Understanding soft clipping

[paulrm]

Okay, so I've built about 12 pedals now and I understand how they work but not really why. So when it comes to soft clipping diodes in the tubescreamer style feedback loop diodes, I understand how they work but not why. I understand that for hard clipping (such as the RAT or Suhr RIOT) the diodes simply cut off the top of the signal to ground. But on the soft clipping, the top of the signal seems to go straight through the diode and the rest goes around the circuit through the distortion pot. I've tried reading electronics books and learning the physics behind it all but I just don't get it. Please can someone explain. Thanks.

[antonis]

Actually, diodes create a voltage difference (Vfd drop) between Op Amp Output and summing point (- Input) so they "restrict" the current flowing through the feedback resistor (gain pot)..

[R.G.]

The easy answer to why they work is in the "corners" of the clipped waveform. Read about Fourier Transforms and the frequency domain content of signals. Sharp corners of clipped waves imply and require that many higher-order harmonics of the underlying signal be created. The human ear interprets a few, low-order harmonics as soft and pleasing, but hears higher order harmonics as harsh and discordant. So any process that makes most of the distortion products be second through perhaps sixth harmonics sounds (and looks) softly clipped. Any process that makes a lot of seventh through whatever harmonics and looks sharp-edged sounds harsh.

Give a read to Russel O. Hamm's paper on tube clipping.

Have you read Distortion 101 and The Technology of the Tube Screamer at geofex.com?

[teemuk]

Sharp corners of clipped waves imply and require that many higher-order harmonics of the underlying signal be created. The human ear interprets a few, low-order harmonics as soft and pleasing, but hears higher order harmonics as harsh and discordant. So any process that makes most of the distortion products be second through perhaps sixth harmonics sounds (and looks) softly clipped. Any process that makes a lot of seventh through whatever harmonics and looks sharp-edged sounds harsh.

This theory is sound but must be put to proper context....

If we assume that distortion is unwanted then we can estimate which type of distortion sounds less disturbing and distinct than another.

If we assume that signal distortion is actually wanted and used as a musical effect then it all falls down to individual preferences of how the distortion should sound like. Soft clipping with very few high order harmonics is nice if you look for subtler overdrive effects, hard clipping with plenty of higher order harmonics is good if you looking for that harsh and brutal distortion, which is perfect for certain types of metal music and alike.

Give a read to Russel O. Hamm's paper on tube clipping.

It's a decent paper. But do note that he doesn't try adress what makes distortion sound nice in various guitar applications. His theory about effects of harmonics is practically based on principles of clean reproduction of vocals.

Do also note that his test setup caused flawed results and overall conclusions: The parts about the article where Hamm claims tubes produce distortion pattern X and transistors distortion pattern Y were debunked pretty much from the start. About every rebuttal to Hamm's paper concluded that distortion characteristics are practically dictated by the circuit architecture, not the active device type.

For example, Hamm tested only certain types of tube mic preamps (all were single-ended) and only certain types of transistor mic preamps (all were push-pull), which is enough to give entirely misleading data about tube vs. transistor characteristics. He did not evaluate "musicality" of purposedly introduced distortion as musical special effect either.

Hamm's paper has idea but you need to read it with grain of salt. I would also recommend reading its rebuttals, such as
- "Transistors Can Sound Better Than Tubes", Dwight O. Monteith Jr. and Richard R. Flowers (discusses design of transistor mic preamp that produces similar "ideal" pattern of distortion as tube preamps tested by Hamm)
- "Comments on 'Tubes Versus Transistors – Is There an Audible Difference?'", Mintz, R. Steven, Journal of the Audio Engineering Society 21 (: 651. (October 1973)

These are much, much rarer and harder to find articles than Hamm's paper. IMO, also much more educative and "on point" in their conclusions.

