Equalization
Canonical Definition
Equalization is intentional frequency-dependent gain shaping used to move a system response toward a defined target within the limits of headroom, driver capability, spatial consistency, and the underlying acoustic cause.
Routing, polarity, delay, crossover, directivity, placement, distortion, rattles, and deep interference nulls must be diagnosed before filter count becomes the strategy.
Classify The Problem First
| Measured feature | Possible action | Required caution |
|---|---|---|
| Broad tonal tilt | Shelf filter or broad low-Q correction | Check target level, channel balance, and driver range first |
| Repeatable resonance peak | Measured peaking cut or notch | Match frequency and width; confirm decay and nearby positions |
| Stable broad deficiency within safe bandwidth | Limited boost may be justified | Account for headroom, power, excursion, and distortion |
| Deep narrow position-dependent dip | Usually fix geometry, phase, delay, crossover, or source contribution | Do not assume boost can fill an interference null |
| Natural driver rolloff | Respect it, reshape it, or change the crossover / driver | Do not force unsafe extension to match a line |
| Level mismatch | Channel gain or trim | EQ is not the correct substitute for overall level |
Cuts And Boosts Are Tools, Not Moral Categories
“Always cut, never boost” is a useful beginner warning against wasting headroom, but it is not a complete engineering rule.
- Cuts: can reduce peaks and preserve headroom, but many cuts may lower the average level and lead to later gain compensation.
- Boosts: can correct a stable response deficiency, but increase voltage, power, excursion, thermal demand, and clipping risk in the affected band.
- Shelves: are often better for broad tonal shaping than a stack of narrow filters.
- Narrow filters: are appropriate for narrow repeatable features, not for chasing every comb-filter ripple.
There is no universal 6 dB maximum boost. The acceptable amount depends on the available margin and the cause, bandwidth, Q, driver, amplifier, listening level, and target.
Headroom Cost Of Boost
| Boost | Voltage multiplier | Theoretical power multiplier at the same impedance |
|---|---|---|
| +3 dB | 1.41× | 2.0× |
| +6 dB | 2.00× | 4.0× |
| +10 dB | 3.16× | 10.0× |
The system may clip, limit, compress, or run out of excursion before those theoretical demands are met. Recheck the full gain structure after meaningful positive filters or shelves.
Measurement Workflow
- Correct routing, polarity, gain structure, crossovers, timing, and obvious hardware faults.
- Define the target curve, match range, microphone positions, smoothing, and seating goal.
- Measure individual drivers and crossover sums before full-system EQ.
- Classify each feature: level, peak, broad trend, rolloff, cancellation, directivity, or noise.
- Apply a small group of purposeful filters.
- Predict the result, but then re-measure the real system.
- Compare nearby positions before retaining narrow high-frequency correction.
- Check headroom, distortion, excursion, and final listening balance.
Automatic EQ Is An Assistant
Automatic filter generation can match a measured trace toward a target, but the software does not automatically know:
- which dips are spatial cancellation
- the driver's excursion or thermal limit
- whether the target level is correct
- which seats matter
- whether a narrow filter will remain useful after small movement
- how much system headroom is acceptable
Limit the match range and boost, preserve manual filters when necessary, and verify predicted changes in both frequency and time domains.
What EQ Does Not Reliably Fix
- broken, noisy, or clipped signal paths
- polarity and routing mistakes
- deep destructive interference at a single position
- poor directivity or lobing across seats
- driver distortion, excursion limits, or mechanical damage
- rattles, buzzes, or panel resonance that requires physical treatment
- missing bandwidth from an unsuitable driver or enclosure
Documentation Standard
For every retained filter, record:
Channel / driver:
Filter type:
Frequency:
Gain:
Q or bandwidth:
Measured cause:
Target or purpose:
Positions checked:
Headroom impact:
Verification trace: