From John Emond Sent Wed, Feb 7th 2018, 03:50
From John Emond Sent Wed, Feb 7th 2018, 03:50
There is along history to this. Back in the 80's we were using the MM5837 to= introduce impairments on the artificial telephone lines (chains of R and Cs= ) we used for product testing. At this time Tim Orr (ex EMS) was using a shi= ft register based noise generator in a kit synthesizer. Cheers, John Monde Synthesizer gives you More www.mondesynthesizer.com > On Feb 6, 2018, at 4:27 PM, Brian Willoughby <xxxxxx@xxxxxxxxxxxx.xxx> wro= te: >=20 > One does not feed digital noise /into/ an LFSR, the LFSR /is/ digital nois= e. >=20 > Let me modsplain this for you. >=20 > The term, LFSR, stands for Linear-Feedback Shift Register. The content of t= he LFSR itself is used as feedback and fed into the binary input. Typically,= the feedback is the eXclusive OR product of two or more bits from close to t= he output. The only catch is that the LFSR cannot start out with all zero bi= ts, or it will always remain zero. The longer the bit chain, the longer the p= attern before it repeats. One can also use just a single output bit at the e= nd of the chain, or consume multiple bits, although adjacent bits are not re= ally unrelated, thus the pseudo-random nomenclature. >=20 > If you were to feed digital noise into an LFSR, it wouldn=E2=80=99t really= be an LFSR any more. >=20 > The nice thing about the LFSR is that it can be implemented in software. I= f you want it to be predictable, that is a better choice, because you can sh= ift only once for each new bit needed. If you want the LFSR output to be unp= redictable, then run it in hardware with its own bit clock so that the seque= nce is not tied to any outside signal. As you probably know, there have been= chips like the MM5837 implementing this since the seventies, if not earlier= . >=20 > Brian >=20 >> On Feb 6, 2018, at 12:41 PM, Kylee Kennedy <xxxxxxxxx@xxxxx.xxx> wrote: >> Also was thinking you were going to mention the Turing Thing from Tom Whi= twell. I have a buddy that uses four of them in his eurorack for just that s= ort of psuedo-randomness.=20 >>=20 >> Another thing to look at is the Buchla voice of uncertainty which is digi= tal noise into an LFSR. So it become more psuedo-random and more musical. >>=20 >> I also like the Lorenz Affect LFO from Non-Linear Circuits called 'Sloth'= for slow evolving LFOs that do weird phase shifts. It's not exactly what yo= u are asking for but I use it similarly to a psuedo-random source. >>=20 >>=20 >> Kylee >=20