Technical

Beyond Frequency | Where Loudspeakers Lose Time — Part II

TCA-M Active Loudspeakers in a New York apartment.

Why Loudspeakers Need Crossovers and Where Timing Breaks Down

A loudspeaker must reproduce more than ten octaves of sound. No single drive unit can do this efficiently or without distortion. Crossovers exist to divide the signal so that each driver operates where it performs best. However, the way this division is implemented has profound consequences for timing.

A crossover, however, shapes not only amplitude but also phase and time. Every filter used to redirect energy between drivers introduces Phase Rotation and Group Delay. In passive crossovers built from coils, capacitors and resistors, these effects are unavoidable and increase rapidly with filter order.

Physical time alignment of drivers at their acoustic centres is often presented as a solution to timing problems in loudspeakers. While this is beneficial, it addresses only one part of the issue. Aligning the acoustic centres can synchronise the initial arrival of sound at a single frequency or at a specific crossover point, but it does not resolve the frequency-dependent Group Delay introduced by the crossover itself.

Conventional passive crossovers, such as fourth-order Linkwitz–Riley designs, are optimised for flat summed frequency response under anechoic conditions, not for Time Coherence. Around the crossover frequency, each branch exhibits a pronounced Group Delay peak. In typical loudspeakers with crossover points between 80 and 300 Hz, this produces several milliseconds of delay before enclosure effects are considered.

In ported loudspeaker systems, this timing error is compounded by stored energy in the Helmholtz resonator formed by the enclosure and port. The port output reaches its maximum precisely where Group Delay is also highest. Published measurements commonly show 30 to 50 milliseconds of delay near the tuning frequency, with some systems exceeding this. By comparison, sealed enclosures usually remain below 10 milliseconds, though still far from constant.

“Bass reflex and bandpass systems often exhibit substantial Group Delay” [in excess of 50 milliseconds]., Audioholics summary (secondary) Audioholics

The combined result is that the low-frequency portion of a transient can arrive tens of milliseconds after the midrange and high-frequency components. This misalignment lies directly within the auditory system’s critical timing window.

Psychoacoustic research shows gap detection thresholds around 0.5 to 5 milliseconds, cross-band synchrony evaluated within approximately plus or minus 10 to 20 milliseconds, and clear perceptual asynchrony beyond about 20 to 30 milliseconds. These values overlap precisely with the musically meaningful microtiming range of roughly 5 to 50 milliseconds, where rhythm, drive and phrasing are encoded.

For many engineers, this realisation comes only after the electronics have been refined. Once bandwidth, noise and linearity are addressed upstream, it becomes clear that the dominant limitations arise beyond the preamplifier. The loudspeaker, and particularly the crossover and enclosure, becomes the defining bottleneck for time accuracy.

When Group Delay vary by Frequency the Fundamental and its Harmonics do Not line up in time

The corresponding Square Wave is distorted and transients less defined

Looking closely at the charts above, we see the effect of a filter with non linear phase. The fundamental and its harmonics are not time aligned because the filter’s group delay varies with frequency, so different frequency components are delayed by different amounts.

Group delay is formally defined as the negative slope of phase with respect to angular frequency. If group delay is constant with frequency, the system is linear phase within the band of interest. If group delay changes with frequency, the system is not linear phase, and multi frequency waveforms will be time smeared.

A square wave can be represented by its Fourier series, meaning it can be built from sine waves at integer multiples of a fundamental frequency. For a 50 percent duty cycle square wave, the spectrum contains odd harmonics only, with amplitudes decreasing as 1 over harmonic number. The steepness of a square wave’s flanks is determined by two factors:

Bandwidth: how many high order harmonics are present and at meaningful amplitude
Time coherence: whether those harmonics arrive together in time, which requires linear phase, or equivalently constant group delay, over the relevant band

This links directly to loudspeakers because each rising or falling edge of a square wave is effectively a step, so the loudspeaker’s step response describes how cleanly it can reproduce rapid waveform transitions. A time coherent loudspeaker produces a step response that rises promptly and settles cleanly, while excess phase rotation, misaligned drivers, or stored energy appears as overshoot, ringing, or a smeared rise.

The charts below show the same sine wave with its first nineteen odd harmonics and the corresponding much cleaner square wave after passing through a linear phase filter.

When Group Delay is Constant the Fundamental and its Harmonics line up in time

If we know timing matters, why don’t most loudspeakers preserve it?

When bass energy lags midband energy by 10 to 40 milliseconds, the auditory system no longer binds these components into a perceived single coherent event. Attacks lose unity, rhythmic articulation softens, and the performer’s timing is replaced by the loudspeaker’s stored-energy behaviour.

Sealed enclosures are often perceived as producing tighter bass than ported designs, partly due to improved time behaviour. However, crossover-induced Group Delay and enclosure-related energy storage remain.

The resulting bass character of these conventional designs has become so familiar that it is widely accepted as normal, even though live acoustic instruments heard in the same room are consistently perceived as more immediate, coherent and physically present.

Because most loudspeakers share similar enclosure and crossover behaviours, listeners and designers alike have become accustomed to their temporal artefacts. What is familiar is easily mistaken for natural. Yet when acoustic instruments are heard live in the same room, their bass transients, decays and spatial cues are consistently perceived as more immediate, coherent and physically present.

The contrast reveals that the limitation lies not in human hearing, but in the playback systems we have learned to accept.

Historically, early radios, televisions and some loudspeakers from the 1950s resembled open frames rather than enclosed boxes. Sound radiated forward and backward, and energy storage was limited. These designs were often described as open and immediate, but they lacked low-frequency extension and sound pressure capability.

