An all-pass filter is a signal processing filter that passes all frequencies at equal amplitude while changing only their phase. Unlike low-pass, high-pass, or band-pass filters, which selectively block certain frequencies, an all-pass filter lets everything through at full strength. Its output has exactly the same magnitude as its input at every frequency, but different frequency components get shifted in time relative to each other.
That might sound like it does nothing useful, but phase manipulation turns out to be a powerful tool. All-pass filters are fundamental building blocks in audio effects, digital reverb, communication systems, and equalizer design.
How It Differs From Other Filters
Most filters people encounter are frequency-selective. A low-pass filter removes high frequencies. A high-pass filter removes low frequencies. A band-pass filter keeps only a specific range. All of these reshape the amplitude spectrum of a signal, making some frequencies louder and others quieter (or silent).
An all-pass filter does none of that. Its amplitude response is unity (a value of 1) at all frequencies. What it changes is the phase relationship between different frequency components. Since any complex signal is really a sum of sine waves at different frequencies, shifting those sine waves in time relative to each other changes the shape of the waveform without changing its frequency content. The signal sounds different or behaves differently in a system, even though a spectrum analyzer would show the same frequency profile before and after.
Phase Shift and Group Delay
The key property of an all-pass filter is its phase response. A first-order all-pass filter shifts phase from 0 to 180 degrees (π radians) across its frequency range. At a specific frequency called the break frequency, the phase shift is exactly 90 degrees. Higher-order all-pass filters produce greater total phase shifts: an Mth-order all-pass filter shifts phase by M × 180 degrees from the lowest to highest frequency.
Group delay describes how much different frequencies are delayed in time as they pass through the filter. For an all-pass filter, this group delay is always positive, meaning the filter genuinely delays the signal rather than advancing it. The group delay varies with frequency, which is what makes the filter useful for phase correction. You can design an all-pass filter whose group delay profile compensates for the uneven delays introduced by other components in a signal chain.
The phase response of a stable all-pass filter decreases monotonically with frequency. It doesn’t jump around or reverse direction. This predictable behavior is what makes all-pass filters mathematically well-behaved and practical to design.
Poles and Zeros in Mirror Positions
If you’ve studied filter theory, the defining structural feature of an all-pass filter is the relationship between its poles and zeros. For each pole located at some point inside the unit circle (in the digital domain), there’s a corresponding zero at the mirror-image position outside the unit circle. If a pole sits at a distance r from the center at angle θ, its paired zero sits at distance 1/r at the same angle.
This conjugate reciprocal arrangement is what guarantees the flat magnitude response. The pole’s tendency to boost certain frequencies is exactly canceled by the zero’s tendency to cut them, leaving amplitude untouched while still affecting phase. Keeping all poles inside the unit circle ensures the filter is both stable and causal, meaning it produces a bounded output and doesn’t require future input values to compute.
Phase Equalization in Communication Systems
One of the most important applications of all-pass filters is fixing phase distortion. When a signal travels through a transmission line, amplifier, or other filter, different frequencies often experience different delays. This non-uniform group delay smears the signal in time, which can cause errors in digital communication or distortion in audio.
Because all-pass filters affect only phase, you can cascade one with an existing system to flatten out the overall group delay without disturbing the amplitude response you’ve already designed. This is especially valuable with certain types of recursive digital filters, which can’t have perfectly linear phase on their own. By adding an all-pass filter in series, you can approximate linear phase (uniform delay across frequencies) over the frequency range that matters.
Audio Phaser Effects
Guitar phaser pedals are one of the most recognizable applications of all-pass filters. The effect works by splitting the audio signal into two paths. One path passes through a chain of all-pass filter stages, which shift the phase of different frequencies by different amounts. When this phase-shifted copy is mixed back with the original, frequencies that end up 180 degrees out of phase cancel each other, creating notches in the frequency spectrum.
The number of all-pass stages determines how many notches appear. A 4-stage phaser produces 2 notches, a 6-stage phaser produces 3, and so on. Analog phasers typically use 4, 6, 8, or 12 stages, while digital phasers can offer 32 or more. The break frequencies of the all-pass stages are swept up and down by a low-frequency oscillator, which moves the notches through the spectrum and creates the phaser’s signature sweeping, swooshing sound.
Digital Reverb Algorithms
All-pass filters play a critical role in artificial reverberation. In the 1960s, Manfred Schroeder developed a foundational algorithm for simulating room acoustics digitally. His design used a bank of parallel comb filters to simulate early reflections, then fed the result through a series of all-pass filters to build up echo density and mimic the dense, diffuse quality of late reflections in a real room.
Natural-sounding reverberation needs an echo density above roughly 1,000 impulses per second. Comb filters alone can’t achieve this without requiring impractical amounts of processing. All-pass filters solve the problem because they increase echo density while maintaining a flat frequency response. The reverb tail doesn’t color the sound by boosting or cutting certain frequencies; it just adds the dense, decaying wash of reflections that makes a space sound real. Nearly every digital reverb algorithm since Schroeder’s original work uses all-pass filters in some form.
Building One With an Op-Amp
In analog electronics, all-pass filters are built using operational amplifiers, resistors, and capacitors. A first-order all-pass circuit needs one op-amp and a handful of passive components. Second-order designs, like the Delyiannis topology, also require only a single op-amp, which makes them economical to build.
The design process starts by choosing a capacitor value, then calculating resistor values based on the desired center frequency and a quality factor (Q) that controls how sharply the phase transitions around that frequency. Higher Q values concentrate the phase shift into a narrower frequency band, producing a steeper group delay peak. These circuits are straightforward enough that they’re standard fare in analog synthesizers, audio equalizers, and signal conditioning systems where precise phase control matters.