Mobo PS fault problem

[Mike Easter]

I have a cheap computer which has some kind of hardware problem which
only surfaces occasionally. I would like to solve the problem
intelligently, not by replacing hardware parts until the problem goes
away. Someone here or somewhere would have me start by replacing the
PS, and then if that didn't work I should replace the CPU and then if
that didn't work I should replace the mobo.

Or, perhaps I could save some time by simply replacing both the mobo and
the PS from the gitgo. Or maybe get rid of the cheap computer and buy a
better one that works all the time.

How about someone simply helping me understand how this mechanism can
work.

I have posted about this issue before here, before I accumulated another
year or so of the occasional misbehavior. I'm going to simplify the
description I provided earlier and focus on one specific behavior/symptom.

With a specific software activity in either a linux installation or a
windows installation, such as scrolling, that is specifically using the
mouse on a webpage's slider to scroll the view down the page, the
computer abruptly shuts down, as if some kind of mobo protection faulted
the powersupply's logic.

This happens the instant the scrolling begins, no time for a heat factor.

After that shutdown, it is not possible to turn the computer back on
with the front case power switch, regardless of waiting period. The
only way to powerup again is to turn off the power supply switch on the
rear of the case, wait about 6-7 seconds, and then turn that switch on
again. Then the front case power switch will work again.

Previous discussion here snurled GG http://snipr.com/ptcm9

From: "Mike Easter"
Newsgroups: alt.comp.hardware.pc-homebuilt
Subject: Power puzzle
Date: Thu, 5 Jun 2008 05:26:05 -0700


mobo ECS 741GX-M Socket 462 for AMD K7 processor - using a Geode NX 1400
with a measly 256 kb cache and 1.4G ram. OSes vary; currently
multiboots WinXPpro, Mint7gnome, Mandriva 2009 kde.

 

[John Doe]

I have a cheap computer which has some kind of hardware problem
which only surfaces occasionally. I would like to solve the
problem intelligently, not by replacing hardware parts until the
problem goes away. Someone here or somewhere would have me
start by replacing the PS, and then if that didn't work I should
replace the CPU and then if that didn't work I should replace
the mobo.

Because that is the intelligent way to troubleshoot hardware
problems.

With a specific software activity in either a linux installation
or a windows installation, such as scrolling, that is
specifically using the mouse on a webpage's slider to scroll the
view down the page, the computer abruptly shuts down, as if some
kind of mobo protection faulted the powersupply's logic.

Sounds like a mouse driver versus Internet connection issue, maybe
something to do with the IRQ. Are you using a dial-up modem? I
suppose the "not possible to turn the computer back on with the
front case power switch" part could be related to BIOS Internet
settings or your cheap mainboard/motherboard.

 

[philo]

I have a cheap computer which has some kind of hardware problem which
only surfaces occasionally. I would like to solve the problem
intelligently, not by replacing hardware parts until the problem goes
away. Someone here or somewhere would have me start by replacing the
PS, and then if that didn't work I should replace the CPU and then if
that didn't work I should replace the mobo.

The best way to solve a problem is substituting parts...
but before you do that I'd run a HD diagnostic and a RAM test.

One thing for sure though...the CPU itself need not be replaced...
they either work or they don't.

BTW: also have a look at the capacitors on the mobo...
if they are swollen or leaky then no sense in looking any further...new
mobo is needed

<snip>

 

[Mike Easter]

Mike Easter

Because that is the intelligent way to troubleshoot hardware
problems.

Maybe if you are running a shop with a plethora of parts and are trying
to sell them whether they are needed or not.

Sounds like a mouse driver

This isn't related to mouse activity, this is related to a particular
kind of screen scrolling activity. I can do all the mouse activity in
the world for hours or days or weeks at a time and nothing bad happens.

versus Internet connection issue,

What? Not internet connection. Screen activity.

maybe
something to do with the IRQ.

Why do you say that?

Are you using a dial-up modem?

No -- ethernet to router to cable modem.

