Coupling cap to clippers: value?

[hans h]

Hi all,

Short question (I think) : with an opamp driving passive clippers to ground, does the value of  the coupling cap prior to the clippers affect clipping or frequency response? Examples:
1. Dist + uses 1uf. Big value is needed to drive volume pot and whatever pedal or Amp that follows. Effect on clipping tone?
2. Bjfe honey bee and dyna red use 1uf. The clippers are followed by a buffer so a large value should not be necessary from a driving point except possibly for the filters between the opamp and subsequent buffer.
3. Barber dirty bomb (schematic at revolution deux) uses 330n. Clippers are followed by a tone stack and volume pot so quite a lot of driving to do, yet the value is substantially smaller than the previous examples.
4. Boss Ds1 uses 470 nf and drives stuff similar to barber dirty bomb.
5. Prco rat uses 4u7 (!) while it doesn't drive that much and has a buffer prior to the output.

So the values differ greatly and large values are sometimes used despite the fact that there is a buffer between the coupling cap and output. Hence my question: does the value have a tonal effect, for example while the pedal is clipping (low to no resistance to by ground)?

[niektb]

The value does have an impact on response but only in combination with the resistors

[ElectricDruid]

1. Dist + uses 1uf. Big value is needed to drive volume pot and whatever pedal or Amp that follows. Effect on clipping tone?

Yes, the cap forms a highpass filter with the resistance to ground.

2. Bjfe honey bee and dyna red use 1uf. The clippers are followed by a buffer so a large value should not be necessary from a driving point except possibly for the filters between the opamp and subsequent buffer.

Exactly! That LPF filter to the next buffer is an effective resistance to ground, as are the feedback networks which go to a virtual ground. The buffer buffers the *filter*, so can be ignored from the point of view of the coupling cap.

3. Barber dirty bomb (schematic at revolution deux) uses 330n. Clippers are followed by a tone stack and volume pot so quite a lot of driving to do, yet the value is substantially smaller than the previous examples.

What does the highpass roll-off work out as? They may have decided to dump some bass to stop stuff getting muddy, or it might be that it rolls off below 80Hz or something, which allows all guitar notes to pass unaffected.

4. Boss Ds1 uses 470 nf and drives stuff similar to barber dirty bomb.

Again, what's the rolloff work out as? Can you guesstimate it, or is the situation complex enough to need a sim?

5. Prco rat uses 4u7 (!) while it doesn't drive that much and has a buffer prior to the output.

Maybe it saved them a part value. If 4u7 is used elsewhere several times, it's cheaper to buy another than get a few of something different.

So the values differ greatly and large values are sometimes used despite the fact that there is a buffer between the coupling cap and output. Hence my question: does the value have a tonal effect, for example while the pedal is clipping (low to no resistance to by ground)?

Yes, the value potentially has a tonal effect. The examples you've given where there's a buffer seem to mostly have other stuff ahead of the buffer first, in which case, the buffer is buffering that stuff, not the cooupling cap.

[antonis]

Only shunt caps (in parallel with clippers, like in Boss DS-1, MXR Dist+, etc..) have an impact in clipping waveform shape..

Stictly speaking, coupling caps DO form a HPF with diode(s) dynamic resistance but let's not dig in deeper..

[m4268588]

If the clip diodes is asymmetric, it affects the transient response (time constant) of the symmetry.
All of your examples are symmetrical clips, so there is no such affect.

Cap is added to the resistance value in series. How affect of the resistance value on clipping.

[hans h]

@antonis that was sort of along the lines I was thinking. Maybe it affects soft clippers to ground like mosfets used as passive clippers. They have more transient behavior as opposed to the harder clippers.

And I indeed need to calculate the r-c to grasp what it is doing in terms of filtering. My question was maybe not worded so clearly. I was primarily trying to ask what the effect of the coupling cap was apart from being a straightforward rc filter.

[antonis]

And I indeed need to calculate the r-c to grasp what it is doing in terms of filtering.

https://physicswave.com/dynamic-resistance-of-diode/

[Elektrojänis]

Stictly speaking, coupling caps DO form a HPF with diode(s) dynamic resistance but let's not dig in deeper..

But then we would leave out a part of the whole puzzle. I think thats the part that needs digging. Not that I would be capable of the analysis, but I'm interested. The regular resistances vs. the coupling cap are pretty straight forward.

[antonis]

Not that I would be capable of the analysis, but I'm interested.

https://en.wikipedia.org/wiki/Diode_modelling#Shockley_diode_model
(especially "Small-signal modelling" paragraph..)

[m4268588]

• Clipped = 1/(2*pi*Rs*C)
• Not clipped　(extremely small signal level)
  = 1/(2*pi*(Rs+Rl)*C)

What more do you need?

[antonis]

What more do you need?

Maybe a not ideal (ON-OFF switch) diode model..

[Elektrojänis]

https://en.wikipedia.org/wiki/Diode_modelling#Shockley_diode_model
(especially "Small-signal modelling" paragraph..)

Well that's a bit heavy on the math, which might be needed for proper analysis. However that's only the diode side. I was kind of wishing for a bit more predigested and maybe something how it will translate to audio.

What more do you need?

