Browse Prior Art Database

Using a Non-Linear Feedback Shift Register to Replace Jumper Pins

IP.com Disclosure Number: IPCOM000101166D
Original Publication Date: 1990-Jul-01
Included in the Prior Art Database: 2005-Mar-16
Document File: 2 page(s) / 55K

Publishing Venue

IBM

Related People

McAnney, WH: AUTHOR

Abstract

This disclosure shows the use of a non-linear feedback shift register as a replacement for jumper input/output pins in VLSI designs.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 78% of the total text.

Using a Non-Linear Feedback Shift Register to Replace Jumper Pins

       This disclosure shows the use of a non-linear feedback
shift register as a replacement for jumper input/output pins in VLSI
designs.

      A maximal-length linear feedback shift register of n stages has
a sequence of length m = 2n - 1, in which the all 0's n-tuple
is missing.  For the application as a jumper replacement, the shift
register must be modified so as to have the all the 0's word and a
sequence length of m = 2n .  Fig. 1 shows such a modified shift
register.  As is well known, the modification consists of ANDing
together the complements of the first n - 1 stages and feeding the
AND output into the Exclusive OR.  The path through the shift
register latches (SLRs) is the scan path.  A new 5-bit word is
generated on each cycle of the Shift A and Shift B clocks.  Fig. 2
shows the sequence generated by the shift register of Fig. 1.  The
sequence generated is called a deBruijn sequence for historical
reasons.

      The output of the nth stage of the shift register is taken to
an observation point.

      Since all 2n possible n bit words are generated by the deBruijn
shift register, any word that occurs when powering on the system is a
legal state.  Furthermore, since the output is at an observable pin,
it is easy to shift out the n bits and determine where the shift
register is in the sequence.  Now determine, using table lookup or
computation, the number of A/B cycles...