Major problems

[Jim Ludwig]

Hi all,
My desktop is an older (about 8.5 years) Gateway P4, 1.8ghz, WinXP that's
come down with a real problem. I generally leave it on 24/7 and it has a
router hooked up for my wireless home network. We use my laptop primarily
because it's so much nicer. So, I haven't used my desktop in a while and
when I went to use it one day last month, the monitor kept blinking out on
me. Every 3 to 5 seconds, the monitor would just go black for about a
second and then come back on. I can even hear a tick from the monitor
similar to that you would hear when you shut your computer down. I never
noticed, but my wife said she even saw the power indicator light blink off
and sometimes it even turned orange from green. Then, I noticed that there
were some anomalies on the screen image. There were some strange artifacts
and some discoloration of different parts of the screen . Also, the cursor
had some dots and lines surrounding it that moved with it whenever I moved
the mouse. It seemed like it was having a difficult time repainting the
screen. At first, I thought the monitor was finally going out, so I swapped
with a spare monitor that I have, but I still have the same problem. Then,
I found that if I turn the system off and leave it off for several hours (or
come back the next day), everything will work fine for a while, but then it
will start misbehaving again. Furthermore, the problem continues to get
worse until the cursor freezes and the whole system hangs. However, even in
the frozen state, the wireless network works, so I just leave it alone so we
can use the laptop. I thought maybe this could be heat related, but both of
my fans (processor and chassis) are running fine. What could be causing
this problem? The video card or the power supply come to mind. But I don't
know how to test either item. Any thoughts about this?
Thanks for any help,
Jim

 

[Richard Urban]

Even though the fans are running, on an 8 1/2 year old system the fins on
the heat sinks may be clogged with dust/dirt. Turn off the computer and
remove the side cover. get a small flash light and look at the heat sink
fins beneath the fan. You should be able to see the individual fins with a
small space between the fins. Look at both the CPU cooler and the video card
cooler fan heat sinks.

If they are clogged you can use a small can of compressed air to blow the
dust out. You can get the cans of air at Radio Shack, Best Buy or any good
computer oriented store.

The next thing to do is to update your video card drivers. You do this by
going to the "manufacturers" web site for either your computer manufacturer
or the maker of your video card. I am betting that you are still running
with the original video drivers.

After you have done this come back and, in this SAME thread, let us know the
results. We can continue from there if need be.

 

[the wharf rat]

know how to test either item. Any thoughts about this?

You test a switching power supply with a special piece of
electronic equipment designed to vary load and voltage while monitoring
waveforms and current. It costs about 10K so you might have to charge
it (nyuck nyuck nyuck). Despite what you may have heard around here you
CANNOT test a switching power supply with a multimeter.

Now, testing the video card is actually easier. You can buy a little
gadget that fits in a PCI slot that will run diagnostics on installed cards
(even if it's on board video it's really a card, just a built in one). Those
setups will only run about $500, plus some training to learn how to use it.

Orrrrrr.... You could get yourself a spare video card and install
it. IMHO you've got a bad video card, could be a simple as the fan on
the thing failing but even so it's not worth fixing. Replace it, even
if that shotgun misfires you'll be glad of the upgrade when you do get
it working again.

Power supply is a reach. You'd have other symptoms: disc errors,
random crashes, if that was it but it's not impossible. Same deal there.
Spend a bunch of time and money testing the old one or spend a little
time and money replacing it. Video card first IMHO. Anything you buy today
will be an upgrade over a 9 year old unit, and run maybe all of 60 bucks.
Careful to get one that fits the slot.

 

[the wharf rat]

Even though the fans are running, on an 8 1/2 year old system the fins on
the heat sinks may be clogged with dust/dirt. Turn off the computer and

Yeah, you're right, I always forget that almost no one cleans these
things PC's should have a little light when it's time to empty the dust
like on a Hoover.

 

[Ray Luca]

My desktop is an older (about 8.5 years) Gateway P4, 1.8ghz, WinXP

XP, huh? And you posted to a Vista group.

Smooth.

 

[Richard Urban]

My Asus Probe screams at me when the fins get clogged due to the temp
increase!

 

[Richard Urban]

Yeah! We threw him a bone!

 

[Dusko Savatovic]

Why don't you connect your monitor to your laptop?
You'll see instantly if monitor is OK.

 

[Dusko Savatovic]

Sorry,
I missed the point where you connected spare monitor.

