Soft vs hard clipping - I feel such a fool

[iep]

Many (many) websites and fairly authoritative texts I have read on guitar effects and clipping techniques have stated that soft (overdrive like) clipping can be achieved by placing clipping diodes in the feedback path of an op-amp (tube screamer) while distortion is achieved by shorting the opamps output to GND via two (or more) back to back diodes (DS-1). I have read this and believed it (and indeed repeated it to others).

The website below is one such source:

Design Your Own Distortion | General Guitar Gadgets

No offence to the author but it turns out this is a load of horse. In fact, either configuration produces identical results. Any real difference is just down to EQ.

Couple of examples.

Soft clipping schematisc and resultant waveforms. See the alleged soft clipping schematic on the left and the alleged hard clipping schematic on the right. Note that output waveforms are identical:





Hard clipping schematics and resultant waveforms. See the alleged soft clipping schematic on the left and the alleged hard clipping schematic on the right (I simply removed the filter caps). Note that output waveforms are identical:





If you are so inclined and do the maths, this is absolutely the expected behaviour. So, how did this daft nonsense get punted around and become accepted as true? Am I missing something?

Cheers,

iep

[potatofarmer]

I dunno, no one noticed the Big Muff's clipping diodes in their feedback loops?

[iep]

Or indeed the back to back clipping diodes to GND in the Boss BD-2 (largely touted as an overdrive pedal). It was the BD-2 that got me looking in to this.

As an aside, an awful lot of rubbish gets written about modding the BD-2. Replacing the non-clipping diodes or changing non-dominant poles and expecting greater bass response...... Rant over.

iep

[pdf64]

I perceive similar claptrap regurgitated regarding overdriven tube circuits, eg power tubes clip softer than pre, primarily even harmonics.

[Mark Hammer]

Oversold and underexplained, perhaps, but not claptrap.

Consider that the diodes in an op-amp feedback loop form a sort of variable conductor. When the signal voltage at the op-amp output reaches the forward voltage of those diodes, they conduct. But the gain of an op-amp is dictated by how much negative feedback from the output is employed. If there is more negative feedback, then the gain is reduced.

So, the action of the diodes in the feedback loop modify the gain of the op-amp, temporarily reducing it. Of course, once the gain is reduced, via the diodes conducting, then that results in less output, which results in the feedback signal dropping below the forward voltage of the diodes. And so on.

The impact is somewhat analogous to how a great many compressors work. The output of their gain stage feeds a rectifier which governs the gain stage. As their output increases, the gain is reduced, and as their output drops, the rectifier turns up the gain to compensate.

This is part of the reason why people will describe the Tube Screamer as sounding compressed.

But does it produce a different quality of clipping?

I think one needs to examine an actual guitar signal, rather than scope a signal generator. Keep in mind that the guitar signal is highly variable, both within a plucked/picked note, and across notes. A signal generator will put out a constant level, and it is a piece of cake for the resulting waveforms of feedback-loop and output-to-ground diode pairs to look identical.

What is missing from that picture is that the feedback-loop pair influence the gain, and resulting waveform, in a dynamic (changing) fashion. And, pairing that with a changing signal, you get different outcomes. In contrast, with the output-to-ground configuration, NOTHING about the gain of the preceding stage changes. The diodes have their conduction threshold, and the preceding gain stage hikes the signal up to some threshold and beyond; end of story. I don't know that such a difference is appropriately captured by the terms hard-vs-soft clipping (which would usually be applied to the manner in which the diodes conduct in response to a steady-level signal), but the two configurations do result in an audibly different outcome, for the reasons given.

Like I say, not claptrap, but generally oversold and under-explained.

[pdf64]

My perception with the TS is that the non-inverting op-amp arrangement plays a significant part in the overall feel, as it allows the clean signal through, mixed with the distorted signal; I find that very apparent with a high output guitar and the amp set clean, or viewed on the scope, as the overdriven part can only get to 0.6Vp-p, whereas the clean portion of the fx signal can get way bigger.
The signal then goes through an extreme low pass filter, so the undistorted bit gets to be very bassy.

If the TS overdrive stage is changed to inverting, with everything else arrange to keep the same gain, level, corner frequencies etc either side of the overdrive stage, it is a rather different effect. Probably due to the clean bassy part of the signal being lost.

EDIT - I think that the sparkledrive takes this a step further, giving some user control of the non-clean signal level.

I'm a little sceptical about the distinction between compression and clipping; maybe if the subsequent signal chain is clean some difference may be noticed, but when used to boost an overdriven amp into heavier overdrive, I doubt it?

[iep]

Hi,
Thanks for the explanation but I'm not sure I agree.

Sure, in the tube screamer topology, the diodes form a rapidly reducing impedance for signals increasing beyond their threshold voltage. At this point Rfb becomes very small compared to Rin and the gain of the stage starts to reduce (stage gain = Rfb/Rin). Once the gain has reduced to a certain point, a stable output voltage is reached. This stable output voltage is basically the diode's threshold voltage. So, basically you just see the voltage threshold of the diode superimposed onto the amp's output. This is exactly the same as you'll see with back to back diodes to GND.

Complexity of waveform has no effect on this.

