24 pin power supplies, stability issues?

  • Thread starter Thread starter Don Burnette
  • Start date Start date
"dino" said:
you can buy a 20-24 pin adapter at most computer stores..I have used that in
most of my builds due to the fact people just don't want to spend the extra
buying an empty tower and adding a $40-50 cheap 24 pin psu, when I am
selling them the tower and PSU for $40.

That passes the ampacity problem, to where the 20 pin adapter
meets the 20 pin ATX power supply. The current flowing through the
two pins on the 24 pin end of the adapter, is flowing through a
single pin where the 20 pin ends meet. You have transferred the
overheat problem to the 20 pin end. Using the adapter hasn't
solved anything, other than being aesthetically pleasing.

For many computer builds, the motherboard load through the
main power connector, on the +12V, does not exceed 6 amps.
While this is "the right way" to build them, it frequently
isn't necessary. The two pins for +12V on the 24 pin
connector can carry up to a total of 12 amps into the
motherboard before they get hot.

ATX_PSU ------------24 pin ----- 24 pin motherboard
^
|
2 pins carries +12V current---+

You can plug a 20 pin connector directly to the 24 pin mobo.
For example, if you had a 6600GT (which has no PCIe power
connector at the end of the card) and three fans, the
4.5 amps drawn while gaming, are easily provided by the 6 amp
limit of the single pin.

ATX_PSU ------------20 pin ----- 24 pin motherboard
^
|
1 pin carries +12V current---+

Using a 20 pin to 24 pin adapter, still forces the current
through the one pin. That pin is where the 20 pin ATX
supply meets the 20_to_24 pin adapter. This adapter
solves nothing, and in fact only adds voltage drop caused
by the increased length of wire. While it looks pretty,
it isn't actually solving any problem. If there was an
overloading problem, the plastic on the ATX PSU end could
be damaged, meaning you might have to replace the power
supply. Which I suppose is marginally better than the
case above, where both the PSU and motherboard plastics
melt. In any case, this is not doing your customers
any good - either you do the math and know the config is
adequately served by the 20 pin supply, connected directly,
or use the 24 pin supply if you think that somehow, the
amps to the motherboard, are going above 6 amps.

ATX_PSU ---20 pin --- 20pin_to_24_pin ---- 24 pin motherboard
^ ^
| |
Only 1 pin here, so Two pins here
this is the hot spot so not as hot

It is pretty easy to see, for a non-SLI board, that
connecting the 20 pin ATX to the 24 pin motherboard will
be OK. But there are components entering the marketplace,
that can negate that easy answer. For example, Matrox now
makes a PCI Express x1 video card, which can be plugged into
any PCI Express slot. Which means a non-SLI motherboard can now
have multiple video cards plugged in, with who knows how much
PCI Express power consumption. Depending on whether the Matrox
product uses +12V or not, could mean that more than 6 amps are
required (i.e. 4 amps for a 6600GT + ??? amps for the Matrox),
and then you need to use a real 24 pin power supply.

That determination becomes more challenging with SLI
systems. If I was crazy, and I SLI a couple of 6600GT
(with no PCIe power connector on the end of the cards), then
8 amps come from the motherboard. If the motherboard has
an EZPlug (a molex connector next to the video cards), then
that molex, plus the 20 pin, is still enough for the job.
However, if the SLI motherboard being used did not have any
auxiliary assistance like that Molex, then a real 24 pin
power supply should definitely be used. If you use your
20_to_24 pin adapter for that silly example of a couple
6600GT PCIe cards, then the adapter will melt where it meets
the 20 pin power supply.

If you cannot work out the expected current, or at least
have not instructed your customer how to do the calculation,
so they know when to replace the supply, you are doing them
a disservice by using that adapter. Sticking a real 24 pin
power supply in there, is an easy way to make friends, even if
it costs a few more bucks. You can even explain to your
customer, why you are doing it - for their own good.