 

[westom]

My desktop is an older (about 8.5 years) Gateway P4, 1.8ghz, WinXP that's
come down with a real problem.  I generally leave it on 24/7 and it hasa
router hooked up for my wireless home network.  We use my laptop primarily
because it's so much nicer.  So, I haven't used my desktop in a while and
when I went to use it one day last month, the monitor kept blinking out on
me.  Every 3 to 5 seconds, the monitor would just go black for about a
second and then come back on.  I can even hear a tick  from the monitor
similar to that you would hear when you shut your computer down.

Just keep replacing parts as the shotgunners recommend. Or you
could save massive time and money learning from one who even designed
this stuff. Your choice. Spend massively swapping perfectly good
parts. Or identify the problem before disconnecting or replacing
anything.

A simple solution takes minutes - but requires a less than $18 tool
from Walmart. The multimeter provides numbers so that the next reply
can provide a definitive answer. So that people with real knowledge
can post something helpful. Your choice. Keep buying parts on wild
speculation - shotgunning - a procedure used by those without
electrical knowledge. Or get the meter to determine in minutes what
is and is not defective.

 

[the wharf rat]

A simple solution takes minutes - but requires a less than $18 tool
from Walmart. The multimeter provides numbers so that the next reply

You can't test a switching power supply with a multimeter, bozo.
Know why? Because your meter gives you a point in time reading for
the circuit but what you really need is a power curve over time for the
unit under varying load.

Jeez. He's got a bad video card. It's not the friggin power
supply.

BTW, here's a GREAT article on switching power supplies and how
to test the things. Good luck doing this with an 18 dollar meter.

http://www.xbitlabs.com/articles/coolers/display/psu-methodology.html

Simple measured voltage at any particular lead is meaningless. You
can tell almost as much just by seeing if the fan is turning.

 

[westom]

    You can't test a switching power supply with a multimeter, bozo.
Know why?  Because your meter gives you a point in time reading for
the circuit but what you really need is a power curve over time for the
unit under varying load.

Demonstrated is insufficient electrical knowledge to understand
these simple engineering concepts. Meanwhile, the OP need only post
those numbers from a meter. Tthe engineer can reply with reams of
useful information. Information obtained in minutes that is not
available even after hours of shotgunning.

How many power supplies have you designed? How many computers have
your designed? Not assembled. Even the most technically naive can
assemble a computer. A computer assembler needs no electrical
knowledge. 'Design' as in read semiconductor manufacturer datasheets
to select the appropriate parts.

We were using digital multimeters in the early 70s to identify both
good and bad switching power supplies. How much engineering knowledge
and experience recommends that multimeter? Same engineering knowledge
that also challenges a shotgunner to explain how electricity works.

A shotgunner tells us digital multimeters could not do what they were
doing 30 years ago. Those who don't know how electricity works also
wildly replace good parts. Shotgunning is recommended when basic
electrical knowledge is missing. The OP was encouraged to shotgun;
to collect a large pile of perfectly good parts. Shotgunning
recommended by those who do not even know how to use a 3.5 digit
multimeter. A device so simple as to be sold in Wal-Mart and for less
than $18.

Amazing how the digital multimeter was identifying good and bad
switching powers supplies even in the 1970s - but somehow cannot do
that today? Those without even basic technical knowledge simply post
wild accusations - hoping their ignorance will not be obvious if they
attack others. Simply note the source: the naysayer who also only
recommends shotgunning - or a $500 BIOS code reader.

A minute with the meter means numbers - so that an engineer can post
something useful. So that money and hours are not wasted shotgunning.

 

[Richard Urban]

You are pushing time intensive testing when all the O/P really has to do is
pull a visual of his heat sinks and fans.

***IF*** his visual turns up nothing - then he may want to start testing.

You are placing the cart before the horse.

Again, as I have stated previously - I seldom have to revert to using my
Fluke 87 to troubleshoot a computer. It usually just isn't necessary, or
productive to do so!

--

Richard Urban
Microsoft MVP
Windows Desktop Experience

You can't test a switching power supply with a multimeter, bozo.
Know why? Because your meter gives you a point in time reading for
the circuit but what you really need is a power curve over time for the
unit under varying load.

Demonstrated is insufficient electrical knowledge to understand
these simple engineering concepts. Meanwhile, the OP need only post
those numbers from a meter. Tthe engineer can reply with reams of
useful information. Information obtained in minutes that is not
available even after hours of shotgunning.

How many power supplies have you designed? How many computers have
your designed? Not assembled. Even the most technically naive can
assemble a computer. A computer assembler needs no electrical
knowledge. 'Design' as in read semiconductor manufacturer datasheets
to select the appropriate parts.

