Flanger, chorus and phaser

Compare three modulation effects with one switch: moving notches in the frequency response, delay line or all-pass chain, LFO and feedback.

Explanation Tasks

Optional views

Block diagram
Notches over time
LFO curve
Readouts

What is happening?

Original plus modified copy

Flanger, chorus and phaser sound very different – from a jet plane to the shimmering chorus guitar to the soft swirl on an electric piano. Technically, however, all three use the same trick: they mix the original signal with a modified copy. Where copy and original are in antiphase, they cancel: the frequency response gets notches. Where they are in phase, they add up.

An LFO (low frequency oscillator, usually below 5 Hz) changes the copy slowly, so the notches move through the spectrum. Because the ear perceives timbre from how energy is distributed in the spectrum, this movement is heard as motion, as a “sweep”. The three effects differ only in how the copy is modified: delayed briefly (flanger), delayed longer (chorus) or rotated in phase by all-pass filters (phaser).

How to read the charts

The “Block diagram” shows the signal path: from the “Input” one branch goes straight to the sum as “Dry” (share 1 − mix), the other through “Delay” or “All-pass” and then joins as “Wet” (share mix). The LFO controls the delay D or the corner frequency fc; the feedback loop “fb” returns part of the wet signal to the start.

“Frequency response at this moment” shows the orange curve for exactly the instant at which the LFO currently stands; the triangles ▲ mark the notches. “Notches over time (two LFO periods)” shows the same as a map: time horizontally, frequency vertically, notches dark, boosts coloured. The vertical line is the current moment.

Try it: press “Play” and watch the notches move in the frequency response and trace their paths on the map.

Flanger: the moving comb

In a flanger the copy is delayed by a few milliseconds. A delay D shifts the phase in proportion to frequency: at frequency f it is 360° · f · D. There is a cancellation where this equals 180°, 540°, 900° …, i.e. at f = 1/(2D), 3/(2D), 5/(2D) … – evenly spaced by 1/D. On a linear frequency axis this looks like a comb (comb filter).

Worked example: D = 1 ms gives notches at 500 Hz, 1.5 kHz, 2.5 kHz … spaced by 1 kHz, 20 notches up to 20 kHz. At 500 Hz the copy is delayed by exactly half a period and, at mix 0.5, cancels the original completely. In the default setting (2 ms ± 1.5 ms) D varies between 0.5 and 3.5 ms; the lowest notch moves between 1 kHz and about 143 Hz. When the LFO changes the delay, all notches contract or spread together.

Try it: set the “Frequency axis” to “linear”, “Depth (± delay)” to 0 and compare “Notch spacing (1/D)” with “Delay now”. The name comes from slowing down one of two synchronised tape machines by touching the reel flange.

Feedback, polarity and mix

Feedback routes part of the copy back into the delay. The copy then arrives several times, each D later. The peaks between the notches become higher and narrower, the sound becomes metallic and gets a pitch of its own (1/D). At D = 1 ms and mix 0.5: with 0.7 feedback the peaks are at +6.7 dB and the notches only at −13.7 dB; with 0.9 the peaks are +14.8 dB high and narrow, and you hear a tone at 1 kHz.

With “inverted” under “Polarity of the wet signal”, peaks and notches swap places; then there is also a notch at 0 Hz and the sound becomes hollow and thin in the bass. The polarity should match the sign of the feedback (negative with inverted), otherwise the effects largely cancel: +0.7 with an inverted wet signal gives only about +1.3 dB and −2 dB.

A common misconception: more “Mix” means more effect. Complete cancellation, however, only happens when original and copy are equally loud, i.e. at 0.5. At 0.3 the notches are only about 8 dB deep. At 1 without feedback you only hear the copy: no notches, just a fluctuating delay – that is vibrato.

Chorus: detuning instead of a comb

Technically, a chorus is a flanger with a longer delay (about 15 to 35 ms) and little feedback. At 20 ms the notches are only 50 Hz apart – several hundred notches that the ear no longer perceives individually as timbre. Something else becomes more important: when the delay changes, the pitch of the copy changes too (Doppler effect). If D gets longer, the pitch drops; if D gets shorter, it rises. The copy sounds like a second, slightly detuned player – hence the name.

