Pros and Cons of OD circuits

[Helmholtz]

I think the above is a helpful contribution to the discussion.

I'd like to add that the slope value is proportional to gain (I'm including gain values below 1 here).
A vertical slope means infinite gain and a horizontal slope means zero gain. Steep slope = large gain, shallow slope = low gain.
Decreasing slope means compression, increasing slope means expansion.

Each straight part of the slope means a constant incremental gain*, i.e. an increase of input signal will be linearlyproportionally transferred within that section.

* With incremental gain I mean dVout/dVin.

[teemuk]

Your list should also include differential and push-pull circuits, in legion of discrete variations as well as integrated types such as OTAs (differential) or linear CMOS inverters (push-pull) These typically generate rather soft and symmetric clipping (though depending on bias and applied feedback) and may or may not have dynamic bias shifting properties.

[uneumann]

Your list should also include differential and push-pull circuits, in legion of discrete variations as well as integrated types such as OTAs (differential) or linear CMOS inverters (push-pull) These typically generate rather soft and symmetric clipping (though depending on bias and applied feedback) and may or may not have dynamic bias shifting properties.

Sure - would these be implementations of the TF's already described, or are the fundamentally different somehow?
I'm familiar with CMOS inverters. I can see how they would have fall into the clip class. Did I miss something else about them?

As for OTAs, I'm not sure what you're getting at. Are you referring to the attack/decay amplitude compression circuits that often use OTAs or is there something else?
They would certainly have dynamic properties. I'm not quite sure you to classify those yet - maybe just under a broad heading of "dynamic" transfer functions - but that feels like a bit of a punt.

[Helmholtz]

Where do you see a fundamental difference (apart from an added folding step) regarding TF between your proposed circuit and the distortion circuit of a tubescreamer?

BTW, with an non-inverting (op)amp circuit the outer sections of the TF will never become horizontal, i.e. until you hit the rails.
Reason is that the (incremental) gain can't drop below 1 because of the added input signal

[teemuk]

As for OTAs, I'm not sure what you're getting at. Are you referring to the attack/decay amplitude compression circuits that often use OTAs or is there something else?

I'm mostly refering to overdriving them. The differential clips very "softly" in open loop. Bonus is that the Iabc input can be used simultaneously for dynamic effects or to alternate between symmetric and asymmetric clipping (use the input signal or full wave rectified input signal as control).

Another type of distortion implementation does not overdrive the OTA but employes Iabc input for instantaneous gain compression that squashes signal peaks softly. (Feed full wave rectified signal to Iabc). The effect can be made very variable by controlling input and Iabc input signal amplitudes. Asymmetry can be introduced by attenuating one of the rectified half waves.

OTAs are very versatile devices. Too bad that they were obsoleted.

[loudthud]

BTW, with an inverting (op)amp circuit the outer sections of the TF will never become horizontal, i.e. until you hit the rails.
Reason is that the (incremental) gain can't drop below 1 because of the added input signal

Gain equation for non-inverting opamp 1+(Rf/Ri)
Gain equation for inverting opamp (Rf/Ri)

It's the non-inverting opamp that the gain can never drop below one. In the inverting opamp, the gain drops to R(diode)/Ri which might be unity depending on the diode, but probably less.

[Mick Bailey]

Another way to apply dynamic control to overdrive circuits is to use a side-chain fed from the clean guitar signal and use this to control a fast-response optocoupler such as the NSLSR32 SR3 and use the resistor side to control clipping (even something as simple as controlling a transistor stage bypass cap). The Hollis Flatline compressor sidechain works well in this application. The result can be an overdrive circuit that responds much better to playing dynamics - expanding the range between clean and distorted.

[teemuk]

And yet another trick to add dynamics is to have clipping diodes referenced to not fixed DC voltage (including 0 VDC) but variable DC voltage, like output of an envelope follower or power supply rail that has distinct "sag" characteristics.

[Helmholtz]

Gain equation for non-inverting opamp 1+(Rf/Ri)

Sorry, typo.
Of course I meant the non-inverting opamp as used with the TS as well as the OP's circuit.
Edited my post above.

Thanks for correction!

[Helmholtz]

I forgot to mention that I like the idea of shunting the "first" diodes with resistors. Without them there would be one gain step less.

Otherwise I've seen many variations of the TS circuit: Using different numbers and mixtures of Si and Ge diodes or Mosfets with or without series resistors in the feedback path.

