Problem reinstalling Windows - think it's a hardware issue

[todd.acker]

I'm having a lot of problems with my pc now. Right now, it keeps blue
screening when it boots into windows. I've tried reinstalling
windows, but it says setup cannot copy the file:xxx. I've read about
this and it points to memory errors. I've swapped out the memory once
and still have this problem. I've run memtest on my original memory
and sometimes it gives me errors, lots of them, and other times it
runs without any errors. I've tried the install on a new hard drive,
multiple OS disks and a different dvd drive. The only thing I can
think of as being the problem is a bad motherboard or power supply.
Is there a way to test the psu? I don't have a spare to swap it out
with. In my BIOS it says my psu fan speed is about 750 rpm, and it
shows it in red. What's it supposed to be? Does anyone else have
any other things I can try? The pc ran fine for about 2 years, then
it did this once. I took it into a pc repair shop and they said
everything tested OK. When I got it home I was able to reinstall
everything OK. About 2 months later, the same thing happened again.

The fact that my pc keeps blue screening and I can't reinstall
Windows, I'm guessing this is a hardware issue.

 

[Paul]

I'm having a lot of problems with my pc now. Right now, it keeps blue
screening when it boots into windows. I've tried reinstalling
windows, but it says setup cannot copy the file:xxx. I've read about
this and it points to memory errors. I've swapped out the memory once
and still have this problem. I've run memtest on my original memory
and sometimes it gives me errors, lots of them, and other times it
runs without any errors. I've tried the install on a new hard drive,
multiple OS disks and a different dvd drive. The only thing I can
think of as being the problem is a bad motherboard or power supply.
Is there a way to test the psu? I don't have a spare to swap it out
with. In my BIOS it says my psu fan speed is about 750 rpm, and it
shows it in red. What's it supposed to be? Does anyone else have
any other things I can try? The pc ran fine for about 2 years, then
it did this once. I took it into a pc repair shop and they said
everything tested OK. When I got it home I was able to reinstall
everything OK. About 2 months later, the same thing happened again.

The fact that my pc keeps blue screening and I can't reinstall
Windows, I'm guessing this is a hardware issue.

The fact that it worked when you got it home, suggests to me that
when they removed some components and reinserted them, it may have
cleaned the contacts and corrected whatever the problem is. I'm not a
big fan of abrasive cleaning methods (like pencil erasers), and
simply reinserting the component into its connector, should clean
it for you. The gold plating on some components is extremely thin,
and can easily be completely removed without too much trouble.

If memtest is giving errors, that isn't acceptable. Memtest is not
really a sensitive test for memory errors, so if you see errors, then
things cannot be very good.

Memory is sensitive to voltage and operating speed, and the timings
you use at that speed. So there are other issues, beside making
bad contact, that can cause perfectly good memory to make errors.

To do maintenance on the computer, make sure all the power is removed.
Turn off at the back, and then unplug it. Pull the memory sticks, and
then reinsert them. Plug in and switch on again.

For memtest, there are a couple versions you can use.

http://www.memtest.org (memtest86+)
http://oca.microsoft.com/en/windiag.asp (microsoft version)

Once you are in Windows, this is a pretty sensitive test of system
stability. This won't identify a particular memory stick, but it
will tell you whether the CPU, Northbridge, and memory sticks, are
all working together error free. This program should run for hours,
without complaint. Using the default settings on my 1GB memory
system, it will use and test about 700MB. It also heats the processor a bit,
as the calculations keep the processor pretty busy. (On my overclock
test machine, this program stops with an error, in a minute or so.
When things are stable again, you should be able to go for four hours
easily.)

http://sp2004.fre3.com/beta/beta2.htm (orthos, based on Prime95 torture test)

There is no point trying reinstalls, when a system is unstable. You
should try to fix the instability first, then reinstall when your
best testing indicates all is well.

