Compressor before Distortion vs Piecewise function generator before Distortion

[Vivek]

In some patents, Tom Scholz extols the virtues of compression before distortion

In fact, almost every SRD product that has distortion stage has also a compressor before it

The compressor evens out the initial high energy attack and effectively boosts the sustain at the end of a note, feeding the distortion stage with a more even signal from start to finish than if it was fed directly with a guitar. This makes the sound more uniform in strength and in harmonic content as compared to guitar directly feeding the distortion stage.


I got thinking that the compression does two things. One in the amplitude domain and one in the time domain.

A piecewise function generator can do something similar to the Compressor in the amplitude domain. It can act as a compressor with zero attack time and zero release time.

But the biggest difference is that the Piecewise function generator will create high levels of distortion.

However if the next stage is going to create huge amounts of distortion anyway, would the added distortion of the Piecewise function generator make a difference ?

There could be some benefits of replacing the compressor with a Piecewise function generator in a distortion generator : Lower cost, no hard to find FET, no set up procedures, easier for the DIYer

So I decided to do a SPICE FFT and Transient analysis of

Compressor feeding a hard clipping Distortion stage

versus

Piecewise Function Generator set to same amplitude graph as above compressor, feeding same hard clipping Distortion stage


I found that the final wave shape, wave amplitude, harmonic content of the above two schematics were fairly similar.



It is interesting to note that very few commercial distortion pedals have a compressor built into them, and almost none have a piecewise function generator before the hard clippers.


Sample function generated by a Piecewise function generator, taken from here:


Compressor output at two different settings:


Compresor curves from Patent 4627094


Keywords : X100, Ultimatum, Sustainor, Distortion Generator, SRD, Rockman, Rockmodule

[amz-fx]

Piecewise clipping:



regards, Jack

[Digital Larry]

Interesting topic and analysis.  The proof is in the playing and goals.  It's very hard to objectively say which is best.  Boston is not John Lee Hooker.  If the absolute consistency of notes is the goal, that's one thing.  If the goal is to have something "live" in your hands, that you can coax different levels of intensity out of with your playing dynamics, that's something else.

As you yell, your voice gains additional harmonics.  We've come to associate distorted guitar sounds with an increase in emotional content from the musician.  It's possible to create a circuit that distorts LESS the harder you play it, and it's a strange experience to play through such a thing.

One thing I have thought would be cool would be a 3-D visual analysis of distortion circuits which represented harmonic content over a wide range of input levels so you could see how different types of overdrives, fuzzes, etc. respond dynamically.  Doesn't seem like it would be so hard to do, but alas I am not a graphics person. 

DL

[Mark Hammer]

Two things, and maybe even three, take place with guitar strings and guitar signal.

First, the amplitude changes, from a sharp peak to a gradual decay and fadeout.
Second, the harmonic content changes just as quickly, with lots at the beginning, and much less as the string settles down.
The third thing is that, because of string imperfections, the amplitude can exhibit some "wiggle" over time.  Not necessarily enough for us to hear, explicitly, but enough to result in ripple, when we try to extract a DC envelope voltage.

Compression before distortion addresses the first and third of those, if somewhat imperfectly.  Depending on the nature of the compression circuit, it can sometimes address the second.

Guitars are mechanical devices, whose string-motion properties are sensed electro-magnetically

Because distortion circuits incorporate assorted headroom thresholds, in addition to filtering at different points, and generate additional harmonic content of what they are fed, the resulting additional harmonic content will depend on both the spectral content and amplitude of the input signal. 

I have long maintained that the principle difference between what we call "fuzz", "distortion", and "overdrive", resides in the duration of that additional harmonic content.  Indeed, the term "fuzz" is essentially an onomatopoeia for the sound produced, the "zz" reflecting the extended presence of harmonic content, well after the initial attack.

