I’ve gone through a bunch of iterations on the idea that are basically the same organization: 3 voices driven by JW GridSeq with roughly overlapping clocks.
This is the latest: http://cornwarning.com/chaircrusher/Chaircrusher-Breakadawn-3.mp3
The variations between all of these patches is determined primarily by the clock ratios of the various GridSeqs and the scale quantization.
This latest one uses what I call the Benta-tonic scale. You get a Pentatonic scale with the 1st, 3rd, 5th, 7th and 10th notes of the scale, giving you a minor 7th scale with the added 5th step. Benta-tonic adds the 3rd step, removes the 10th and adds the 11th. This introduces two 1/2 step intervals – between the 2nd and Minor 3rd and, the major 7th and the root note.
Which is maybe more music theory than you care to read. The short version is it has the feel of the Pentatonic scale but adds a couple of dissonant notes.
As for the rhythm: If you use the clock multipliers x4 and x1 on Gridseg for the ‘left’ and ‘down’ clocks, it will give you an orderly step advance through the Gridseq grid which repeats once per 4/4 measure (i.e. 16 16th notes).
This goes a bit wonky by using /2.5:x1.5, /2.5:x2.5 and /4:x1 for the three voices. For example, dividing by 2.5 give you a step down every 10 16th notes & multiplying by 1.5 gives you a step right every dotted quarter note. When you go away from /4:x1 for a gridseq, it gives you odd note trigger intervals, because it will skip down a row before playing every note in the row.
This combination of stepping patterns gives it a random feel without being random at all. The combination of the three sequencers will repeat over some time period, but the math to figure that out is more complicated than I want to work out.
Patch file here