Importance of IV knee graph of a Diode in soft clipping applications

[Vivek]

From all above

and from that famous old article

https://web.archive.org/web/20190831161652/http://www.bteaudio.com/articles/TSS/TSS.html

I understood : The output waveform of the noninverting clipper consists of two components:
(1) the amplified and clipped version of the input waveform
(2) plus the unamplified input waveform.


I also feel compliance resistors allow part of original signal to go through


My question is

What is the difference between :

A) noninverting clipper with no compliance resistors
and
B) Inverting clipper with compliance resistors in series with the diodes

[iainpunk]

My question is

What is the difference between :

A) noninverting clipper with no compliance resistors
and
B) Inverting clipper with compliance resistors in series with the diodes

they can sound exactly the same, but the inverting stage can clip hard, and the non inverting stage can't.
the systems both respond differently to capacitor tone shaping,
the inverting one can go down to -inf dB where the non inverting can only go down to 0dB
i prefer noninverting due to easier math... you can throw almost anything in the nfb loop, calculate the response and mirror it around the 0db axis. the math is harder with inverting stages imho.

cheers, Iain

[Vivek]

I for one like diodes in the NFB path of an inverting opamps better than a non inverting because I find it easier to tune the clipping characteristics.

i prefer noninverting due to easier math. imho.

cheers, Iain

I'm lost !!!!!!

[iainpunk]

its both valid, and up to personal preference,
he prefers noninv. for tuning the specific clipping sounds, like knee and hardness.
i prefer inv. for easier tuning the freq. response, and i care less (not at all) about the specifics of clipping, like knee and hardness.
i think his method is more usefull for you, since it allows for really fine tuning and perfecting the clipping itself.

cheers

[Fancy Lime]

Hmm, funny. I find inverting stages easier math-wise. Well, neither kind is rocket surgery. But for me, the ability to go below 0db is often useful when cramming filter functionality into clipping stages. One big caveat, though: if you want now noise and high input impedance, you need a non-inverting opamps stage. So the very first stage should usually be non-inverting to lift the signal out of the noise floor. For an amp in a box, I would make the first stage something similar to a tube screamer but with a higher corner frequency, maybe 1.5kHz, LEDs, possibly asymmetrical, and probably a bit less gain. Next a bass control and maybe mid control, then an inverting clipping stage and after that the treble control. At the very end I would stick a very primitive pseudo-speakersim in the form of a two to four pole Butterworth(ish) low pass filter at somewhere around 6kHz ( adjust to taste). Remember, this is just where I personally would start tinkering but there is more than one way to pickle a herring. You do you! We'll TRY to help, of course, but reading my own text here I wonder if that is really all that helpful...

Cheers,
Andy

[Steben]

I for one like diodes in the NFB path of an inverting opamps better than a non inverting because I find it easier to tune the clipping characteristics.

i prefer noninverting due to easier math. imho.

cheers, Iain

I'm lost !!!!!!

There is no spoon mate

[Fancy Lime]

I for one like diodes in the NFB path of an inverting opamps better than a non inverting because I find it easier to tune the clipping characteristics.

i prefer noninverting due to easier math. imho.

cheers, Iain

I'm lost !!!!!!

There is no spoon mate

True!

I thought some more about my and Ian's different perceptions on this.  I think, which is easier depends on whether you think in volts or in amperes. Non-inverting is easier to reason about in voltages whereas inverting is easy in current. As long as there are only resistors involved, that does not make a real difference. But with diodes, one or the other may be easier to wrap your head around, depending on how you are wired.

Cheers,
Andy

[iainpunk]

Hmm, funny. I find inverting stages easier math-wise.

well, i think when you use 'hard number math' its kinda equal, but when you are in to 'freestyle-interpretive math' non-inverting is easier because you can just ''flip over'' the response of what you put in the feedback loop.
put a mid scoop in the NFB? you get a mid boost and the frequency's above and below that still have a gain of 1!
put in crossover diodes? you get clipping!
put in clipping? you get gating! *
put in a mid boost? you get a mid scoop!
dislike the big muff midscoop, but like the functionality? put it in a NFB!!!!!!! **
the only thing you have to keep in mind is phase margins and such, but phase margins also apply to inverting stages.

But for me, the ability to go below 0db is often useful when cramming filter functionality into clipping stages.

you can as well in a non inverting gain stage, but it needs another active element to get that to work, and the phase margins may become a bit sketchy...

Remember, this is just where I personally would start tinkering but there is more than one way to pickle a herring. You do you! We'll TRY to help, of course, but reading my own text here I wonder if that is really all that helpful...

this is exactly the message Vivek needs to hear.

cheers, Iain

* not clean gating, but what i'd like to call ''soft dirty gating'' the gain under the threshold is 1 and there is crossover distortion at play.
** make sure that the pot's range is limited on the treble boost side, use a 5k6 resistor with a 100k pot. this keeps the treble boost shelved. omitting the resistor makes the op amp try to boost the high frequency's WAY to much (near infinite)

[Vivek]

Thanks everyone

Learnt a lot from this thread, and still learning !!!

