Re: Help - can't re-install windows?


"Johnny B Good" <jcs.computersbutt@xxxxxxxxxxxxxxxx> wrote in message
news:3130303037373036497F91E097@xxxxxxxxxxxxxxxxxxx
The message <glnoi0$fe6$1@xxxxxxxxxxxxxxxxxxx>
from "John Brown" <ha_lf@xxxxxxxxxxxx> contains these words:

====huge snip====

Many thanks John for the very useful info...and yes bios version found
it as
you said....
....currently doing a further clean XP installation..what fun :-)

Well, they do say that "Practice makes perfect". ;-)

I'm sure most of us in this news group have "Been there, done that, got
the T-shirt and the medal... etc", I know I certainly have! Usually I've
done this sort of exercise as a form of diagnostic to work out where the
problem is (or, sometimes, as a method to bypass the need for a step I
forgot to include in the first attempt - partition images and cloning
tools proving their worth in this case ;-)

BTW, do you remember that I mentioned a problematical PSU a few post
back? Well, I've sorted out the problem by replacing all the connectors
and their wires with another set retrieved from a scrapped PSU.

The wires used were ok, it seems to have been a connector crimp issue,
a very consistent high resistance connection on, as far as I could tell,
every single crimp!

The job of rewiring took quite a bit of time and the first test started
ok. I went straight into the cmos setup to monitor the voltages in the
PC Health section. The 5v was showing 4.86v - a vast improvement over
the 4.30v it had previously reported. However, the PSU, very quietly and
without fuss, shut itself off after about 5 to 10 minutes and needed the
mains shutting off before I could get any reaction from the PS button -
just a quick twitch from the CPU fan, It had all the hallmarks of either
an overload (unlikely) or overvolt mediated shutdown.

I double checked that I hadn't confused the PG, PS and 5VSB lines and
meter tested to prove I hadn't shorted any lines to ground or bridged
them before opening up to check my soldering. After re-sweating all the
connections and reassembling it, I had no more problems so I suspect I
might have had a dry joint or three that went high resistance with
temperature rise.

I've had it running on the test bench for the past hour or so without
incident (even including a bit of vibration testing). The 5v line now
shows a steady 4.89v suggesting that some of the ground and/or 5 volt
wires were dry jointed in my initial repair. The fact that the 20 pin
MoBo connector now runs cool, pretty well confirms that the crimp
connections were the original cause of the instability symptoms.

I mention this, since a lot of the weird problems experienced with PCs
that have proved to respond to a PSU replacement (often an expensive one
being replaced by a cheap 'n' cheerful test unit) could well be due to
bad connector crimps.

I think this information could prove useful to those of us who hang
onto some of their scrapped PSUs, have a multimeter and soldering iron
to hand and hate to see an otherwise perfectly good expensive (and just
out of warranty) PSU go to waste.

I know the two or three hours repair time involved might make it look
economically unviable, but it can save a lot of downtime and get you the
remaing 3 or 4 years of service life out of a hard to source/expensive
replacement. In my case, I was getting the remaining 5 or 6 years worth
of service life since it had hardly accumulated more than a score of
runtime hours on the workbench.

For anyone interest in testing for high contact resistance crimps, you
just need to unscrew the PSU from the case and insulate the PSU's case
from the PC case (keeping all the wires connected) and measure the
voltage difference between the two. This will only test the ground
wires, but it should be enough to give you the hint as to whether the
connector crimps are the issue.

For reference, the test MoBo in this instance is a socket A board with
Athlon XP2500+ Barton cored CPU, slightly undervolted, 1 stick of 1GB
ddr400 ram, a cdrom drive and a floppy. This represents about a 90 watt
load running memtest86.

Thinking on, a quick and dirty way to determine whether or not there is
a high resistance crimp problem is to finger test the 20 pin connector
for noticably elevated temperature (isolating the metalwork of the PSU
from the PC case will make this effect more pronounced if you're seeing
a systematic crimp failure).

I was seeing about half a volt difference between the MoBo ground and
the PSU case. The BIOS health monitor was showing the 5v line voltage as
4.30v, presumably I was losing the other 0.2 volts in the 5 volt crimps
with a fair load on the 3.3v rail). Having repaired the PSU, I now only
see a 0,05v difference in the grounds which looks more like what I might
expect with good crimp connections.

I would say, as a rough guide, expect to see about 0.05 volt drop per
hundred watts of loading (the grounds would normally be bonded via the
metalwork of the PC and PSU cases when properly assembled, bypassing
most of this ground return volt drop). Readings higher than 0.10volt per
hundred watts loading should be regarded with suspicion.
=======================

Wonder if you have any thoughts on this one...

...yesterday when doing a clean install of Windows got the message during
the Set up..something like...
...there's a problem with your hard drive and Windows has been shut down to
prevent damage to your compute...you may have a virus or the drive may be
corrupted..suggest do CHECKDISC /F and check for hard drive corruption and
then restart the computer...

....so I cut my losses and unconnected that drive (nice new 750gb Western
Digital) and reconnected another (a 5 year old Maxtor)..and started
over...all has gone well...Windows clean installed...great.
So now I'm wondering what to do with the other because I'd obvious rather be
using the new disc.
If I now disconnected the Maxtor, reconnected the WD and try again with a
reformat and reinstallation of Windows...if it goes ok then I'll just wipe
the Maxtor later and if it doesn't I can presumably just reconnect the
Maxtor and expect things to be in working order exactly as it is now..is
that how you would see it?
Thanks,
John


.

Relevant Pages

  • Re: Help - cant re-install windows?
    ... do you remember that I mentioned a problematical PSU a few post ... The wires used were ok, it seems to have been a connector crimp issue, ... shows a steady 4.89v suggesting that some of the ground and/or 5 volt ...
    (uk.comp.homebuilt)
  • Re: 2.6.8-rc2 crash(s)?
    ... > I was watching the psu voltages via gkrellm, and was seeing the 5 volt ... > at the psu connector on the mobo cause I suspect the supply itself is ... > hit a drive power connector since there are spares on this psu, ...
    (Linux-Kernel)
  • Re: 4 pin molex replacement?
    ... The Molex connector is likely to remain a standard for some time, ... aggravated the volt drop issue by almost doubling the volt drop that ... it turned out to be the actual crimp contact resistance ...
    (uk.comp.homebuilt)
  • Re: 4 pin molex replacement?
    ... The Molex connector is likely to remain a standard for some time, ... aggravated the volt drop issue by almost doubling the volt drop that ... it turned out to be the actual crimp contact resistance ...
    (uk.comp.homebuilt)
  • Re: [OT] ATX-PSU and amperage on connectors...
    ... Here are the wire sizes according to my copy of the 2002 Canadian ... then the 4pin HDD connector need already ... Ltd" 300Watt PSU which has ... And there will go your efficiency. ...
    (Debian-User)