Paul
 
Bill said:
You have a very good power supply. Why buy another one?

Another person just posted about an adaptor for the 20 to 24-pin
connectors. That sounds like a better idea than buying a whole new
supply.


Well, I guess the only reason would be as an attempt to get rid of my
locking up problems. It does not lock up now as much as it did, since
disabling the cpu spread spectrum, but it still does occasionally. It has
now gone a couple of days without a lockup. And there are no error or
warning messages in event viewer pointing to the cause...
It appears some believe, with it just being the 20 pin connector plugged
into the mb, that it could be underpowered? If indeed that is the cause, I
am not sure just getting an adaptor would do the trick.

But, according to the MSI K8N Neo4 Plat mb manual, using just the 20 pin
connector should be fine.

Ironically, it has not locked up yet while playing a 3d game, only when
browsing the internet or reading newsgroups, but that is what I do the
majority of the time.
 
Could also be a driver problem. Before spending tons of money, make sure
your bios is flashed to the most recent version and use the latest
motherboard drivers. Also make sure your using the most recent drivers for
your vid card and sound card too.

--
XP2600@171 [email protected]
PC3200 Samsung 512mb, SB Live OEM
AIW9600XT, A7N8X-X
WD120gb + 80gb HD 8mb buffers
Plextor PX-712A, Liteon 1693S 16X Dual Layer
Pioneer DVR-110D 16X - 4X Dual Layer
Thermaltake Lanfire, 420 Watt PS
ViewSonic 19" A91f+ CRT
Micrsoft Sidewinder Precision 2 Joystick

Overall Score-2066, cpu_score-2926
in 3DMark2005 basic 1078X768, No AA
 
Hey guy's I'm not sure what the problem is but I can assure you that it is
not the power connector.

We are going to start looking at other componants and software (drivers and
such).

Have you tried running Msinfo32 (system information) and looking for trouble
spots?

Or for that matter looking in device manager for any problems?
 
Yes, no trouble spots.

I am going on about 3 days now without it happening, keeping fingers
crossed.. I have to go out of town however today for a couple of days.

If it happens again, I may see about updating the bios and mb drivers. I am
running the bios which came with the board, 1.0 a I believe, and the
motherboard drivers that came on the MSI install cd.

Thanks all,

Don
 
I just recently assembled a new Athlon dual core system .
Athlon XP4400+ dual core cpu
MSI K8N Neo4 Plat Nforce4 mb
Saphire Radeon X850XT PCIe video card.
2gb of Corsair LL matched ( 2 ea 1gb) ddr ram
2 ea 250 gb WD Sata Hard drives
I am using an Antec True Power 530 watt power supply, but it is the 20 pin
( plus 4 ) connector, and the power connector on the motherboard is the
newer 24 pin connector ( plus 4). Someone posted here recently that there
might be some stability issues when using the 20 pin in the 24 pin
connector? I know MSI 's manual says it should be fine.

Simple test. take out one 1gig ram stick and disconnect one 250gb HD.
See if the system works now.

If problems solved, you're 530 is simply too weak.

Yeah they upped all that +12volts stuff but you notice the PS wattage
didn't increase ? Something for nothing ? Not likely.
 
Simple test. take out one 1gig ram stick and disconnect one 250gb HD.
See if the system works now.

I think that would indicate a memory stick is bad more than anything.
If problems solved, you're 530 is simply too weak.

Yeah they upped all that +12volts stuff but you notice the PS wattage
didn't increase ? Something for nothing ? Not likely.

While it's true you don't usually get something for nothing, the power
needs are barely enough to tax a good 530w supply.

I have a system with more power drain than the above listed setup, and
my power supply is a similar yet less capable Antec SmartPower 450w
unit. My system typically floats around the 60% mark of full power
depending on the activities being performed. Do the math involved in
calculating total power consumption. I seriously doubt his power supply
requirement is in question.

The power supply may be defective, but that's completely different from
not being capable of providing enough power.
 
