overdub
ModAbleton MoveBehringer Model D

schwung-model72

sd88me/schwung-model72 · ★ 0

Idle Foundry's CA-72, a circuit-modelled homage to the 1970s Minimoog Model D synthesizer ported as a Schwung Module for Ableton Move.

README from GitHub, GPL-3.0

Model-72 for Schwung

A circuit-modelled 1970s monosynth for Ableton Move, as a Schwung sound generator.

Model-72 is a port of CA-72, Idle Foundry's free, open-source software model of the Minimoog Model D (about 1972-73, the CA3046 "old" oscillator board). CA-72's sound is derived from the instrument's own circuit: every board was transcribed from its service documents into SPICE netlists and simulated, the real-time models were derived from those simulations and tested against them, and the result was calibrated against a hardware instrument. It is an exceptionally accurate and excellent-sounding digital emulation, and it is the reason this module exists.

Model-72 puts that engine on Move as a native Schwung module, its panel laid out as Move's knob pages, and makes it run in real time on Move's CPU. It plays in a Signal Chain slot like any Schwung synth, with no computer and no other software.

Model-72 is an unofficial, independent port. It is not affiliated with or endorsed by Idle Foundry Ltd., Moog Music Inc. or Ableton AG. "Minimoog" is a trademark of Moog Music Inc., used here only to say what CA-72 models. The module's name carries no company's name or logo.

Pages

The root page is the preset browser: turn the jog wheel to choose a sound, click to open the pages. Its knobs are the playable eight: Cutoff, Emphasis, Contour Amount, Filter Decay, Loudness Decay, Glide, Mod Wheel, Level.

PageKnobs 1-8
OscillatorsOsc 1 Range · Osc 1 Wave · Osc 2 Range · Osc 2 Wave · Osc 2 Freq · Osc 3 Range · Osc 3 Wave · Osc 3 Freq
MixerOsc 1 · Osc 2 · Osc 3 · Noise · Ext In / Feedback · Noise Type · Drive · Bass Comp
FilterCutoff · Emphasis · Contour Amount · Attack · Decay · Sustain · Filter Mode · Filter Mod
LoudnessAttack · Decay · Sustain · Level · Auto Gain, and Auto Gain's correction in dB (a readout)
ControllerTune · Glide · Glide Switch · Decay Switch · A-440 · Kbd Control 1 · Kbd Control 2
ModulationMod Mix · Osc Mod · Osc 3 Control · Mod Wheel · Bend Range · AT > Mod
VoiceQuality (LO / HI, experimental) · Mode · Voices · Entropy · Note Priority · Retrigger
StereoWidth · Inner · Detune · Played As · Placement

Every switch on the panel is an enum: Off / On, or its named positions (FILTER MODE LO / HI, NOTE PRIORITY Low / Last / High, PLAYED AS Scatter / Double ...). Everything is saved with the set.

The Move panel

It is CA-72's instrument with a few changes for a box with pads and eight knobs:

  • No source switches. OSC 1, 2, 3 and NOISE play while their volume is above 0. At 0 the module switches the source off, exactly as the panel's switch does (it costs nothing then); on the instrument a volume of 0 is already about -140 dB, so nothing is lost.
  • PATCH is always wired. The output is cabled into EXTERNAL INPUT, the classic feedback mod, and the Ext In / Feedback volume is how much: 0 is unpatched. EXTERNAL INPUT takes no other audio.
  • AT > Mod. Pad pressure (polyphonic or channel aftertouch) turns the modulation wheel. The higher of the pressure and CC 1 wins, and the Mod Wheel knob adds as before. Move has no wheels: pitch bend and CC 1 come in over USB MIDI.
  • Presets. CA-72's 19 factory sounds and an Init, converted to this panel: a switched-off source comes in at volume 0, and Ext In / Feedback is 0 unless the sound used PATCH (Feedback Funk does).
  • One output control. The instrument's VOLUME stays at 10 and Level (-30 to +12 dB) sets the output; a preset's lower VOLUME is folded into its Level (each within 0.005 dB of the original).
  • Dropped: MAIN OUT (always on).

Performance and accuracy: the hybrid approach

CA-72's full circuit voice is far too heavy for a small ARM board in real time. CA-72 also has a "light" model (its Potato mode), whose laws are fitted to the circuit model. Model-72 uses both, as a switch:

  • QUALITY LO: CA-72's light voice. MONO, POLY (2 to 10 voices) or UNISON, with ENTROPY, stereo, DOUBLE, DRIVE and AUTO GAIN. Light enough that a full chord or a ten-voice unison fits.
  • QUALITY HI: one voice, the hybrid. The light oscillators, mixer and contours drive the circuit's own filter: board 4's transistor ladder, solved in real time, through its real control node, in place of the light filter. In measurements it stays within 0.2 to 0.6 dB of CA-72's full circuit model on most patches (third-octave bands), and the filter self-oscillates in tune to within a cent. DOUBLE is off in HI.