[Kipper4]

Ha maybe this will help
No Science but sounds

[PBE6]

Exactly as antonis said, the diode in the feedback loop ensures there is a voltage drop between output and inverting input. Since the opamp keeps the non-inverting input and the inverting input the same, there is always a voltage drop between the output and the incoming signal, or:

vout = vin + vD

The value of vD will vary depending on the amount of feedback current present and how it partitions itself between the diode and the feedback resistor for a given input, but the net result of all this is that soft clipping is an equal mix of the unamplified input signal and the diode hash. That's why it never gets as saturated sounding as a hard clipper no matter how much gain you apply.

[R.G.]

This theory is sound but must be put to proper context....

As must everything to do with the human perception of music and tone. But it is helpful to beginners to give them a simple place to start for their own research.

If we assume that distortion is unwanted then [...]
If we assume that signal distortion is actually wanted and used as a musical effect then it all falls down to individual preferences of how the distortion should sound like.[...]

As does all musical *production* as opposed to *re-production*.   There can be no disagreement about matters of taste.

re: Hamm Paper

It's a decent paper. But do note that he doesn't try adress what makes distortion sound nice in various guitar applications. His theory about effects of harmonics is practically based on principles of clean reproduction of vocals.

But it does indeed address issues of harmonics and their harshness versus non-harshness. And when that's overlaid on a beginner trying to understand distortion for musical *production* it's a great starting point.

Do also note that his test setup caused flawed results and overall conclusions [...] much, much rarer and harder to find articles than Hamm's paper. IMO, also much more educative and "on point" in their conclusions.

The audio world is filled to bursting with claims and counterclaims, special cases and outright misdirection. The only good way through this morass is to read - and read and read and read, and relate what you read to what you hear, however your own personal tastes color your hearing, and to keep an open mind to what other people may hear - even in the same literal audio waves.

It's important to lead beginners, not to drown them.

[antonis]

Ha maybe this will help
No Science but sounds

Rich, I honestly can't discern audible differences between the different diode settings..
Maybe you recordered them in a "nearby audio" order - and my ears need an in depth cleaning too..

(despite of that, I really liked the 1N(914 or 4148..??) and Yellow LED pair...)

[tca]

Rich, I honestly can't discern audible differences between the different diode settings..
Maybe you recordered them in a "nearby audio" order - and my ears need an in depth cleaning too..

(despite of that, I really liked the 1N(914 or 4148..??) and Yellow LED pair...)

After 10,000 hours of listening hours you will be able to identify the various forms of distortion and how they correlate to each other.

ah, and soft clipping is over rated.

[Kipper4]

Antonis Thats not my video. I just found it on you tube.

[ashcat_lt]

The diodes act as voltage-dependent resistors.  As the voltage across them tries to get bigger, their effective resistance decreases.  This resistance is in parallel with the gain pot such that when it is very large (because the voltage across them is very small), the gain pot dominates the overall value.  Likewise, when the diode resistance gets very small (because voltage across them tries to be large), the total will be closer to that of the diode itself.  This total parallel resistance is the "top resistor" in a voltage divider in the feedback loop.  Smaller resistances (diode dominates) here means less attenuation of the feedback, and therefor less gain from the opamp.  As the diode resistance gets bigger, there is more attenuation in the feedback and the opamp gives more gain.  Bigger voltage in means less gain through, but it never quite gets down to unity, and can't go below (until the opamp hits the rails).

In hard clipping, the diodes (parallel to whatever load is attached) are the "bottom resistor" in a voltage divider.  All of the gain has already been applied.  As the voltage tries to get bigger, the diode gets smaller and you get more attenuation, which is kind of the same thing as less gain, except in this case, it is all less than unity.

Don't know if that helps, but that's how it works.

[dschwartz]

Rich, I honestly can't discern audible differences between the different diode settings..
Maybe you recordered them in a "nearby audio" order - and my ears need an in depth cleaning too..

(despite of that, I really liked the 1N(914 or 4148..??) and Yellow LED pair...)

After 10,000 hours of listening hours you will be able to identify the various forms of distortion and how they correlate to each other.

ah, and soft clipping is over rated.

I agree, I'm  surprised how marketing and articles praise soft clipping and even harmonics, but most of the best tones are well tailored hard clipping with odd harmonics of low order.