The boxes that shaped
everything in between.

Variable-mu tube — gain reduction via shifting tube bias. Attack as fast as 0.2 ms; six preset release time constants. The smoothness comes from tubes responding to envelope rather than discrete gain-control elements.

T4 optical attenuator — electro-luminescent panel + dual cadmium-sulfide photo resistors sealed in a vacuum-tube-sized canister. One resistor controls gain reduction, the other drives the VU meter. The lag in the luminescent panel's response is what gives the LA-2A its famously slow, program-dependent compression character.

FET as voltage-variable resistor for gain reduction; I/O transformers; Class A line-level amplifier; T-pad resistive attenuator drops input voltage to FET; 1108 circuit preamp uses bipolar Darlington pair.

VCA-based stereo bus compression with specific gain-reduction character. The "glue" comes from the way the VCAs respond to programme material — subtle, fast, and consistent across the stereo image.

VCA gain reduction using the Blackmer Gain Cell. RMS level detection + feed-forward gain reduction — smoother than feedback designs of the era. Percussive transient response with very fast attack.

Diode-bridge gain reduction — diodes mounted in a bridge configuration with carefully selected control voltages. Pioneering use of this topology in pro audio; significantly reduced distortion vs earlier diode-based designs. Distinct sonic character from VCA / FET / optical / variable-mu.

Passive LC (inductor-capacitor) filter section; the negative-feedback amplifier circuitry was licensed from Western Electric, but the filter design was Shenk's own. The "Pultec trick" — simultaneously boosting and cutting at the same low-frequency band — creates a slight dip just above the boost peak, producing more low-end without muddying.

Combines API's proprietary 2520 discrete Class-A/B op-amp with Proportional Q filter topology. As gain increases, bandwidth narrows: ±2 dB = ~3 octaves wide, ±12 dB = ~1 octave wide. Gentle EQ moves are broad and musical; surgical moves are surgical — both behaviours from one circuit.

Class A discrete transistor amplification. Marinair input transformer (collaboration with Marinair Radar Company, late 60s/early 70s). LO1166 gapped output transformer — designed by Rupert Neve in 1964 for his first transistorised console (Phillips Records), spec still proprietary at Neve HQ. Inductor-based EQ with handcrafted UK-made inductors.

Fully parametric — independent controls over frequency, bandwidth (Q), and gain on each band. Conceived around 1966 by Macneal as a sweep-tunable EQ avoiding inductors and switches.

Entirely passive filter network in the EQ signal path — only metal-film resistors, film capacitors, and hand-wound inductors sculpt the sound. Zero active circuitry in the EQ; a clean tube amplifier stage restores gain afterwards.

1 m × 2 m thin steel plate suspended in tubular steel frame under controlled tension. Driver transducer induces sound waves; one (mono) or two (stereo) pickups capture wave energy. Decay time controlled by a mechanical damping pad moved closer to / further from the plate via remote.

TMS9900-class processors + custom DSP boards. Multiple algorithms (Plate, Chamber, etc.) selectable from front panel. Famous four large coloured paddle controls for parameter adjustment — pre-screen-era UX design.

Two-part architecture: 4U mainframe with audio processing + dedicated console-top remote for parameter control. 2 inputs / 4 outputs. Interchangeable program cartridges for different spaces (chambers, plates, rooms).

18-bit / 48 kHz sampling, 98 dB dynamic range. Two simultaneous reverb engines that can run in parallel in the same box.

Microprocessor-controlled DSP with 9 reverb programs (Hall, Plate, Chamber, Ambient, Non-Linear — the gated-reverb program). Front-panel parameter editing; internal converters.

16-bit linear converters at 31.25 kHz sampling rate (compromised to keep cost down — 20 Hz–12 kHz on processed signals; full bandwidth in bypass). 30 preset algorithms: 4 reverb / 6 chorus-flange / 4 pitch + 14 others (gates, comps, vibrato, panning, freeze, gated reverb). MIDI program change + parameter control + footswitch trigger.

Long springs in a tensioned chassis with carefully matched transducers and pickups + electronic conditioning + adjustable damping. Two independent channels (dual mono or stereo configuration); each signal has individual decay time adjustable 2.0 to 4.5 seconds.

Tape loop in cartridge format; moving record head as the core innovation — variable delay without changing tape speed. Plus dedicated playback head + erase head. The preamp section became famous in its own right (many players ran signal through an Echoplex with delay off for the colour alone).

Free-floating 1/4" tape loop in tape chamber (no cartridge). Single capstan transport with three playback heads + multi-tap selector — produces the characteristic syncopated delay pattern. Spring reverb built-in (the RE-101 was identical minus the spring tank).

Steel-alloy drum spinning over 4 playback heads. 100 super-fine stainless-steel recording wires wrapped around the drum exterior — each 0.1 mm diameter (pencil-line thin). 6× 12AX7 tube preamps. Each playback head selectable independently — multi-tap echo before "multi-tap" was a marketing term.

Continuous loop of 1/4" tape passing three replay heads + one record head; each head switchable independently. Feedback loop sends output back to record head for runaway/oscillating echo. Wow + flutter from worn capstans is part of the signature.

Digital pitch-shift via buffer-splice technique — splice points were audible as "glitches", a quirk later solved by the H949. Plus delay + simple modulation. Red 7-segment readout.

MicroPitch mode — proprietary single-sideband modulation for very small pitch shifts (0.1% resolution, 4-digit display) — used as a precision tuning + tempo tool. Reverse mode digitally emulates cutting tape into ~6-inch segments (~400 ms), reversing each, then splicing in original order. RAM-based memory architecture. Unique flange + randomized-pitch effects.

2U 19" rack. 11 effect categories × 100 presets — diatonic stereo pitch-shift, multi-tap delays, modulation (chorus / flange / doubler / etc.), reverb algorithms, modular + upgradeable cards. MIDI implementation for external control.

Adds phase-shifted harmonics + intermodulation distortion + slight compression. Produces perceived brightness + "air" that simple EQ cannot replicate — psychoacoustic enhancement: the ear hears more clarity without signal energy being simply boosted.

Dolby A: 4-band split-frequency companding. Input split via filters at 80 Hz, 80 Hz–3 kHz, 3 kHz high-pass, 9 kHz high-pass. Each band compresses on record + complementarily expands on playback → ~10 dB NR at low end, up to 15 dB at 15 kHz. Dolby SR (1986): more bands, sliding action, ~24 dB NR.

Dual-channel musical noise gate with separate range + release controls + frequency-conscious key filter (high-cut + low-cut on the side-chain detection). External key input enables one signal to trigger the gate on another — the snare-triggers-ambient-mic technique that defined 80s drums.

Splits incoming voice (modulator) into 16 frequency bands; each band's envelope controls the corresponding band of a carrier signal (typically a synth pad or sawtooth). Mic/line inputs, internal oscillator + noise source, voiced/unvoiced detector, slew + freeze control.

BBD (Bucket Brigade Device) analog charge-coupled delay lines for very short delay times, modulated subtly. Distinct from typical chorus: no pitch modulation, only spatial widening + slight time offset between channels. Result is "wider stereo" rather than the warbling vibrato most choruses produce.