Well, got back in from out of town this evening, booted up the computer, and
was reading emails and newsgroups, and blammo, locked up again. Once again,
no warning or errors in event viewer.

Ok, since I am using the motherboard drivers I installed off the MSI cd that
came with this board, my next step is I am going to search for updated
drivers and install if any, and update my bios if an update is available,
and see if that does any good.
Btw, is there any advantage to getting the motherboard drivers say from
Nvidia rather than MSI?

Thanks,


Don
 
Don said:
Well, got back in from out of town this evening, booted up the
computer, and was reading emails and newsgroups, and blammo,
locked up again. Once again, no warning or errors in event
viewer.

Ok, since I am using the motherboard drivers I installed off the
MSI cd that came with this board, my next step is I am going to
search for updated drivers and install if any, and update my
bios if an update is available, and see if that does any good.
Btw, is there any advantage to getting the motherboard drivers
say from Nvidia rather than MSI?

Please don't top-post, and please do snip.

Sounds to me more as if you have snaky hardware, possibly a
marginal memory chip (unless you have ECC, which would eliminate
that possibility) or a cold soldered joint on the MB, or a marginal
socket connection. You can try removing, inspecting contacts, and
reseating to (possibly) handle the sockets. A replacement
identical MB, without memory or cpu, would probably not be
expensive.

Bad connections are likely to be affected by temperature. You can
try spraying refrigerant, or heating with a hair dryer, and see if
things become worse.

--
"If you want to post a followup via groups.google.com, don't use
the broken "Reply" link at the bottom of the article. Click on
"show options" at the top of the article, then click on the
"Reply" at the bottom of the article headers." - Keith Thompson
More details at: <http://cfaj.freeshell.org/google/>
Also see <http://www.safalra.com/special/googlegroupsreply/>
 
Well, got back in from out of town this evening, booted up the computer, and
was reading emails and newsgroups, and blammo, locked up again. Once again,
no warning or errors in event viewer.

Ok, since I am using the motherboard drivers I installed off the MSI cd that
came with this board, my next step is I am going to search for updated
drivers and install if any, and update my bios if an update is available,
and see if that does any good.
Btw, is there any advantage to getting the motherboard drivers say from
Nvidia rather than MSI?


Yes, nothing from MSI should ever be installed unless MSI is
the only source for (whatever). The original drivers should
not be installed, nor newer ones from MSI. This minimizes
(or eliminates) any residual driver residue on the OS
installation and ensures the newest driver possible is used.

There are rare exceptions, but those are noteworthy (noted
and can be found via google search). Generally these are
not even worth considering until after the newer nVidia
driver is used and it is found that some function isn't
supported (yet), at which point the supplimental driver is
added. Usually one isn't needed.

You might check MSI's site for bios updates/related notes to
see if anything seemingly relevant has been patched on newer
bios versions.

MSI may also have the easiest (to set correctly or use, I
can't be sure of it since I don't have your board here)
hardware monitoring software. Try to keep an eye on the
temperatures and voltages to see if there's any correlation
between them changing and the problem.
 
kony said:
Yes, nothing from MSI should ever be installed unless MSI is
the only source for (whatever). The original drivers should
not be installed, nor newer ones from MSI. This minimizes
(or eliminates) any residual driver residue on the OS
installation and ensures the newest driver possible is used.

There are rare exceptions, but those are noteworthy (noted
and can be found via google search). Generally these are
not even worth considering until after the newer nVidia
driver is used and it is found that some function isn't
supported (yet), at which point the supplimental driver is
added. Usually one isn't needed.

You might check MSI's site for bios updates/related notes to
see if anything seemingly relevant has been patched on newer
bios versions.

MSI may also have the easiest (to set correctly or use, I
can't be sure of it since I don't have your board here)
hardware monitoring software. Try to keep an eye on the
temperatures and voltages to see if there's any correlation
between them changing and the problem.


