In article: <9506231915.aa26619@punt2.demon.co.uk> carlott@interport.net writes:
:
:  While the topic is on pseudo random noise generators (well sort of). Does
: anybody have a simple explanation of why specific feedback points are used
: in shift register designs.
:  In the TTL cookbook (an excellent book) Don Lancaster shows the feedback
: points for 2 to 31 stage shift registers (for some odd reason he skips 19).
:  I know the whole point is to have the largest possible sequence before
: repeating, but if you did not have this chart how would you calculate them
: for yourself?

Horowitz and Hill in 'The Art of Electronics' state that the maximum length
sequence is 2^m - 1, where m is the number of bits in the shift register
(obvious) and go on to state 'the criterion for maximal length is that the
polynomial 1 + x^n + x^m be irreducible and prime over the Galois field' (not
so obvious). m being the number of bits, n being the feedback bit.
Interestingly these things are symmetrical, i.e. instead of using the nth bit
you can use the m-n bit to the same effect. And you can use more than two
taps, eg. a 24 bit tapped at 17, 22 and 23 will also give a maximum length
sequence of 1,677,215 cycles. They also list some interesting properties of
maximal length shift registers. Worth looking up.

The advantage of using a PIC would be that you could control the 'clock'
frequency and the sequence length in software, thus shaping the output power
spectrum to your needs - something that the MM5437 couldn't do easily.

Karel
--
++++++++khladky@khdesign.demon.co.uk++Manchester M14 6BF England++++++++