Granular [Tetra] Grid

Core Techniques:

  • Chaotic cross modulation with magnets (Hail the donut)
  • Complex circular FM tones from Cš-L
  • Rhythmic gating over index ratios, wavefold depth and signal amplitudes
  • Granular /reverberant spatialisation using arbhar
  • TZFM as tone generator and processor simultaneously
  • Harmonic chord voice using Harmonàig/Saïch

There are several unusual techniques used in this patch. The result being a very interactive, playable and texturally rich performance instrument.

There is a lot of chaos involved on both the audio and control generation sides so some luck is required when improvising with the intent of landing on a regular “groove”. Many of the patching methods translate beautifully to more harmonically based patches.

In this instance I very much embraced the noise!

There is a little bit less linearity to the signal flow, but mixing via a matrix (Lìon) allows for relative ease and adaptability of the configuration.

In a nutshell, the key players in this patch are:

• Cš-L (primary sound source)
• arbhar (granular processing/spatialization)
• Harmonàig/Saïch (harmonic element)
• Manifold Research Centre – Tetragrid (hail the donut)

The Tetragrid is the first module from Manifold Research Centre. MRC is the creation of Aimo Scampa, a long time Instruō colleague, and the developer behind the Lìon (and many more modules to come!).

The Tetragrid is fundamentally 6 oscillators configured as 6 CV/gate sources. Connecting the triangle touch points on the playing board with magnet pieces or a finger will interconnect modulations between these 6 oscillators. Placing magnets in sweet spots can produce consistent repeating rhythmic gates from the various outputs. Patching in just the right way can produce some extremely musical ostinatos and grooves!

The individual sound sources are fundamentally pretty straight forward so let’s dive in there:

The Cš-L is producing most of the audio content.

I have implemented circular FM between the voices controlled via the INDEX control. The INDEX also defines a level of cross mixing via the wavefolder symmetry inputs.

This means each FINAL wavefolder output is a folded sum of both oscillators’ sine waves. Their frequencies are a product of the complex FM algorithm that is defined by the FM attenuators and INDEX position.

I use 4 gate signals from The Tetragrid to interact with the Cš-L. Two CV sources are attenuated via Tanh[3] channels and control the 1V/octave inputs. Two farther signals from The Tetragrid are patched to the Cš-L’s wavefold depth CV inputs. The attenuverters of these are relatively low and introduce jumps in timbre periodically by these gate signals.

When the 4 Cš-L controlling gate signals produced by The Tetragrid stabilise on something periodic the result is a rhythmic ostinato (repeating pattern) that plays from both FINAL outputs. 

The tones are constant coming from the module although depending on the index ratio of cross modulation, and intervals applied at the 1V/octave inputs, there can be natural breaks in the tones as frequencies land in sub or ultrasonic territories. Both of these extremes get cleaned out by AC coupling and antialiasing filtering as they go through the arbhar’s hardware codec.

Before summing the two Cš-L signals to mono there is a latching switch stage that is applied to the two parallel FINAL tones.

A 5th gate signal is used from The Tetragrid to control a Cèis ADSR module. The envelope output is not actually utilised in this patch but the timing of the Attack, Decay and Release stages and their accompanying gate outputs are.

The Decay and Release stage gates are used to toggle the latching states of a Tàin module’s switches which pass or mute the Cš-L voices separately.

The timing of A, D, and R stages of the Cèis add a further level of rhythmic timing control to the patch in the form of signal alteration. Simply pressing the buttons on the Tàin can jog the timing and invert the rhythmic metre of the ostinato.

The switched Cš-L signals now sum via the Lìon matrix mixer. Parallel mono mixes are produced of the Cš-L tones and parsed to two destinations.

One of these mono branches is processed farther by means of the Neóni oscillator in its Through-zero frequency modulation configuration.

This complex voice from the Cš-L is actually utilised as a modulation signal applying through-zero frequency modulation (TZFM) to the Neóni’s core. The FM input AC coupling is engaged and the sine output is used as the signal output. 

The Neóni is now serving the obscure dual purpose of tone generator and complex waveshaper simultaneously.

When the modulation voice/signal sits in comfortable audible frequency range territory, the TZFM effectively wavefolds the amplitude adding a unique harmonic distortion. The “carrier frequency” as defined by the fundamental pitch of the Neóni (the coarse/fine pitch knobs) will define the level of wavefolding/distortion applied to the throughput audible modulator signal.

If the frequency of the tones land on something in the low frequency/subsonic territory the “tone generation” element comes into play.

When modulated into LFO ranges these DC signals and particularly slow movement voltages will cause the carrier pitch to approach stalling/0Hz. As the modulation signal moves with an increasing rising or falling vector this translates to an audible swing up in frequency (or down in negative frequency) of the carrier tone.

When the modulator signal is a combination of slower/static voltages and flurries of movement the sonic result from the carrier signal can become audible lower frequency rumbles, rhythmic “wobble” bass tones and percussive kicks/toms.