I
suppose the "not possible to turn the computer back on with the
front case power switch" part could be related to BIOS Internet
settings or your cheap mainboard/motherboard.

How would the bios settings play a part? I can describe all of those
settings including some of them I don't understand if anyone is intereseted.

This mobo has integrated graphics. There are graphics related bios
settings that I don't understand and the manual is not helpful to me.

 

[Sjouke Burry]

Maybe if you are running a shop with a plethora of parts and are trying
to sell them whether they are needed or not.


This isn't related to mouse activity, this is related to a particular
kind of screen scrolling activity. I can do all the mouse activity in
the world for hours or days or weeks at a time and nothing bad happens.


What? Not internet connection. Screen activity.


Why do you say that?


No -- ethernet to router to cable modem.


How would the bios settings play a part? I can describe all of those
settings including some of them I don't understand if anyone is intereseted.

This mobo has integrated graphics. There are graphics related bios
settings that I don't understand and the manual is not helpful to me.

I would start testing memory, google for memtest86+

 

[Mike Easter]

The best way to solve a problem is substituting parts...

I would rather come up with an intelligent hypothesis which can be
tested if possible. For example, I have discovered one such test, the
scroll test. I would like a more specific test.

There are other 'events' which are occasionally associated with the
fault, but the scroll is pretty reliable. Nowadays I page down webpages
instead of scrolling. Paging down doesn't cause the problem.

but before you do that I'd run a HD diagnostic

This is not related to hdd activity. I can install an OS without problems.

and a RAM test.

Realize that this problem is very intermittent. I can do all kinds of
ram intensive activity for hours or days and nothing happens; but if I
specifically scroll a webpage, then something faults.

One thing for sure though...the CPU itself need not be replaced...
they either work or they don't.

I know that the cpu works and I know that cpu/s have caches, and this
one has a little cache. I also know that under certain circumstances
something going on between something important on the mobo and the ps
stops working right and I also know that the cpu is something important
on the mobo. As is just about anything complicated or even simple on
the mobo.

BTW: also have a look at the capacitors on the mobo...

I did a look around a while back and I'll be doing that again.

 

[Mike Easter]

I would start testing memory, google for memtest86+

It is on hiren's boot cd. Now running.

 

[Mike Easter]

It is on hiren's boot cd. Now running.

Altho' the test is still running, I'm looking at a preliminary error.

It sez:

Monospace for columns
Tst - Pass - Failing Address - Good - Bad - Err Bits -
Count Chan
3 - 0 - 000437847d4 1079.2MB - 20202020 - 20203020 - 00001000 - 1

I don't have the case open right now, so I can't see the ramsticks, but
I seem to recall that one of them is 1G. So I would interpret the
ramtest as saying one of the tests so far shows an error in the 1G chip.
It is currently halfway thru' test 4.

 

[Paul]

I would rather come up with an intelligent hypothesis which can be
tested if possible. For example, I have discovered one such test, the
scroll test. I would like a more specific test.

There are other 'events' which are occasionally associated with the
fault, but the scroll is pretty reliable. Nowadays I page down webpages
instead of scrolling. Paging down doesn't cause the problem.


This is not related to hdd activity. I can install an OS without problems.


Realize that this problem is very intermittent. I can do all kinds of
ram intensive activity for hours or days and nothing happens; but if I
specifically scroll a webpage, then something faults.


I know that the cpu works and I know that cpu/s have caches, and this
one has a little cache. I also know that under certain circumstances
something going on between something important on the mobo and the ps
stops working right and I also know that the cpu is something important
on the mobo. As is just about anything complicated or even simple on
the mobo.


I did a look around a while back and I'll be doing that again.

In addition to running memtest86+, you can give Prime95 a try as well.

Start Prime95 and use its torture test. See how long it'll run
without a problem.

http://www.mersenne.org/freesoft/

Prime95 verifies you have sufficient cooling for the CPU, as
well as testing both the CPU and memory.

In a Windows environment, you can use this for video testing.

