Synthesizer module · 0.1.2 · Released 6 November 2026 · Free · GPL-3.0
Highstead
CA-74
A synthesizer modelled from the circuit of a 1970s expander module: one voice, like the module, or one whole module for each note with POLY.
No purchase, no account. VST3, CLAP, AU and standalone.
Listen
Factory presets of the first release.
PWM Strings
Four held chords, two bars each, at 104 BPM, the last resolving from suspended.
Notch Pad
Four held chords, two bars each, at 100 BPM.
Band-Pass Funk
An eight-bar sixteenth-note bass line in E at 104 BPM.
Huck Finn
Four low notes, each held two bars, at 100 BPM.
Leap
An eight-bar brass chord riff in F at 112 BPM.
Ocean
One key held for eighteen seconds.
Rendered offline from the 0.1.0 release at QUALITY HI. No effects, and each preset at its own level.
The module
One board, transcribed from the schematic and simulated in ngspice, then derived into a voice that runs in real time. It is calibrated by default to a current recreation of the module, measured part by part.
The 12 dB state-variable filter. NOTCH sweeps from low-pass through notch to high-pass, or switch to band-pass. As on the module, the filter does not follow the keyboard.
- SYNC
- Hard sync between the two oscillators.
- MODULATION
- Sends the envelope or the LFO to each oscillator’s frequency or pulse width.
- PULSE WIDTH
- From 10 to 90 %.
- MIXER
- Each VCO blends from saw to pulse into the filter; EXT takes your host’s side chain or noise.
- ENV 1
- Attack, decay and sustain; drives the VCA and can modulate VCO 1.
- ENV 2
- Attack, decay and sustain for the filter and VCO 2.
- LFO
- The module’s own; the mod wheel brings it in on the oscillators.
The patch points are left out, so EXT on the modulation switches does nothing. The module has no velocity input, so velocity doesn’t change the sound. MPE pressure and timbre can move the modulation or the cutoff.
Voices
One voice like the module, or one whole module per note. The strip under the panel holds what the module never had.
- MONO · POLY · UNISON
- POLY gives each note a whole module, 2 to 10 voices; UNISON stacks them on one note.
- ENTROPY
- Part variation and drift for each voice, from 0 to 100 %.
- SCATTER · DOUBLE
- Place voices across the stereo field, or double each note into a detuned pair.
- DRIVE · AUTO GAIN · LEVEL
- Up to 24 dB into the filter, with AUTO GAIN taking back the added loudness.
- PRIORITY · RETRIGGER
- Which held key sounds, and whether each key restarts the envelopes.
- CALIBRATION
- In settings, for each instance: MEASURED (the hardware reference), MEASURED with perfect tuning, or DRAWN, the circuit as drawn.
Quality
Three voices for three jobs. The window beside the QUALITY slider shows which is running: a potato battery, a copper coil or a reactor.
A light voice, about a twelfth of HI’s work, for many voices on ordinary computers.
The circuit model, for playing.
About three times HI’s work. Render with it, or play it live with few voices.
Presets
29 factory presets, levelled to about −18 LUFS so none wins by being louder. Eight are set-ups from the module’s 1975 instruction manual. Search by tag, star favourites, save your own, and step through them from a MIDI controller.
Specs & download
- Formats
- VST3, CLAP and standalone; Audio Unit on macOS
- Systems
- macOS 11+, Windows 10 or 11 (64-bit), Linux x86-64
- Voices
- MONO, POLY (2–10), UNISON
- Stereo
- WIDTH, INNER, DOUBLE with DETUNE
- Expression
- MPE pressure and timbre to modulation or cutoff
- MIDI
- MIDI Learn; preset pads
- Presets
- 29 factory presets at about −18 LUFS
- Quality
- LO, HI and ULTRA
- Licence
- GPL-3.0-or-later
- Version
- 0.1.2, released 6 November 2026 · release notes · source code (tar.gz)
- macOSmacOS 11 or later, Apple silicon or Intel
- WindowsWindows 10 or 11, 64-bit
- Linuxx86-64: Ubuntu 22.04, Debian 12, Fedora 36 or later
Every file, the checksums and the Windows ZIP are on the release page. After installing, rescan plug-ins in your host: Highstead appears under Idle Foundry.
Before you install
- The Windows installer isn’t signed yet: choose More info, then Run anyway.
- There is no minimum specification yet. Ten voices at HI have only been measured on fast desktop computers; LO is a twelfth of HI’s work.
- ULTRA is about three times HI’s work. Render with it, or play it live with few voices.
Modelled from the Oberheim SEM (Synthesizer Expander Module), SEM-1A board of about 1976–78. Oberheim and SEM are trademarks of their owner. Highstead is not affiliated with, sponsored or endorsed by them. The names are used only to say which module’s published circuit the model was derived from. Highstead is free software from Idle Foundry Ltd., under the GNU General Public License, version 3 or later.