Low or high: why the same distortion sounds different across the neck

Play a fifth at the first fret and the same fifth at the twelfth through the same pedal at the same drive, and the pedal does the same thing to both. Every product it makes lands at m·f1 + n·f2, so moving the interval up an octave moves every product up an octave with it: the same grid, the same roles, the same order structure, the same levels. Nothing in a clipper knows what register you are in. If the two chords sound different — and they do — the difference is downstream of the pedal, in two places: your ear’s filters, and your speaker’s passband. They pull in opposite directions, and neither is on a PedalScope record.

The ear’s filters are the same width all the way down the neck

Hearing sorts sound into filters whose width — the equivalent rectangular bandwidth, ERB — follows 24.7 × (4.37·f/1000 + 1) Hz. That is about 30 Hz wide at the bottom of the bass, about 100 Hz wide by 700 Hz, and only then starts growing in proportion to frequency. Musical intervals are proportional at every register. So the same interval spans fewer filters the lower you play it:

(Centre taken as the geometric mean of the two notes.)

Below roughly one filter’s width, two notes stop being heard as two pitches: they land in one filter and interact there as a fluctuation in loudness — beating, roughness — rather than as a chord. An open bass fifth is inside one filter before any distortion happens. Add a pedal and its products land in that same filter, so they arrive as more roughness and less definition, not as added notes. Play the same fifth at the twelfth fret and the products separate into distinct pitches you can point at.

That is the real change with register. Not how much intermodulation you get — that is the same — but whether you hear it as texture or as notes.

The products move up; the ear’s sensitivity is not flat

Distortion products carry energy upward. A low note sits where hearing is some 30–40 dB less sensitive than at its peak; the note’s fifth- and seventh-order products land at 1–3 kHz, where it is most sensitive. A product 60 dB below the note is a very different thing to hear at 2 kHz than the same 60 dB would be at 80 Hz. The record quotes dBc against the note; the ear does not.

The speaker throws away the bottom — including the “growl”

The other edge cuts the opposite way. A guitar speaker in its cabinet is typically down 3 dB around 80 Hz and falling fast below it, and the loudest product a clipped fifth makes, the sub-octave 2f1 − f2, sits below the lower note:

This corrects something easy to believe, and something the Growl tile’s name invites: that the sub-octave is the growl that makes a distorted fifth sound huge. In the DI signal and on the record it is exactly that — on a symmetric clipper it is the loudest product on the page. Through a real cabinet, for a chord rooted below about A2, most of it never reaches the air. What it does instead is consume power-amplifier headroom and cone excursion. The sub-octave genuinely reproduces for roots around A2–A3 and above; below that, the tile is telling you what the pedal made, not what the room will hear.

The top edge has the same effect in reverse. At A4 + E5 the seventh-order products sit at 3.3–4.4 kHz, at the cabinet’s roll-off, so high-register distortion progressively filters itself and sounds thinner and cleaner than the record says it is.

So: low or high?

Neither, uniformly. The character of the damage changes, and each register has its own worst case.

Low — the bass register. Everything crowds into one or two filters. What fails is definition: mud, a chord whose pitches you cannot place. This is where intermodulation is most destructive, because what it destroys is intelligibility rather than adding a wrong colour.

Middle — the power-chord register, roots around E2 to A3. The fifth- and seventh-order products land at 1–4 kHz, at the peak of the ear’s sensitivity and inside the cabinet’s passband. What fails is fizz and harshness. This is also where distorted guitar gets most of its character, so it is the register where intermodulation is most audible without being most harmful.

High. Products resolve cleanly, and the grid points that sit off any real note — the seventh, eleventh and thirteenth, 31, 49 and 41 cents from the nearest pitch — stand out as identifiable wrong notes rather than blending into roughness. What fails is sour intervals, and this is the register where a listener can name what is wrong.

Bass against guitar

Two things make bass the harder case; one makes it easier.

Harder. The crowding above is the whole story for chords, and a bass third is worse than a bass fifth. At a just 5 : 4 the product grid’s fundamental is f1 / 4, so three product families land below the low note — for E1 + G♯1 that is 10.3, 20.6 and 30.9 Hz, all infrasonic, all invisible, all eating excursion. Ranking the common intervals by the denominator of their reduced ratio ranks their tolerance for distortion: the octave (2 : 1, nothing below the root), then the fifth (3 : 2, one family below), the fourth (4 : 3, two), the major third (5 : 4, three). That ranking holds in every register; what gets worse as you go down is that the below-root families leave the speaker’s band altogether.

Easier. Bass is mostly played one note at a time, and a single note is a special case where intermodulation is nearly free. The “two tones” are the string’s own partials, already in whole-number ratios, so every product m·f1 + n·f2 lands back on the harmonic series the note already had. Distorting a perfectly harmonic tone redistributes energy within its own series; it cannot invent a new pitch. That is why a distorted single bass note sounds fine and a distorted bass third does not — and why PedalScope measures the two cases with two different instruments: Harmonic Distortion for one note, Chord IMD for two.

The escape is only approximate, because real strings are not perfectly harmonic. A string’s n-th partial sits at f1 · √(1 + B·n²); with B ≈ 2 × 10⁻⁴, typical of a wound low string, the tenth partial is about 17 cents sharp. Products built from high partials therefore miss the series by tens of cents and beat against it — the same mechanism as the probe’s own lattice, where a product near a note but not on it produces slow beating rather than reinforcement. Wound low strings are the worst offenders.

It is also why bass overdrive is built differently from guitar overdrive. The standard design — split the signal, keep the lows clean, distort only the upper band, recombine — is a direct answer to the filter-width problem: distorting only above roughly 200 Hz puts every product in a register where it resolves. Guitar amplifiers do not need the circuit; the guitar equivalent is a playing convention. Fifths and octaves low on the neck, thirds higher up.

What the record can and cannot tell you

Everything above the pedal’s output — the filters, the cabinet, the room — is absent from a PedalScope record on purpose: the record is a measurement of the pedal, and it would be a different instrument if it tried to include your speaker. What the record gives you is the part that does not change with register: which products the circuit makes, on which grid points, at what level against the notes. This page is the part that does change, so you can read one against the other. See also What a measurement can’t tell you in the guide.

With thanks to an independent reviewer whose analysis of the register question this page is built on, including the correction about the sub-octave and the cabinet. Every figure was recomputed for this page from the formulas given.