After that slightly less ideal diode model I'd like to find out how it will affect the frequency balance and the clipping knee at different frequencies... And maybe how it works when there are several frequencies present. We have to remember that the diodes turn on and of (or change their resistance depepending the voltage) all the time inside every signal cycle. That means the cutoff frequncy would theoretically be modulated by the signal itself. And the clipping doesn't really clip each frequency present in the signal separately... It clips the combined signal. To me that interaction seems like very complex one, but then again more complex effects might be just too subtle to matter in real life.

Maybe i need to get back to circuit simulation at some point and see if I can get anything out that might be relevant. On the other hand, it would be very odd if I could find anything significant tha hasn't been analyzed to death already in such a common circuit snippet. Not that it would stop me though.

[m4268588]

I'm not write to cutoff frequency.
It's unmeaning to be interpreted this as a dynamic HPF.
Do you see what this simulation suggests? Try other diode models.

Maybe a not ideal (ON-OFF switch) diode model..

Don't need that.  And, no mention of imaginary numbers is needed now.

[Fancy Lime]

For easy understanding of frequency response, I find it useful to pretend that everything is a resistor. Some "resistors" depend on variables. Caps are frequency dependent resistors (called reactance), diodes are voltage dependent resistors with an exponential I/V curve, in this mental model. This approach runs into serious limitations for many applications but works quite well for audio signals.

The coupling cap (CC) acts like a large resistor (R) for low frequencies (LF) and a small resistor for high frequencies (HF). This means it will pass little current (I) at LF and a lot of I at HF. Diode clipping to ground means we have a voltage divider between the CC plus any additional series R, and the resistance of the diodes. The resistance of the diodes drops lower as the I pushed through them increases. The voltage (V) across the diodes is whatever the I/V curve of the diode says it is for that I.

TLDR: with a small coupling cap and diodes with a large soft knee, low frequencies clip at a lower threshold than higher ones, letting through more unclipped treble, meaning the sound gets muddy and tinny at the same time. Not normally what we are after. But with the usual diodes, say 1N4148, the effect should be very subtle unless you use a cap that is really way too small (say <100n). With hard-knee diodes like LEDs, the effect is even less pronounced. In either case it can easily be compensated by minor adjustments to the frequency response before and after the clippers.

HTH
Andy

[antonis]

It's unmeaning to be interpreted this as a dynamic HPF.
Do you see what this simulation suggests? Try other diode models

I can't see any HPF (static, dynamic or whatever) 'cause I can't see any capacitor..
(what I see is the clipping threshold forn a 1kHz signal..)

[Rob Strand]

The bottom line is there's no direct equivalence between the Cs + Rs + diode clipper and a HPF feeding an Rs + diode clipper.  (s = series parts)    If you think about the input waveform below +/-0.6V (the diode clip voltage) then the Cs + Rs isn't actually filtering, the diodes aren't clipping and the signal passes straight through.   

If we have a HPF feeding an Rs + diode clipper the HPF is present all the time even when the signal is below +/-0.6V.

If we start with a base-line circuit of Cs + Rs + diode clipper with C=100n and Rs=1k then feed +/- 2V input then you can roughly match-up the output level and the amount of distortion for input frequencies 100Hz and 1kHz with following circuit:

Mental equivalent circuit:    HPF  -->  R + diode clipper

HPF:    C=100n, R=5k1           ; actual values here not important only the cut-off f3 = 312Hz
Gain stage: 1.2
Clipper:  Rs=1.8k

It's by no means an exact match and we can't expect it to be.
I only did a rough estimate and in fact the match isn't so great at 300Hz.

However it shows the difficultly matching the real  Cs + Rs + diode clipper  circuit with
a circuit that separates out the high-pass filter function.

If you cranked the input level above 2V I suspect the whole equivalent circuit would need to
be revised and no doubt the filter cut-off would be higher than 312Hz; in fact approaching the formula m4268588 posted.

This one improves the match at 300Hz but still isn't there.

HPF:    C=100n, R=4k3           ; actual values here not important only the cut-off f3 = 370Hz
Gain stage: 1.4
Clipper:  Rs=2.2k

[antonis]

@hans h & Elektrojänis: That was what I tried to steer clear of when said " let's not dig in deeper"..

[Fancy Lime]

@hans h & Elektrojänis: That was what I tried to steer clear of when said " let's not dig in deeper"..

Antonis, have you learned nothing about your fellow forum members in all those years? 

[antonis]

Antonis, have you learned nothing about your fellow forum members in all those years? 

Andy, there is no fellow forum member with the glitz of snow123..!!

[Elektrojänis]

@hans h & Elektrojänis: That was what I tried to steer clear of when said " let's not dig in deeper"..

I think this has been interesting so far. As with any system with any complexity, there are a lot of ways to analyze it and they all bring in a different piece of the same puzzle.

I still think I eventually need to simulate it myself... Or just breadboard something and feed it some signals.

TLDR: with a small coupling cap and diodes with a large soft knee, low frequencies clip at a lower threshold than higher ones, letting through more unclipped treble, meaning the sound gets muddy and tinny at the same time.

And this reminds me of some experiments I made long ago with cap in series with feedback clippers... That gets very boomy quite quick and the distortion generated on the higher frequencies get masked quite severely by the bass if I remember it correctly.