Frequency-dependant clipping?

[dano12]

I've been wondering about something. Would different types of diodes clip differently/better depending on the frequency range of the signal?

In other words, would it be worth experimenting with a circuit that splits up the input guitar signal into  4 or 5  focused frequency ranges, and then feed each into a different type of clipper.

Tell me I'm smoking crack here....

[Taylor]

Craig Anderton designed just such a device, except using the same clipping type for each band, around 1980 called the Quadrafuzz. Schematics and other info are available on the PAIA website as it was distributed as a DIY circuit.

It would be quite easy to build that up and then do the usual clipper swap rotary on each band, or breadboard a few types to see what you think. IMO it's definitely worth playing with GE vs SI diodes, and some other more exotic stuff like MOSFETs. Not sure the difference between 1n4001 and 1n4148 is noticeable enough to my ears.

Apart from the different diodes and extra control, you get a very different sound by clipping bands separately - the major result is a loss of intermodulation, so chords become much clearer and there's less nasty hash and beating.

Full disclosure, I am involved with a commercial device based on this concept, designed with Craig Anderton's input. So I have done this kind of experimenting and can say there's definitely a world of fun to be had with these ideas. The trick is just reining in all the controls possible with something like this. Then again, I've seen some of your boxes with 40 knobs, so maybe there's no need to keep it simple - a man after my own heart.

[dano12]

Craig Anderton designed just such a device, except using the same clipping type for each band, around 1980 called the Quadrafuzz. Schematics and other info are available on the PAIA website as it was distributed as a DIY circuit.

It would be quite easy to build that up and then do the usual clipper swap rotary on each band, or breadboard a few types to see what you think. IMO it's definitely worth playing with GE vs SI diodes, and some other more exotic stuff like MOSFETs. Not sure the difference between 1n4001 and 1n4148 is noticeable enough to my ears.

Apart from the different diodes and extra control, you get a very different sound by clipping bands separately - the major result is a loss of intermodulation, so chords become much clearer and there's less nasty hash and beating.

Full disclosure, I am involved with a commercial device based on this concept, designed with Craig Anderton's input. So I have done this kind of experimenting and can say there's definitely a world of fun to be had with these ideas. The trick is just reining in all the controls possible with something like this. Then again, I've seen some of your boxes with 40 knobs, so maybe there's no need to keep it simple - a man after my own heart.

Ah yes, the quadrafuzz. That would be a great starting point.

Has your commercial device shipped yet? Intrigued....

The trick is just reining in all the controls possible with something like this.

Lol, I am that beavis guy. Moar knobs

[R.G.]

One interpretation of the original post is "Do different diodes clip different frequencies better or worse?"

The short answer, for guitar audio, is "Not to any significant extent." There are differences in the internal turn on/off speeds of diodes, but the do not have much effect in the audio range, being mostly so fast that audio looks like slowly-varying DC to them. However, there are some power rectifiers which have such big junctions that their turn-off time at the upper end of the audio range may be slower than others. This is mostly not noticeable when clipping low power audio.

[Paul Marossy]

One interpretation of the original post is "Do different diodes clip different frequencies better or worse?"

That was the way I read it.

The short answer, for guitar audio, is "Not to any significant extent." There are differences in the internal turn on/off speeds of diodes, but the do not have much effect in the audio range, being mostly so fast that audio looks like slowly-varying DC to them. However, there are some power rectifiers which have such big junctions that their turn-off time at the upper end of the audio range may be slower than others. This is mostly not noticeable when clipping low power audio.

What kind of power rectifiers would that be? Not like a 1N4001, 1N4002, 1N4003, etc. is it?

[R.G.]

To some extent. the 1N4000 series is a little slower than signal diodes like the 1N914/4148. In general, the higher the current rating, the more heavily doped the junction is and the greater its area. Bigger area translates right into junction capacitance, and heavily doped translates into a long recombination/sweep out time. For the little 1A guys, it's not a big deal. For 10A rectifiers, it can be very noticeable on higher frequency sine waves.

[Paul Marossy]

To some extent. the 1N4000 series is a little slower than signal diodes like the 1N914/4148. In general, the higher the current rating, the more heavily doped the junction is and the greater its area. Bigger area translates right into junction capacitance, and heavily doped translates into a long recombination/sweep out time. For the little 1A guys, it's not a big deal. For 10A rectifiers, it can be very noticeable on higher frequency sine waves.