Will do, thanks Kony. The only thing I have noticed, is the 3.3v shows
outputting 3.10 volts, don't know if that would be related or not. I do have
an adjustment for that on the front panel that connects to the power supply,
but I have not messed with adjusting it yet.

I will uninstall the MSI drivers and install the latest Nvidia drivers
sometime this weekend. There is one updated bios out there, but I think the
only thing it corrects is how it reports on something, but I will probably
go ahead and update to it as well. I have only used liveupdate to update the
bios once, on a previous mb, it worked fine but I am always leery of doing
it ( don't have a floppy on this system).

I loaded the optimized defaults for the bios yesterday, in preperation for
possibly updating the bios. Ya know, I am not sure, I ever really loaded the
optimized defaults when I first assembled this system, heck that may help in
itself.

My temps are fine, idles pretty constant around 34c, and under load around
40-42c.


Thanks,
 
"Don Burnette" said:
Will do, thanks Kony. The only thing I have noticed, is the 3.3v shows
outputting 3.10 volts, don't know if that would be related or not. I do have
an adjustment for that on the front panel that connects to the power supply,
but I have not messed with adjusting it yet.

I will uninstall the MSI drivers and install the latest Nvidia drivers
sometime this weekend. There is one updated bios out there, but I think the
only thing it corrects is how it reports on something, but I will probably
go ahead and update to it as well. I have only used liveupdate to update the
bios once, on a previous mb, it worked fine but I am always leery of doing
it ( don't have a floppy on this system).

I loaded the optimized defaults for the bios yesterday, in preperation for
possibly updating the bios. Ya know, I am not sure, I ever really loaded the
optimized defaults when I first assembled this system, heck that may help in
itself.

My temps are fine, idles pretty constant around 34c, and under load around
40-42c.


Thanks,

If this is a TrueControl power supply, you should put the adjustment
device in place. Apparently the power supply will run on the low
side, with the adjustment device disconnected. That could account
for a 3.10 reading on the 3.3V supply. I would adjust the knobs
while using a voltmeter to verify them. You may be able to touch
the metal of the pin within the 20 pin connector, while it is
connected to the motherboard. Clip the ground of your voltmeter
to a safe place, like an I/O screw on the back of the computer,
so it won't fall into the computer. Then, you'll only need
one hand to probe the backside of the 20 pin connector. A
Molex drive connector gives access to +5V and +12V, but the
main 20 pin will be needed to get at a +3.3V pin.

While it would be fun to trust the hardware monitor chip on the
motherboard, for these voltage readings, the hardware monitor
seldom seems to be correct. You will occasionally see reports
where someone checks their outputs with a voltmeter, only to
find the power supply has virtually "perfect" outputs, while
the monitor chip is reading some other values. So, with the
adjustment plate in place, using a voltmeter is the best way
to set it up.

The adjustment is probably not critical to system operation,
and if you just set the rail adjustments to their "middle"
value, that could very well be good enough. Using the
voltmeter is for the peace of mind knowing it is set up
right.

The only thing I've heard of, that becomes unstable with
a low 3.3V, is some older ATI AIW video cards. Some users
of those have noted in the past, that if 3.3V was low,
there was reduced stability.

Paul
 
On the driver issue there are pros and cons of each.

First the drivers included on the MSI CD are written by the hardware mfg
(i.e..Nvidia). they are customized to the hardware that is installed on the
MB and may have features not always supported by the generic drivers.
However I have not often seen either versions of the drivers cause problems.

On the other hand the support offered to the MB mfgs is more limited than
what is offered to comparable stand alone products. This means that if your
MB is more than a year or two old (from production date, not from when you
bought it) then the drivers from the component mfg. will most likely be the
most bug free.

As for someone's comment that you should never use the drivers from the MB
manufacturer, that is ludicrous, I have set up thousands of machines and I
always use the MB drivers that come on the included CD and have yet to have
a serious problem.

