Re: Unstable A7N8X-DLX 2.0 system, Power Supply was culprit
- From: Paul <nospam@xxxxxxxxxx>
- Date: Fri, 08 Feb 2008 19:02:45 -0500
sdlomi2 wrote:
"Kyle" <me@xxxxxxxxxxx> wrote in message news:o6CdnYyDgpcvRzTanZ2dnUVZ_rCtnZ2d@xxxxxxxxxxxxxxAm writing this for the benefit of anyone that might take the time toGlad you found/fixed problem, Kyle. IIRC, that mb also likes ps to deliver 28 amps on the 3v rail, kinda high as compared to many mb's of its era. I'm still using 2 of those--an -8-X/Rev 2.0 and an -8x-DeLuxe 2.0. They've never given me a problem, and have spoiled me. Were it not for fixing friends' and neighbors' (sometimes mutually exclusive!) boxes, I'd have forgotten even the "basics" and missed out on bunches of fun.
read this post
My trusty old A7N8X-DLX 2.0 based system has been behaving badly over
the last couple of weeks, primarily crash to desktop and BSOD crashes
while playing a CPU and graphics intensive game (COD4 to be precise),
and sometimes even a system reset. Researching each of the STOP codes
produced (OS is winxp) provided little comfort, something critical was
going wrong, bugchecks, minidumps, errors in critical system drivers,
etc. I even heard the voice error message "CPU System Failed", or
whatever that little witch says, while playing COD4 and I'd never
before heard BIOS messages when winxp or win2k was up and running
(system is dual-boot).
Normal desktop use did not cause problems, but intense gaming programs
did. I began to suspect my vid card, vid card drivers or some other
odd source for the crashes. I was just about to install a new video
card driver update when I decided to check system voltages. My PS is
a decent brand (Enermax 350W) that has a few years on it, and I was
quite surprised to see Speedfan's graph of the output voltages. The
5v rail was down to about 4.65-4.70v at startup, and during gaming the
voltage dropped a bit more, apparently causing CPU Vcore regulation
problems if not vid card problems (vid card is an ATI 9800 Pro). I
checked/unplugged/replugged all connections to the PS but none seemed
loose.
Much as I don't like to disassemble a system, I proceeded to pull out
the PS and open it up in the hope of finding a pot to adjust output
voltages. Much to my surprise, there were no pots to adjust that I
could find. So, I thought, I might as well clean up the dust/crud
inside the PS since I had it opened up (it was a bit dirty, not really
bad compared to some of the machines I've fixed for others) and do
some bench voltage probing on the PS. Cleaning dust and dirt from the
fans always makes me feel like the PS will run a bit cooler, even tho
this particular system lives in my basement where the ambient temps
are always cool.
After cleanup, the PS 5v unloaded output was 5.11v. Hmmm, me thinks
that looks good, so I reinstall the PS in my rig, and low and behold,
under heavy load, the 5v rail now regulates to no lower than 4.85v
during heavy loads. Moral of the story, crud/dust/fuzz inside the PS
apparently can affect regulation circuit performance sufficiently to
impact output voltages in a negative fasion.
I've seen dirt/dust/crud affect a mobo before (fixed IDE interface
problems once by brushing the dirt from a mobo, which fixed read/write
problems to a HD that checked good in other systems) but not seen such
with a PS before, so I thought I'd share this story.
Best regards,
--
Best regards,
Kyle
ATB, sdlomi2
I've measured the power on my A7N8X-E Deluxe, with a clamp-on DC ammeter,
and these are the numbers I got. The reason the 5V current is so high,
is because the processor draws power from it. (Some S462 motherboards
from that era, use a 2x2 ATX12V power connector for the processor, so
they wouldn't have a large +5V component.)
3.3V @ 5.18A
5V @ 16.6A
12V @ 0.53A
Test conditions were running nultiple copies of Prime95, in Linux.
I've also tested 3DMark2001SE in Windows, and the 3.3V on the A7N8X-E
rose to 6.4A or so. Which could be flowing to the video card. Since the
video slot power connections, come through the main power cable, there is
room for higher numbers. But the 3.3V rail on AGP slots, has an
upper limit for allowed current. So there are some limits as to
how much larger that consumption could get.
The video card I used, has a separate Aux connector, and that adds more
5V consumption (another 5 amps).
By comparison, my P4C800-E Deluxe draws 3.3V @ 14.4A, and that is
because the memory runs from 3.3V (seems to use what looks like linear
regulation). Power varies, depending on whether one or two memory channels
are active. On that system, the 5V consumption is very low, only 0.5A.
The rules on expected current consumption, change with the generation of
design. On modern motherboards, +12V is used for the processor, so the
lower rails won't see that loading. Modern memory has pretty low power
consumption. And that leaves a 10-20W Northbridge and lower power
Southbridge, to suck up power from the lower rails. For modern boards,
I'd assume 50W, coming from 3.3V or 5V, would be enough to cover it.
Is the 28A number a measured value ?
Paul
.
- Follow-Ups:
- References:
- Prev by Date: Re: Unstable A7N8X-DLX 2.0 system, Power Supply was culprit
- Next by Date: M2N-SLI Deluxe Raid setup questions
- Previous by thread: Re: Unstable A7N8X-DLX 2.0 system, Power Supply was culprit
- Next by thread: Re: Unstable A7N8X-DLX 2.0 system, Power Supply was culprit
- Index(es):
Relevant Pages
|
Loading