Worked example: the pitch changes by the factor 1 − dD/dt, where dD/dt is the rate of change of the delay D, i.e. how many seconds of delay are added per second. With a sine LFO the fastest change is depth · 2π · rate, so at 3 ms and 0.8 Hz it is 0.003 s · 2π · 0.8 Hz ≈ 0.015. That is 1200 · log₂(1/(1 − 0.015)) ≈ 26 cent. The flanger with 2 ± 1.5 ms and 0.25 Hz only reaches about 4 cent; a chorus with a 2 Hz rate already reaches about 66 cent, which sounds clearly out of tune.

Try it: in the chart “Delay and pitch (dashed) over time” the pitch deviation is largest where the delay rises or falls most steeply, and zero at the turning points. With “Triangle” as the LFO waveform, by contrast, it jumps between two fixed values. Several “Voices” with offset LFOs make the effect denser and the notches shallower.

Phaser: rotating phase instead of delaying

The phaser does not delay; it sends the copy through a chain of all-pass filters. An all-pass passes all frequencies at the same level but rotates the phase – up to 180° per stage, depending on frequency. At its corner frequency fc it rotates by exactly 90°. Notches appear where the whole chain rotates by 180°, 540° … With N stages there are only N/2 notches, and they are not evenly spaced. That is why a phaser sounds softer and less “tonal” than a flanger.

Worked example: each stage rotates by 2 · arctan(f/fc), so the notches lie at f = fc · tan((2k+1) · 90°/N), with k = 0, 1, 2 … as the notch number. With 4 stages and fc = 800 Hz: 800 Hz · tan 22.5° ≈ 331 Hz and 800 Hz · tan 67.5° ≈ 1.93 kHz. With 8 stages there are 4 notches at about 159 Hz, 535 Hz, 1.2 kHz and 4 kHz. With “Depth (± octaves)” = 1.5 the two notches of the 4-stage phaser move between about 117 Hz and 5.5 kHz.

Try it: switch “All-pass stages” between 2 and 12 and count the ▲ in the frequency response.

Harmonics, microphones and stage

Switch on “Show harmonics of a sawtooth”. The grey bars are the harmonics without the effect, the blue ones after the effect. Worked example with a 110 Hz fundamental and a static flanger at 4.545 ms (1/220 s): the notches lie at 110, 330, 550 Hz … exactly on the odd harmonics. What remains is 220, 440, 660 Hz – again a sawtooth, one octave higher.

Comb filters also appear unintentionally: two microphones on the same guitar with a 34 cm difference in distance give about 1 ms of time difference and thus notches at 500 Hz, 1.5 kHz … When the guitarist moves, the comb moves – a natural flanger. Remedy: align the tracks in time or mix one microphone clearly lower.

In the mix, chorus usually goes on individual instruments, on bass only above about 200–300 Hz, and you check in mono regularly. On stage, a guitar with phaser or flanger can get quieter in time with the LFO when notches move through its fundamental range.

Limits of the model

The model shows the frequency response as if the LFO stood still at each moment. This is a good approximation as long as the effect changes slowly. The all-passes are ideal first-order analogue all-passes; real devices often have unequal stages, saturation and other feedback points. The delay cannot go below 0.05 ms, so a “through-zero” flanger is not possible. No sound is played – only the frequency response and the pitch modulation are shown. The readouts measure on a grid between 20 Hz and 20 kHz; very dense notches (chorus) appear shallower there and on the map than they really are.

Tasks

Practice questions

Questions from studio, stage and mastering – to think about, discuss and check with the module.

1. Two microphones record the same guitar at different distances. In the sum it sounds thin, and when the guitarist moves it starts to “flange”. What is happening, and what helps?

Show answer

The time difference between the microphones acts like the delay of a flanger: 34 cm path difference is about 1 ms, i.e. notches at 500 Hz, 1.5 kHz … When the source moves, the delay changes – a natural flanger. Remedies: align the tracks in time (delay or shift, phase alignment plug-ins), the 3:1 rule for distances, mix one microphone clearly lower (shallower notches) or use only one microphone. Simply flipping the polarity only helps with very small time differences.
2. Why does a chorus on a stereo sum or a bass often cause problems, and how is it used instead?

Show answer

The chorus slightly detunes the copy and creates dense notches. In the bass this weakens the fundamental and makes it unsteady, and in mono (summing wide stereo chorus signals) whole ranges can audibly flutter or cancel. It is therefore common to put chorus on individual instruments (guitar, pads, electric piano), on bass only above a crossover frequency (often 200–300 Hz), keep the mix moderate and check in mono regularly. Stereo chorus with antiphase LFOs on left and right makes the sound wide, but is the least mono-compatible.
3. On stage, a guitarist using phaser or flanger should not “disappear”. What do you watch with rate, feedback and level?

