In 1956, Curt Knoppel wired one channel of a Heathkit stereo amplifier kit incorrectly. When he ran audio through both the correct and incorrect channels simultaneously, the sound became brighter, more detailed, more present. He did not know why. Nearly two decades later, he brought the discovery to what would become Aphex Systems, and in March 1975 the Aural Exciter 402 became available to recording studios on a rental basis: $30 per minute of finished recording time.
Studios paid it. The effect was that useful.
Understanding why requires looking at what an exciter actually does, because it is not what most descriptions suggest.
It Does Not Boost High Frequencies
The common explanation is that exciters "add high-end" or "bring out air." That sounds like EQ, but the processes are different. An equalizer boosts frequencies that already exist in the signal. An exciter synthesizes frequencies that were not there.
The difference matters practically. If a recording lacks definition because the high-frequency content was never captured cleanly, boosting 12kHz with an EQ amplifies whatever exists at 12kHz, including the noise floor. An exciter instead takes the existing signal and uses nonlinear distortion to generate new harmonics above and around the fundamental content. You are not turning up the volume of something; you are creating something new.
The psychoacoustic consequence: the auditory system interprets harmonically rich signals as more present and resolved. When a sound has a full harmonic series extending into the upper registers, the brain identifies it as more clearly defined. Exciters exploit this directly.
The Four-Stage Architecture
A harmonic exciter has four functional stages:
High-pass filter. The input signal is split. A high-pass filter, typically a second-order Butterworth with a cutoff between 1 and 6kHz depending on the unit, isolates the upper frequencies. The low end bypasses the exciter path entirely and recombines at the output. This prevents low-frequency content from being distorted in a way that generates intermodulation products across the spectrum.
Nonlinear saturation. The filtered high-frequency content passes through a saturating nonlinearity. The shape of the curve determines what harmonics are generated. A symmetric saturator, such as a hyperbolic tangent, generates only odd harmonics: 3rd, 5th, 7th, and so on. An asymmetric saturator, where the positive and negative halves of the waveform are processed differently, generates both odd and even harmonics.
Even harmonics have a specific perceptual quality: they sit at octave-related intervals above the fundamental (the 2nd harmonic is one octave up, the 4th is two octaves up) and the ear interprets them as consonant extensions of the source. Odd harmonics, particularly the upper ones, create more edge and grit. The original Aphex circuit used an asymmetric nonlinearity to keep the result on the smoother, more musical side.
Level detection. A rectifier feeding into a lowpass filter with a cutoff around 10Hz tracks the envelope of the incoming signal. This controls the gain driving the saturation stage, making the harmonic generation dynamic rather than static. Loud transients drive the nonlinearity harder; quiet passages produce fewer new harmonics. The effect follows the program material rather than sitting as a fixed EQ curve.
This is where the Aphex's "Transient Discriminate Harmonics Generator" name originates: it responds differently to transients than to sustained content.
Mix. The processed signal is added back to the original dry path. Because the exciter synthesizes harmonics that extend above and around the existing content, it creates perceived brightness without adding gain that registers on a loudness meter. That was one of the selling points of the original unit: engineers could treat a recording without nudging the mix level.
Why Not Just Saturate Everything
The high-pass filter before the saturation stage is essential. Without it, distorting the full-bandwidth signal generates intermodulation products throughout the spectrum: nonlinear distortion of two simultaneous frequencies f1 and f2 produces components at f1+f2, f1-f2, 2f1-f2, and related combinations. At low frequencies, these intermodulation products quickly become mud.
By restricting the saturation to only the high-frequency portion of the signal, the exciter keeps intermodulation products above the frequency range where they cause problems, while the generated harmonics extend further into the upper registers. The result is additional high-frequency content that relates harmonically to what was already there, rather than new artifacts scattered across the spectrum.
Tube vs. Solid-State vs. Digital
The original Aphex 402 was a tube unit. Vacuum tubes have a transfer curve that produces asymmetric saturation by their physical nature, which is why tube amplifiers generate primarily even harmonics and are associated with perceived warmth. Only six 402 units were ever built. Three are currently known to exist. The Aphex 602 that followed was solid-state: replicating even-harmonic emphasis required designing the circuit's transfer curve deliberately asymmetric, rather than relying on thermionic physics.
Modern digital implementations have the most control over this. The waveshaping curve can be defined precisely, the filter cutoff adjusted per instance, the envelope detector tuned to specific attack and release time constants. The Waves Aphex Vintage Aural Exciter models the original circuit behavior. Tools like iZotope Neutron's Exciter module work from generalized psychoacoustic principles without attempting to replicate a specific hardware transfer function.
A Contrast: The BBE Sonic Maximizer
The BBE Sonic Maximizer is often grouped with exciters but works by a different mechanism. It does not generate new harmonics. Instead, it applies frequency-dependent delays: frequencies below 150Hz are delayed by approximately 2.5ms, frequencies between 150Hz and 1200Hz by approximately 0.5ms, and the high-frequency content passes through without delay. Dynamic processing is applied to the high-frequency band to enhance transients.
BBE's premise is that most loudspeakers introduce progressive phase lag at lower frequencies, causing perceived smearing, and the BBE pre-corrects for this. The output sounds clearer not because new content was added, but because the phase relationships between frequency bands were adjusted.
Whether the BBE mechanism works as described is debated; phase relationships affect stereo imaging more than mono clarity, and many of the perceived improvements could be attributed to the dynamic enhancement on the high-frequency band. What is not debated is that the BBE and the Aphex produce different results on the same material, because they are doing different things.
When an Exciter Makes Sense
An exciter is the right tool when:
A recording lacks definition but is not noisy (boosting EQ would also raise the noise floor proportionally). Vocals or acoustic instruments need air that the microphone did not capture cleanly. An instrument is getting masked in a dense mix and EQ moves have not resolved it. You need perceived brightness without affecting the mix's integrated loudness.
EQ makes more sense when the frequency content exists and just needs level adjustment. Full-spectrum saturation makes sense when you want harmonic enrichment across the entire spectrum, including warmth in the low-mid range, which exciters deliberately avoid by high-passing before the saturation stage.
Knoppel's Heathkit
That miswired amplifier was doing all four things: the amplifier's inherent frequency response was acting as a filter, the miswired circuit was introducing asymmetric distortion, the amplifier's operating point was providing dynamic behavior, and combining the two channels performed the wet/dry mix.
Knoppel had inadvertently built an exciter from a construction kit. It took until 1975 to productize, another few years to make it cheap enough that studios could buy rather than rent, and a 1979 patent grant to formally describe what was happening. The four-stage architecture has not changed.