About
Update 16 Jan 2024: i finally built a much better wavefolder - build that instead. This design is for acrchival purposes only, divide all my good words about it by four while reading! Its sound is quite one-dimensional compared to the Lyssa Wavefolder. Here's to me thinking of a completely original design someday...
What it says on the lid: a wavefolder! More precisely, six similar 'weak' wavefolders stacked in series for a huge fold, like an origami crane or something. It is actually a frankensteiner of the Kassutronics discrete VCA and my own quad wavefolder, which is itself a frankensteiner of Ken Stone's "wave multiplier", Serge TWS and whatever else i saw on the internet on the topic. It uses a simple joy of diode clipping to form nonlinearities in the input-output voltage transfer, or more simply, makes your sinewave a funky mess, for a cost of a few cheap parts. While it is a little bit underwhelming in its range of sounds, what it does - it does well.
On the top, there is a huge LEVEL knob. It has two associated CV inputs, each with a dedicated orange attenuator. It controls the level at which the mixdown of the two inputs - 1 (AC) and 2 (DC), with the white attenuators - hits the wavefolding part of the circuit. The louder it gets, the more folds it 'reaches', hence the most harmonically complex the sound becomes.
This comes with a downside - when the LEVEL is low, the output is usually somewhere from silence to quiet, slightly altered version of the inputs. This may be handy to use this unit as an interesting wavefolding VCA of a sort, but the BLEED control allows to fade in as much of the dry, unprocessed input mixture, as you wish, to make it feel more like the input is gaining harmonics without changing in volume too much.
The module has six wavefolding subcircuits stacked one after another, and apart from having inputs and outputs, each has a special bias voltage point, 'around' which its nonlinearity is folding. They all share a slightly complicated network which sets their voltages. It is controlled with the SHIFT control and its CV input (blue knobs) and the SYMMetry control (which, frankly, does close to nothing most of the times). The MODE switch switches between two shift modes - sometimes it barely makes a difference, sometimes the sound becomes opposite.
Finally, at the bottom, there are three 'wet' outputs 2F, 4F and 6F, after 2, 4 and 6 folders respectively. First sounds like a gentle folding VCA, second is a full-fledged lush fold, and the third goes into waveshaping and extreme harmonic enrichment territories. One of them can be patched into the "to sum" input to the right of them to be mixed down with the dry inputs, post BLEED knob. 6F is normalled into the to sum input for instant operation. This system can be replaced with a three-position ON-ON-ON switch, rotary or slider, for quicker routing. On the other hand, this system allows, for example, to crossfade between 2F and 6F through a separate module, and then use that as the input added to the dry signal.
Overall, this is a good module that is a bit too complicated for what it does, but what it does - it does really well. Can go from lush to harsh, has a lot of sweet spots, and definitely makes the nice clean waves into an obliterated mess.
Schematic
I am not going to explain this schematic in big detail, for a simple reason: half of it is magic beyound my understanding, half of it is the same dual-inversion summator in different poses and at different angles. So, i'm going to give a brief, slightly chaotic descriptioni of what's going on, which may be just enough for you to understand it and build one, or maybe use its ideas in your own crazy wavefolder project.
This design takes the quad wavefolder's core subcircuit to a whole new level, where it actually works: it is shown in the red rectangle. After playing with its values for a while, i found out, that overamplifying the signal by lowering R1 leads to a bigger "fold", or harsher nonlinearity, but has to be compensated with a bigger compensating amplification afterwards. This circuit also showed interesting behaviour when the voltage 'against' which the diodes do their nasty foldy deed is slightly altered around ground. This summed up, i decided to stack up a bunch of these in pairs, each pair more 'nonlinear' then the previous one, and have some funny network that shifts the 'folding points' (BIAS) around. This is the core circuit of this module - six identical wavefolding units, one after another.
The initial voltages for the 'shift network' are created with IC3B/C - a usual double-inverting summator kind of a deal, which sends a non-inverted sum of the SHIFT and SCV to all odd folders (1,3,5), and a MODE switch selects to send either the inverted or the non-inverted sum to all even folders (2,4,6). The voltage is then divided down and down with 10K resistors both for odd and even side, until it reaches the 50K symmetry control potentiometer. This thing 'skews' the whole network more towards the odd or even side: sometimes it doesn't do anything, but sometimes it changes the character of the unit drastically. If you don't want to have it, replace it with two 22K resistors from each side to ground.
The input to the wavefolder chain is passed through a VCA, depicted at the top of the page. This sort of wavefolder design can have two sorts of voltage control that would animate it: either process the volume of the signal before it so that it 'reaches' more folds the louder it gets, or apply some weird biases inside the circuit (surely more experimental). I went with the first one, as i already experimented with the quad wavefolder and attenuating its inputs and it works really well. There's a lot of VCA designs out there, but i always had an itch to try the Kassutronics discrete VCA, and this was the perfect case, so i went with a somewhat beefed up version of it: 2 audio inputs (one AC and one DC coupled), 2 CV inputs, all with dual-inverting summation (inverting amp configuration has a bonus of virtual ground at its - terminal, allowing for way less crosstalk). It works fairly well even with me not giving a single heck about actually matching the transistors, so i bet it will sound even better if you actually do that. You can read about how it works through the link. I still don't get why is T3 'reversed', but hey, it does the job.
Finally, the audio inputs of the VCA, fed to the wavefolding chain, shifted by the shift and symmetry controls, are mixed down with the dry, pre-VCA mixture of the audio inputs. The bleed control sets how much of the dry signal bleeds throuhg, hence the name. The more you add it, the more it sounds like "original wave getting more funky harmonics", instead of "weird VCA that goes from silence to folds to more folds to distortion, maybe" (which also has its users).