The fascinating parallel comes when mixing and/or comparing the raw modulator signal (Cš-L voice) and the result of TZFM. When the Cš-L voice is inaudible for periods due to the tones landing subsonically, these dead spots get sonified by the Neóni generating audible carrier tone content.

This patching configuration is technically very complex and can be a bit of a mind stretcher conceptually when coming from more traditional FM algorithm implementations… but in practice it actually feels very intuitive to dial in musical sweet spots.

(Definitely a Neóni use-case to keep on hand for experimental processing of audio signals.)

The now-processed branch of the Cš-L voice gets summed in with the raw Cš-L voice sum at the Lìon.

I use a couple of tricks to mix at preferred amplitudes.

The Cš-L sources from the Tàin patch to inputs 1 and 2 of the Lìon. These branch via ζ to the Neóni which then re-enters the matrix at input 6.

The new combined mix which progresses the patch’s audio signal path comes from output γ.

Nodes are patched for the raw Cš-L/Tàin tones at α,1 and β,1. Purple patch pins are used which attenuate the signals to 1/4 amplitude.

As nothing is patched directly from outputs α or β, a farther amplitude reduction to 1/3 is applied to the γ output sum.

The Neóni source is pinned at node γ,6 with a blue pin which attenuates to 1/2 amplitude.

The signal branch going to the Neóni is pinned at ζ,1 and ζ,2 again with purple pins quartering the amplitudes. (The amplitude is then halved again at the Neóni’s FM attenuator).

It’s really better not to bother meticulously calculating attenuation and mix ratios when using the Lìon in practice! Experimentation is what it was designed for. If attenuation is desired then pick a different pin, or cascade an output.

Trial and improvement!

This brings us to the arbhar stage:

The signal from γ is the dry source which patches to the ONSET input jack (which internally normals to the input).

A self patched øchd serves as a rich modulation source to various parameters on the arbhar. 

LFO1 – self modulates the øchd.

LFO2 – Grain direction probability >%<



LFO5 – MOD (Reverb)

LFO6 – Grain windowing


I use most of these modulated parameters in essentially a set-and-forget configuration. I will occasionally manually adjust the length parameter to pull the arbhar back from its wavetable range or to nudge the grain length in a preferred direction.

The Dry/Wet knob is the primary performance control of the arbhar as well as a [1]f fader used to attenuate LFO5 on its way to the MOD CV input.

The arbhar is set to its internal reverb mode which is controlled via the MOD CV.

When CV is at 0V there is no reverb applied. With positive CV up to +5V the reverb mix will be brought up. When negative CV is applied down to -5V the wet reverb is applied while the dry throughput is reduced and removed.

The polarity of CV is effectively randomised by the øchd LFO while the reverb level/depth can be manually controlled via the [1]f. 

Dynamic reverb throws are achievable by manually riding the MOD CV amplitude.

The outputs of the arbhar are patched to inputs 3 and 4 of the Lìon which then output from δ and ε with unity gain (black) pins at δ,4 and ε,3.

These outputs end the audio signal path as they go to the cuïr. The outputs could also have been patched directly from the arbhar outputs but integrating them into the Lìon matrix opens up some extra possibilities in the form of feedback patching.

I use a purple pin (1/4 amplitude) to occasionally route an output from the arbhar back into the γ signal sum which patches to the arbhar input.

Feedback in this configuration can add an interesting resonance with a Karplus-Strong like timbre.

This occurs due to the slight latency occurring from the signal digitisation and processing through the arbhar.

When combined with throws of the reverb, strong metallic resonances can be introduced to the patch.

It adds an interesting character to the tonality/texture but can creep up quite aggressively in amplitude depending on the voice’s character.

I actually placed a tanh[3] channel directly before the arbhar input. This benefits both the occasional feedback pin patching from running away too much, as well as controlling the widely variable amplitude ranges coming from the Neóni voice.

The final part of the patch is the harmonic element.

This is simply my go-to Harmonàig voice.

The Root 5th and 7th chord tones control the Saïch (oscillators (‘) and (”) doubling the root). There are some interval offsets applied to the voicing by the Saïch.

The Harmonàig is defined in Lydian mode and manually played in Performance mode.

The Saïch output is patched through an I-ō47 utilising the Notch output as a low-pass output with the ability to mix in some high-pass content as needed.

The Harmonàig gate output triggers a Cèis ADSR which opens the cutoff frequency.

An LFO (øchd) sitting at ~4Hz adds a musical vibrato to the voicing.

This chord voice is then summed to the γ signal at γ,5 with a unity gain pin. The input gain fader of the I-ō47 defines the level in the mix that goes to the arbhar.

With the harmonic voice summing in pre-arbhar the chordal content can then be sampled into the arbhar buffers as well as being included in reverb throws (which sounds great!).

There are a wide range of capabilities in this patch. 

With a larger system there are many aspects that could be developed farther and utilised with more precision control to be a more curated performance piece.

I developed the patch to be very hands on and responsive within a minimal set of parameters. The full range of arbhar buffers can be accessed via a Tàgh sample & hold. This expansion opens up a lot of opportunity to use a broader pallet of source audio to be captured into the arbhar.