PSU type compatibility?

Do you have a question? Post it now! No Registration Necessary.  Now with pictures!

Threaded View
I know of 5 PC: PSU types, based on the connectors:-

1. with 2 side-by-side, single row sockets that plugged to the  
   PC single row of pins.
2. with 2-rows of 10 sockets to 20-pins of PC.
3. above "2" plus 2x2=4-way socket yellow-black wired.
4. above "3" plus extra 2x2=4-way socket which can
  slot onto 2x10=20-way socket to fit 24-pins of PC.
5. as above "4", but with one-piece 2x12=24-way socket .

I seem to remember that I once ran a PC without the  
   yellow-black 2x2=4-way socket feeding the 4 pins.
Is that possible?

And that I ran a 24-pin MOBO with a 20-pin PSU?

Clearly, many of the wires supply in parallel, to cater for
the increased current, as the PC evolved, and the PCB tracks
couldn't be increased in width.

I bought a spare MOBO, which can handle SATA & IDE, and
which you can't buy new here. After testing it with my PSU of my
normal SATA & IDE MOBO, I wanted to connect it as a standby.

So I plugged a "type 3 above" PSU, which had been running
an old IDE only MOBO.

The CPU-fan didn't run although I "reset" the same pins as for
the successful test.

After a minute, an electolytic capacitor exploded.
Later someone with a fancy PSU tester said the PSU was dud.

What happened?

The 20-sockets PSU had been running the old IDE-only MOBO,
but the 2x2=4-way socket yellow-black hadn't been used.  
No MOBO pins.

If the 2x2=4-way yellow-black supply had been drastically
faulty, the old IDE-only MOBO, would not have detected the
problem. Not connected.

I guess?

Re: PSU type compatibility? wrote:
Quoted text here. Click to load it

Verify the 2x2 has the wires in the right place ?

I'm not sure there's a good reason for the electrolytic to explode.

An electrolytic will explode if you apply a reverse potential
across it. That shouldn't have happened in this case.

An electrolytic can fail if you exceed the working voltage
That could happen, for example, if the 12V rail shorted
to the 5V rail, lifting the 5V rail to 12V.

If you draw overcurrent from the supply, that increases
the ripple current flow in the output capacitors. But
I don't know if ripple current through the ESR of the
capacitor, will cause it to explode like that. I don't
really know what a ripple current failure looks like.

Power supplies can have OCP and OVP, but they can be
set so high that they don't trigger. And in the case
of a cheap power supply, the feature may be missing
entirely. A $20 PSU might not have OCP and OVP.

If it was my supply, I would take it outdoors, remove
the four screws from the top of the PSU, remove the
cover and take a picture. Don't poke anything in
there, just look. Then put the cover back on (so
no one else can poke in there either). Write "blown"
on the lid, so when your neighbor finds it in the
garbage can, they won't be plugging it in :-)

You may be able to tell from which capacitor is
blown (one of the tiny ones on the output side),
as to what the failure mechanism was.

If the explosion was really loud (caused temporary
deafness), it could be the primary side capacitor
that failed. That's the bigger one in there, and those
don't normally blow.


Re: PSU type compatibility?

Quoted text here. Click to load it

The 2x2 is 'keyed' and can't be misconnected?
This 2black+2yellow wires wasn't used in the previous PC, which
worked ok with this PSU.
Quoted text here. Click to load it

WoW!! Thanks, nice info.
Quoted text here. Click to load it
Yes, but how could any reverse or excess voltage exist, when the
PSU was tested/confirmed: running the old PC previously.
Unless it was from the previously unused 2x2.

You didn't confirm that using a 20-way-socket on a 24-way MOBO
can't cause damage, because the extra 4 are in parrallel to spread
the increased current, as the PC industry used more current?

Quoted text here. Click to load it

If I get a new PSU, it may be destroyed by the now  
destroyed MOBO?

What incremental tests can I do?

== TIA.

Re: PSU type compatibility? wrote:

Quoted text here. Click to load it

You can run a 24 on a 20, or a 20 on a 24, as long
as the alignment is correct. To do a 24 on a 20, where
the 24 is a solid strip, there cannot be any components
in the way, in the overhang area. (The Playtool site has
pictures of these cases.)

If you plug a 20 PSU into a 24 motherboard, there is only
one yellow wire and not two. If you plug in two NVidia
6600 video cards (2x4=8amp load), this is more than you
might recommend for a single 6amp wire. So yes, 20 onto
24 works, but don't use too many video cards. They draw
12V from their PCI Express slot.

