Why is distortion mostly only Clipping or Crossover ?

[Vivek]

Most distortion pedals implement clipping

Some implement crossover distortion.

Is there any advantage sonically in implementing transfer functions that distort the middle range ?

I would initially think that would make a vibrant sound with changing harmonic content dependant on volume.

( I saw some multidiode clippers that implement progressive clipping (Slope of transfer function flattens out with each diode kicking in) , but only one schematic that changes slope upwards in the middle of the transfer function)

[teemuk]

Back in the 90's I has this waveform editing software called "Cooledit 96" that allowed to process the waveforms with user adjustable transfer functions. (You could draw one yourself too). IME, there was very little difference between "S" curve of traditional clipping and "zig zag", all distortions sounded remarkably alike.
Why make stuff complicated when there's not much to benefit? Clipping and crossover are easiest schemes to introduce distortion. 

[Vivek]

Thanks, Sir Teemu K !!


I was thinking of low distortion situations, if clippers get into play only when a signal is above a particular level. it might be possible to have transfer functions that start distorting from low signal all the way to high signals.

[Mark Hammer]

But you need to think about how musicians use distortion, and what the communicative properties of distortion are.

Music and speech/language are inextricably connected.  We may think they are two different things, but we know from the study of early communication in and to infants that:

a) raspiness is used to convey "urgency" (e.g., if my cry becomes raspy, then my diaper is REALLY bothering me) and intentionality;
b) pitch and volume are generally associated with properties like giving and receipt of comfort, frailty or need, and so on (e.g., descending pitch, volume, and more space between notes "means" that everything is fine, now);
c) spacing within utterances, and what one might call "vocal cleanup", as amplitude and other indicators of urgency decline, is also a key parameter.

So the near constant search for distortions that preserve dynamics, and "touch-sensitive" distortions, is a reflection of the player's wish to have raspiness they can deploy "on demand", the way that we can use properties of our voice to convey intention and subtext.  As such, distortion of equivalent intensity or degree, no matter how hard or soft one picks, and no matter what register one is playing in, simply runs counter to the way that music employs timbre, pitch, and pitch direction.  It becomes a bit like trying to communicate in ALL CAPS WITH NO PUNCTUATION ANYWHERE ALL THE TIME NOT EVEN BETWEEN SENTENCES DO YOU GET WHAT I MEAN I HOPE YOU DO.  In other words, holding back, and spooling out, properties of sound, is fundamental to the way we use instruments to "express" ourselves.

Though crossover distortion has its place, clipping is the more "emotionally sensitive" form of distortion since it accompanies variations in pick attack.

[Vivek]

Thanks

But normally clipping transfer curves seems to be 2 zone = no distortion till signal reaches some value

And then distortion after that

So the proposal to have a zigzagy transfer curve means that some distortion exists even at low signal levels, and we can design more distortion for slightly more signal and much higher distortion for much higher signals.

Which means that the schematic of nonlinearity all over the transfer curve could lead to more expression and control than 2 zone clippers.

[Mark Hammer]

Right.  The idea is to have some element to your tone that you can introduce, hold back on, and vary, as easily as you can modulate your voice.

[iainpunk]

Back in the 90's I has this waveform editing software called "Cooledit 96" that allowed to process the waveforms with user adjustable transfer functions. (You could draw one yourself too). IME, there was very little difference between "S" curve of traditional clipping and "zig zag", all distortions sounded remarkably alike.
Why make stuff complicated when there's not much to benefit? Clipping and crossover are easiest schemes to introduce distortion.

if you have FL studio the same kind of thing is standard in it, its called FL Waveshaper, absolutely my favorite distortion plugin. it can also do asymetric, and other cool things.

Vivek, search for wave folders, i'm currently working on a wavefolder fuzz thing. the coolest thing is that you can get really high amount of distortion with low gain, and  it changes sonically dependent on volume,
check this thread
https://www.diystompboxes.com/smfforum/index.php?topic=125698.msg1198277#msg1198277
there are some sound samples of me playing an open D chord with different volume control settings.

So the proposal to have a zigzagy transfer curve means that some distortion exists even at low signal levels, and we can design more distortion for slightly more signal and much higher distortion for much higher signals.
Which means that the schematic of nonlinearity all over the transfer curve could lead to more expression and control than 2 zone clippers.

in my experience, that sounds like shit, and feels really weird to play.
the volume drops choppy instead of smoothly, which feels capricious and independent of your playing, like a tremolo that doesn't have a stable oscillator.

later i'll share a sound sample and screenshot of my FL studio plugin doing that

cheers, Iain

[iainpunk]

https://soundcloud.com/iain-schmaloer/stair-transfer-shape


it sound like a fizz that's unrelated to the clean signal is mixed in... not so nice.

