sfcs: digital noise generators

About

This module is ~~ripped off~~ heavily inspired by the Doepfer A-117 DNG, basically - i didn't manage to grab one that was selling for dirt cheap on local used euro stuff market, so i HAD to compensate by designing my own, even weirder one. By itself, Doepfer's DNG is pretty weak, because it has no CV over the linear feedback shift register based noise rouce's clock, and the 808 tones are fixed. So i gave that 'module with noise sources mainly for drums' idea my own spin: it's almost the same as the A-117, but it consists out of three sections.

On top, we have a high frequency pulse source, which is a voltage-controllable oscillator that outputs only a narrow pulse, which also may change its duty cycle in relation to CV. A toggle switch selects ranges - very high frequencies in upper position and audable/upper CV frequencies in lower. Its only purpose is to clock the feedback noise section below, which is based on the same principle as the 117's digital noise, but uses the CD4015 dual 4-bit shift register as a base, instead of a much rarer 60-something-step out of the box CMOS shiftreg. This makes the noise less noisy and more tonal, and it sounds more like something you may find in early arcades as a noise sound - and i'm fine with that. Theoretically, it can be clocked externally, but the 4015 turned out to be really picky on what is it clocked with, so i had little success clocking it with other modules. (most probably you can cram some comparator/amplifier section before the 4015 to accomodate for that! i believe in u)

Finally, at the bottom, we have the tonal noises generator section. The COW output is not affected by the CV input - it's just a mix of two squarewaves that form a fourth from a certain frequency. It can be passed through a VCA triggered by a transient, and then a bandpass filter, to get a percussive sound fairly close to the 808 cowbell. Below it, there's the XOR output - it's a very abrasive source of digital noise, that can be modulated by the CV input. It throws in 2 more oscillators into the mix, xors each with one of the cowbell squarewaves, and then those xor outputs are xored yet again to obtain a very metallic, ear-grinding sound. It is very sharp and cuts through the mix nicely, and has a somewhat 606-y mood to it.

Generally, the main aim of this module is to provide noises for shaping them into drum sounds, but it also may be used as a pseudo-random trigger sequencer by running the feedback noise section at fairly low speed, or as a noisy horrorish backdrop by smearing the XOR hat noise source with heavy reverbs.

Schematic

The ciruit is built entirely out of CMOS chips, so you may think of it as of a tiny prepatched lunetta synth inside a module. It uses the CD40106 as a base for 5 oscillators, then the CD4030 XOR and the CD4015 dual shiftreg for getting a metallic hat noise mix and also making a feedback path for the LFSR. On the top section is a somewhat weird version of the 40106vco by synthmonger - i added a resistor in series with the diode to widen up the pulse so it could clock things, and also changed the resistor values near the current sink transistor to give a more steady range. It outputs a sharp, narrow pulse, and its frequency can be controlled with external CV linearly.

Right below is a very, very simple XOR-based linear feedback shift register (LFSR) based on the 4030 and the 4015. Read up on LFSRs to learn how they work, but essentially when clocked this thing generates a finite (looping) pseudorandom stream of 0s and 1s (or gates, if you want) - and when clocked at appropriate frequencies, it sounds like a cyclic noise. At frequencies above those it gets obvious that it loops, at frequencies below it is a set of repeating clicks (can be used as a trigger sequencer!) I added a switch to select between the two different feedback paths - they have slightly different pattern characters and lengths.

Finally, down below is the tonal noises generator part. The generation of squarewaves is handled by the most basic 40106 oscillators - two of them are fixed (cow mix), two are voltage-controllable by vactrols (LDR-LED combinations). The cowbell oscillators' values were picked by ear, hence i labelled R12 and R13 X and Y ohms respectively. For me it was about 9K for the lower pitched one, and 4.7K + 10K trimpot for the higher one. To achieve this sound, these oscillators should run at about 395 and 585 Hz respectively. Tune the upper one (with the trimpot) to be a fourth above from the lower one. They get mixed down by a passive summator R14-R15 and that's the COW output. The XOR output is derived by XORing each COW oscillator with one of the VCOs, and then XORing the results of those pair-wise XORs. The result is a very metallic, abrasive noise which heavily depends on the frequencies of the two VCOs and can be voltage controlled.

Media

Sweeps of the LFSR noise output. On higher frequencies it gets obviously cyclic and tonal, but at lower frequencies it is noisy enough for happiness with an arcade aftertaste.

Pulse source ran at slow speed, and the feedback noise output becomes a pseudorandome patterned trigger sequence generator. Here, it strikes a resonant VCF.

Demo of the cow output

Demo and frequency sweep of the xor output

A simple patch with the COW and XOR outputs shaped with transient envelopes by an LPG and a VCA respectively

Full demo of the module - everything except the bassdrum is made out of its outputs.

Pictures

To top

Digital Noise Generators