Question On 'ATX 12V connector'

[Jethro]

I have an old P4B533-VM motherboard that I have resurrected into a
tower w/PSU that lacks the square so-called ATX 12V connector plug,
and so, cannot connect up that component on my motherboard.
Even so, the resurrected machine is working just fine.

So I am wondering - will this system run okay this way? Will this
come back to haunt me when (if) I try something I so far have not? Or
should I opt for another PSU with such a plug?

Thanks

Jethro

 

[Mike Walsh]

If I recall correctly the square 12V connector was needed for the high power consumption of Pentium 4 processors. If the old power supply can supply enough current through the standard ATX power plug you should not have any problems. A utility that monitors power supply voltage will tell you if the voltage drops because of excessive load.

 

[kony]

I have an old P4B533-VM motherboard that I have resurrected into a
tower w/PSU that lacks the square so-called ATX 12V connector plug,
and so, cannot connect up that component on my motherboard.
Even so, the resurrected machine is working just fine.

So I am wondering - will this system run okay this way? Will this
come back to haunt me when (if) I try something I so far have not? Or
should I opt for another PSU with such a plug?

Thanks

Jethro

Intel made the recommendation to power their P4 from the 12V
rail via the 4 pin connector. The recommendation was
because there is lower current needed at the higher voltage
to deliver the same power to the CPU VRM subcircuit, and by
using a separate smaller 12V connector that connector could
be placed in a higher density/populated area of the
mainboard with the main ATX connector remotely located,
easing power distribution losses across the board copper,
and removing the need to reserve a large copper area
connecting the main ATX connector across the board, else
having to position that connector in an area that many saw
as suboptimal.

Asus among others often continued to produce boards that
used 5V power instead of 12V to power the CPU VRM subcircuit
(which is the circuit that steps-down this voltage in one or
more stages to the CPU vCore voltage) and so long as the
current required by the processor wasn't excessive, or the
board was designed to minimize the effects of the scenarios
mentioned above, that can work fine.

The other factor is that whether the board uses 5V or 12V
for CPU power, marked the beginning of a departure in which
era of the ATX spec was optimal for PSU selection... that
the PSU would either have a lot of 5V current and needed
only a few amps on 12V (less than 5A actually, in a modest
system with few hard drive(s)), or a lot of 12V current
which could additively be an additional,

[(CPU Wattage Spec) / ~ 90% VRM Efficiency ] / 12V = Addt'l
Amps needed on 12V rail


There are a few rare boards which have a shared 12V
connection between the ATX connector's single pin and the
CPU 12V connector. A continuity meter test would find these
rare boards. Since yours doesn't have the 4 pin 12V
connector, odds are much higher that it uses more 5V current
and the selection of PSU should depend on the PSU's bias for
more 12V current or mostly 5V current, regardless of whether
that PSU has the 4 pin connector that you can't use.
However, if you were to buy a new PSU today, that PSU would
remain most viable in the future (reuse on a newer system)
if it did have the 4 pin connector and a fairly high 12V
current rating.

 

[Paul]

I have an old P4B533-VM motherboard that I have resurrected into a
tower w/PSU that lacks the square so-called ATX 12V connector plug,
and so, cannot connect up that component on my motherboard.
Even so, the resurrected machine is working just fine.

So I am wondering - will this system run okay this way? Will this
come back to haunt me when (if) I try something I so far have not? Or
should I opt for another PSU with such a plug?

Thanks

Jethro

In a picture of the P4B533-VM, I see it has the ATX12V connector, implying
Vcore runs from +12V.

The main 20 pin connector has one 12V pin. The pin is rated at 6 amps
or so, safely.

If you had a P4 3.06GHz/FSB533/512KB cache processor (the fastest
one specified on the CPUSupport page), that draws 82W. If the Vcore
converter is 90% efficient, (82W / 12V) * (1/0.90) = 7.6 amps.

Now, if you run the processor with a 100% loading program, and don't
have the 2x2 ATX12V connector in place, all that current has to come
through the single 12V pin on the main connector. If the fan headers
are providing 0.5A for up to three fans, the total current so far
is 8.1 amps through a 6 amp pin.

After some period of use, if you disconnect the main power connector,
the nylon shell may be deformed or the pin overheated and oxidized.
In some cases, it can even be difficult to get the two connector
halves apart.

Using a supply with a 12V connector, provides more paths for the
current to safely flow. So change out that supply and use a proper
one. Otherwise, the motherboard connector and the PSU connector
could be damaged on the 20 pin one.