[iainpunk]

Thanks everyone

Learnt a lot from this thread, and still learning !!!

you're welcome! i enjoy these kinds of threads the most tbh. low pressure and just throwing up some ideas, opinions and thoughts.

cheers, Iain

[ThePracticalPeasant]

and from that famous old article
https://web.archive.org/web/20190831161652/http://www.bteaudio.com/articles/TSS/TSS.html

Sorry to resurrect; I felt I'd get some value out of this but I always worry when I see the only working reference being on webarchive; And given that the article is "famous", I took a moment to make 3 PDF documents out of the link/references for my own use... (and please let me know if this is an inappropriate offer on the forum) ...would be happy to make them available.

[brett]

Hi.  My memory is a bit rubbish...but 40 years of experience says...
YES the knee is similar in shape (I vs V) for EVERYTHING from a Schottky diode clipping at 0.3 V to an LED clipping at 3V.
The difference in tone is 99% due to the V threshold for conduction.  At a nominal gain (say 10), a 0.1V input signal is lipped to blazes with a Schottky, not at all with an LED. 
End of story....fairly much...
I put up a post with data about this about 15 years ago...
There was a TINY exception with a 1N914/1N4148, with some early conduction, possibly due to the metal (gold?) doping of the junction.
Sorry if I'm repeating what was said above.  I'm not up for wading through tonnes of muddy thinking to find a few pearls of wisdom.  Lazy old me!
Cheers, and happy building!  Never give up.

[Vivek]

TLDR

I am trying to say that

If two different clippers are

A) Standardised for clipping percentage
( for example, 6 Vp signal clipped with diode of Vf 0.6V and 25Vp signal clipped with diode of Vf 2.5V)

and

B) The output is standardised for volume
(both above clippers adjusted to have 1Vp final output)


Then the differences in most diodes is negligible.

[Fancy Lime]

I recently saw a pedal that called the switch between 1N4148 and LED clipping "mid boost" and almost wet myself laughing while watching the YouTube reviews waxing poetic about how effectively it boosts the mids. So I say, the only thing that really matters, is how you sell or label it.
Kidding aside, yes when normalized to gain, most diodes sound the same. You have to design for "riding the knee" by scaling current, if that is what you want. One exception worth mentioning, apart from the slightly softer knee on 1N4148, is reverse breakdown of low voltage Zeners. These are much softer, which can be useful.

Andy

[amptramp]

For those who are looking for a diode drop somewhere between silicon and red LED, there are infrared LED's available and they are included in any optoisolator.  This raises the possibility that you could use the LED in an optoisolator as a circuit element as the diode and use the phototransistor output to drive a LED to show when clipping starts.  LED's are sensitive to ambient light - if you mount one on the outside of the enclosure, it will have different characteristics in high and low ambient light and it will be modulated by AC lighting, causing powerline hum even if you are running from a regulated supply.

[Steben]

Hi.  My memory is a bit rubbish...but 40 years of experience says...
YES the knee is similar in shape (I vs V) for EVERYTHING from a Schottky diode clipping at 0.3 V to an LED clipping at 3V.
The difference in tone is 99% due to the V threshold for conduction.  At a nominal gain (say 10), a 0.1V input signal is lipped to blazes with a Schottky, not at all with an LED. 
End of story....fairly much...
I put up a post with data about this about 15 years ago...
There was a TINY exception with a 1N914/1N4148, with some early conduction, possibly due to the metal (gold?) doping of the junction.
Sorry if I'm repeating what was said above.  I'm not up for wading through tonnes of muddy thinking to find a few pearls of wisdom.  Lazy old me!
Cheers, and happy building!  Never give up.

How they sound is a bit personal. I mean what some call a difference some call a small difference and some no difference.
Both are true: most diode behaviour follows the same Shockley law. But, the internal resistance is neglected in this law, like comparing 2 germanium diodes vs a silicon diode.
And stacking diodes raises the combined resistance. The lower the impedance of the signal, the more important this is.
But since a diode is an element following shockley's law with an internal resistance, diode graphs can be mutually mimicked with combinations of .... diodes and resistors.

[amptramp]

If you connect a diode in a feedback loop of an op amp, you can get an ideal diode that can be used as a clamp circuit that stops the circuit from going in a selected direction, either positive or negative.  The diode goes from the output to the inverting input and the clamp voltage is set by the voltage at the non-inverting input.  You can put a resistor in series with this to simulate diode resistance.  The circuit input comes through a resistor connected to the inverting input and the output is taken from the inverting input, making it a clamp circuit that prevents the voltage from exceeding the clamp voltage.  By varying the clamp voltage, you can set the forward voltage and the resistor simulates the diode resistance.



The schematic shows specific devices and voltages, but the op amp, diode and operating voltages can be changed.

[Steben]

but the inverting stage can clip hard, and the non inverting stage can't.

It can when the opamp hits the rails And it is one of the pitfalls of "soft" clippers. Running large signals into non-inverting clipper circuits ruin a lot. While hard clippers retain their character.
The ratio of input signal to soft clipping treshold makes for a great sounding or meh sounding device.

But as far as diodes go you are right of course.
Inverting stages can "tailor" the clipping points, just as a passive clipper to ground can. It is like making a clipper "active" just as in tone control circuits.
Non-inverting stages can bring in the unity gain clip limitation. Which can be useful. But as I said, adding hard clipping before a soft clipper helps to retain the character.

Do not forget the plexi holy grail of (hard) rock sound is mainly a hard clipped negative feedback power amp. It is why it is compared to other sounds so easy to simulate.