Compression before an overdrive, coupled with modest gain, will indeed result in a "warmer" and more consistent drive sound.  One of the virtues is that players can not always pay scrupulous attention to the strength of their picking or strumming, such that a performer can maintain the desired tone across lots of variation in their picking/strumming.

As players, builders, and designers, we would do well to always remember that a guitar signal is not a function generator, and always consider what the net effect of a circuit between guitar and amplifier will do, given the variability of the mechanical device we call an electric guitar.

[iainpunk]

> piecewise transfer curve

so basically its soft clipping before hard clipping.
i think that a good alternative is the CMOS gainstage's transfer graph


i have that Xenos Overdrive circuit that employs this exact transfer curve, and yes, its great to put in fron of any distortion or overdrive, but the thing is, youre adding harmonics to a signal that then gets filtered and sees gain and clipping, so youll distort the already un-natural harmonics present depending on the aggressiveness of the bass cut before the clipping stage. were this filtering not present, i'd say your idea makes sense, but in practice, its a very different beast. it has more harmonics on the attack, and less in the sustain, making the attacks more broken sounding and dulling the sustain... kind of the opposite of what most people want from their ''compressor > Distortion'' sound.

cheers

[Ben N]

There are those of us who think overcompressed lead tones are the bane of our age, but to each their own, I guess.

[RANT ALERT (Enter at your own risk):
I was recently at a wedding (a context in which those overcompressed tones through digital amps/sims have been near universal IME), and, lo & behold, the guitar player was wielding a vintage-style sunburst Les Paul, and was playing pretty much clean! No amp in sight, as is standard these days. I snuck up to the stage and saw that his signal chain, gain wise was a Boss SD-1 with the gain set low & level high for lead boost, going through a Strymon Iridium on the Fender setting ("round"). Along with that, the dude was taking short lead breaks and not overplaying, but he sounded great--and like a guitar, of all things.
[Rant concluded.]

Sorry about the rant, and carry on with the enlightening technical discussion. And, BTW, what in the FFF is a "piecewise function generator" besides some math geek gizmo, i.e. in a music context?

[puretube]

I never liked the monotoneous, undynamic, compressed "Rockman"-sound ...

[Mark Hammer]

Not to take anything away from Tom Scholz's musicianship (probably better than my own), but he is first and foremost an electronic engineer, and likely to view signal-processing through that lens.  If Jeff Beck ever decided to stop touring and design FX, I suspect his approach would be very different than Scholz's.

Some of us value variability and serendipity more than consistency and control.

[iainpunk]

There are those of us who think overcompressed lead tones are the bane of our age, but to each their own, I guess.

[RANT ALERT (Enter at your own risk):
I was recently at a wedding (a context in which those overcompressed tones through digital amps/sims have been near universal IME), and, lo & behold, the guitar player was wielding a vintage-style sunburst Les Paul, and was playing pretty much clean! No amp in sight, as is standard these days. I snuck up to the stage and saw that his signal chain, gain wise was a Boss SD-1 with the gain set low & level high for lead boost, going through a Strymon Iridium on the Fender setting ("round"). Along with that, the dude was taking short lead breaks and not overplaying, but he sounded great--and like a guitar, of all things.
[Rant concluded.]

Sorry about the rant, and carry on with the enlightening technical discussion. And, BTW, what in the FFF is a "piecewise function generator" besides some math geek gizmo, i.e. in a music context?

piecewise function generator or a device with a piecewise transfer function, basically is a fancy multistage soft clipper.
the gain is linear in the middle, but at a certain threshold, the gain gets reduced to a lower amount, like soft clipping, but less immediate, after another of those thresholds, the gain is reduced yet again.

if you were to take the smooth transfer curve of the CMOS inverters i posted before, but instead of it being smooth, its divided up in a few linear portions.
hes basically suggesting a Tube Sound Fuzz/hot tubes/red llama/darkglass style pedal at this point.

cheers

[Ben N]