Paul

 

[Jethro]

In a picture of the P4B533-VM, I see it has the ATX12V connector, implying
Vcore runs from +12V.

The main 20 pin connector has one 12V pin. The pin is rated at 6 amps
or so, safely.

If you had a P4 3.06GHz/FSB533/512KB cache processor (the fastest
one specified on the CPUSupport page), that draws 82W. If the Vcore
converter is 90% efficient, (82W / 12V) * (1/0.90) = 7.6 amps.

Now, if you run the processor with a 100% loading program, and don't
have the 2x2 ATX12V connector in place, all that current has to come
through the single 12V pin on the main connector. If the fan headers
are providing 0.5A for up to three fans, the total current so far
is 8.1 amps through a 6 amp pin.

After some period of use, if you disconnect the main power connector,
the nylon shell may be deformed or the pin overheated and oxidized.
In some cases, it can even be difficult to get the two connector
halves apart.

Using a supply with a 12V connector, provides more paths for the
current to safely flow. So change out that supply and use a proper
one. Otherwise, the motherboard connector and the PSU connector
could be damaged on the 20 pin one.

Paul

Hey guys -
What do you think of using an adaptor connecting a normal 12V
connector to the square 12V ATX connector? There are many for sale.
Thanks
Jethro

 

[larry moe 'n curly]

Hey guys -
What do you think of using an adaptor connecting a normal 12V
connector to the square 12V ATX connector? There are many for sale.

That will take care of your problem for sure, if there is one, but
maybe you don't need it if the +12V rail at the CPU voltage regulator
MOSFETs is virtually the same as the +12V rail at the unused disk
drive power connectors. I would also check the yellow +12V wire on
the 20-pin mobo connector, and if it's not very warm you don't need an
adapter.

I had a BioStar mobo that lacked the 4-pin square connector but
powered the CPU from the +12V rail anyway, and with an old AMD XP1800+
(probably 0.13 micron) the yellow wire didn't overheat, and there was
hardly any voltage droop at the CPU voltage regulator.

 

[kony]

Oops, disregard my last post. For some reason I was reading
the above to mean that the board, not the PSU, lacked the
12V plug (socket).

 

[Paul]

Hey guys -
What do you think of using an adaptor connecting a normal 12V
connector to the square 12V ATX connector? There are many for sale.
Thanks
Jethro

Sure, you can make an adapter, or buy one. Just make sure that the power supply
has enough amps available on 12V, to power the system. When a supply is old
enough not to have the ATX12V, chances are it has powerful 3.3V or 5V rails,
and the 12V is probably the weaker rail on the thing. Total up the amps
you expect the load to draw, then check the label on the side of the
supply to see if it is strong enough.

Paul

 

[Jethro]

Sure, you can make an adapter, or buy one. Just make sure that the power supply
has enough amps available on 12V, to power the system. When a supply is old
enough not to have the ATX12V, chances are it has powerful 3.3V or 5V rails,
and the 12V is probably the weaker rail on the thing. Total up the amps
you expect the load to draw, then check the label on the side of the
supply to see if it is strong enough.

Paul

This PSU is in an old Gateway case, and the is not standard - and so,
none of my extra PSUs I have laying around will fit. Therefore
upgrading the PSU would be a task.

The PSU is a ATX93-3405.

I expect to run one CD optical drive, one FDD, one 40GB HDD, one
dial-up PCI modem, one USB printer.

Jethro

 

[Jethro]

This PSU is in an old Gateway case, and the is not standard - and so,
none of my extra PSUs I have laying around will fit. Therefore
upgrading the PSU would be a task.

The PSU is a ATX93-3405.

I expect to run one CD optical drive, one FDD, one 40GB HDD, one
dial-up PCI modem, one USB printer.

Jethro

Surprisingly, the PSU says it is only 90W!!!!! System has been left
on for days with no problems.

Jethro

 

[Guest]

Surprisingly, the PSU says it is only 90W!!!!! System has been left
on for days with no problems.

I have a Gateway Celeron system with the 90W supply, and it's been
running a 1 GHz Tualitin CPU, a pair of 7200 RPM HDs, CDRW, and DVDRW.

 

[Jethro]

I have a Gateway Celeron system with the 90W supply, and it's been
running a 1 GHz Tualitin CPU, a pair of 7200 RPM HDs, CDRW, and DVDRW.

