Confirm that the sound system is powered and connected. Start with all faders down. This workflow will guide you through a controlled ring-out process to identify and eliminate feedback frequencies.
The ring-out technique involves intentionally bringing the system to the edge of feedback. Keep your hand on the fader at all times and pull back immediately if feedback becomes loud or sustained. Protect your ears and your loudspeakers. Start with gain well below unity and increase in small increments (1-2 dB at a time).
Audio feedback is the most disruptive problem in live sound reinforcement. That piercing howl or low-frequency rumble can derail a performance, damage equipment, and harm hearing. Understanding the physics behind feedback and mastering systematic elimination techniques separates professional sound engineers from amateurs. This guide covers everything from the acoustic theory to practical ring-out procedures and filter design.
Feedback occurs when an electroacoustic system forms a closed loop: a microphone captures sound, the signal is amplified, a loudspeaker reproduces it, and the reproduced sound travels back to the microphone. If the total loop gain at any frequency exceeds unity (0 dB) and the phase shift around the loop is a multiple of 360 degrees, oscillation occurs at that frequency. This is a direct application of the Nyquist stability criterion from control theory.
The loop gain is the product of several factors: microphone sensitivity and directivity at the angle toward the speaker, preamplifier and channel gain, loudspeaker sensitivity and directivity toward the microphone, and the acoustic coupling between speaker and microphone (which depends on distance, room reflections, and air absorption). The frequency where feedback occurs first is where these factors combine to produce the highest loop gain. This is almost never the frequency where the system sounds loudest to the audience — it is determined by the complex interaction of all elements in the chain.
The ring-out (also called “ringing out the monitors”) is the standard method for identifying feedback frequencies in a live sound system. With the microphone in its performance position and the sound system at operating levels, the engineer slowly increases the gain until the first frequency begins to ring — sustaining with a tonal quality after transients have passed. This frequency is noted, a narrow notch filter is applied to suppress it, and the gain is increased further to find the next feedback frequency. The process continues until the system achieves adequate gain-before-feedback (GBF).
A well-executed ring-out typically identifies 3 to 6 primary feedback frequencies per microphone channel. Each notch filter applied increases the GBF by 1 to 3 dB. After treating 5-6 frequencies, a total GBF improvement of 6 to 10 dB is typical. The goal is not to eliminate all possible feedback frequencies — it is to push the GBF high enough that the system operates with adequate headroom during the performance.
A notch filter (also called a band-reject or narrow parametric cut) reduces the gain at a specific frequency with minimal effect on surrounding frequencies. Three parameters define a notch filter: center frequency, depth (gain reduction in dB), and Q factor (bandwidth). The Q factor is inversely proportional to bandwidth: Q = f_center / bandwidth. A Q of 10 at 1 kHz has a -3 dB bandwidth of 100 Hz (about 1/7 octave). A Q of 30 at the same frequency has a bandwidth of only 33 Hz (about 1/21 octave).
For feedback suppression, the optimal approach uses the narrowest possible filter with the minimum depth required to suppress the feedback. Start with a Q of 15-20 and a depth of -3 to -4 dB. If feedback persists at that frequency, increase the depth to -6 dB before widening the filter. Depths beyond -6 dB are rarely needed and suggest the frequency has a broader resonance problem that may benefit from physical treatment (microphone repositioning, loudspeaker aiming, or acoustic absorption).
Automatic feedback suppressors (AFS) detect feedback frequencies in real time and apply notch filters automatically. While convenient, they have limitations: they can misidentify tonal musical content (sustained notes, organ pipes, feedback guitar) as feedback and incorrectly notch it. They also add latency and may react slower than a skilled engineer. Most professional sound engineers use automatic suppressors as a safety net — preset a few “fixed” notches during soundcheck (via manual ring-out) and allocate a few “live” filters for the automatic system to handle unexpected feedback during the show.
SonaVyx combines the best of both approaches: its real-time feedback detector uses spectral peak tracking with Q analysis and prominence detection to distinguish feedback from musical content, while the engineer retains full control over whether to apply each recommended filter. This hybrid approach provides the speed of automatic detection with the judgment of manual intervention.
The most effective feedback prevention happens before any EQ is applied. Microphone technique is paramount: position the microphone as close to the sound source as possible to maximize the direct-to-reverberant ratio. Use directional microphones (cardioid, supercardioid, hypercardioid) with their null axis pointed at the nearest loudspeaker. Loudspeaker placement should maximize the distance from microphones and use directional speakers aimed at the audience, not the stage. Monitor positioning for wedge monitors should exploit the microphone's rear rejection: the monitor should fire into the back of the capsule for cardioid patterns.
In-ear monitors (IEMs) eliminate stage monitor feedback entirely by removing the acoustic coupling between monitors and microphones. This is the single most effective feedback prevention measure for performers. Acoustic treatment on stage (absorption on back walls, side fills angled appropriately) reduces the reverberant field that contributes to feedback. System gain structure should be optimized so that each component operates within its optimal range without requiring excessive gain at any point in the chain.
Using graphic EQ for feedback suppression is inefficient because each band has a fixed bandwidth of about 1/3 octave (Q ~4.3). This removes far more signal than necessary — a parametric notch at Q=20 is roughly 6 times narrower and removes 6 times less content. Over-filtering with too many or too deep notches makes the system sound thin and unnatural. If you need more than 8-10 notches per channel, the problem is physical, not electrical. Ringing out at performance volume from the start risks speaker damage and hearing injury — always begin well below operating level. Forgetting to ring out with all microphones open simultaneously misses feedback frequencies that only appear when multiple open microphones combine their loop gain.
SonaVyx's Feedback Elimination workflow guides you through each step methodically, tracks identified frequencies, recommends appropriate filter parameters, and documents everything for future reference. The before/after comparison verifies that your notch filters have improved the gain-before-feedback without degrading the system's overall frequency response.