Skip to content
MINISYNTON.

From the MGW-111-A to the plug-in

Source material

The functional model is based on the supplied scans of Grzegorz Wodzinowski's article, “Minisynton — a musical synthesizer for everyone”, and its corrections. The supplied original article pages are 3, 4, 5, 7 and 8, plus page 19 of the follow-up explanations. Page 6 was not supplied. The corrections identify the original publication as issue 12/83.

This is an independent implementation, not an endorsed product of the original author, Steinberg or Cockos. The historical scans are not redistributed in the public downloads.

Signal path

Keyboard and pitch memory → transposition → VCO → three-waveform mixer → VCA → output. An envelope controls the VCA, while two independent low-frequency oscillators provide vibrato and tremolo. Additional capacitors in the hardware create glissando. The original is monophonic and has no VCF, noise generator or separate filter envelope.

Keyboard, memory and glissando

The hardware keyboard uses PR1…PR48 to set individual note voltages. US1 provides threshold detection; the transistor network and C6 store the control voltage. The stored pitch remains after the key is released. Switching in C3…C5 slows voltage changes.

The plug-in stores the current and target frequency, then applies exponential smoothing in the frequency domain. This is an idealized RC-like response in the hardware's linear voltage-to-frequency control path. It does not model capacitor leakage. Last-note priority is a deliberate MIDI implementation choice, not a claim that the article specifies that exact multi-key algorithm.

Tuning and transposition

US4 and its surrounding components, together with P6/P7, scale and shift the keyboard voltage. The calibration procedure indicates a linear frequency/voltage relationship, rather than the familiar 1 V/octave convention. The article gives approximately 1.6 Hz per millivolt and a VCO control range from +10.2 mV to +9.8775 V. The correction changes the printed upper frequency to 15,804 Hz, not 1,580 Hz.

MIDI notes use equal temperament:

f = 440 × 2^((note − 69 + 12 × octave + cents/100 + pitch_bend)/12)

Pitch bend in this expression is measured in semitones. This frequency becomes the glide target. Transpose and fine tuning replace voltage adjustments. The software's approximate 16 Hz–16 kHz limit is not a precise measurement of a hardware specimen.

Oscillator and mixer

US5 is an integrator and US6 a comparator, with T11 switching the oscillator. The available outputs are square at terminal 35, triangle at 36 and an approximated sine at 37. The converter around T12 and the diode/resistor network shapes the sine in segments. The calibration procedure requires a 1:1 square-wave duty cycle.

The digital waveforms share phase and frequency. The square uses PolyBLEP; the triangle uses a frequency-limited finite sum of odd harmonics. The oscillator runs at twice the host sample rate and is followed by a fourth-order low-pass decimation filter. This reduces aliasing but does not guarantee its complete elimination at extreme settings.

Sine character adds small third and fifth harmonics. It is not a measured recreation of the hardware's piecewise transfer curve. P11…P13 and the hardware mix resistors are represented by three waveform levels. A summed level above 100% is normalized for predictable headroom; that normalization is a software design choice.

VCA and envelope

The hardware VCA uses T8…T10. Its digital counterpart multiplies the waveform mix by the envelope and amplitude modulation. The exact nonlinear transistor gain characteristic is not simulated.

Although the scanned diagram uses “ASDR”, the text describes attack, held sound, and release. P9 controls attack and P8 release; the middle stage lasts while the key is held. The implementation therefore uses a linear attack and release with full sustain during the held note, not an invented independent decay stage and sustain-level control.

Legato preserves the envelope across overlapping notes. Optional retrigger uses a short 2 ms fade-down before a new attack. That behaviour is a MIDI extension.

Vibrato and tremolo

US7/US8 generate the low-frequency signals. T13/T14 drive lamps whose light changes the resistance of FR1/FR2. One path affects pitch and the other amplitude. The article specifies approximately 4–8 Hz and separate depth controls.

The plug-in uses two independent sine LFOs in that range. It omits lamp thermal inertia and photoresistor memory and nonlinearity. Depth in cents or percent is a practical musical control, not a one-to-one mapping of a measured hardware potentiometer position.

Corrections taken into account

  • Keyboard PR1…PR48 are 100 kΩ; they form a keyboard divider, not 48 separate oscillators.
  • FR1 and FR2 are RPP130 photoresistors.
  • R19 connects directly between the output of US3 and its inverting input.
  • D1 and D2 must be connected in the opposite direction to the erroneous drawing.
  • The upward line from US2 in the power-supply drawing is redundant.
  • PCB corrections concern the ground connections of R31/R33/R34, connecting C3/C4/C5 to R15 rather than R13, interchanged T8/T10 labels, and C1/C2 versus C7/C8 markings.
  • The page 7 calibration text should read 10 kΩ, including a 10 kΩ trimmer replacing R12/R13, rather than the erroneous 10 Ω.
  • Page 8 corrections include 15,804 Hz, P9/P8 rather than P9/P10, PR8, and PR2/PR49.

These corrections inform the functional interpretation. The erroneous PCB drawing was not converted blindly into an electrical netlist.

What is not modeled

There is no ±15 V supply simulation, test-point calibration, resistor trimming, thermal drift or aging. There is no X–Y joystick panel or separate manual always-open VCA mode. MIDI velocity, sustain, pitch bend, automation, software presets and selectable legato are extensions.

Envelope time ranges, modulation depth ranges and sine colour are software design decisions where no direct measured mapping is available. A more exact emulation would require oscillator waveforms, spectra, VCA transfer curves and optical-modulator response measurements from a physical instrument.

The result is a working functional interpretation of the documented synthesizer, not a certified component-by-component or sound-identical copy.