We were using digital multimeters in the early 70s to identify both
good and bad switching power supplies. How much engineering knowledge
and experience recommends that multimeter? Same engineering knowledge
that also challenges a shotgunner to explain how electricity works.

A shotgunner tells us digital multimeters could not do what they were
doing 30 years ago. Those who don't know how electricity works also
wildly replace good parts. Shotgunning is recommended when basic
electrical knowledge is missing. The OP was encouraged to shotgun;
to collect a large pile of perfectly good parts. Shotgunning
recommended by those who do not even know how to use a 3.5 digit
multimeter. A device so simple as to be sold in Wal-Mart and for less
than $18.

Amazing how the digital multimeter was identifying good and bad
switching powers supplies even in the 1970s - but somehow cannot do
that today? Those without even basic technical knowledge simply post
wild accusations - hoping their ignorance will not be obvious if they
attack others. Simply note the source: the naysayer who also only
recommends shotgunning - or a $500 BIOS code reader.

A minute with the meter means numbers - so that an engineer can post
something useful. So that money and hours are not wasted shotgunning.

 

[Charlie Tame]

Just keep replacing parts as the shotgunners recommend. Or you
could save massive time and money learning from one who even designed
this stuff. Your choice. Spend massively swapping perfectly good
parts. Or identify the problem before disconnecting or replacing
anything.

A simple solution takes minutes - but requires a less than $18 tool
from Walmart. The multimeter provides numbers so that the next reply
can provide a definitive answer. So that people with real knowledge
can post something helpful. Your choice. Keep buying parts on wild
speculation - shotgunning - a procedure used by those without
electrical knowledge. Or get the meter to determine in minutes what
is and is not defective.

If you knew what you were talking about the first thing you'd realize is
that monitors also fail, so are you suggesting that a person who may be
completely untrained start probing the voltages in the back of a monitor
with an $18 meter? And please tell how one does that without
disconnecting anything?

 

[westom]

You are pushing time intensive testing when all the O/P really has to do is
pull a visual of his heat sinks and fans.

30 seconds with the multimeter is time intensive? Only to those who
cannot and never used one. He also said fans are working. Do you read
before posting? Since all computers must be perfectly happy even in
a 100 degree F room, then heat (in any properly constructed computer)
should not be a problem in a 70 degree room. When heat creates
problems in a 70 degree room, find and replace the defective hardware;
not install more fans as shotgunners so often advocate.

Heat is a diagnostic tool. The naive want to cure that defect with
more fans - as if any computer needs more than one chassis fan.
Hardware that is defective (and gettting worse) will fail when
warmer. First identify the failure; then replace the part. As with
a multimeter, a soution that takes maybe five or ten times less labor
- and no pile of perfectly good and removed parts.

Meanwhile, long before diagnosising anything else, the informed tech
first knows the power 'system' is defintively good. Not just good.
Good without any doubt. Only a multimeter can provide that
knowledge. Shotgunner always means doubts.

Charlie Tame - read what was posted or demonstrate minimal hardware
knowledge. Nobody is poking a meter on anything dangerous -
obviously. No dangerous parts are exposed to a human or the meter.
Otherwise shotgunners could not do their thing. Computers are
constructed so a that even shotgunners can swap parts. You should
have known that even without reading before posting.

How does the OP get immediate and informative replies from the few
who actually know this stuff? Numbers from a meter obtained in a
minute and then posted here. Those numbers report computer functions
even unknown to shotgunners.

 

[Richard Urban]

I see that you are one of those who knows a tad about electricity but don't
have the common sense to look for ***AIR BLOCKAGES*** in the cooling fins.

I guess 30 seconds with a multimeter will tell you that you have to blow out
the detritus from the heat sink fins.

<snicker> <snicker>

--

Richard Urban
Microsoft MVP
Windows Desktop Experience

You are pushing time intensive testing when all the O/P really has to do
is
pull a visual of his heat sinks and fans.

30 seconds with the multimeter is time intensive? Only to those who
cannot and never used one. He also said fans are working. Do you read
before posting? Since all computers must be perfectly happy even in
a 100 degree F room, then heat (in any properly constructed computer)
should not be a problem in a 70 degree room. When heat creates
problems in a 70 degree room, find and replace the defective hardware;
not install more fans as shotgunners so often advocate.

Heat is a diagnostic tool. The naive want to cure that defect with
more fans - as if any computer needs more than one chassis fan.
Hardware that is defective (and gettting worse) will fail when
warmer. First identify the failure; then replace the part. As with
a multimeter, a soution that takes maybe five or ten times less labor
- and no pile of perfectly good and removed parts.

