I think that is incorrect; I think that a clean install does NOT
necessarily create a new MBR.

You have some options when you do an install; you can use an existing
partition (with either a "full" format or a "quick" format) or you can
delete and/or create new partitions. But NONE of those actions replaces
the MBR, which is a portion of the drive itself and not of any one
particular partition (each partition has it's own boot record, but those
are separate from the MASTER boot record). I don't know that ANY of the
options when doing an install from an XP CD rewrites the master boot
record, although several of them will rewrite the partition boot record.

The only two ways that I know of to be sure that you get a new MBR are:

1. Run the old MS-DOS FDISK program using the /MBR option .... this
explicitly rewrites the MBR

2. Run a program the blows away the MBR, which will leave the drive
without an MBR, forcing it to be recreated to install any OS. The
program that I use to do this is "ZAP", which was an IBM program that
was, at one time, originally on the IBM (later Hitachi) web site, but it
seems to have been removed. It writes binary 00's to the first several
cylinders of the drive.


BillW50 wrote:

>
> Yes a clean install does recreate a new MBR. But something could be
> modifying it. As it seems to stick for awhile and then poof, right?
>
>> I reset the BIOS to its defaults at one point to make sure that wasn't
>> the problem, but it made no difference.
>
> Most BIOS uses flash memory. And sometimes flash fails. Thus either the
> settings won't stick very long or not at all. A weak RTC battery can
> also cause similar symptoms, but it is rare it does.
>
> I just worked on an EeePC 702 that couldn't read the SSD (solid state
> drive). And I couldn't flash the BIOS either. Pulled the SSD out and
> popped a new one and everything worked again. I could even flash the
> BIOS. Weird that the BIOS needs a working drive just to flash the BIOS.
> Who would have guessed that one?
>
>> Considering that two drives (one old, one brand new) both behaved the
>> same way, I too tend to suspect the IDE controller.
>
> Yup, you are probably right. Although I wouldn't write it off yet. ;-)
>