Taming harsness from clipping diodes.

[davepedals]

A really dumb question:  For taming a bit of harness from diodes, does it matter which direction the two diodes are pointing in relation to the cap?  In other words does this also work if first diode next to cap has cathode pointing down and the other up?  I have a few projects I'd like to add a cap to with the diodes oriented differently, thus the reason my question.  I can't see where it would matter but I would rather not heat the iron up until sure.  This is borrowed from the Briggs mod cot50:  Quite often I confuse myself with the simplest things!

[Mark Hammer]

They are only drawn that way because if one drew them as they really are - on top of each other - no one would be able to read the schematic.

So, no, it doesn't matter what apparent order they seem to be in.  If it was drawn diode-cap-diode, diode-diode-cap, or cap-diode-diode, it would be the same thing.  Don't let pictures throw you off.

[davepedals]

Thanks, Mark . I didn't think it would matter but just wanted to be quite certain before I started soldering / soldering components.

[ElectricDruid]

As Mark said, it doesn't matter. If you're trying to tame the harshness, have a look at this article too:

http://sound.whsites.net/articles/soft-clip.htm

The ESP pages are great. One of my favourite sites. Always excellent stuff.

HTH,
Tom

[davepedals]

Great info, thanks for the link, Tom!  It still baffles me to this day the fact that I have built stereo FM transmitters yet many times the simplest of things will go right past me. I tend to over-think things way too much.

[Mark Hammer]

I won't profess any deep (or even shallow) knowledge of things like fourier transforms and such, but it goes almost without saying that harshness of clipping is a function both of what you feed the clipping subcircuit, and what you "let out" of it.

So imagine something like the near-standard 10k in the signal path with a diode pair to ground immediately after.  Let's reconceptualise that 10k as potentially made up of smaller resistances.  So let's make it a 4k7 and 5k1 in series (almost 10k), followed by the diodes.  A cap to ground from the junction of the two resistors will provide lowpass filtering and alter what the clipping element gets.  A cap to ground in parallel with the diode pair will also provide some lowpass filtering (in tandem with the 5k1) and alter what the result of the clipping is.  Between the two loci of filtering, they will change the audible harshness of the clipping.

[ashcat_lt]

A cap to ground in parallel with the diode pair will...

...create a "which came first" scenario.  It sort of filters both the "input" and "output" of the clipping diodes.  The relationship between these components is pretty complex.  If you put a resistor in series after the diode junction, and then the cap-to-ground, it becomes a bit more clear cut.

[bloxstompboxes]

I won't profess any deep (or even shallow) knowledge of things like fourier transforms and such, but it goes almost without saying that harshness of clipping is a function both of what you feed the clipping subcircuit, and what you "let out" of it.

So imagine something like the near-standard 10k in the signal path with a diode pair to ground immediately after.  Let's reconceptualise that 10k as potentially made up of smaller resistances.  So let's make it a 4k7 and 5k1 in series (almost 10k), followed by the diodes.  A cap to ground from the junction of the two resistors will provide lowpass filtering and alter what the clipping element gets.  A cap to ground in parallel with the diode pair will also provide some lowpass filtering (in tandem with the 5k1) and alter what the result of the clipping is.  Between the two loci of filtering, they will change the audible harshness of the clipping.

I do remember Fourier Series and La Place Transforms. However, there is no way in hell I could do one today. I remember them taking up 2 or more pages just to answer one. Oh and if you made a mistake... you wanted to kill yourself. I did feel a sense of fulfillment when finishing one correctly though. Has been ages since then.

[PRR]

> does it matter which direction the two diodes are pointing in relation to the cap?

For our purposes (small audio systems), wires are Perfect Conductors. So things along a wire are "in the same place". Electrically the situation is this:


In practice we can't really superimpose parts (and drawing them this way is confusing). So we connect them with wire, which is the same effect. Being perfect, it does not matter where along the wire, or (within reason) how long the wire is.

We do have trouble with long wire and/or high frequency. Ideally my lamps would plug directly to the dam generator. But I live back from the road 500 feet. If I turn-on heavy loads (well-pump) the lights sag. That much wire, the imperfection is apparent. Long audio wires tend to pick-up crap from wall-outlets or even their own outputs. At radio and computer frequencies, feet or inches (even millimeters) of wire may matter.

[davepedals]

Nice explanation.

[Perfboard Patcher]

I'm not a believer in the harshness from clipping diodes.    As a matter of fact diodes clip pretty smoothly.
The 47nF cap doesn't deal with the harshness from the clipping, it functions as a pre-distortion high end roll-off.