PSU fans only spin as fast as is needed to cool the supply. The
BIOS doesn't really know what a decent fan speed is. You could have
a high efficiency (80%+ efficiency) power supply, with a 140mm fan,
and that might only need to spin at 750RPM.

Feel the air coming out of the PSU. Does it feel cool or boiling hot ?
If it is boiling hot, that will reduce the potential life of the
capacitors inside the supply.

Did the fan on the PSU always spin at that speed, or is it spinning
a lot slower than it used to ? Does the PSU have a fan speed
adjustment know (only a few supplies have those) ? Does the fan
make noise, run erratically, or otherwise no longer perform like
new ?

If you look in the BIOS hardware monitor, there may be voltages
listed for the major supply rails. There could be 3.3V, 5V, and
12V measured in there. You don't want the voltages to be off my
more than 5% of nominal. On the 12V rail, below 11.4V would be
too low. A bit less than that, and some disk drives start to
complain.

The BIOS hardware monitor is not that accurate. If the hardware
monitor measured values leave any doubt in your mind, then you
find a multimeter, switch to DC volts, use a 20V scale perhaps,
and measure the rails by poking the backside of the main
ATX power connector. You can clip the ground lead of the
multimeter to a shiny piece of metal on the computer chassis,
such as a screw on an I/O connector. Then you'll only
have one probe from the multimeter to manipulate, to make your
readings. Use the volts/ohms holes on the multimeter, for your
measurements, and not the current hole. Some meters have more than
two holes on the front, for the test leads.

HTH,
Paul

 

[taa800]

Thanks Paul. I used the memtest program initially and now I'm trying
the microsoft version. I'm still getting memory errors. I reseated
the memory and my video card. That's really all I have in the pc. Do
you think it could be something with my motherboard then? I've tried
putting the memory in another slot, and it still does the same thing.
However, it fails at different points in the microsoft version when I
moved it to a different slot. If I had an extra motherboard or psu
lying around, then I would just swap them out. Since I don't, I'd
have to buy one, then return it if it doesn't solve my problem. Do
you have a recommendation on which one I should try first? I'm still
unsure of the direction I should take in resolving this. The bottom
line is that my memory tests fail, but for some reason, I just don't
think it's the memory.

 

[Paul]

Thanks Paul. I used the memtest program initially and now I'm trying
the microsoft version. I'm still getting memory errors. I reseated
the memory and my video card. That's really all I have in the pc. Do
you think it could be something with my motherboard then? I've tried
putting the memory in another slot, and it still does the same thing.
However, it fails at different points in the microsoft version when I
moved it to a different slot. If I had an extra motherboard or psu
lying around, then I would just swap them out. Since I don't, I'd
have to buy one, then return it if it doesn't solve my problem. Do
you have a recommendation on which one I should try first? I'm still
unsure of the direction I should take in resolving this. The bottom
line is that my memory tests fail, but for some reason, I just don't
think it's the memory.

Right now, I don't have any details on your hardware at all.

What is the motherboard make and model number ? What kind of
processor and RAM are you using ?

I'd probably try bumping up the memory voltage by 0.1V and
test again. Or relax the memory timings a bit.

To list the current timings and clocks, use this:

http://www.cpuid.com/cpuz.php

Paul

 

[taa800]

I have an Asus A8V board with an Athlon 64 3500 processor. I have one
1 gb Corsair value select memory. I have a Seasonic 380 watt power
supply and Sapphire Radeon 9600 video card. I just tried a new power
supply and it ran for longer than normal (about 5 minutes), but still
gave me a blue screen. I don't know how to change my memory
settings. I see things in my bios, but nothing obvious that would
make me feel safe to change the settings.

After I swapped out my psu and it blue screened, when I rebooted the
pc, it gave me a cmos checksum error, and nothing works now. If I
remember correctly, that means my battery is bad???
Again, any help is appreciated.