> piecewise transfer curve

so basically its soft clipping before hard clipping.
i think that a good alternative is the CMOS gainstage's transfer graph


i have that Xenos Overdrive circuit that employs this exact transfer curve, and yes, its great to put in fron of any distortion or overdrive, but the thing is, youre adding harmonics to a signal that then gets filtered and sees gain and clipping, so youll distort the already un-natural harmonics present depending on the aggressiveness of the bass cut before the clipping stage. were this filtering not present, i'd say your idea makes sense, but in practice, its a very different beast. it has more harmonics on the attack, and less in the sustain, making the attacks more broken sounding and dulling the sustain... kind of the opposite of what most people want from their ''compressor > Distortion'' sound.

cheers

Ah, Iain, I think I missed this before because you and I were posting at about the same time.  Assuming this approach is generally beneficial, maybe the answer then is not to stack pedals, but to design a combined circuit from the ground up. I don't know how practical that would be, but it might make an interesting challenge to create a super-smooth-sustaining distortion on these principles.

[ElectricDruid]

A piecewise function generator can do something similar to the Compressor in the amplitude domain. It can act as a compressor with zero attack time and zero release time.

I don't agree with this statement, and the second sentence is a good part of the reason why. A compressor makes *slow* changes to the amplitude, which ideally eliminates or at least reduces the signal distortion. Something that makes *instantaneous* changes introduces lots of signal distortion and is no longer a compressor but rather a transfer function.

The important difference is the timescales. A compressor operates over many cycles of a waveform. A transfer function is a direct relationship in-against-out for a single cycle of a waveform.

[StephenGiles]

Not to take anything away from Tom Scholz's musicianship (probably better than my own), but he is first and foremost an electronic engineer, and likely to view signal-processing through that lens.  If Jeff Beck ever decided to stop touring and design FX, I suspect his approach would be very different than Scholz's.

Some of us value variability and serendipity more than consistency and control.

Jeff Beck seems to achieve effects not otherwise available from electronics, by simply thumping different parts of his Strat with his hand!! Of course he knows just where to thump and how hard - without graphs!!

[Vivek]

A piecewise function generator can do something similar to the Compressor in the amplitude domain. It can act as a compressor with zero attack time and zero release time.

I don't agree with this statement, and the second sentence is a good part of the reason why. A compressor makes *slow* changes to the amplitude, which ideally eliminates or at least reduces the signal distortion. Something that makes *instantaneous* changes introduces lots of signal distortion and is no longer a compressor but rather a transfer function.

The important difference is the timescales. A compressor operates over many cycles of a waveform. A transfer function is a direct relationship in-against-out for a single cycle of a waveform.

Brother Tom, the differences are exactly what I was trying to understand

Does a "compressor before distortion" rely on the time based effect of a compressor or the amplitude based effect ?

Suppose there is a compressor that can dynamically shrink a 1vp guitar signal to 200mvp and it feeds a severe distortion stage that chops all inputs above 20mvp into square waves, will any evidence of the time based action of the earlier compressor be visible in the output of the final distortion stage ?


How huge is the difference between feeding 200mvp Sine waves and 200mvp peak squares wave into a distortion stage that chops all inputs above 20mv into square waves ?


#thingsthatkeepmeawakeatnight

[teemuk]

A compressor makes *slow* changes to the amplitude, which ideally eliminates or at least reduces the signal distortion. Something that makes *instantaneous* changes introduces lots of signal distortion and is no longer a compressor but rather a transfer function.

This sums it up very well. Instantaneous compression is synonymous to distortion so there's instant introduction of harmonic distortion. Gradual compression slowly reduces gain so it results to distortion only at very first transients and later on the signal is overall just attenuated. This is not only different structure of harmonic distortion but different dynamic reactions to signal's envelope.

[teemuk]

Not to take anything away from Tom Scholz's musicianship (probably better than my own), but he is first and foremost an electronic engineer...

Very famous professional  musician, studio technician, sound and electronics engineer. I trust that he had a lot of musicianship to design several methods to get the cranked amp sound at low volume levels. He is one of the pioneers of, say, using attenuators, cabinet emulations, amp emulations overall and easy MIDI control.

The Rockman tone was pretty influential in the 1980's, though I admit it is indeed very "processed sounding" by today's standards. Back then, however, that was the tone many bands sought after.

So what did Jeff Beck innovate?

[Vivek]

A compressor makes *slow* changes to the amplitude, which ideally eliminates or at least reduces the signal distortion. Something that makes *instantaneous* changes introduces lots of signal distortion and is no longer a compressor but rather a transfer function.

This sums it up very well. Instantaneous compression is synonymous to distortion so there's instant introduction of harmonic distortion. Gradual compression slowly reduces gain so it results to distortion only at very first transients and later on the signal is overall just attenuated. This is not only different structure of harmonic distortion but different dynamic reactions to signal's envelope.

Yes

But the question I had was

Does any characteristic of a compressor show through once it is passed through a distortion stage that makes square waves of anything that's more than 20mvp ?

[Mark Hammer]

Not to take anything away from Tom Scholz's musicianship (probably better than my own), but he is first and foremost an electronic engineer...

Very famous professional  musician, studio technician, sound and electronics engineer. I trust that he had a lot of musicianship to design several methods to get the cranked amp sound at low volume levels. He is one of the pioneers of, say, using attenuators, cabinet emulations, amp emulations overall and easy MIDI control.

The Rockman tone was pretty influential in the 1980's, though I admit it is indeed very "processed sounding" by today's standards. Back then, however, that was the tone many bands sought after.

So what did Jeff Beck innovate?

My earlier comment wasn't meant to disparage Tom Scholz or electronic engineers.  Rather, anyone tackling any sort of problem/challenge views it through the lens of their training and perspective.  That's not good or bad, merely human.  I have a propensity to view signal processing through the lens of my own training, which is psychological.  That ignores the electrical properties somewhat.  Similarly, viewing a musical signal on a scope can result in one zooming in too closely on the electrical properties, to the neglect of the broader context.

One of the things that compressors do for guitarists - at least when used responsibly - is to unharness timbre from amplitude.  Picking style and attack elicit different spectral properties from strings (as mechanical devices), but are normally accompanied by changes in amplitude.  Reducing variation in amplitude, via compression (or simply peak limiting), allows the player to vary what they want to extract from the string, without having to pay attention to amplitude and dynamics that may be unwanted at that moment.  And, as we nearly all know, what you feed a clipping circuit alters what you get out of it.

As for Jeff Beck, I think he innovated music....and maybe wristbands.

[teemuk]

Does any characteristic of a compressor show through once it is passed through a distortion stage that makes square waves of anything that's more than 20mvp ?
Yes. The envelopes of instantly compressed (iow "clipped") and gradually compressed signals are different and this reflects in how those signals distort by clipping.

[puretube]

One huge benefit of the Tom Scholz compression-units is their ability to offer any of the BBD-circuits that often follow them, a fairly high and constant input-level which results in a good S/N-performance.

[Mark Hammer]

Yes

But the question I had was

Does any characteristic of a compressor show through once it is passed through a distortion stage that makes square waves of anything that's more than 20mvp ?

If the gain, clipping and filtering of the distortion stage is severe enough, probably not.  This is also why sticking a compressor AFTER similarly severe gain and clipping makes little sense.  If the clipping has virtually eliminated any dynamics from the signal, there isn't very much for the rectifier/envelope-follower in the compressor to respond to.

However....there is a universe of both compression settings and distortion-circuit settings that will permit both compressor attributes to be audible (and likely observable and measurable) when the compressor feeds the distortion, as well as when the distortion feeds the compressor.

I realize your example of "square waves of anything that's more than 20mvp" is an exaggeration, used to make a point.  Realistically, there's not much in our universe, if used in typical fashion, that would do such a thing.