Mixing and gain
How to Eliminate Feedback in a Church
Quick answer
Ringing out a sanctuary typically returns 6 to 12 dB of gain before feedback: about 3 to 6 dB from four narrow parametric cuts at the ringing frequencies, and another 6 dB every time you halve the distance from the talker mouth to the microphone capsule. Four cuts is the honest limit.
Feedback is not a mystery and it is not a personality trait of your building. It is a number. Every open microphone in a room has a measurable level at which the loop gain from capsule to loudspeaker to capsule reaches unity, and everything you do to fix feedback is an attempt to move that number a few decibels further away. The useful question is never "why does it squeal", it is "how many decibels of margin do I have, and what is the cheapest decibel left to buy".
In a typical 200 to 400 seat sanctuary with a lavalier on the pastor and the mains hung 8 feet in front of him, the realistic answer is that ringing out the system buys you 6 to 12 dB. Four narrow parametric cuts are worth roughly 6 to 8 dB of that. Moving the capsule from 16 inches to 4 inches off the mouth is worth 12 dB on its own, which is why the first thing to check is never the equaliser. Before you touch anything, work out what level the room actually needs with the SPL loss by distance calculator, because a system being pushed 8 dB harder than the room requires will ring no matter how it is tuned.
On this page
What number are you actually chasing?
Gain before feedback is the difference, in decibels, between the level you can get out of the loudspeakers with a given microphone open and the level at which the system rings. A church with a handheld on a stand 3 feet from a talker and mains 10 feet away might have 4 or 5 dB of margin, which means one enthusiastic reader stepping back from the microphone starts the ring. The same room with a headset at 1 inch from the corner of the mouth has 25 dB of margin and nothing rings all year.
That is the entire subject in two sentences. Distance from the source to the capsule sets your margin, the loudspeaker pattern and where it is aimed sets how much of the system output arrives back at the capsule, and equalisation shaves the last few decibels off the specific frequencies where the loop is hottest. In that order. A room that needs 12 dB of EQ correction to stay stable does not have an EQ problem, it has a microphone placement problem, a loudspeaker aiming problem covered in speaker placement, or a reverberation problem covered in reverberation time for churches.
The second thing worth knowing before you start is that the room changes. A sanctuary full of people is roughly 2 to 4 dB more absorptive at speech frequencies than the same room empty, and the reverberant field that feeds the microphone drops with it. A system rung out in an empty room on a weekday is being tuned in the wrong room. Tune it empty to get close, then verify with people in the seats and expect to give back a couple of decibels of the margin you thought you had.
The five frequencies that ring first in a sanctuary
Rooms are individual, but the frequencies that ring in a church are not random. Six ranges account for the large majority of first rings, and knowing them means you find the frequency in one pass instead of sweeping blind for twenty minutes.
Four narrow cuts averaging 5 dB return 6 to 8 dB of gain before feedback in a typical sanctuary, and the 4 kHz cut on a lavalier or headset alone accounts for 2 to 3 dB of that.
| Frequency | What it usually is | Cut and width | Margin returned |
|---|---|---|---|
| 250 Hz | Boundary buildup from a solid wooden pulpit or a hollow lectern front | 4 to 6 dB, Q of about 4 | About 2 dB |
| 500 Hz | A width mode in a narrow sanctuary of roughly 30 to 45 feet across | 4 dB, Q of about 6 | About 1.5 dB |
| 1 kHz | A hard plaster or glass rear wall throwing energy back at the platform | 3 to 5 dB, Q of about 8 | About 1.5 dB |
| 2.5 kHz | The presence peak of a handheld dynamic sitting near a floor wedge | 4 to 6 dB, Q of about 6 | About 2 dB |
| 4 kHz | The high frequency rise of a lavalier or headset capsule, the single most common ring | 6 dB, one third octave | 2 to 3 dB |
| 6.3 kHz | Horn output on axis reaching hanging choir microphones | 3 to 4 dB, Q of about 8 | About 1 dB |
Margins are what one cut typically returns in a room of 200 to 500 seats with point source mains. They do not add cleanly: the fifth and sixth cuts usually return well under a decibel each, which is why the discipline of stopping at four matters more than the precision of any single cut.
How to ring out a church sound system
- Set gain structure before you touch an equaliser. Every channel should be sitting at roughly -18 dBFS average with peaks near -10 dBFS and the fader near unity. A channel with the preamp low and the fader pushed to the top has the same loop gain but far worse noise, and it makes every later measurement misleading. Work through the procedure in the gain structure guide first.
- Empty the room and kill everything else. House music off, other microphones muted, HVAC running as it will run during the service because a blower adds 3 to 5 dBA of masking noise that changes what you hear. Bypass every dynamics processor, every effect and every existing graphic EQ curve so you are tuning the room and not last year mistakes.
- Put the microphone where it will actually live. Not in your hand at the console. On the stand, at the height and angle the pastor or the singer uses, with the loudspeakers where they hang. The capsule position is the variable that matters most, so tuning with the microphone in the wrong place produces a curve that is wrong for the real position.
- Raise the master slowly until it just starts to ring. Move in 1 dB steps and stop at the first sustained tone. Do not let it build to a scream: a ring that is 10 dB into feedback smears across two or three frequencies and you will chase the wrong one. Note the master position, because this number is your starting margin and every step from here is measured against it.
- Find the frequency, then cut it narrow. A real time analyser on a tablet finds it in two seconds, and a sweep with a parametric band boosted 6 dB at a Q of 8 finds it in twenty. Once you have it, invert the boost into a cut of 4 to 6 dB at the same narrow width. A wide cut of 6 dB at a Q of 1 removes an octave of the preaching voice to solve a single tone.
- Repeat four times, not fourteen. Raise the master again, find the next ring, cut it. After the fourth cut you are typically returning under a decibel per cut and starting to hollow out the sound. If four cuts have not made the system usable, the problem is placement, coverage or the room, and more filters will not rescue it.
- Back off 6 dB and call that your working level. A system rung out to the edge of feedback is a system that feeds back the moment somebody cups a handheld or a reader leans away. Six decibels below the ring point is the standard working margin. If that leaves the back row too quiet, you have an amplifier or coverage problem, so check the watts you actually need.
- Verify with a full room and save the scene. Run the same microphones at the next service and watch for the ring returning at a different frequency, which is normal as bodies absorb the high end. Save the result as a named scene on the console so a volunteer cannot lose six weeks of work with one recall, and write the cut frequencies on a card in the booth.
Distance is worth more than any equaliser
The inverse square law says level changes by 20 times the log of the distance ratio, so every doubling of distance costs 6.02 dB and every halving returns it. This is the cheapest gain in the building and it costs nothing but a conversation with the person holding the microphone.
Moving a talker from 16 inches off the capsule to 4 inches returns 12 dB of gain before feedback, which is more than four parametric cuts will ever give you.
| Mouth to capsule | Level relative to 2 inches | Where you see this |
|---|---|---|
| 1 inch | Plus 6 dB | Headset boom at the corner of the mouth, the most stable position in a church |
| 2 inches | Reference, 0 dB | A handheld held properly, almost touching the grille |
| 4 inches | Minus 6 dB | A handheld held politely, and a well placed gooseneck on a low lectern |
| 8 inches | Minus 12 dB | A gooseneck aimed at the chin, a lavalier high on the lapel |
| 16 inches | Minus 18 dB | A lavalier low on a jacket, or a pulpit microphone the preacher has drifted away from |
| 32 inches | Minus 24 dB | A stand microphone at reading distance. Expect to fight this one every week |
| 6 feet | Minus 31 dB | An area microphone over a platform. This never works for speech reinforcement |
Figures are free field inverse square loss referenced to 2 inches, so a real room with a reverberant field will show slightly less loss at the far distances and slightly more forgiving behaviour at the close ones. The ranking does not change.
Microphone choice and pattern
After distance, pattern is the biggest lever. A cardioid rejects about 25 dB directly behind the capsule, a supercardioid rejects best at about 126 degrees off axis and has a small rear lobe, and a hypercardioid trades a slightly bigger rear lobe for the tightest front pattern. The practical consequence is that a cardioid wants the loudspeaker directly behind it, while a supercardioid such as the Shure BETA 58A Supercardioid Dynamic Vocal Microphone or the AUDIX OM5 Hypercardioid Handheld Dynamic Vocal Microphone for Stage Singers wants the loudspeaker slightly off the rear axis, because its null is not at the back. Aiming a supercardioid dead away from a wedge is a classic reason a change of microphone makes feedback worse rather than better. The polar pattern chart lists the rejection angles for each.
For a preaching voice, the ranking by gain before feedback is unambiguous: headset first, then lavalier, then gooseneck, then handheld on a stand. A Shure SM35 Performance Headset Condenser Microphone - TQG headset sits about an inch from the corner of the mouth and delivers 12 to 18 dB more margin than a lavalier on the same person, which is the single largest improvement most churches can make for under two hundred dollars. If the objection is visual, and it usually is, the flesh-toned earset option and the trade-offs are covered in lavalier versus headset for preaching and in the microphone for a preaching pastor page.
For sung vocals, a Shure SM58 Pro XLR Dynamic Vocal Microphone at $109.00 is stable because singers work it close. The moment a vocalist backs off to 8 inches, you have handed back 12 dB, and no console setting recovers it. That is a coaching problem, not a gear problem, and it belongs in your volunteer training rather than in a purchase order.