Most contemporary loudspeaker systems remain assemblies of loosely coupled elements. Passive crossovers, external cabling and separate power amplifiers each contribute their own Phase Rotation, Group Delay variation and level-dependent behaviour. The resulting Transfer Function is complex, unstable and impossible to fully characterise or correct.

Taken together, these loosely coupled elements form a complex and unstable transfer function.

Phase rotation, group delay variation, enclosure energy storage and room excitation all contribute to the final acoustic output, yet none are controlled as part of a unified system. Once timing errors and stored energy are introduced inside the loudspeaker, they cannot be removed downstream. This is why no amount of optimisation in electronics alone can restore temporal accuracy once the loudspeaker itself becomes the dominant source of distortion.

Psychoacoustic research consistently shows that the temporal resolution of human hearing exceeds the timing accuracy of typical loudspeakers by a wide margin. Timing governs groove, synchrony across frequencies, and whether early reflections are fused with the direct sound or perceived as separate events.

The challenge is therefore not to recreate an original sound field, which is physically impossible once a performance is recorded and replayed elsewhere. The challenge is to preserve the temporal relationships that allow the brain to recognise sound as coherent, expressive and real. When these relationships are preserved, reproduction ceases to draw attention to itself and becomes a convincing window into musical intent.

Early Attempts at Restoring Time

By the early 1980s, it was already clear to a small group of engineering friends that the problem did not end at the music source or preamplifier stage. Even with wide-bandwidth, DC-coupled analogue electronics exhibiting excellent linearity, the moment the signal entered a conventional loudspeaker the time relationships embedded in the recording were disrupted.

Passive loudspeaker crossovers were the obvious next target. While a well-designed passive network could achieve a near-flat anechoic frequency response, it could never be phase linear.

This led to the development of analogue phase linear active crossovers placed before the power amplification stage. By dividing the signal at line level and feeding each band into a dedicated power amplifier directly coupled to its driver, it became possible to improve both electrical control and acoustic integration. With careful design and crossover slopes limited to a maximum of 12 dB per octave, simulations showed that near-linear phase behaviour was achievable without instability.

Three-way, and later four-way, analogue active systems were built and evaluated. The improvement over passive systems were immediate and unmistakable. Transients became clearer, bass articulation improved, and musical timing felt more intact.

Yet even with these advances, a fundamental obstacle remained unresolved.

Cover image for: Early Attempts at Restoring Time — Active Analogue 4-way Phase Linear 12dB/Oct Crossover - conceived ca. 1983

The Unfinished Problem

Despite optimised electronics and active phase linear crossovers, conventional loudspeaker enclosures still imposed their own time errors. Stored energy in sealed and vented cabinets, delayed port output, and mechanical resonances introduced low-frequency group delay that could not be corrected electrically. The bass system itself became the dominant source of temporal distortion.

This was particularly evident when listening to acoustic instruments. Even when reproduced in the same room, loudspeakers failed to convey the immediacy, spatial coherence and physical presence of live sound. The difference was not subtle, and it could be recognised instantly, even when heard briefly through an open door or window.

These early experiments represented a decisive step forward, but they also revealed the scale of the challenge. Restoring timing through electronics alone was not enough. The loudspeaker had to be treated as a complete electro-mechanical-acoustic system, including drivers, enclosures and radiation behaviour, not merely as a load at the end of a signal chain.

Why the Time Dimension Remained Missing

For decades, loudspeaker development continued to advance primarily in the steady state frequency domain. Measurements became more refined, distortion figures fell, and modelling tools improved. Yet the time domain remained largely uncontrolled because it spanned multiple physical domains that were rarely addressed together.

True time coherence requires simultaneous control of frequency response, phase behaviour, driver alignment, enclosure energy storage and room interaction. Passive systems, by their nature, fragment this transfer function across amplifiers, cables, crossover components and mechanical structures. No amount of optimisation in one area can compensate for uncontrolled behaviour in another.

Patented Folded Dipole Air Velocity Transducer Bass System

Because most loudspeakers exhibit similar forms of group delay and transient smearing, listeners and designers alike have become accustomed to their sound. What is familiar is accepted as natural, even when it differs profoundly from live acoustic experience.

Only with the emergence of high-resolution Digital Signal Processing (DSP), sufficient processing power and the ability to measure and model complete systems end to end did it become possible to address the time dimension properly. Not as an abstract ideal, but as a practical design goal grounded in psychoacoustic reality.

This realisation forms the bridge to the TCA-M Active Loudspeaker. Not as an incremental improvement, but as a system conceived from the outset to restore what had been missing all along: precise control of timing across the entire audible range, without compromise.

Once timing is lost inside the loudspeaker, it cannot be corrected downstream.

In this Part II of our Beyond Frequency series we have concluded that:

Physical driver alignment alone cannot resolve frequency-dependent timing errors
Passive crossovers inevitably introduce Group Delay within the musically critical band
Boxed bass systems and reflex ports store energy precisely where timing matters most
These errors fall directly inside the brain’s sound event integration window
Once timing is lost inside the loudspeaker, it cannot be corrected downstream

This leads to a fundamental question: what if the entire system could be treated as a single, controlled Transfer Function?

“another advantage of active speaker technology, particularly when implemented digitally. With DSP and proper drivers, amplifiers, and enclosures, a designer can overcome the limitations that plague purely passive designs, especially in terms of bass extension.”

Monthly Column, by Doug Schneider for SoundStage! HIFI | Dec 01, 2025