Speaker Crossovers: What They Do and Why They Are Hard

Why a speaker needs one

No single driver covers the audio band well. A cone large enough to move air at 30 Hz becomes directional and starts breaking up into uncontrolled modes long before the top of the band. A dome small and light enough to reach 20 kHz cannot survive the excursion that low frequencies demand for more than a few seconds.

So the band gets split, and each driver is fed only the part it handles cleanly. The network that does the splitting is the crossover, and the frequency where responsibility passes from one driver to the next is the crossover point.

Passive and active

A passive crossover sits between the amplifier and the drivers, inside the cabinet, and works at full power. It needs large inductors and capacitors, it dissipates some of the amplifier's output as heat, and its behaviour is entangled with the drivers' own impedance curves, because those form part of the filter. It has the enormous practical advantage of needing exactly one amplifier and one pair of cables.

An active crossover sits before the amplifiers and works at line level, with one amplifier channel per driver. It can be built to a precision a passive network cannot approach, and because it operates on the signal rather than on power it can also apply delay and correction per driver, which is what allows the acoustic centres of physically offset drivers to be aligned in time. The cost is more amplifiers, more channels and a fixed design that cannot be used with anything else.

Order, slope and what each one trades

Filter order sets how steeply the level falls beyond the crossover point: first order is 6 dB per octave, second order 12 dB, third 18 dB, fourth 24 dB.

A shallow slope leaves a wide band where both drivers are producing sound together. That overlap is where the two sources, separated by real distance on the baffle, interfere with each other, so response changes noticeably with listening height and with angle. Shallow filters also ask more of each driver, since the tweeter still receives substantial energy well below the crossover point.

A steep slope narrows the overlap, protects the tweeter and gives cleaner power handling near the crossover point. It also introduces more phase rotation, uses more components, and pushes each driver harder right up to the point where it stops, which can leave a breakup mode only just suppressed.

Neither is correct in the abstract. What matters is that the acoustic result, the drivers' own responses combined with the electrical filter, comes out the way the designer intended.

Phase, and why Linkwitz-Riley became the default

Filters shift phase as well as level, and the two halves of a crossover have to sum back to a flat response through the region where both are contributing.

A Butterworth alignment is 3 dB down at the crossover point, so two of them sum to a 3 dB bump rather than to flat. A Linkwitz-Riley alignment, built as two cascaded Butterworth sections, is 6 dB down at the crossover point, and the two halves sum flat with both drivers radiating in phase. That is the reason it became the standard choice for both passive and active designs, and why fourth-order Linkwitz-Riley in particular is so common: it sums flat without needing one driver inverted.

The crossover region is where speakers are judged

The awkward part is that a two-way speaker usually has to hand over somewhere between 2 and 3 kHz, which is close to the region where hearing is most sensitive and where the ear is best at detecting small irregularities. The handover also happens between two drivers at different points in space, so the combined output varies with vertical angle: sit or stand outside the design window and the response through that region changes.

This is why a speaker that measures well on axis can still sound different in a real room, and why off-axis behaviour is worth as much attention as the on-axis curve.

What else the network is doing

Splitting bands is only part of the job. A well designed passive crossover is usually also compensating for baffle step, the gradual loss of low-frequency output as a cabinet stops reinforcing into half space; matching levels, because tweeters are almost always more sensitive than the mid-woofer they sit beside; conforming the impedance the amplifier sees, so the filter behaves as designed; and notching out a driver's breakup peak, which may be far above the crossover point.

That is a lot of overlapping requirements met with one set of parts, and it is why crossover design is the part of speaker engineering that separates good from ordinary.

Two-way against three-way

A two-way design has one crossover region and fewer parts, and its mid-woofer has to cover everything from the bass up into the presence region, crossing where the ear is least forgiving.

A three-way design can put its handovers where hearing is less critical, perhaps a few hundred hertz and around 3 kHz, and let a dedicated midrange driver own the vocal band without also having to move air for the bass. The price is a second crossover region, a taller cabinet and a more expensive network.

Neither is automatically better. A well executed two-way beats a compromised three-way easily, and the number of drivers on the front of a cabinet is not a specification.

What to listen for

Play a solo voice and move your head up and down slowly. A design with a well managed crossover region holds its tonality through that movement; one with a difficult handover will change audibly. Then listen off to the side. The direct sound barely changes, but what the room returns to you does, and that is most of what makes a speaker easy or tiring to live with.