How to tune a PA, step by step — what actually matters

Why the equaliser comes last

A measurement shows the sum of everything: the loudspeaker itself, where it stands and where it points, how sub and tops play together, and what the room does with all of it. The equaliser can only change part of that — the part that is the same at every seat. A filter acts in the front row exactly as it does in the back row.

The big mistakes, though, are the ones that differ from seat to seat: too loud at the front, too quiet at the back, too much bass in the middle, a hole at the side. A filter changes those by zero decibels. Floyd Toole puts it in one sentence: an equaliser cannot change a loudspeaker's directivity. And for Bob McCarthy, who shaped the tuning of large systems for decades, aiming always comes before equalising — never the other way round.

The five steps, ordered by effect

Next to each step is how many decibels are at stake — calculated, not guessed. The number falls from top to bottom. At the very bottom is the step most people start with.

1. 1

   Place and aim the tops

   up to 18 dB front to back

   Direct sound loses 6 dB every time the distance doubles. Front row at 2 m, back row at 16 m: three doublings, an 18 dB difference — if the box sits at ear height. Raise it to 2.8 m above the ears and it is down to 13.5 dB, because the front row is now further away. Tilt it towards the back rows and the front row sits off the main axis and gets even less. And the coverage angle on the data sheet is the angle at whose edge the level is already 6 dB below on-axis. Whoever sits there hears 6 dB less — with nothing wrong at all; that is the definition.

2. 2

   Place the subwoofers

   up to 15 dB from seat to seat

   Two subs left and right cancel wherever their path difference reaches half a wavelength — at 60 Hz that is 2.86 m. Even if one of the two arrives 3 dB quieter there, the dip sits 10.7 dB below a single sub and 15.3 dB below what the same two would give in phase. Put both together in the centre and the striped field is gone at every seat.

3. 3

   Sub and tops: polarity and timing

   +6 dB or a hole

   In the crossover region both play the same note. Equally loud and in phase, they give 6 dB more than either alone. At 60° offset that is still 4.8 dB, at 90° 3 dB, at 120° nothing at all — and at 180° they cancel. At 100 Hz, 180° is 5 ms or 1.72 m of path. Check polarity first, then set the delay from the measured arrival time, not from the tape measure.

4. 4

   Decide the bass share and the target curve

   a decision worth up to 12 dB

   This is not a correction but a decision, and it belongs before the equaliser — otherwise the EQ works towards a target that has not been set yet. How far it reaches is shown by a compromise two FOH engineers described in Live Sound International in 2019: flat from 200 Hz to 20 kHz, and below that 12 dB more sub centred on 60 Hz. For a speech that would be far too much. So decide what the system is playing today, and set the sub level to match.

5. 5

   Equaliser

   0 dB between seats

   Only now. It evens out what is equally wrong at every seat: the box's own voicing and individual room resonances in the bass. It does nothing about the difference between front and back, centre and side. McCarthy's rule of thumb for a single system: usually three or four bands, never more than six, rarely narrower than half an octave. If you need twelve, you skipped one of the four steps above.

How to measure so the measurement holds up

Microphone: an omnidirectional measurement mic with a calibration file, on a stand in the audience area — not right in front of the box, because there the woofer and tweeter have not yet summed into the sound the audience hears. Height: where the ears will be during the show, usually standing head height, lower for a seated audience. Point the capsule at the loudspeakers.

Several seats: never equalise from a single position. Above 343 Hz the wavelength is shorter than a metre; at 1 kHz it is 34 cm. A hand's width further over you measure a different fine structure. So measure at several seats in the audience and equalise against the average. REW recommends the dB average for this, because it gives peaks and dips equal weight.

Window and smoothing: a frequency-dependent window of 15 cycles — the way REW does it — is 1.5 ms long at 10 kHz. In that time sound travels 51 cm, so in the treble you see almost nothing but the loudspeaker. At 100 Hz it is 150 ms and 51 m: there you see the whole room, and that is exactly how it should be. For curves that are going to be equalised, REW recommends variable smoothing, from 1/48 octave below 100 Hz up to 1/3 octave above 10 kHz.

The point where the room stops having individual resonances is the Schroeder frequency: 2000 × √(reverberation time ÷ room volume). A parish hall of 1,200 m³ with 1.2 s of reverb comes out at 63 Hz. Below it the equaliser may work narrow, against a spatially averaged measurement. Above it only broad — Toole calls narrow corrections there a significant risk of making a good loudspeaker worse.

How to tell the equaliser is doing the wrong thing

The most common mistake is boosting a dip. If it comes from a cancellation — two paths taking each other out — the filter raises both, and the dip stays. Pat Brown describes the case with a dip between two drivers: on axis it is gone afterwards; off axis, where most people sit, a peak appears.

The test takes a minute: move the mic 30 cm and measure again. If the dip moves or disappears, it was interference — placement and delay fix that, not a filter. If it stays at every seat, it is real. With peaks it is the other way round: cutting costs no headroom. Hence the old rule: cut narrow, boost wide and little.

What AURIX does at this point

AURIX measures sub and tops separately with a sine sweep, derives arrival times, delay and polarity, and proposes filters against the target curve you choose. It cuts by up to 12 dB, boosts the sub by 3 dB at most and the tops by 5 dB at most — and never below a box's natural roll-off or into a cancellation. When it finds one, it states the path difference in metres instead of boosting it. Alongside that it lists the room modes, where a narrow filter does belong.

Above the Apply button it shows how good the measurement is, from 0 to 100 %, with reasons. A blocked measurement is refused on the first press. What AURIX does not do for you are steps 1 and 2: placing and aiming stay manual work. And AURIX measures at one position — so check the result with the RTA at a second one.