Back to engineering projects
// PRJ_006 · Analog Electronics & Audio

3-Key Analog
Synthesizer.

A polyphonic analog synthesizer built across two breadboards with zero microcontrollers. Three voices are summed through a UA741 op-amp, shaped by an ADSR envelope driving JFET VCAs, filtered through a switchable HP/LP stage, and amplified by an LM386 into an 8 Ω 2 W speaker.

Project type
Analog electronics mini-project
Role
Analog circuit design · ADSR & filter stages
Core systems
Oscillators · VCAs · ADSR · filter · LM386
Focus
Pure analog audio synthesis
Jump to gallery
Clean-wired synthesizer breadboard with binding posts, three potentiometers, op-amps, and the output speaker
// Clean rewire · full synth on a single breadboard
Voices
3 polyphonic summed via UA741
Filter
HP / LP switchable -3 dB at 500 Hz · 10 dB gain
Envelope
ADSR 74LS02 SR latch · 2N5457 VCA
Output
LM386 → 8 Ω 2 W ∼0.5 W typical, 6.25 W theoretical max
// 01 — Problem

A musical instrument made entirely of op-amps.

The brief was to build an analog synthesizer that produces distinct musical pitches when keys are pressed, mixes those voices, filters them in real time, and amplifies the result through a speaker — all without a microcontroller. Frequency comes from RC products, polyphony comes from a summing op-amp, character comes from an envelope generator and a filter, and loudness comes from a dedicated audio amp IC.

Design constraints

  • Low-voltage DC supply (9–12 V); dual ±12 V for op-amps
  • Analog electronics only — no microcontrollers or DSP
  • Op-amps used for mixing, filtering, and envelope/effects
  • Momentary push-button keys, like a real keyboard
  • Impedance-matched output to an 8 Ω 2 W speaker
  • Real-time tuning via potentiometers and switches
  • Must fit on 1–2 breadboards (forces a 3-key count)

Design selection

Three concepts were scored 1–5 on sound variety, real-time control, and harmonic capability: a mono synth with filter (3/15), a poly synth with filter and knobs (11/15), and a poly synth with ADSR + filter + knobs (15/15). The third concept won on every axis, so the final build adds envelope shaping on top of polyphony and tunable filtering.

// 02 — System Flow

From key press to shaped speaker output.

Each key gates a voice through a JFET VCA; the ADSR envelope decides how that voice grows and fades over time. All three voices land at a summing op-amp, the sum passes through an HP/LP filter, and the LM386 lifts the result up to speaker level.

RC oscillators set the pitch

Each voice uses an op-amp RC oscillator where output frequency is set by f = 1/(R·C). Pitches follow twelve-tone equal temperament with A4 = 440 Hz, so a note n semitones from A4 sits at F = 440 · 2^(n/12).

Momentary keys trigger ADSR

Pressing a DPDT key starts the envelope: a capacitor charges through R_A (attack), partially discharges through R_D to a sustain voltage set by R_S, and releases through R_R when the key is let go. A 74LS02 NOR-gate SR latch with a comparator transitions the stages.

JFET VCA imprints the envelope

The DC envelope voltage drives a 2N5457 JFET biased in its ohmic region, varying its channel resistance and modulating the AC oscillator signal. The result is the same waveform with attack, decay, sustain, and release shaping baked in.

UA741 summing op-amp mixes voices

The three shaped voices feed an inverting summing amplifier — resistors 10 kΩ, 8.45 kΩ, and 6.65 kΩ into the virtual-ground node, with a 10 kΩ feedback. V_out = −R_f(V1/R1 + V2/R2 + V3/R3), dual ±12 V supply.

HP/LP filter trims the timbre

A dual op-amp first-order filter on the LM358 sits between the summer and the amp. An SPDT switch picks high-pass or low-pass; a potentiometer tunes the corner. Designed for −3 dB at 500 Hz with 10 dB gain — R_f = 2.2 kΩ, C = 0.33 µF.

LM386 drives the speaker

The LM386 is wired for a gain of 200 to lift the small mixed signal to speaker level. At a 0.1 V input that's 20 V peak in theory; in practice the supply caps the swing — comfortable in the half-watt range into the 8 Ω 2 W speaker.