Ha ha, if you make room on a PCB for a 10A rectifier diode on your own PCBs, I guess you could mess with that and see what kind of results that yields (if anything audible).

I honestly can't tell much of a difference between a 1N4001 and a 1N4148 type. But I only hear up to about 18kHz these days...

[amptramp]

I took the OP's question to be a little different.  You can use, say, a DoD 250 with antiparallel diodes to ground and instead of connecting them directly to the signal path, connect them through a capacitor or inductor or tuned circuit.  A capacitor in series with the pair of diodes would reduce the clipping at low frequencies without affecting others.  This would get rid of the muddiness of clipping low frequencies.  Imagine a signal input with a fairly large fundamental at low frequencies and harmonics riding on this wave.  If you clip the fundamental, the harmonics disappear entirely during low frequency extremes of the waveform because the signal stays at the clipping level.  AC coupling to the clipping diodes can alleviate some of this.

[Paul Marossy]

I took the OP's question to be a little different.  You can use, say, a DoD 250 with antiparallel diodes to ground and instead of connecting them directly to the signal path, connect them through a capacitor or inductor or tuned circuit.  A capacitor in series with the pair of diodes would reduce the clipping at low frequencies without affecting others.  This would get rid of the muddiness of clipping low frequencies.  Imagine a signal input with a fairly large fundamental at low frequencies and harmonics riding on this wave.  If you clip the fundamental, the harmonics disappear entirely during low frequency extremes of the waveform because the signal stays at the clipping level.  AC coupling to the clipping diodes can alleviate some of this.

That sounds like a really cool idea to experiment with. 

[WhenBoredomPeaks]

I took the OP's question to be a little different.  You can use, say, a DoD 250 with antiparallel diodes to ground and instead of connecting them directly to the signal path, connect them through a capacitor or inductor or tuned circuit.  A capacitor in series with the pair of diodes would reduce the clipping at low frequencies without affecting others.  This would get rid of the muddiness of clipping low frequencies.  Imagine a signal input with a fairly large fundamental at low frequencies and harmonics riding on this wave.  If you clip the fundamental, the harmonics disappear entirely during low frequency extremes of the waveform because the signal stays at the clipping level.  AC coupling to the clipping diodes can alleviate some of this.

I remember posting a BMP related topic about this idea, i even breadboarded the Muff to mess with the clipping caps but then instead of 100k pots the guy in the shop sold me 100 ohm pots so the project is delayed until my next big parts order.

[DDD]

There are some opinions that to get a great sound one one should clip low frequencies (LF) asymmetrically and high frequencies (HF) symmetrically.
Also there are some opinions that LF don't need to be clipped heavily but HF should be clipped significantly... and so on depending of the style one plays.
So, if you build the following:
- input buffer with two outputs (1 FET or 1 BJT)
- two filters (LF and HF) with the volume control on their outputs (2 OpAmps)
- two separate clippers - symmetrical and asymmetrical (2 OpAmps)
- simple toggle switch swapping filters' outputs at the clippers' inputs
- mixing potentiometer connected to the filters' outputs
- mids tone control after the mixing pot (1 Opamp)
you'll get the possibility to try very interesting combinations for the cost of 5 OpAmps and 1 transistor totally.
*** Low intermodulation distortion is a bonus.

[mills]

I took the OP's question to be a little different.  You can use, say, a DoD 250 with antiparallel diodes to ground and instead of connecting them directly to the signal path, connect them through a capacitor or inductor or tuned circuit.  A capacitor in series with the pair of diodes would reduce the clipping at low frequencies without affecting others.  This would get rid of the muddiness of clipping low frequencies.  Imagine a signal input with a fairly large fundamental at low frequencies and harmonics riding on this wave.  If you clip the fundamental, the harmonics disappear entirely during low frequency extremes of the waveform because the signal stays at the clipping level.  AC coupling to the clipping diodes can alleviate some of this.

I like doing this.  I've done it in a few times in DOD250 kinda circuits, and trying it right now in a feedback loop.  I experimented a bit and usually go with LED's and a capacitor for the lows and silicon diodes for the highs.  Clearer bass and crisper clipping.  lots of variables to play around with.