And By the way Top Posting makes more sense then having to read every post
over and over and over again to get to the comment on the thread. If I want
to read more I will.
 
On the driver issue there are pros and cons of each.

First the drivers included on the MSI CD are written by the hardware mfg
(i.e..Nvidia). they are customized to the hardware that is installed on the
MB and may have features not always supported by the generic drivers.
However I have not often seen either versions of the drivers cause problems.

On the other hand the support offered to the MB mfgs is more limited than
what is offered to comparable stand alone products. This means that if your
MB is more than a year or two old (from production date, not from when you
bought it) then the drivers from the component mfg. will most likely be the
most bug free.

As for someone's comment that you should never use the drivers from the MB
manufacturer, that is ludicrous, I have set up thousands of machines and I
always use the MB drivers that come on the included CD and have yet to have
a serious problem.

Don't be ridiculous. Fact is, the OEM drivers are older
versions of the drivers that are released in newer versions
for the very reason that there is benefit.

You have not set up thousands of machines and tested them
against the myriad problems possible. Maybe you installed
the driver, woohoo, that is not even close to knowing what
problems there might be- and maybe you even tested or used
one or more such systems extensively, but it cannot mean you
then have more knowledge than this one static platform in
the particular use. In short, you're just plain wrong.
Millions of experienced users have know for a long time that
installing the oldest driver from the manufacturer is most
likely to be problematic. Perhaps blissfull ignorance is a
better way to put it.


And By the way Top Posting makes more sense then having to read every post
over and over and over again to get to the comment on the thread. If I want
to read more I will.

Then snip out the irrelevant portions at least.
 
Ok, still having stability issues on my new build:

MSI K8N Neo4 Plat
Athlon X2 4400+
2 ea 250 gb WD Sata hard drives
ATI X850XT PCIe video card
Audigy2 ZS soundcard
2gb Corsair XMS LL pc3500 ( 2ea x 1gb), running in dual channel
Windows XP Sp2 completely updated

I updated the mb drivers with the latest from the Nvidia website about a
week ago.
I disabled cpu spread spectrum
I am running at optimized defaults, no overclocking as of yet.

My system locks up every few days, usually when using Outlook or Outlook
Express. I have yet to have it lock up while gaming, however, I only game a
couple hours a week at most. Most of my time is spent emailing and reading
newsgroups.

I get no error or warning messages in event viewer whatsoever. Just a lock
up, nothing works, not evening ctrl-alt-del, I have to power down by holding
the power switch in a few seconds, then restart, and it is fine.

I have the Nvidia ide sw drivers installed with the motherboard drivers,
wonder if this could be the problem, wonder if it would be worth
uninstalling the mb driver and re-installing and not installing the Nvidia
ide sw drivers?

I am also tempted to go ahead and order a new power supply that has the
extra 4 pin connector that goes on the end of the 20 pin connector just in
case...



Don Burnette




Could also be a driver problem. Before spending tons of money, make
sure your bios is flashed to the most recent version and use the
latest motherboard drivers. Also make sure your using the most recent
drivers for your vid card and sound card too.

--
 
I'm using a gigabyte board and use the ide driver and I'm not having any
lockups myself. Did you update your bios first before doing your fresh
install? Also, the proper order for doing a fresh install is:
1. OS
2. ServicePck 2
3. Motherboard drivers
4.Vidcard/soundcard Drivers

It also appears that when doing a fresh install, you may want to pull all
your PCI devices out as it apparently causes interference with the IDE
driver(according to Gigabyte), just like Win 98. I also disable my onboard
sound and lan and install drivers after everything else.
 
Thanks Van,

I was able to solve my stability issues, by uninstalling the motherboard
drivers, and reinstalling and then NOT installing the Nvidia IDW SW driver.

Not sure why it was having that affect on my particular system, but without
the Nvidia IDE SW driver, my system has been rock solid.

Don Burnette
 
Back
Top