What a feedback eliminator actually buys you
Say this plainly, because it costs a sale and it is true: an automatic feedback suppressor buys a few decibels and it does not fix a microphone pointed at a loudspeaker. A unit like the dbx AFS2 Advanced Feedback Suppression Processor with Full LCD Display or the cheaper Behringer FEEDBACK DESTROYER PRO FBQ2496 Automatic and Ultra-Fast Feedback Destroyer/Parametric EQ with 40 FBQ Filters and 96 kHz Audio Performance places very narrow notches, typically one twelfth of an octave, at frequencies it detects as ringing. In a system with sensible placement that is worth roughly 3 to 6 dB, which is genuinely useful for a roaming handheld during a kids talk or a wireless lavalier on somebody who will walk in front of the mains.
What it cannot do is create margin that placement threw away. Point a cardioid at a horn from 4 feet and the suppressor will place notch after notch, each one removing a slice of intelligibility, until the channel sounds like a telephone and still rings. The honest rule is to allow a suppressor four or five live filters and treat the sixth as a warning that something physical is wrong. If you are shopping, the feedback eliminator roundup compares them, but spend the money on a headset first.
The same caution applies to a 31 band graphic equaliser such as the dbx 231s Dual Channel 31-Band Equalizer. A graphic band is roughly one third of an octave wide and interacts with its neighbours, so pulling three adjacent sliders down to fix one ring removes far more of the voice than a single parametric cut would. Use a graphic for broad room shaping and a parametric for rings. A processor such as the dbx DriveRack PA2 2x6 PA Complete Professional Loudspeaker Management System, Rack Mountable with Android and iOS Control. Black gives you both plus the crossover and limiting, which is why it ends up in so many church racks.
How many open microphones can you afford?
The number of open microphones matters as much as any single one. Every doubling of open microphones costs about 3 dB of system gain before feedback, so a platform running eight open channels has given away 9 dB compared with running one. This is the arithmetic behind the old rule that you mute what is not being used, and it is why a service with four readers each on their own handheld is harder to keep stable than the same service with one handheld passed along.
On a digital console this is easy to solve without a volunteer thinking about it. Gate the choir and lectern channels, set the threshold so ambient platform noise stays closed, and the console effectively runs two open microphones during the sermon instead of twelve. On an analogue desk the answer is a mute group on a footswitch and a rehearsed habit. Either way, count your open microphones before you blame the room, and if you are still specifying the console, the mixer channel calculator will tell you how many channels you actually need including growth.
Once feedback is under control, the next step is a full system tune, which sets the house curve, the delay times and the limiters rather than just the stability margin. That sequence is in how to tune a church sound system.
Sources
Frequently asked questions
How much gain before feedback should I expect after ringing out a room?
Expect 6 to 12 dB in a typical 200 to 500 seat sanctuary, made up of roughly 6 to 8 dB from four narrow parametric cuts and whatever you gain by moving the microphone closer. If ringing out returns more than about 12 dB, what you really fixed was a placement or aiming problem that was costing you margin, and the equaliser simply exposed it.
How narrow should a feedback cut be?
Use a Q of 6 to 10, which is roughly one sixth to one tenth of an octave, and cut 4 to 6 dB. That is narrow enough to remove the ringing tone without audibly thinning the voice. A one third octave graphic band is about three times wider and pulls neighbouring sliders with it, so it removes far more of the speech range to solve the same single frequency.
Why does my lavalier always ring at 4 kHz?
Lavalier and headset capsules are built with a deliberate high frequency rise, usually 4 to 8 dB somewhere between 3 and 6 kHz, to compensate for being off axis to the mouth and buried under a collar. That rise sits exactly where most loudspeaker horns are also most efficient, so the loop gain peaks there. A one third octave cut of about 6 dB at 4 kHz is the standard fix.
Will a better microphone stop feedback?
Only if it changes the distance or the pattern. Moving from a lavalier to a headset gains 12 to 18 dB because the capsule goes from 8 or more inches to about 1 inch from the mouth. Moving from one handheld to a more expensive handheld gains almost nothing, because the distance is identical and the pattern difference is worth a few decibels at best.
Do I tune the room empty or full?
Both. Ring the system out empty because you cannot run feedback tests during a service, then verify at the next full service and give back a couple of decibels of margin. A full sanctuary is roughly 2 to 4 dB more absorptive at speech frequencies, so the ring point moves and sometimes moves to a different frequency entirely. Save both as separate console scenes.
Is a feedback eliminator worth buying for a church?
It is worth about 3 to 6 dB, which matters for roaming handhelds and guest speakers who walk in front of the loudspeakers. It is not worth buying instead of a headset microphone or instead of fixing loudspeaker aim. Treat it as insurance on the unpredictable channels, allow it four or five filters, and investigate physically if it wants more than that.
Researched, not professional advice. This page is compiled from published manufacturer specifications, operator manuals, FCC rules, published standards and owner-review consensus, not hands-on testing. Sound system design, rigging loudspeakers overhead, and any electrical work are jobs for a qualified professional: have flown loudspeakers and their attachment points signed off by a structural engineer or a certified rigger, and have all wiring done by a licensed electrician to your local code. Wireless microphone rules change, so confirm the current FCC position before buying. As an Amazon Associate we earn from qualifying purchases.