stacking clipping parts - soft knee

[Lino22]

Guys, i understand how hard clipping works and i understand that by stacking clipping parts there is more headroom for signal and so the clipping happens on harder notes.

What i don't get is this - which of the stacked parts have the major impact on the softness of the knee.
The Ge diodes are soft, MOSFETs are soft, Si diodes are a bit harder but allow more voltage through before they start clipping.





Is there any reason for mixing those parts? Ge, FET, Si ...
I mean i have a MOSFET clipper and i want more headroom, shall i add Ge diodes to keep the knee soft, or i can afford Si diodes?

I did an ear check for Si+Si+FET / Ge+Ge+FET and apart from the obvious volume difference, i didn't hear any tone difference.

[antonis]

I mean i have a MOSFET clipper and i want more headroom, shall i add Ge diodes to keep the knee soft, or i can afford Si diodes?

Add more Mosfet clippers..!!!
(or a series resistor..)

As for the "knee", parallel caps are your friends..

[Lino22]

As for the "knee", parallel caps are your friends..

I am not sure i understand

(or a series resistor..)

Will the series resistor work even in a feedback - for soft clipping?

[antonis]

Compare a hard clipping Si diodes pair with and without a 1nF to 10nF cap in parallel..
(like in BOSS DS-1 distortion..)

[antonis]

Will the series resistor work even in a feedback - for soft clipping?

Why not..??
(current through resistor times resistor's value determines extra headroom..)

[GibsonGM]

Note in the clipping stage linked below how there is a 1n cap in parallel with the clipping diodes. This affects the knee. Experiment with the value of this cap!       

https://www.electrosmash.com/images/tech/mxr-distortion-plus/mxr_distortion_schematic_parts.png

[Lino22]

Do we use the cap trick also in the feedback clipping? Probably yes, right? I remember seeing them there.

[antonis]

The cap you refer on serves as part of the feedback loop LPF..
(its value depends on feedback reistor value..)

Clipping diodes cap is usually set in series with clipping pair..
(ala Big Muff style..)

[Lino22]

Ah, ok, so when i do hard clipping, i use a cap in parallel, when i do soft clipping, i use a cap in series - i mean a cap to make the knee softer?

[Mark Hammer]

I can be rehabilitated, but I have my doubts about the relevance of diode conduction "knees" to clipping tone.  Not that such conduction-rate differences don't exist, but that the speed differences are well above the usable bandwidth of most guitars.  Keep in mind that such diodes were being developed for use in high-speed switching systems DECADES before their application to guitar sound.

From what I understand, an important determinant of the clipping function of a diode is the current of the signal hitting it.  If we go back 30 0r 40 years, one would often see the inclusion of small-value resistors (i.e., often less than 100R) to form a voltage divider, with so-called "hard clipping" diodes connecting one of the resistors to ground.  The result apparently "softened" the clipping.

[Lino22]

Well i did an A/B ear test, and i can't hear any difference between Ge and Si diodes in what is now a soft clipping stage. I was just asking because i can't picture it in my head when the clipping elements are mixed.

[GibsonGM]

That's why it's often best to simply use your ears, Lino!    The cap in parallel with hard clipping diodes is intended to shunt to ground some of the harsh high frequency content the diodes introduce to the signal...this might be taken as 'softening the knee', but like Mark - I tend to believe that we don't really hear that knee very much.  Current seems to have more effect. 

[Lino22]

Ears are definitely the best judge

I will do an A/B ear test for silicon versus Mosfet. I know there were designs that had MOSFET in the soft clipping stage (Mosferatu), accompanied by Ge diodes to get some extra headroom. I just wonder if i can hear any difference.

[GibsonGM]

I've always found that I can tell  Si vs. Ge....and LED from both.   Mosfet is more subtle, but I think sometimes seems smoother, more tube-like, but that is subjective.

What I can REALLY tell is if I stack diodes....adding a Ge vs. a Si to an existing clipping pair.  It seems I can hear the difference in level, but not actually the QUALITY they may or may not give.   For example, 2xGe in series is to me much like 1XSi, in this use.    Try them both

[Mark Hammer]

Well i did an A/B ear test, and i can't hear any difference between Ge and Si diodes in what is now a soft clipping stage. I was just asking because i can't picture it in my head when the clipping elements are mixed.

To beat an already dead horse, my view is that what we call "soft" and "hard" clipping is really the difference between single and double-clipping.

When diodes are in the feedback loop of an op-amp, the signal gain is essentially limited to whatever the +/- forward voltage of the diodes is.  With a pair of Si diodes in the feedback loop, it doesn't really matter HOW much gain you set it for, because the maximum signal amplitude will not exceed much beyond +/-600mv, which well below the headroom limits of the op-amp.