There is something else that needs to be considered. Imagine the situation when a rather dark sounding pickup is being used and in order to obtain a good clean tone the eq on the amp is set up really boosting the highs, something like 0dB@400Hz (400Hz as reference) , 8db@100Hz, 30dB@4kHz. Now when an overdrive/distortion pedal is inserted in front of the amp you wind up having 30dB of treble boost after the distortion pedal. This eq could make it sound harsh, but it's unfair to blame it on the diodes. A solution to this problem would be to relocate the treble boost, i.e. 10db treble boost pre-distortion and 20db treble boost post-distortion (as an example) instead of the 30db post-distortion.

Lately Pete Thorn did a YT-video on overdrive/distortion pedals for darker or brighter sounding amps, you may want to check that one out.

cheers

[davepedals]

Roll-off may have been a better term for me to use in the first place, because that's what I really mean. Those shrill high notes really get to me.  I'll rarely turn the trouble up very much on my amps , choosing a more warm sound . That's really what I'm trying to get by using the capacitor.

[stallik]

Roll-off may have been a better term for me to use in the first place, because that's what I really mean. Those shrill high notes really get to me.  I'll rarely turn the trouble up very much on my amps , choosing a more warm sound . That's really what I'm trying to get by using the capacitor.

Hey Dave, you'll probably give the credit to autocorrect but I really like your choice of the word trouble in this context

[davepedals]

Good one , Kevin! And it's true... Treble can be trouble when too shrill ! Apparently my speech to text on this tablet understands only about eight out of every 10 words I say.

[jubal81]

Can't see your pic, but I looked up the COT50 schematic and there's no resistor between the output cap and clipping diodes. Try something between 10K and 47K there. It's going to reduce the volume, but even out the clipping signature quite a bit and give it more symmetry.

[amptramp]

What you have not shown is that the diodes are not ideal, they have an impedance and there is an impedance in series with the source of the signal that allows the two impedances - the source and the diodes - to act like a variable potentiometer to ground with the wiper position being set by the instantaneous impedances.  If you have a low source impedance in comparison to the diode impedance, the diodes do not do all that much and the variation of the original waveform is not that great.  Therefore, there is very little in the way of harmonics being generated.  If the source impedance is high and the diode impedance is low and the signal strength at the input to the source impedance is high enough to cause clipping to begin rapidly, if the diodes have identical characteristics, you get an increase in odd-order harmonics.  An input signal that sounds like a flute would provide an output that sounds like a clarinet where the sine-wave flute input gets converted to a square-wave clarinet output.

Odd-order harmonics are considered more harsh than even-order harmonics that appear to give a "warm" tone to the sound.  As a way to get more favourable harmonic content, some people use different diodes for up and down with different threshold voltages or a resistance in series with one of the diodes to give asymmetrical clipping which gives some even-order harmonics.

The capacitor across the diodes acts with the impedance of the diodes in parallel with the source to roll off the harmonics that are generated in the clipping process.  You select the capacitor to give the amount of rolloff you need.

[Digital Larry]

Odd-order harmonics are considered more harsh than even-order harmonics that appear to give a "warm" tone to the sound.  As a way to get more favourable harmonic content, some people use different diodes for up and down with different threshold voltages or a resistance in series with one of the diodes to give asymmetrical clipping which gives some even-order harmonics.

The ultimate in asymmetrical clipping is a half wave rectifier which many will recognize as the basis for the "octave fuzz".  These can be super-harsh!  IMO harshness is independent of even/odd considerations and mostly a matter of how "sharp" the waveform corners are.  The sharper the corners, the higher harmonics will be generated.  Harshness can be controlled by low pass filtering which does not affect even/odd balance.  I will agree that 2nd harmonic sounds "warmer" because relatively speaking it is boosting the amount of energy at low frequencies.

[pinkjimiphoton]

remember too, you can add resistance to the equation.

you can put a resistance before or after the cap, and use it as a hi or lo pass filter to "tune" the frequencies the clipping occurs

you can also put a resistance before or after the diodes as well, like jack orman's warp controls. the combination of the two techniques can give you some great results.

i dunno if i'd go as big as 47n on a clipper in a guitar circuit, i'd shoot for more like 470p to 4.7n. 47n will roll off a lot of the highs and upper mids and can make the circuit sound kinda dead.

jmo's, ymmv

check my monkey balls project. you can also add noise reduction and an actual tone control too depending on the circuit by adding a feedback path with a pot and a couple caps from the input of the stage to the center of the diode clipper.
in the monkey balls it takes the signal from before the second stage's input cap, goes to the center of a 250kpot and then there is a tiny and a small cap on either side. this goes to an led clipper (in this case) with a 470p can already there.

by doing this, i can use feedback on the signal for presence, and have some control over it by "faking" a variable cap. in the case of my circuit, the caps were like 10p on one side and 3.3n on the other... the 10p is more to protect the pot from any dc, but it does work... lets ya sweep a nice tonal range, with the added bonus of not killing all your gain... feedback can be additive i guess.

the bitch is it won't work on every circuit. so i'm still kinda monkeying with the idea, but the point is you can control the clipping and the tone with one knob this way and it works pretty good.

peace