Xref: utzoo comp.unix.wizards:10685 comp.unix.questions:8979 comp.periphs:1159
Path: utzoo!attcan!uunet!lll-winken!lll-tis!ames!ncar!tank!oddjob!gargoyle!att!chinet!les
From: les@chinet.UUCP (Leslie Mikesell)
Newsgroups: comp.unix.wizards,comp.unix.questions,comp.periphs
Subject: Re: Formatting disk theory
Message-ID: <6378@chinet.UUCP>
Date: 25 Aug 88 14:01:25 GMT
References: <619@philmds.UUCP>
Reply-To: les@chinet.UUCP (Leslie Mikesell)
Organization: Chinet - Public Access Unix
Lines: 25

In article <619@philmds.UUCP> hulsebos@philmds.UUCP (Rob Hulsebos) writes:
>.. But the 0xE5 pattern is still
>used, which is not correct: double-density disks must be formatted with
>the 0xD6B6 pattern.
>
>A different pattern is necessary because data is stored differently. The
>0xD6B6 is some kind of worst-case pattern for MFM, so flaws on the disks
>can usually be found.
>
The worst-case pattern is only needed for testing the disk for flaws.  The
idea is that the pattern that puts identical-polarity magnetic fields
together surrounded by opposite-polarity fields is the worst because
the identical-polarity bits will try to move apart.  The reason the pattern
is different for FM and MFM is that a clock pulse is put between each
data bit under FM recording but not MFM (thus the "double-density" of data).

Personally, I have always thought that it would be a good idea to 
go back and re-write the sectors after testing using a best-case
(alternating fields) pattern since that would be less likely to
degrade over time.  Of course stored data overwrites this pattern
anyway, but there are often partial clusters beyond the end of
a file (even partial sectors if the operating system does a read-before-
write) that contain the original formatting pattern and will cause
trouble if a disk error occurs.

Les Mikesell