[uneumann]

Where do you see a fundamental difference (apart from an added folding step) regarding TF between your proposed circuit and the distortion circuit of a tubescreamer?

TS is designed for use with guitar level signals - call it 0.5-1v peak. That makes the op amp a clipping stage (type 4) since zero-crossings are amplified linearly before the diodes conduct and the waveform peaks are compressed by the diodes at the peak input levels. The net effect is a flattened waveform. See https://www.electrosmash.com/tube-screamer-analysis for examples.

I was describing my circuit in the context of a larger input. If you look at the transfer curve and the triangle wave example I posted, you'll see that my inputs are much larger than the diode conductance requires. At those higher levels that circuit acts as a unity gain stage with zero crossing distortion (type 3). The two diodes and resistors are use to smooth the nonlinear portion near the zero crossing.

This is an example of why it's important to consider the input signal amplitude. The TS and the posted circuit are similar, but their behaviors in the context of different input levels are different.
I placed my circuit into a later section of an amp where the signals are about 10x the guitar level (5-10v peak). At that level the stage operates at near unity gain for most of the waveform, but the zero-crossings are nonlinear.
The signal level context matters.

[uneumann]

Thanks for the comments about envelope dynamics and OTAs.
Are there examples of OTA OD pedals or amp circuits you can point to? Most web references I found mostly focus on reducing distortion in OTAs
I did a quick search for envelope pedals - there are more than I was expecting to find - those pedal makers are busy.
Envelope control usually implies linear gain control, but at some point a fast envelope compression and OD compression may be perceived as similar. Food for thought.

[teemuk]

Are there examples of OTA OD pedals or amp circuits you can point to?

Peavey VT & Musician series
JMF Spectra amps series (ones with distortion)
Lab Series & Pearce amps, A.R.T DST and SGX
Fender Rumble V3 series

[Helmholtz]

TS is designed for use with guitar level signals - call it 0.5-1v peak. That makes the op amp a clipping stage (type 4) since zero-crossings are amplified linearly before the diodes conduct and the waveform peaks are compressed by the diodes at the peak input levels. The net effect is a flattened waveform. See https://www.electrosmash.com/tube-screamer-analysis for examples.

The TS distortion stage doesn't clip, the output waveform never gets really flat. Just look at its triangle response. So the term clipping amplifier isn't appropriate.

I was describing my circuit in the context of a larger input. If you look at the transfer curve and the triangle wave example I posted, you'll see that my inputs are much larger than the diode conductance requires. At those higher levels that circuit acts as a unity gain stage with zero crossing distortion (type 3). The two diodes and resistors are use to smooth the nonlinear portion near the zero crossing.

Your circuit is linear around crossing having an initial gain of 12/1 +1 = 13.
Diodes kick in at about +/- 0.4V output.

No doubt signal and gain levels matter and change the shape of the output signal.
Nevertheless to me your circuit is a variation of the TS theme adapted to larger signals.

In my job I had to read and evaluate lots of patent applications. The essential question always was : What is the level/merit of invention?

[uneumann]

The TS distortion stage doesn't clip, the output waveform never gets really flat. Just look at its triangle response. So the term clipping amplifier isn't appropriate.

Your circuit is linear around crossing having a gain of 12/1 +1 = 13.
Diodes kick in at about +/- 0.4V output.

No doubt signal and gain levels matter and change the shape of the output signal.
Nevertheless to me your circuit is a variation of the TS theme adopted to larger signals.

FLAME ON:
For a guy who claims misunderstandings are due to being a non-native speaker, you seem to really get hung up on terminology.
Fine - you don't like my terminology. However, it is irritating when you incorrectly try to correct people. It also seems to cause you to miss the big picture in all of this.
A DCCF or diode clipper doesn't produce a truly flat top either - yet everyone understand those to be clipping circuits.
There are degrees of clipping. The world has shades of grey. Some may call it compression, or slight clipping or rounded flattening - it's all the same thing and people (except you) seem to understand it.
FLAME OFF

As for the TS circuit - yes - my circuit is similar to the TS circuit and linear until the diodes kick in. I never claimed it something different.
My only point is that is that in the large signal context it behaves nothing like the TS pedal does in the context of a guitar input. You just don't seem to get this.
The waveform below is produced with a 10v peak triangle wave input. The peaks are not compressed/clipped/whatever. Only the zero region appear nonlinear relative to the whole waveform.