// 03 — Subsystems

Five analog stages, one signal path.

Voice oscillators

Each voice is an op-amp RC oscillator whose frequency comes from f = 1/(R·C). With C = 0.01 µF and target A4 = 440 Hz, R works out to 227 kΩ — resistors are sized per note from the twelve-tone formula.

ADSR envelope generator

R_A / R_D / R_S / R_R are potentiometers so attack, decay, sustain, and release are all tunable live. A capacitor C_Env stores the envelope state; a comparator + 74LS02 SR latch flips between charging and discharging paths when the key is pressed and released.

JFET VCA

A 2N5457 N-channel JFET runs in its ohmic region as a voltage-controlled resistor. The envelope's DC level varies the channel resistance, which scales the oscillator's AC signal — transferring envelope shaping onto an audio waveform.

UA741 summing op-amp

Inverting summer with input resistors 10 kΩ, 8.45 kΩ, and 6.65 kΩ (different values let each voice carry a different weight), 10 kΩ feedback to ground via the virtual short. Dual ±12 V rails keep peaks well clear of clipping.

HP / LP filter

Dual first-order filters built on a single LM358 dual op-amp. An SPDT switch routes the signal through either the high-pass or the low-pass leg; a potentiometer trims the corner frequency on the fly. Targeted: −3 dB at 500 Hz with 10 dB gain.

LM386 audio amplifier

Audio-amp IC dedicated to driving small speakers from a 4–12 V supply. Configured for a gain of 200 (Av = 200) with the LM386's standard input/decoupling network feeding the 8 Ω 2 W speaker. Frequency response ∼5 Hz to 250 kHz covers the audio band comfortably.

UA741 op-amp LM358 dual op-amp LM386 audio amp 2N5457 JFET 74LS02 NOR / SR latch 1N4149 diode DPDT momentary keys SPDT filter switch Dual ±12 V supply Twelve-tone equal temperament Multisim Breadboard prototyping
// 04 — Validation & Math

Tuning by hand, verified on paper.

Because there's no firmware to print a value, every stage is validated either with a Multisim simulation or with hand calculations against datasheet numbers, then tuned in real time on the breadboard with the pots and the SPDT switch.

Filter design (HP & LP)

  • Target: 10 dB DC gain → 20·log(1 + R_f/R_1) = 10 dB → R_f/R_1 = 2.16
  • Pick R_1 = 1 kΩ → R_f = 2.16 kΩ (standard value 2.2 kΩ)
  • Cutoff: f_c = 1/(2πRC) at 500 Hz → with R = 1 kΩ, C = 0.32 µF (standard 0.33 µF)
  • Same math applied symmetrically for the high-pass leg

LM386 output power

  • Gain configured for A_v = 200; V_in = 0.1 V → V_out = 20 V peak (theoretical)
  • Practical clip-limited at ∼10 V peak → 7.07 V RMS → P = V²/R = 49.98/8 ≈ 6.25 W
  • Within the speaker's 2 W rating: drive at 2 V RMS → P = 4/8 = 0.5 W comfortable
// 06 — Outcome

A working synth with real-time character.

The finished synth plays three distinct pitches polyphonically, with attack/decay/ sustain/release shaping each note and a switchable filter trimming the timbre. Six potentiometers and an SPDT switch on the front of the board mean a player can sculpt the sound while pressing keys — the whole point of the design.

What worked

  • Building the LM386 amp on its own board first let us debug audio in isolation
  • SR-latch + comparator cleanly separates attack/decay/sustain/release stages
  • JFET VCA in the ohmic region transfers the DC envelope onto AC audio without distortion
  • SPDT switch + pot on the filter gives a satisfying real-time control surface

What I'd change

  • Move from breadboard to a soldered PCB to drop the noise floor and add more keys
  • Add an LFO stage for vibrato and slow filter sweeps
  • Switch to a 2nd-order Sallen-Key filter for a steeper, more musical roll-off
  • Replace the discrete-resistor oscillators with a tunable RC ladder for finer pitch control