That's sure good to know. I see on the web that there are stronger
wattage PSUs of this model, the highest being 250W. I am making this
system for my son in CA, so I sure don't want it to break down on him,
Thanks
Jethro

 

[kony]

That's sure good to know. I see on the web that there are stronger
wattage PSUs of this model, the highest being 250W. I am making this
system for my son in CA, so I sure don't want it to break down on him,
Thanks
Jethro

I could be wrong but would not expect this GW PSU to last
long term powering those parts. Have you yet ran a CPU
stress test such as Prime95's Torture Test and measured the
12V rail voltage level? If that PSU is old enough to not
have the 4 pin 12V connector it is unlikely to have much 12V
current capability.

 

[Jethro]

I could be wrong but would not expect this GW PSU to last
long term powering those parts. Have you yet ran a CPU
stress test such as Prime95's Torture Test and measured the
12V rail voltage level? If that PSU is old enough to not
have the 4 pin 12V connector it is unlikely to have much 12V
current capability.

No, but I will try to find said program and run it

Thanks

Jethro

 

[Jethro]

I could be wrong but would not expect this GW PSU to last
long term powering those parts. Have you yet ran a CPU
stress test such as Prime95's Torture Test and measured the
12V rail voltage level? If that PSU is old enough to not
have the 4 pin 12V connector it is unlikely to have much 12V
current capability.

I downloaded and tried it. No errors were found after several hours
of testing.

Jethro

 

[larry moe 'n curly]

No, but I will try to find said program and run it

Can you borrow/steal/buy a Kill-A-Watt?

www.amazon.com/P3-International-Kill-Electricity-Monitor/dp/B00009MDBU

This swell device can tell you how much power the computer draws from
the wall outlet, and about 75% of that number is a pretty good
estimate of what the PSU is putting out.

I have a 90W from a Gateway that looks like this inside:

http://farm1.static.flickr.com/56/170384710_c2066d4a2b_b.jpg

It's fairly decent quality and is made by Newton Power (notice the
small vertical circuit board) but looks like a Delta (even has a Delta
DNA100xx chip inside). The bigger of the two transformers runs at
about 120 KHz, meaning it can put out more power despite being smaller
than a transformer that runs at the more common 60 KHz. If you
upgrade the PSU be sure to get good quality because some stuff out
there is real junk with exaggerated power ratings. I've seen some so-
called 300W PSUs with input filter capacitors smaller than the 220uF
ones in this 90W Newton.

 

[kony]

Can you borrow/steal/buy a Kill-A-Watt?

www.amazon.com/P3-International-Kill-Electricity-Monitor/dp/B00009MDBU

This swell device can tell you how much power the computer draws from
the wall outlet, and about 75% of that number is a pretty good
estimate of what the PSU is putting out.

I have a 90W from a Gateway that looks like this inside:

http://farm1.static.flickr.com/56/170384710_c2066d4a2b_b.jpg

It's fairly decent quality and is made by Newton Power (notice the
small vertical circuit board) but looks like a Delta (even has a Delta
DNA100xx chip inside). The bigger of the two transformers runs at
about 120 KHz, meaning it can put out more power despite being smaller
than a transformer that runs at the more common 60 KHz. If you
upgrade the PSU be sure to get good quality because some stuff out
there is real junk with exaggerated power ratings. I've seen some so-
called 300W PSUs with input filter capacitors smaller than the 220uF
ones in this 90W Newton.

While that is good for a 90W PSU, it doesn't necessarily
tell us if it (his) has sufficient 12V current, nor if it
will remain regulated enough with a lesser 5V rail loading.

I have one of these very low wattage SFX Gatway supplies in
a system as well, think it started out as a Celeron 500
system, but it isn't a Newton (might be Astec, I can't
remember), isn't built as well and barely has any capacity
on the 12V rail.

 

[kony]

I downloaded and tried it. No errors were found after several hours
of testing.

The primary purpose for the test was to produce a high load
on the 12V rail to see if the PSU was maintaining the same
12V and 5V levels as measured by a multimeter at the plug
(to the highest loaded parts, insert the multimeter probes
into the back of the 4-pin connector (for 12V) and ATX
connector (for 5V)).

Having the CPU not produce errors is of course a
prerequisite, but the concern was about excessive rail droop
and ripple (and in PSU, associated cap ESR heating) wearing
out the caps in the PSU or on the motherboard, or related
voltage fluctuations before and especially after capacitor
wear, potentially doing damage to other things.

Perhaps we should back up a bit, what other consumers of 12V
power do you have in the system like fans, hard drives?
What is the TDP, thermal design power for your specific CPU?
What is the 12V current rating on the label of this PSU?