3Dmark 2001 SE (Build 330)
http://majorgeeks.com/3Dmark_d99.html

In Linux, I'd have a harder time cooking up a free test.
GLXGears is not exactly stressful. There are games I can run
in Linux (I've run Quake3 Arena), but you need the files for
stuff like that. I copied the Quake3 map files from a Mac
CD, to be able to run the game in Linux.

Paul

 

[philo]

I would rather come up with an intelligent hypothesis which can be
tested if possible. For example, I have discovered one such test, the
scroll test. I would like a more specific test.

There are other 'events' which are occasionally associated with the
fault, but the scroll is pretty reliable. Nowadays I page down webpages
instead of scrolling. Paging down doesn't cause the problem.


This is not related to hdd activity. I can install an OS without problems.


Realize that this problem is very intermittent. I can do all kinds of
ram intensive activity for hours or days and nothing happens; but if I
specifically scroll a webpage, then something faults.


I know that the cpu works and I know that cpu/s have caches, and this
one has a little cache. I also know that under certain circumstances
something going on between something important on the mobo and the ps
stops working right and I also know that the cpu is something important
on the mobo. As is just about anything complicated or even simple on
the mobo.


I did a look around a while back and I'll be doing that again.

Looks like a tough problem...
since the screen moves the same whether you scroll down or page down...

it looks like it may be mouse related or a possible problem with the
PS/2 port.

I'd try a different mouse...and go so far as to trying a serial or USB mouse

 

[Mike Easter]

Mike Easter wrote:

Ram test running with one test error so far.

Looks like a tough problem...
since the screen moves the same whether you scroll down or page

down...

No, the video/screen effect is quite different. All kinds of
things/pixels are going off the screen and all kinds of things/pixels
are crawling/scrolling onto the screen when you scroll. When you page
the entire screen content just 'flips'/transitions all at the same time.
Try it.

it looks like it may be mouse related or a possible problem with the
PS/2 port.

I'd try a different mouse...and go so far as to trying a serial or USB
mouse

I really don't suspect anything with the mouse. If I weren't running
the memory test right now I would figure out how to keyboard or arrow
scroll a page and maybe trip the error without using the mouse.

 

[Mike Easter]

It sez:

Monospace for columns

Evidence is mounting:

Tst - Pass - Failing Address - Good - Bad - Err Bits -
Count Chan
3 - 0 - 000437847d4 1079.2MB - 20202020 - 20203020 - 00001000 - 1
5 - 0 - 000437847d0 1079.2MB - 20000000 - 20001000 - 00001000 - 10
6 - 0 - 000437847d4 1079.2MB - 20000000 - 20001000 - 00001000 - 11

Now starting test 7.

Fry's will replace this 1 G stick, I'm sure. Whenever I buy such stuff,
I scan the receipt and I can find the filed images more easily that way
than saving the originals. Not only can I find them more easily, but
the original receipts fade severely with time, whereas my fresh printout
is easily read.

If you think /that/ receipt handling is odd, you should hear my
redundancy routine for rebate processing.

 

[John Doe]

Maybe if you are running a shop with a plethora of parts and are
trying to sell them whether they are needed or not.

You are suggesting that I am trying to sell parts?

If you have all the answers, Dude, why do you have the problem?

Spare parts are useful for troubleshooting hardware and even
software problems, Dude. Anyone who regularly builds/upgrades his
(or her) own PC should keep spare parts on hand.

This isn't related to mouse activity

With a specific software activity in either a linux installation
or a windows installation, such as scrolling, that is
specifically using the mouse on a webpage's slider to scroll the
view down the page, the computer abruptly shuts down

Considering the fact that it is your writing, what part of "that
is specifically using the mouse on a webpage's slider" don't you
understand?

I feel your frustration... Good luck.
















--

 

[John Doe]

I would rather come up with an intelligent hypothesis

Yes, try to gather enough data to develop a lead.

which can be tested if possible.

That can include swapping parts.

For example, I have discovered one such test, the scroll test.