 

[Paul]

I have an Asus A8V board with an Athlon 64 3500 processor. I have one
1 gb Corsair value select memory. I have a Seasonic 380 watt power
supply and Sapphire Radeon 9600 video card. I just tried a new power
supply and it ran for longer than normal (about 5 minutes), but still
gave me a blue screen. I don't know how to change my memory
settings. I see things in my bios, but nothing obvious that would
make me feel safe to change the settings.

After I swapped out my psu and it blue screened, when I rebooted the
pc, it gave me a cmos checksum error, and nothing works now. If I
remember correctly, that means my battery is bad???
Again, any help is appreciated.

No, that doesn't mean your battery is bad.

First of all, *promise* me you won't flash the BIOS ? I can see
where this is headed, and before I can finish writing this post,
you'll probably flash the BIOS and completely fry it. That is
because the onscreen prompt will suggest that as a course of
action, and that is not a good idea when the computer is
unstable. To flash a BIOS, the computer should be perfectly
stable, as you cannot afford a crash in the middle of flashing
the BIOS. So now is not the time for that.

You're going to reset your CMOS. Either use your paper manual,
or preferably, download the A8V manual from here, because it
will be easier to read on the screen of your backup computer.

http://dlsvr03.asus.com/pub/ASUS/mb/socket939/a8v/e1994_a8v.pdf

The only thing critical at this time, is switch off the PSU
at the back, and then unplug the computer. Then follow the
procedure on PDF page 40 "Clear RTC RAM". (Unplugging the computer
before working inside, should become a habit. It guarantees
no voltages are present inside.)

When the Clear CMOS procedure is complete, the computer should
be returned to defaults. You should be able to enter the BIOS
and set up any custom settings. The most critical thing there,
is if you were using any special settings for your boot disk,
now would be the time to restore the settings.

To expose some useful settings in the BIOS, to work on the
original complaint, see PDF page 83 and set "AI Overclocking" [Manual].
When the manual setting is used, you'll have the opportunity
to set the Vdimm memory voltage. A good place to start, is 2.7V.
The normal value is 2.5V for DDR333 or slower type memory, or
2.6V for DDR400 (PC3200) memory. Sometimes, a little extra,
like 2.7V, will reduce infrequent errors, if they are present.
It is even possible, that the "Auto" setting for Vdimm, is
using 2.7V anyway.

Paul

 

[taa800]

Actually, the error was a bad bios checksum. It then looks for a
floppy or cd. I cleared the cmos and still get the same error. I
would say that the only thing I can do now is put in a floppy to
restore the bios. I did flash it at the beginning of all my
problems. Everything flashed successfully. I'm at a loss now because
of this error. The only thing I can think of is to flash it again.

 

[taa800]

Ok, so I reseated my memory and I am not getting the BIOS checksum
error anymore. I changed my voltage in the BIOS. I didn't see one
for Vdimm. I saw one for DDR so I changed that to all the settings
possible, nothing worked. I did see one called Vlink. But that only
had a setting for 2.5 and 2.6. I still can't get into Windows.

 

[Paul]

Ok, so I reseated my memory and I am not getting the BIOS checksum
error anymore. I changed my voltage in the BIOS. I didn't see one
for Vdimm. I saw one for DDR so I changed that to all the settings
possible, nothing worked. I did see one called Vlink. But that only
had a setting for 2.5 and 2.6. I still can't get into Windows.

OK.

So how is memtest doing ?

Still seeing errors ?

Vlink is probably the interface used between the Northbridge and Southbridge.
So that could be considered to be one of the chipset voltages. Your
memory is connected right to the processor, so a memory error is not
dependent on that. It might have an effect on reading a disk though.

Paul

 

[Paul]

OK.

So how is memtest doing ?

Still seeing errors ?

Vlink is probably the interface used between the Northbridge and
Southbridge.
So that could be considered to be one of the chipset voltages. Your
memory is connected right to the processor, so a memory error is not
dependent on that. It might have an effect on reading a disk though.

