
non-polarised 2-terminal devices is designed to use a 4-pin header].
And on the PC that I use there, I noticed the 4-wire-socket hanging
from the PSU, which I'd never had connect to any [old] PC, and used
to assume was for some new-kiddy-external-device.

Although that is a 20 pins PSU, as you know the extra 4 pins seem
only for spreading the PCB-track-current, and I had brought the
MoBo, to test with the intended, but failed, new PSU purchase;
so I tried the MoBo with this PSU.

It beeped and worked and reset-power-down correctly![/QUOTE]

A couple observations.

There are now 20 pin and 24 pin ATX power supplies. The
24 pin ones, a four pin section unhooks, allowing the
connector to also work with 20 pin motherboards. So the
24 pin power supplies, are actually 20+4 pin connectors.
The four pin section, has four different colors wires on it.

That four pin section, should not be confused with the
CPU power cable. The CPU power cable is square, has two
yellow wires and two black wires, and is called ATX12V.
It is a 12V source for the CPU to use. The regulator located
next to the CPU socket, converts that 12V, into the voltage
the CPU needs.

For most all modern motherboards, you need to connect the
CPU power cable, for it to work. And connecting a four pin
cable, with all different color wires, would be a deadly
mistake. The CPU uses the connector with only yellow and
black wires on it.
So I made a flat-bottomed aluminum water cup, for CPU-cooling
and managed to boot a spare IDE. I shows <Nvidia..> in 'graphic
mode' without loading a graphic driver from disk.

When I got back to my normal location, the tests <powered down>
before completion, although I had increased the cpu-cooling.
Perhaps because the ambient had increased. It's mid summer in
S. hemisphere. My mate, who I gave the flimsy PSU to, thinks the
cpu is shutting-off the PSU due to over temperature, but I think
the PSU is shutting off due to overload.

The MoBo's got a chip with substantial finning, which gets hot,
which I guess is an advanced video processor, using much more
wattage that the old video daughter-boards.

So far, I've not been able to confirm if the MoBo can r/w
IDE <-> SATA, which was the whole reason for buying it.

I suspect that this 2nd-hand BoMo has some functionality broken.
Like I expected it to automatically 'use' a USB-keyboard.
Perhaps that needs to be set in the BIOS?

Before I spend more to buy a 'bigger' PSU, how would I test
if the present power-down after 3 minutes is caused by the PSU?

== TIA.

Modern computer motherboards, have thermal protection for the
CPU. If the CPU gets too hot, there is a hardware signal that
shuts off the power. It causes the PS_ON# signal on the main
ATX cable, to go to the "off" state. In many ways, what
happens is similar to the user pressing the power button on
the front of the computer.

Your CPU cooling solution described above is non-standard.
You need a cooler, capable of removing the heat rating
of the CPU you bought. A flat-bottomed aluminum water cup,
if filled with cold water, might keep the thing cool for
five to ten minutes. But for long term cooling, the cooler
needs "lots of square inches of aluminum" and "lots of
airflow over that surface area of aluminum". That helps
transfer the heat, into the air stream. Convection cooling
with a too-small surface area, will result in the CPU overheating
in a relatively short period of time.

So you need to shop for a CPU cooler with an attached fan.
The best way to do this, is to identify the socket type
on the motherboard (S478, LGA775, LGA1155, LGA1156, LGA1355,
LGA2011, S754, S939, AM2, AM3, G34, and so on). Then, shop for
a cooler designed for the particular CPU socket. Some
coolers support as many has half a dozen different sockets.
The kit may come with several different clamps to hold
the cooler in place.

The person who sold you the CPU, might also have the
cooler in hand. When you buy a brand new retail CPU,
many (but not all) of those, come with a cooler and
fan. It's right in the box. Someone selling a used CPU,
could keep the cooler and attempt to sell it separately,
but that's being a bit silly. It should be sold with
the CPU.

*******

I don't know of a reliable way to identify if your
power supply is switching itself off. But based on
your description of the flat-bottomed aluminum water cup,
my guess is your CPU was overheating, and it switched
off the power supply to protect itself. The function
is built right into the CPU, and requires no software
to work. The signal comes from the CPU, saying the
CPU is too hot, and that logic signal goes into the
same chip as the one that listens to the front power
button.

*******

For the LGA775 socket, Intel has a video about installing
the CPU (16MB download). It takes a few moments, before
the sound track starts. You can see the cooler in this
case, is a simple aluminum cooler without a heatpipe.
The cooler they show, is adequate for the job.
Your CPU socket might not be LGA775, but I'm not
aware of any movies for the other sockets. This movie
was provided, because LGA775 was the first land grid
array processor, and people were not familiar with the
lack of pins on the bottom of those.

http://web.archive.org/web/20070208...ww.intel.com/cd/00/00/24/12/241209_241209.wmv

The only thing I can add to that movie, is the need to
remove the old thermal paste, and add some fresh
paste. Thermal paste, placed between the cooler
and processor, helps aid heat transfer. This is
an example of a popular thermal paste. One
tube lasts a long time. I still have my original
purchased tube of paste and there is still some left.
A dot of paste, about the size of a grain of rice,
is roughly enough paste for the job. Don't get the
paste in the socket, or on the CPU pins.

http://www.arcticsilver.com/cmq2.html

They also make a cleaner, but I just use alcohol
when cleaning mine. And while a fluid like
gasoline might seem like a good idea, it
can attack plastics, and is not recommended.
Isopropyl alcohol is pretty safe, by comparison.
It doesn't clean that well, but here at least,
is readily available.

http://www.arcticsilver.com/arcticlean.htm

Paul
 
Back
Top