Cheers,

iep

[Chuck H]

OP is addressing SS devices. I can only speak to similar "claptrap" about tube circuits. The first one being hard vs. soft clipping... Common knowledge is that power tubes clip soft and preamp tubes clip hard. IME this is NEVER true and, in fact, the opposite is in many cases. Then there's the bogus about odd and even harmonics. More common knowledge is that clipping power tubes make even order harmonics and preamp tubes make more odd order harmonics. Also bunk. And, also, in many cases the opposite is true. The exception being the noteworthy degree of duty cycle mismatch in most push/pull tube output circuits that usually only happens when clipping very hard. And even then, with more even order harmonics from the power amp, it's still not more than you typically get from the preamp and the clipping isn't "softer".

This doody and the OP are just some of the internet lore about tone that afflicts intelligent analysis for the layperson. I think it starts with someone's impression of what's happening getting tossed at a wall of equally unknowledgeable gear heads and whatever suits perceptions sticks. Then it gets repeated as fact because of wide broadcast.

Pete,
The only contention I would have with clipping as it relates to compression would be, again, WRT tubes amps. Where power supplies and operating points can change in a way that creates an actual pumping effect analogous to the attempted power in dynamic circumstances. So there is actual, non clipping related compression happening as well.

And, in this light, just as drinking the KoolAid from some internet site can be shown as perception rather than reality, so can circuit analysis with a fixed sine wave. How an amp behaves under dynamic circumstances is just as important, if not more so, that what a standing wave form looks like on a scope. With response times and high/low operating conditions sometimes even skewed so far from a standing wave form as to be incomparable.

But we need to work with the tools we have and, for me, gleaning any specific tonal information from scope shots with an actual guitar in use or from pulse triggering is just impossible. With the exception that I can plainly see round or square edges when the opposite might have been true of a standing wave or I can recognize the difference at a given signal level when a time constant related parameter is involved that is changing operating conditions for that level under dynamic circumstances as compared to static. Other, more technically savvy members here are much better with dynamic signal interpretation on scopes. I read the posts and assimilate what I can (not always much).

[pdf64]

The only contention I would have with clipping as it relates to compression would be, again, WRT tubes amps. Where power supplies and operating points can change in a way that creates an actual pumping effect analogous to the attempted power in dynamic circumstances. So there is actual, non clipping related compression happening as well.

I think this happens in pre-amps, tube and with SS too, eg the fuzzface etc exhibits signal strength related bias shift that seems to produce a compression effect.

[iep]

Ahhhhhhhhhhhhhhhhhh, now I do feel like a fool. I have been missing something fairly obvious (thanks to pdf64 for helping me spot it). The TS uses a non inverting opamp configuration. I'd been daftly considering the inverting case.

In the inverting case, once the output voltage hits roughly Vthreshold for the diode, Rfb_total (Rfb in parralel with Diode) tends to zero and the output essentially tracks the voltage over the diode (hence identical to back to back diode clipping).

However, in the non inverting case, because the gain on the non inverting stage is 1+Rfb/Rin rather than -Rfb/Rin, the gain of the stage tends to one rather than zero. As pdf64 says, at this point it becomes more like a buffer amp. The distortion is essentially created by the signal dependent gain of the stage and the relatively abrupt change from one gain to another. I guess this is what Mark Hammer was getting at when referring to the compressor like behaviour and it certainly better mimics tube distortion. Sorry for misunderstanding.

By way of recompense, I'll attach a couple of plots to highlight the differences in the waveform and how an inverting amp configuration can be set up to produce a similar effect.

Non - Inverting amp and resultant waveform (actual soft clipping):






Below is the equivalent inverting amp circuit. You simply add a resistor in series with the diodes so that you get a reduced gain change when the diodes start to conduct.






Note that the gain discontinuity (where the diodes start to conduct) occurs sooner in the inverting case, this is due to the inverting input of the amp being fixed in the inverting case. The upshot might be 'cleaner' sounding peaks.

Hope this is all helpful to somebody elses understanding.

Cheers,

iep

[pdf64]

RG's 'The technology of ...' pages are well worth a looky (and re-looky if it's been a while) "The Technology of ... " Series of Effects Articles

[guitician]

It's better to work with your ears on things like sounds, kind of like using your eyes when making things of a visual nature...

A non-clipped audio amp, still has harmonic distortion, some as high as 10%, and depending on whether it's odd, or even will determine whether you can stand listening to it.

[Mark Hammer]

I'm not sure that the inverting-vs-noninverting op-amp makes that much of a difference in the manner in which clipping occurs. Whatever the case, we do concur that whereas feedback diodes change the gain of the op-amp in response to a changing signal, ground-path diodes do not modify the gain of the op-amp stage preceding them, such that the clipping produced is purely/largely a function of the signal amplitude at the op-amp output (whether inverting or noninverting), and the forward voltage of the diodes.

I say "largely", because the diode conduction is also a function of the current and the current-limiting resistor just ahead of it.

Now, while series-resistors ARE used in conjunction with feedback diodes (and I incorporated this into a simple drive unit I made: The Aefea Drive - a simple one-chip diversion https://www.youtube.com/watch?v=sWiRd-N2_ik ), their impact is not quite so much on when and where the diodes conduct, but rather on what their effect on gain is when they do.

[pdf64]

A non-clipped audio amp, still has harmonic distortion, some as high as 10%, and depending on whether it's odd, or even will determine whether you can stand listening to it.

How have you come to that viewpoint?

[guitician]

I've heard it I guess.