A matrix mixer with output-to-input feedback self-oscillates into an instrument with no source

A matrix mixer connects multiple inputs to multiple outputs through a grid of level controls, each pot setting how much of one input reaches one output. When output channels are patched back into input channels, the mixer enters feedback: signal circulates, builds, and transforms through the loop. Even with no external sound source, once the loop gain exceeds 1 the circuit self-oscillates and sings — this is ‘no-input mixing,’ pioneered by David Tudor (e.g. ‘Rainforest IV’) and carried on by later musicians. The character of the feedback is shaped by loop gain (feedback-pot levels), equalization (which frequencies sustain versus die out), phase relationships, and any time-based effects in the path: a delay or reverb in the loop adds instability, rhythmic patterns, and evolving timbral complexity. The patch is not pre-programmed; the sound emerges through careful, continuous level adjustment, making the matrix itself a performance instrument with each pot a musical parameter.

Examples

A 3×3 matrix: outputs 1 and 2 feed back into inputs 1 and 2 while output 3 drives the speaker; slowly open the feedback pots while adjusting a graphic-EQ insert and the system starts to sing. Or route one send to a speaker, another to a toy’s clock, a third to a delay that loops back — the result evolves with no further input.

Assessment

What minimum condition must be met for a no-input feedback matrix to produce sound? Explain how a graphic EQ inserted in the feedback path shapes the musical character, and how placing a digital delay in the loop differs from using it as a linear send-return effect. Name one parameter that would most dramatically change the feedback character.