Two similar-but-different nonlinear amplifiers

[earthtonesaudio]

This image shows two similar circuits which work very differently.  The only thing that is changed is the placement of the pair of diodes.

Circuit A on the left should resemble a stripped-down version of the clipping stage in a Tube Screamer type of distortion effect.  The placement of the diodes in parallel with Ra causes the gain to be reduced for large signals.

Circuit B moves the diodes so they are in series with Rb.  This has the opposite effect, meaning gain is increased for large signals.


Most of the time when people experiment with clipping diodes it just involves the type or number of diodes, but there is plenty of sonic territory to be explored from changing their position, or whether they're in series or parallel with other components, etc.

I think I'll experiment with combining circuit A and B into one stage.  If the two pairs of diodes are exactly matched, in theory they would cancel each other exactly and the circuit would perform as if there were no diodes present at all.  However I don't have a matched quad of diodes, so I'm interested to see how it works with a random assortment.

[petemoore]

  Looks very intrigueing, thanks for posting, I'm interested to hear about results..

[Brymus]

Somebody did this a few months ago, IIRC ,Scruffy ?
He used a pot to change between the two types of clipping agaian IIRC but it was very nice circuit and did exactly what you are talking about.
I wish I could remember what it was called,he is in the UK I am pretty sure if that helps narrow it down.
I remember I really liked the sound of his amp and guitar ,if I could just access that archive in my brain... 

[Rob Strand]

You can can build pairs of these where one circuit completely cancels out the distortion produced by the other.

Sig -> Non-linear "Compress" -> Distorted Signal -> Non-linear "Expand" -> Clean Signal

or

Sig -> Non-linear "Expand" -> Distorted Signal -> Non-linear "Compress" -> Clean Signal

In the first case the Distorted signal is squashed, whereas the in the second it is stretched.

Unfortunately if you place any frequency dependent circuits in the middle it no longer cancels out.
Also in practice it's hard to match diodes.  Best done in spice.

[earthtonesaudio]

When I use the two in series (compress first then expand), I do not get complete cancellation.  With all the same type of diode, it is sort of like a very weak TS sort of distortion.  Very prominent dry signal with fuzz in the background, much attenuated.  With a lower-threshold diode in the compression section, there is very little to no expansion.  With the lower-threshold diode in the expander, it is difficult to control.  The circuit starts to behave more like a noise gate.

Interestingly, when I "combine" both arrangements of diodes into a single op-amp stage, the effect appears to disappear altogether.  Perhaps my diodes are well matched by chance (which is possible since they're all from the same batch/reel) or maybe there's something else going on. 

[Rob Strand]

The distortion cancellation thing is easier to get right with inverting opamp configurations.
For perfect matching you also have to scale the resistor properly in one of the circuits.

[Quackzed]

you could make rb a pot, and soften the 'expander' clipping... smoothing out the background fuzz... have controll between very little crossover dist and full crossover(based on the diode thresholds ).

[Taylor]

Mouser still stocks a quad diode array, UC3611. Now, I'm not sure what the point of this is if we're aiming for cancellation of the 2 effects (purely academic, I guess). But the expansion technique definitely seems to be underutilized in distortions.

[earthtonesaudio]

Yeah I'm not really interested in perfectly matched processes.  There are compander ICs for that anyway.  Just some cool little bits of circuit to think about.

[slacker]

Somebody did this a few months ago, IIRC ,Scruffy ?

I think that was my hard/soft clipping idea you're thinking about. That just gave you TS or Distortion Plus style clipping though, version B here is something completely different.

B looks pretty interesting, just simmed it up quickly, and if the input signal is below the voltage where the diodes start conducting you get no gain, so play quietly and you get completely clean then hammer away and you get filth  
Splitting Rb into 2 separate resistors, one for each diode, might be interesting.

[caress]

Just some cool little bits of circuit to think about.

you're very good at recognizing those!   

[edvard]

B looks pretty interesting, just simmed it up quickly, and if the input signal is below the voltage where the diodes start conducting you get no gain, so play quietly and you get completely clean then hammer away and you get filth   

Hmmm... kinda like the Boss Dyna-Drive?
Interesting...

[earthtonesaudio]

B looks pretty interesting, just simmed it up quickly, and if the input signal is below the voltage where the diodes start conducting you get no gain, so play quietly and you get completely clean then hammer away and you get filth   

Hmmm... kinda like the Boss Dyna-Drive?
Interesting...

Maybe.  I've never seen a schematic for that one, but the description is about right.  I would imagine in most cases you'd want to add diode clipping to circuit B if used as a distortion, because most op-amps sound pretty bad when driven into clipping.

Maybe LEDs in the feedback loop and Si diodes from (-) to Vref.  That would give you dynamic expansion for signals up to a certain level, but then the LEDs would keep it from getting huge.

[Earthscum]

Use a pot in place of Rb, both outer lugs connected to the diodes, and the wiper connected from the feedback path. One-knob wave shaping. (schem B)

[edvard]

B looks pretty interesting, just simmed it up quickly, and if the input signal is below the voltage where the diodes start conducting you get no gain, so play quietly and you get completely clean then hammer away and you get filth   

Hmmm... kinda like the Boss Dyna-Drive?
Interesting...

Maybe.  I've never seen a schematic for that one, but the description is about right.  I would imagine in most cases you'd want to add diode clipping to circuit B if used as a distortion, because most op-amps sound pretty bad when driven into clipping.

Maybe LEDs in the feedback loop and Si diodes from (-) to Vref.  That would give you dynamic expansion for signals up to a certain level, but then the LEDs would keep it from getting huge.

Yep, all I've ever seen is gut shots.
Lots of SMD...
I think your idea would work nicely, some sort of gain/dynamics control for Rb, or wave-shaping like Earthscum said.
This could get interesting...

[Lurco]

http://www.audiodesignguide.com/doc/ac_logarithmic_amplifier.jpg

[slacker]

very cool

[earthtonesaudio]

http://www.audiodesignguide.com/doc/ac_logarithmic_amplifier.jpg

I like it!