Show answer

If notches move through the guitar's fundamental range (about 100–400 Hz), it gets quieter and louder in time with the LFO; with high feedback the level varies even more, because peaks of over 10 dB sweep through the spectrum. A slow rate matched to the tempo (e.g. one period per bar) sounds musical, a fast one more like vibrato or tremolo. Many guitarists compensate the level loss with the effect level; at the desk a light compressor on the channel helps catch the fluctuation. In the mix the effect often seems stronger than alone – better use it sparingly.

Tasks

Predict first, then try it out, then open the solution. “Load starting point” sets the matching initial values.

1. The comb

The flanger stands still (depth 0) with 1 ms delay and no feedback. At which frequencies are the notches, and how far apart are they? Then set 2 ms.

Load starting point

Show solution

At 1 ms the notches lie at 500 Hz, 1.5 kHz, 2.5 kHz … spaced by 1 kHz = 1/D, 20 notches up to 20 kHz in total. At 500 Hz the copy is delayed by exactly half a period and cancels the original completely. At 2 ms everything halves: notches at 250, 750, 1250 Hz …, spacing 500 Hz, 40 notches. The linear axis shows the even comb.
2. Feedback

The same flanger with 70 % feedback. How high are peaks and notches? Raise it to 0.9. Then set −0.7 and invert the wet signal.

Load starting point

Show solution

With 0.7 there are peaks of +6.7 dB at 0 Hz, 1 kHz, 2 kHz … and notches of −13.7 dB at 500 Hz, 1.5 kHz …; with 0.9 the peaks become +14.8 dB high and narrow – the sound gets a metallic tone at 1 kHz. With −0.7 and an inverted wet signal peaks and notches swap places: notches at 0 Hz, 1 kHz, 2 kHz, peaks at 500 Hz, 1.5 kHz. Because the lowest frequencies are cut too, it sounds hollow.
3. From flanger to chorus

A chorus with one voice: 20 ms ± 3 ms, rate 0.8 Hz. Read the notch spacing and the pitch deviation. Raise the rate to 2 Hz and then switch to three voices.

Load starting point

Show solution

The notches are only 50 Hz apart – several hundred notches that the ear no longer perceives individually as timbre. Instead, the pitch of the copy varies by about ±26 cent (for the flanger with 2 ± 1.5 ms and 0.25 Hz only about ±4 cent). At 2 Hz it is about ±66 cent – that already sounds like vibrato or out of tune. With three voices the copies no longer cancel completely anywhere: the notches become shallower and irregular. The “Deepest cut” readout then shows about −32 dB; it measures on a frequency grid, so the narrowest notches actually go a little deeper.
4. Phaser

A phaser with 4 all-pass stages stands still at 800 Hz. Where are the notches? Switch to 8 stages and compare the spacing with the flanger from task 1.

Load starting point

Show solution

With 4 stages there are only 2 notches, at about 331 Hz and 1.93 kHz (800 Hz · tan 22.5° and tan 67.5°): there the chain rotates the phase by 180° and 540°. With 8 stages there are 4 notches at about 159 Hz, 535 Hz, 1.2 kHz and 4 kHz – the spacing is equal neither in Hz nor in octaves. The flanger, by contrast, has 20 evenly spaced notches. With the LFO running (depth 1.5 octaves) the two notches of the 4-stage phaser move between about 117 Hz and 5.5 kHz.
5. Flanger and harmonics

A 110 Hz sawtooth passes a static flanger with 4.545 ms delay (1/220 s). Which harmonics disappear? What happens with an inverted wet signal?

Load starting point

Show solution

The notches lie at 110, 330, 550 Hz … – exactly on the odd harmonics, which are cancelled completely. What remains is 220, 440, 660 Hz …: again a sawtooth, but one octave higher. With an inverted wet signal the even harmonics disappear; the odd ones remain as in a square wave – the sound becomes hollow. With the LFO running, the notches move through the harmonics and the timbre changes constantly – the typical flanger “jet” sound.

Settings

Sound

Level calibration (dB SPL)

MIDI keyboard

Flanger, chorus and phaser