In contrast, run the op amp at gains of >50 or so, and whatever leaves the op-amp to hit a pair of diodes to ground, has already reached the headroom limits of the chip before it ever hits the diodes, making it in effect "double" clipping.  It's not harsher sounding because the diodes are going to ground.  It's harsher sounding because the op-amp has already added harmonic content that wasn't there in the original signal.

[antonis]

When diodes are in the feedback loop of an op-amp, the signal gain is essentially limited to whatever the +/- forward voltage of the diodes is.  With a pair of Si diodes in the feedback loop, it doesn't really matter HOW much gain you set it for, because the maximum signal amplitude will not exceed much beyond +/-600mv, which well below the headroom limits of the op-amp.

With all the respect, Mark..

Gain set determines unclipped signal waveform slope (gradient)..
e.g. for infinite gain, unclipped sides are parallel hence suared wave output (considering diodes abrupt clipping)..

[Mark Hammer]

When diodes are in the feedback loop of an op-amp, the signal gain is essentially limited to whatever the +/- forward voltage of the diodes is.  With a pair of Si diodes in the feedback loop, it doesn't really matter HOW much gain you set it for, because the maximum signal amplitude will not exceed much beyond +/-600mv, which well below the headroom limits of the op-amp.

With all the respect, Mark..

Gain set determines unclipped signal waveform slope (gradient)..
e.g. for infinite gain, unclipped sides are parallel hence squared wave output (considering diodes abrupt clipping)..

Thanks for catching that.  I used the wrong word.  Signal level is limited to whatever the diode forward voltage is.  That's one of the reasons why the Boss SD-1 is "louder" than the TS-9; it uses a 2+1 diode trio, which raises the maximum signal amplitude on half the wave.

Gain can be set to whatever the heck you want via the feedback loop and ground leg.  Naturally, it will boost the dickens out of the softest signals and raise the overall perceived loudness, but nothing is going to get bigger than whatever the voltage is that makes the diode conduct.

[Rob Strand]

hat i don't get is this - which of the stacked parts have the major impact on the softness of the knee.
The Ge diodes are soft, MOSFETs are soft, Si diodes are a bit harder but allow more voltage through before they start clipping.

Is there any reason for mixing those parts? Ge, FET, Si ...
I mean i have a MOSFET clipper and i want more headroom, shall i add Ge diodes to keep the knee soft, or i can afford Si diodes?

Softness is the curvature around the clipping point
https://www.diystompboxes.com/smfforum/index.php?topic=125795.msg1200675#msg1200675

I would call germaniums soft but I wouldn't call MOSFETs soft.   The other difference is germaniums have a lower clip voltage than silicons whereas MOSFETs (connected as MOSFETs) have a higher clip voltage than silicons.   (In your circuit  the MOSFET wired as a diode.  The clip voltage is about the same as silicons and these clip harder than silicons.)

If you switch between diodes you get changes between softness and clipping levels which will only confound the experiment.

You can add series diodes to match the clip levels, as antonis mentioned.

It's difficult to separate the concept of headroom and clipping levels.

A soft clipper starts distorting on smaller signals.   If we define headroom as the level of clean then a soft clipper will have less headroom.   However when you push a soft clipping the distortion is less aggressive and so you can drive it harder before you get harsh distortion and from that definition you can say it has more headroom.   A soft clipper has a wider grey area of where the distortion starts and when distortion is harsh.

If you define headroom as the harsh clip level then if you use two of each type of diode in series the output level is simply doubled.   However the underlying softness and hardness of clipping characteristic has not changed.  In fact this situation is identical to putting a x2 amplifier after the clipper.

If you do experiments at 9V you will get confounding results because if you double the output level any opamps following the clipping may or may not clip.  If they clip then they will be adding a different distortion to diodes.

Another issue is if you set the distortion knob to say 12 O'Clock the gain stage will have a certain amount of gain.   If you double the number of diodes the clip levels increase.  However since the gain has not changed the doubled diodes are driven less hard and it sounds like less clipped, just louder.   That can happen when going from silicons to MOSFETs or LEDs.

[amptramp]

There is a consideration of what circuit it is used in.  If you have antiparallel diodes to ground, you get the voltage division between the output impedance of the amplifier that is driving it forming the upper part of a volume control with the impedance of the diode that is conducting forming the lower part.

If you use the diode pair in the feedback of an inverting amplifier, you get the same effect but this restricts the output levels to the diode drop, so it is unlikely that you will get any op amp clipping.

If you use the diode pair in the feedback of a non-inverting amplifier, the amplifier will always pass the original signal with a gain of 1, so the clipping will not seem as harsh.  This is what the Tube Screamer does.