Nonsymmetrical Clipping

[Sean Costello]

At 10:46 PM 9/9/96 +0200, you wrote:
>JH wrote:
>
>>Background:
>>As I have posted some time ago, the famous SSM2040 filter
>>sound is partly due to its nonsymmetrical clipping, when the
>>ota output (!) is overdriven. 
>
>Thank you for this insight!
>
>Do i take it nonsymmetrical clipping means that either the lower or
>the higher halfwave clips earlier than the other?

Nonsymmetrical clipping implies that the clipping of the waveform halfves
are not equal; one half may be clipped more than the other half, or the type
of clipping may differ between the halves.  For example, a typical fuzz box
circuit uses back-to-back diodes to achieve the "fuzz" sound (these diodes
are usually either in the feedback loop of an op-amp, or simply connected
between the signal and ground).  If only one diode is present, only one half
of the waveform will be clipped; this waveform will have more even harmonics
present than the waveform clipped by back-to-back diodes (which will sound
more or less like a square wave, depending on the input signal and the type
of diodes).  Silicon and germanium diodes clip differently, with the onset
of clipping in a germanium diode being "softer" (more gradual); thus, having
a silicon diode back-to-back with a germanium diode will result in
nonsymmetrical clipping.

>If so, my question is are there differences in sound depending on
>whether the high or low halfwave clips first?

Depends on the input signal, I guess. 

>Would it be generally desireable to have the option of nonsymmetrical
>clipping as an effect?

I think so.  I'm forwarding something I wrote in a message to A.S.P. (Romeo)
a while back:

>The SSM 2040 is reputed to have good distortion.  I wonder if the quality
of the distortion is due to the inverting gain stages.  The stuff I have
read in the diy archive suggests that each stage has asymmetrical
distortion, but that the inversion of the signal in each gain stage makes
the distortion symmetrical (as the non-clipped side of the signal would be
clipped by the next stage).  Undoubtedly both sides of the signal would be
distorted in this situation, but this does not mean that the distortion is
symmetrical.  I know that there is a tube-emulation circuit that clips one
half of the waveform, amplifies it, then clips the opposite half of the
amplified waveform; this results in asymmetrical distortion.  Asymmetrical
distortion sounds more "tube-like" than symmetrical distortion, in that more
even harmonics are found in the output signal (as opposed to the square-wave
type signal found in symmetrical distortion); even harmonics are generally
regarded as more pleasing to the ear.  Tubes run in class A mode (such as
the Vox AC-30, the Fender Champ, and the Matchless amps) have lots of even
harmonics in the output, and a very musical sound; the Fuzz Face also uses
asymmetrical clipping for an output rich in even harmonics.  It seems to me
that the inversion of each gain stage in the 2040 would result in a signal
that was clipped, amplified and otherwise modified, than clipped on the
other side repeatedly, resulting in a complex waveform rich in even
harmonics.  The 2040 emulation circuits I have seen (I am including the
Electronotes 4-pole in this category) seem to use noninverting OTA stages;
this would result in asymmetrical clipping, but of a different type than
that in a circuit that uses inverting stages.  Perhaps the ideal circuit
would use inverting OTA stages, and smaller resistor values at the imput and
between the stages to obtain higher clipping levels.  

Unfortunately, I haven't heard anything about traditional OTA op-amps (3080,
13600/13700) clipping asymmetrically.  Many other things could be wrong in
the above statement (this should be a warning on all of my messages - I was
an Anthropology major, and am learning electronics on the fly).  Anybody out
there who knows what they are talking about have any ideas?

Later,

Sean