>Hi Jean,
>
>>In fact, it was a bad soldering contact onto the mainboard. While cold, the PC
>>worked fine. But when hot, after a few hours, due to dilatation, the solder >did not get
>>in contact and the PC hanged up sometimes. Sometimes because it appeared only >if
>>the monitor was placed onto the CPU case causing box deformation ...
>
>mmmm ... interesting... however, I had also changed the motherboard of
>this unit just before changing the hard disk.
>
>Just recently I've noticed that it was using CHS(?) instead of LBA
>in the CMOS configuration. Do you think this could be the problem?

This indicates the method of addressing raw sectors on the drive; LBA (Logical Block Addressing) is needed to access large drives without the use of drive overlay software such as EZ-Drive or Disk Manager. CHS stores the Cylinder, Head and Sector limits in CMOS in the format originally used by the IBM AT BIOS; at this point, it's really only there for legacy support.

This should only affect what happens during the system boot process IAC, since the BIOS isn't used once the boot loader has taken over for Win9x or WinNT except in cases where DOS compatibility mode is needed to access motherboard features. if it were an issue at all, the drive would not be able to be booted.

In addition to the possibility of a cold solder joint as mentioned, I'd also check to make sure that the cooling system for the CPU is working properly (the cheap fans used on some Pentium cooling systems tend to get gummed up and stop working at times. Cooling systems for anything faster than a 486/66 are necessary, and all Pentium-class processors need more than just a heat sink to provide adequate cooling.