Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Redundant Index Code for Disk File

IP.com Disclosure Number: IPCOM000099564D
Original Publication Date: 1990-Feb-01
Included in the Prior Art Database: 2005-Mar-15
Document File: 2 page(s) / 87K

Publishing Venue

IBM

Related People

Conway, JL: AUTHOR [+2]

Abstract

A scheme is described that enables index marks encoded on the servo surface of a disk file to be reliably detected, with a minimum amount of logic, in the presence of errors caused by disk defects and noise. Multi-bit index information is spread over a number of sectors prior to index and requires coincidence of selected delayed bits to be detected. The problem is crucial where an external index is required and the sector counter is not accessible. The scheme gives a high level of protection against errors for low logic count and cost.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

Redundant Index Code for Disk File

       A scheme is described that enables index marks encoded on
the servo surface of a disk file to be reliably detected, with a
minimum amount of logic, in the presence of errors caused by disk
defects and noise.  Multi-bit index information is spread over a
number of sectors prior to index and requires coincidence of selected
delayed bits to be detected.  The problem is crucial where an
external index is required and the sector counter is not accessible.
The scheme gives a high level of protection against errors for low
logic count and cost.

      A way of encoding digital data on the servo surface of a disk
file, with detection of errors due to noise and other disk defects,
is described in (*). With such digital data, it is generally more
important to detect the fact that an error occurred than to correct
it, on the assumption that no information is better than bad
information. In the case of index, it is necessary to ensure that
index is always produced at the correct time on every revolution;
otherwise, errors in the handling of customer data may occur.  In
previous files (e.g., 9335) the sector counter is used to generate
the index pulse; the index mark encoded on the disk is only required
to reset this counter in the first place. Subsequently, if the index
mark is not decoded from the disk, then index is still generated from
the counter so that the rest of the file can operate correctly.

      On later files, the sector counter is embedded in a vendor chip
which requires an external index to be provided. So the scheme
outlined above cannot be used. It would be possible to insert an
independent  sector counter, just to check the index mark.  But this
would be wasteful of logic. So a coding scheme has been devised which
will produce index at the correct time, even with two errors in the
index fields.  (The 9335 code will tolerate only one error.)  The
servo format allows three bits in each servo sample to be used for
the purpose of encoding index.  But if there is an error in any part
of the field on a particular sample, then all three bits will be
unusable.

      The three bits on each servo sample used to encode index are
called I, J, and K.
    K is set to 1, two samples and five samples before index.
     J is set to 1, three samples and four samples before index.
     I is set to 1, one sample before index.
     I, J, K are set to 0 on all other samples.

      If an error is detected, then I, J and K are all set to
0 on that sample.

      In the logic which decodes index, J is delayed by two and by
three sample times, and K is delayed by one and by four sample times.
 These delayed outputs are compared together with I (which is not
delayed):
           Sample         I    J-2  J-3  K-1  K-4       Total
             ...