The circuit filter is built on a background thread when the module loads (it takes some seconds); LO plays straight away, and HI plays the light filter until it is ready.

HI is experimental on Move. Measured on a Move with Model-72 alone in the chain, LO takes about 0.2 ms of each 2.9 ms audio block (7 %), HI about 0.8-0.9 ms (26-31 %). With HI the whole audio frame runs close to its budget and overruns occasionally, which can be heard as clicks. Use HI for one sound on its own, and LO everywhere else.

Features

  • The whole instrument: three oscillators (range, waveform, frequency), the mixer with noise (white or pink), the filter and both contours, glide, the modulation wheel and pitch bend, FILTER MODE, keyboard control, NOTE PRIORITY, RETRIGGER, A-440, as on CA-72's panel.
  • MONO, POLY or UNISON, 2 to 10 voices, ENTROPY (every voice a little different, as every real instrument is).
  • Stereo: WIDTH, INNER, DETUNE, SCATTER / DOUBLE, EVEN / CENTER.
  • FILTER DRIVE, AUTO GAIN (with a readout of its correction) and LEVEL.
  • Two mods that are not the instrument's:
    • BASS COMP: raises the filter's input with resonance so the bass holds as EMPHASIS rises (at EMPHASIS 8 the bass below the cutoff measured about -17 dB with it against -30 dB without). It is not DRIVE, which raises the input by a fixed amount whatever the resonance.
    • PATCH: the output cabled into EXTERNAL INPUT (above).

Not ported from CA-72: LOCK, MIDI Learn, audio into EXTERNAL INPUT, and AUTO GAIN's per-sound measurement (CA-72's average curve stands in, as it does there until a sound is measured).

Installing

Needs Schwung on the Move. Download model-72-module.tar.gz from the releases and unpack it into Schwung's sound generators:

scp model-72-module.tar.gz ableton@move.local:/data/UserData/
ssh ableton@move.local 'tar -xzf /data/UserData/model-72-module.tar.gz -C /data/UserData/schwung/modules/sound_generators/'
ssh root@move.local reboot

Then add Model-72 to a Signal Chain slot. (scripts/deploy.sh [host] --reboot does the same from a build.)

Status: 1.0.0 runs on a Move: the presets, pages, saved presets and LO tested; HI plays but may click (above). Reports are welcome.

Building

Needs Docker (the Rust engine and the aarch64 cross-compiler run in engine/Dockerfile's image):

scripts/test.sh                       # contract checks + host simulation of the plugin under ASan/UBSan (x86)
scripts/build.sh                      # aarch64 build -> build/model-72-module.tar.gz
scripts/deploy.sh [host] [--reboot]   # install on a Move (default move.local); a new dsp.so needs a reboot
engine/ca72/CA-72's real-time model, vendored; src/light.rs is the only file changed (engine/VENDORED.md)
engine/ca72-mpc/The engine: voices, MODE, ENTROPY, stereo, AUTO GAIN, the hybrid HI voice; fitting and bench examples. engine/params.json and presets.json are its parameters and factory sounds
tools/gen_schwung.pyThe pages, names, ranges, the converted presets and module.json (generates src/schwung_meta.h)
src/schwung_plugin.cThe Schwung plugin: the engine on a worker thread, the derived switches, AT > Mod, presets
test/host_sim.cDrives the plugin the way Schwung's chain host does
vendor/schwung/Schwung's plugin API header

Licence and credits

Model-72 is free software under the GNU General Public License, version 3 or later (LICENSE), as CA-72 is. The complete corresponding source of every release is this repository at its tag.

PartAuthorLicence
CA-72: the circuit and light models, the sound, the factory presets, the stereo and voice-character code (engine/ca72/, engine/ca72-mpc/src/character.rs, stereo.rs)Copyright 2026 Idle Foundry Ltd. (ca-72, commit 3f99076, 0.2.0; stereo placement from plugin-kit 4c0f0f3)GPL-3.0-or-later
The port: engine/ca72-mpc (hybrid HI voice, voices and modes, AUTO GAIN, BASS COMP, PATCH), the changes to light.rs, the Schwung modulesd88me, with Claude (Anthropic)GPL-3.0-or-later
Schwung plugin API headerCharles Vestal and contributorsMIT
libm (Rust, statically linked) and Rust's standard libraryThe Rust projectMIT / Apache-2.0

THIRD-PARTY-NOTICES.txt has the notices; VENDORED.md lists the vendored code. Model-72 contains no Moog Music, Idle Foundry or Ableton trademarks or logos beyond the attribution above.

If you like this, the thanks belong first to Idle Foundry for CA-72 and for releasing it as free software.