And the more expensive the video card gets, the less power
it draws from the slot. And the more power it draws from the
2x3 or 2x4 end connectors. An expensive video card draws
2 amps from 12V, from the slot. You could easily
run two expensive cards from a 20 pin PSU, but of
course, you would be entirely lacking the number of
2x3 or 2x4 connectors, so the application would be a
failure. You won't find a 20 pin PSU with four
PCI Express power connectors on it.

But if you have an el-cheapo build, an Intel CPU where the
GPU is inside the CPU, you can easily run the 24 pin
motherboard in that case, off a 20 pin, with nothing
to worry about.

Quoted text here. Click to load it

Absolutely that can happen. That is called a "domino failure",
where damage is passed from item to item. Some posters
to the newsgroups ruin three items before they
determine "maybe I should stop".

This is one of the recognized dangers of being a home
repair person. Most of the time, you figure it out and
fix it. But sometimes, instead it becomes very expensive
and you curse your luck.

In your case, you can:

1) Use an ohmmeter and check the readings between rails.
    Is 3.3V shorted to 5V ? Is 5V shorted to 12V ?
    Are any of them shorted to ground ? You will need a
    lesson from someone, in interpreting dynamic readings.
    If the multimeter reading rises and rises, you're charging
    a capacitor. That's an example of a response pattern.

2) Open the power supply, and visually figure out which rail
    the failed cap was on. Then, when ohming out the motherboard,
    you use that information, to select what wires to start
    checking for short circuit conditions.

And I measure risk with symptoms. Plenty of people have
a PSU quit, when the OCP or OVP kicks in and the circuit
stops. They can mix and match components at home, and
learn stuff. If a second combination causes the PSU to
quit, that's a data point. There's no explosion.

Whereas, you've had an explosion. That means there is a
definite risk factor, compared to the guy where the PSU
just switched itself off and played dead. If your symptoms
are bad enough, you might never use either item again.
This is your call to make.

And while it may not seem fair, I sometimes recommend
people take their broken kit to a computer shop. And
have it blow up some of their gear. Which is the alternative
if you simply must get the box working again. You can tell
them exactly what happened, so they can weigh the danger
to their collection of scrap equipment. I've not heard
of a computer shop refusing to test something, but there
is always a first time.

And there are certainly "safe" pieces of test gear.
There are professional power supply testers, that
are fully protected. And there are bed of nails
motherboard testers, that can survive insults on
the production line. And if you do manage to
blow one of those, the circuit cards can be
individually removed and replaced. So at the
factory, they "have no fear", because they
have very expensive test equipment.


We can take this supply, and perform a thought experiment on it.

Look on the right hand side of the diagram. Now, short +5V to
-12V. The 5V circuit has much lower impedance, and it "wins".
It charges C28 to 5V. But, the potential is applied the
wrong way around. C28 is reverse biased. The supply driving
+5V is muscular and won't take no for an answer. C28 explodes.
So if I traced the wire loom on the 20 or 24 pin, saw which
colored wire was joined to that exploded C28, I might start
checking how the -12V wire, got shorted to something with a
more positive potential. On the motherboard end, I check from
-12V to +3.3V, -12V to +5V, looking for signs of a dead short
(a rail-to-rail short, rather than a rail-to-ground short).


Re: PSU type compatibility?

In article <>, nobody wrote:  

Quoted text here. Click to load it

I believe this <2+2=4> is just an extra/parallel supply for the CPU ?

Quoted text here. Click to load it
My bad description, of placing the 2 statements together?
The exploded-elko is on the MOBO.

Quoted text here. Click to load it
Quoted text here. Click to load it
Yes I agree with that analysis: driving 5V reversed across a 12V elko
may 'expect' to explode it.

Again, the events were:-
The now-damaged MOBO, tested OK on my usual PSU.
Then when I tried to confirm with the PSU of the 486 MOBO, the
elko exploded.

My thinking is:--
The PSU worked with the old/486 PC, where the <4 yellow-black>
leads were not used/confirmed, and would not have been detected
if faulty.
First confirm THEORETICALLY that plugging a 20-way connector & the  
<4 yellow-black> connector is OK for a 24-way MOBO?

Yes, I'll trace the exploded elko, to see what rails it's across.


Re: PSU type compatibility? wrote:

Quoted text here. Click to load it

Yes, 20-way connector plus ATX 2x2 CPU power,
work fine on 24-way motherboard with ATX 2x2 CPU power input.

The pictures on the playtool site should show
such things being done. Either combo works.
20 in 24, or 24 in 20.

The second one doesn't always work out, because
sometimes there is a capacitor in the way, that bumps
against the right-hand end of the connector (the overhang).


Site Timeline