[PD] audio rate shift register abstraction with Pd vanilla

[Samuel Burt]

Hi, all.

I'm having a problem with an abstraction I've created to produce a shift
register at the audio rate.

Here's a picture of the patch and a help patch explaining how it works.
https://raw.githubusercontent.com/TricksterSam/Personal-Pure-Data-Development/main/shiftregister~.png

Also, here's a Github link if you want to download and try it.
https://github.com/TricksterSam/Personal-Pure-Data-Development

*Quick description:*
The abstraction shiftregister~ is an object that generates 8 random bits
set to 1 or 0 (output as signals). The bits cycle when a (gate) signal in
the left inlet decreases in value. A signal in the right inlet sets the
probability that the first bit will flip. When the right inlet is 0, the
bits will cycle unchanged. At 0.5, the cycle will be completely random. At
1, the cycle will flip every bit creating a 16-stage sequence.
   In my help patch, I distinguish the outputs by multiplying them each by
a power of two and combine them to form an 8-bit number which is then
sonified as a note sequence by the [makebeep] subpatch.

*My problem:*
I've noticed that as I increase the cycle frequency into the kHz range,
some instabilities occur. In one instance, I've locked the cycle with a 0%
chance for a bit to flip. When I increase the frequency high enough, the
cycle will either flatten out to a single value, or I've observed the
pattern change so only one value is high while all the others are low.
   In addition, if I run phasor~ at 1/8 of the cycle speed, I expect it to
see the phasor synced to the cycle from the shift register. At lower
frequencies this works fine, but at higher frequencies the phase starts
drifting.

*How it works:*
I'm using sixteen [samphold~] objects in a loop. Odd numbered [samphold]~s
receive a gate (square wave created from a phasor) while even numbered
[samphold]~s receive an inverted gate.
   Is there some kind of delay introduced by [delwrite~] [delread~] even
when the delay is set to 0ms? Is there any delay created by the [samphold~]
objects?
   Maybe someone can help me make sense of this, or has a better way of
producing audio rate shift registers. Please, let me know!
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.puredata.info/pipermail/pd-list/attachments/20220222/bded8e33/attachment.htm>