The inability to scroll without crashing is a symptom, not a
cause.

Realize that this problem is very intermittent. I can do all
kinds of ram intensive activity for hours or days and nothing
happens

Don't look now, Dude, but your RAM is failing in your other reply.

 

[Franc Zabkar]

I have a cheap computer which has some kind of hardware problem which
only surfaces occasionally.

I have posted about this issue before here, before I accumulated another
year or so of the occasional misbehavior.

If you've been dropping memory bits for over a year, as your testing
elsewhere in this thread suggests, I wonder how many of your data
files can be trusted. A corrupt executable file will fail during
execution, but a corrupt data file may remain undetected.

- Franc Zabkar

 

[Mike Easter]

Evidence is mounting:

Tst - Pass - Failing Address - Good - Bad - Err Bits -
Count Chan
3 - 0 - 000437847d4 1079.2MB - 20202020 - 20203020 - 00001000 - 1
5 - 0 - 000437847d0 1079.2MB - 20000000 - 20001000 - 00001000 - 10
6 - 0 - 000437847d4 1079.2MB - 20000000 - 20001000 - 00001000 -

11

This result eventually turned into additional repetitions of the lines
for test 5 & 6.

But...

I removed and replaced the 1G ram stick with a new one, which new stick
I decided to test because the testing disk was still in there. I
couldn't accomplish much of a test because the problem seems worse, and
the machine shuts down before I can run the test very long, more than a
few minutes. This happened several times.

The problem also seems worse in normal operation, such as booting up an
OS. It has shut down abruptly without doing any scrolling 'test' -
symptom elicitor - at all, while just sitting there booted.

Now I've removed that new ram stick and I'm running the ram test on the
other ram.

In the short time that I was running the ram test on the new ram, it
hadn't shown any errors yet.

 

[Mike Easter]

I removed and replaced the 1G ram stick with a new one, which new stick
I decided to test because the testing disk was still in there. I
couldn't accomplish much of a test because the problem seems worse, and
the machine shuts down before I can run the test very long, more than a
few minutes. This happened several times.

Now I've removed that new ram stick and I'm running the ram test on the
other ram.

In the short time that I was running the ram test on the new ram, it
hadn't shown any errors yet.

Naturally the .5G oldest stick tested fine. Then I removed the .5G good
stick and replaced it with the newest 1G stick which was acting so badly
when I added it to the .5G in the 2nd slot.

Currently the newest 1G is running the ram testing OK without errors on
test #4 pass 12%.

This mobo is supposed to be able to run a total of 2G with up to 1G in
either of the 2 slots.

 

[Mike Easter]

Currently the newest 1G is running the ram testing OK without errors on
test #4 pass 12%.

All tests in memtest86+ completed without errors with the new 1G
ramstick only.

During the testing, the machine shutdown once. Then the next testing
was run more than an hour and there were no shutdowns.

Summary. Variable machine shutdowns, depending on/ influenced by/
unknown factors. Current new 1G ram tests normal. Previous diagnostic
information about ram errors was mitigated by ram replacement with a
good ramstick which appeared to worsen the problem The new replacement
ramstick tests normal without errors.

Next I'll try adding the known good .5G ramstick to see if more ram
increases the likelihood of shutdowns.

 

[Mike Easter]

So... my humble experience is... the bad cap thing can in fact show RAM
as bad when it is not. Don't ask me how because I don't know. I can
sympathize with the OP's frustration if RAM swaps are giving him a
confusing result if there are bad caps... *especially* when the tests
support the theory of replacing them.

I'm not ready to post the confusing mixture of reports right now, but my
preliminary opinion is that something other than bad ram can give bad
ram results to the memtest86+ tests.

I'm seeing a very confusing picture in which ram which tested good over
here is testing bad over there or rather vice versa.

 

[SteveH]

A number of years back my beloved K7S5A box was giving me all sorts of

Ah but you also had another problem an ECS K7S5A, one of the biggest POS
known to mankind.