IRL it sounds a bit different, but not too much.

cheers, Iain

[ThermionicScott]

To piggyback on Mark's excellent synopsis, I feel like you are looking for the "vocal fry" of guitar distortion.  Ever notice how some people always speak in a grumbly, raspy way at low or high volumes because they never give their vocal cords enough breath support?  It drives me up the wall whenever I have to hold a conversation with such an offender.  Perhaps it's because some part of my brain is signalling an alarm for urgency, when the actual content of the conversation is completely benign.

[PRR]

.....I has this waveform editing software called "Cooledit 96" that allowed to process the waveforms with user adjustable transfer functions. (You could draw one yourself too). IME, there was very little difference between "S" curve of traditional clipping and "zig zag", all distortions sounded remarkably alike....

Agreed.

[Vivek]

IME, there was very little difference between "S" curve of traditional clipping and "zig zag", all distortions sounded remarkably alike.

Why do we make big fuss about Ge versus Si versus LED diodes as clippers. I really feel that there is no real tonal difference except that they chop the input at different places.

I'm quite convinced that a 3.5v wave chopped at 0.35 is same as 6.5V wave chopped at 0.65 is same as 25V wave chopped at 2.5V, and the differences in slope of the different  diodes is inaudible.

[teemuk]

Why do we make big fuss about Ge versus Si versus LED diodes as clippers.

Maybe because it's an extremely easy parameter to tweak and (as is) provides quite an audible result?
Why are simple circuits more popular than very complex ones?

One needs only rudimentary knowledge of circuit theory in order to "modify" by diode swapping whereas other more advanced circuit tweaks actually require deeper knowledge of electronics in general. e.g. It's more common knowledge how to change a diode to another one than how to change gain of a gain stage.

[iainpunk]

IME, there was very little difference between "S" curve of traditional clipping and "zig zag", all distortions sounded remarkably alike.

Why do we make big fuss about Ge versus Si versus LED diodes as clippers. I really feel that there is no real tonal difference except that they chop the input at different places.

I'm quite convinced that a 3.5v wave chopped at 0.35 is same as 6.5V wave chopped at 0.65 is same as 25V wave chopped at 2.5V, and the differences in slope of the different  diodes is inaudible.

in a non inverting gain stage, if they are in the feedback loop, they decide the ratio clean to dirt.
in other situations, it doesn't matter much, especially when you take in account the large impact a small tonal tweak in pre-distortion the filter has.

i always use 4148's unless i need several clipping thresholds, then ill use some Ge or red LED's.
i use them not because they sound special, but because they have a nice ratio to one-another.

cheers, Iain

[daverdave]

Both bit crushers and sample hold amplifiers introduce distortion within the middle range. It might not be exactly what you're looking for but it should give an idea of how it sounds.

[ElectricDruid]

+1 what Iainpunk said about doing a few searches around waveshapers and wavefolding. There's a load of stuff in the synth world including the theory of what shapes give you what harmonics. In the synth world the volume is usually tightly controlled though, so it's a bit different, but the theory still stands.

The biggest difference is between mirror-symmetric transfer functions (even harmonics only) and rotationally-symmetric transfer functions (odd harmonics only). Other transfer functions that are neither produce some mixture of harmonics.

[Rob Strand]

There's a load of stuff in the synth world including the theory of what shapes give you what harmonics. In the synth world the volume is usually tightly controlled though, so it's a bit different, but the theory still stands.

It's easy to have some intuition about how wiggles emphasize harmonics.   Normally the wiggles are drawn tilted like Reply #1.  If you subtract off the linear part you end up with no tilt.  The linear part just amplifies the original input signal.  The remaining part is just horizontal wiggles.   If you multiply horizontal wiggles by the input signal it's an amplitude modulator.    Loosely speaking if there's two wiggles per cycle it doubles the frequency.   It should be fairly obvious wiggles can emphasize certain harmonics.      For varying levels it's probably more of a span of frequencies.

[FiveseveN]

Just wanted to remind everyone that this exists:

IMHO the less "conventional" wave tables are also less musically useful. I'd wager wave folding is more popular in the synth world because you have more control over the input signal.

[Mark Hammer]

Wavefolding IS more common in the synth world, basically because oscillators can provide a steady-state signal.  Guitar signals, on the other hand, don't sit still, either in terms of amplitude, pitch, or harmonic content, so it is never clear what is the product of folding.  Of course, if one manipulated a guitar signal to emulate "sitting still" (i.e., compression and near constant waveform), then wavefolding becomes a more achievable goal.

[iainpunk]

i like using wavefolding for guitar and bass, it has some cool harmonic movement when the note decays, but i only like a single stage, multiple stages tend to get more in to envelope filer territory than fuzz territory.

cheers, Iain