Paul

Also, as another test, try booting a Knoppix (knopper.net) or Ubuntu
(ubuntu.com) CD. Both sites provide a 700MB ISO9660 file, which you
burn to a CD. You can boot a computer, right from the CD. I use
that for testing overclocks on my computer. (Rather than use the
hard drive with Windows on it.) Knoppix displays text messages during
the boot sequence, making it easier to see if the software is
having trouble with some hardware. You can also open a terminal
windows when the OS is finally booted, and type "dmesg | more"
and reviews the messages there. That command reads as "dmesg piped
to the more command", and prevents the text from scrolling out
of view, so you can watch it a page at a time.

Linux can mount a FAT32 hard drive, and do read/write on it. So you
can actually fiddle around with the contents of a Windows drive.
For an NTFS formatted drive, out of the box, the Linux CD will mount
that read-only, which means you can look but not touch. There is at
least one package for Linux, that allows read/write to NTFS, but
that would likely be a bit more work to set up and I've never tried
it.

For the hard drive, you can go to the hard drive manufacturer's web
site, and get their drive test software. It should boot from a
floppy, and it might be an idea to give the drive a test. With the
one I've played with, the test software only uses reads, so won't
wipe the drive. It is possible some software would attempt writes,
but you should be prompted before the test software does that.

Paul

 

[taa800]

I haven't been trying memtest anymore, because I still can't boot into
Windows. I kind of use that as my test. I'm not sure I understand
your direction in the help you are suggesting to me. Is this to
verify that I don't have a memory problem? From what I've read
elsewhere, even if I get memtest errors, it doesn't mean that my
memory is bad. I think I'm ready to swap out my motherboard and see
if that fixes the problem.

 

[Paul]

I haven't been trying memtest anymore, because I still can't boot into
Windows. I kind of use that as my test. I'm not sure I understand
your direction in the help you are suggesting to me. Is this to
verify that I don't have a memory problem? From what I've read
elsewhere, even if I get memtest errors, it doesn't mean that my
memory is bad. I think I'm ready to swap out my motherboard and see
if that fixes the problem.

Well, I'm trying to find something that works, to get started.

Memtest86+ (memtest.org) runs from a floppy, and the purpose of that
is two-fold. It tests all the memory (because there is no OS present).
It won't corrupt your Windows hard drive (because the hard
drive is not used at all during the test).

My other suggestions are similar. The Linux test disks are
intended to provide an alternate OS to use as a test of the
hardware. Since those bootable CDs mentioned, do not need to
use the hard drive, then the Windows install on the hard
drive cannot get corrupted.

My test procedures are intended to do a couple things:

1) If we get lucky, identify a single defective component.

2) By difference testing, determine whether a problem may
be hardware or software. For example, on one of my older
motherboards, both Windows and Linux had problems, so I could
see my motherboard had a hardware problem. If Windows had
a problem and not Linux, then I know the software on
my Windows drive must be screwed up.

If the problem was purely a memory issue, memtest86+ may be
enough to allow playing with it.

1) You can change the timing values for the memory. If
the memory was rated CAS2.5, you could go to the
BIOS and try CAS3. That is called "slacking off the timing",
by allowing more cycles for a memory operation to complete.

2) You can reduce the clock speed of the memory, by changing
from DDR400 to DDR333, for example. If the memory bus and
installed memory are not operating up to par, they may work
better with a reduced clocking rate.

3) You can adjust the memory voltage. If memtest86+ only sees
a couple errors per pass, then increasing Vdimm might be
enough to fix it.

The thing is, if you don't have some kind of test procedures
to provide intelligence about your computer, how will you
know the replacement motherboard is doing its job ?

Waiting for Windows to crash is not a very sensitive test
that all is well. Some tests, like the Orthos one, are
excellent at bringing errors to the surface. If a stressful
test like Orthos or Prime95 runs error free, then more casual
usage of the computer should be error free as well.

HTH,
Paul