Meanwhile, long before diagnosising anything else, the informed tech
first knows the power 'system' is defintively good. Not just good.
Good without any doubt. Only a multimeter can provide that
knowledge. Shotgunner always means doubts.

Charlie Tame - read what was posted or demonstrate minimal hardware
knowledge. Nobody is poking a meter on anything dangerous -
obviously. No dangerous parts are exposed to a human or the meter.
Otherwise shotgunners could not do their thing. Computers are
constructed so a that even shotgunners can swap parts. You should
have known that even without reading before posting.

How does the OP get immediate and informative replies from the few
who actually know this stuff? Numbers from a meter obtained in a
minute and then posted here. Those numbers report computer functions
even unknown to shotgunners.

 

[+Bob+]

Heat is a diagnostic tool. The naive want to cure that defect with
more fans - as if any computer needs more than one chassis fan.
Hardware that is defective (and gettting worse) will fail when
warmer. First identify the failure; then replace the part. As with
a multimeter, a soution that takes maybe five or ten times less labor
- and no pile of perfectly good and removed parts.

I'd be interested in knowing how you measure heat levels and determine
case airflow with a multimeter.

Many cases have poorly designed airflow. When you start cranking up
applications that push the CPU & video chips to the limit then the
case cooling becomes inadequete. The only solution is to add more
input area or set up push-pull fans. Since case modification (input)
is beyond the skills of most users, adding a fan is a reasonable
alternative.

 

[westom]

I'd be interested in knowing how you measure heat levels and determine
case airflow with a multimeter.

Obviously you don't and don't have to. One spinning 80 mm fan means
more than sufficient airflow. A fact demonstrated in better computers
(ie HP, Dell, etc) and obvious by doing simple thermal calculations.
As long as a hotspot does not exist (due to a poorly trained computer
assembler), then one spinning 80 mm fan is more than sufficient. How
to find defects created by that computer assembler? Operate the
computer in a 100 degree F room.

When the chassis is open in a 70 degree room, zero chassis fans are
more than sufficient.

A chassis is poorly designed only when the computer assembler does
not do his job - creates those hotspots using inferior workmanship.

If CPU and graphics card generated so much heat, then the accuser
posted numbers to prove that. He does not for good reason. Too much
heat and "More Fans!" (the Tim Allen joke) are myths promoted by those
who learned only from urban myths.

If a computer in an open chassis or with a spinning 80 mm fan has a
heat problem in a 70 degree room, then the computer has defective
hardware. That computer with dust balls in a 100 degree room works
just fine if the semiconductors are not defective AND the computer
assembler knew what he was doing. Heat is a powerful tool to even
find defective semiconductors before those defects grow to completely
crash the computer. A tool never learned when junk science cures
symptoms by doing what Tim Allen mocked: more fans!

 

[the wharf rat]

One spinning 80 mm fan means more than sufficient airflow.
A fact demonstrated in better computers (ie HP, Dell, etc)

So no Dell computer uses more than a single 80mm fan? Think
carefully now!

If a computer in an open chassis or with a spinning 80 mm fan has a
heat problem in a 70 degree room, then the computer has defective
hardware.

One time I build a gaming system that crammed three graphics
cards, two 18GB SCSI drives, and a whole gigabyte of PC-100 RAM (PARITY
ram at that) into a mini-tower case with a little handle on top so I could
carry it to parties. That thing ran so hot it'd set off smoke alarms.

There's a reason Ghu, the great, gave us the Dremel.

Heat is a powerful tool to even
find defective semiconductors before those defects grow to completely
crash the computer.

And if they weren't defective before you put the heat gun on 'em
they sure as hell are now, right, genius?

 

[+Bob+]

So no Dell computer uses more than a single 80mm fan? Think
carefully now!



One time I build a gaming system that crammed three graphics
cards, two 18GB SCSI drives, and a whole gigabyte of PC-100 RAM (PARITY
ram at that) into a mini-tower case with a little handle on top so I could
carry it to parties. That thing ran so hot it'd set off smoke alarms.

Apparently he's never heard of the phrase "upgrade". If the engineer
who designed the case did his job right, then the fan should be
sufficient to handle the equipment in the case. As soon as you add or
upgrade, the ventilation is in danger of being out of spec (long
before we hit your extreme gaming machine level). This is especially a
problem with companies like Dell, HP, etc who often design without a
CPU fan and depend on the cooling of a large heat sink by the case fan
to keep the CPU cool.

Not to mention... many home or SOHO machines run in environments that
are consistently over 80-90 degrees in the Summer months. That throws
a big hook in his "one fan at 70 degrees" strategy.