Has this distortion idea been used?

[Sirsnitty]

This is my first post, and I'm rather new to the diy effects game. I've tinkered with electrics for brief lengths of time on and off for about 15 years. I've lurked these forums long enough and I figure it's about time to take part.

I'm familiar with how a simple clipping circuit works, and its shortcomings in that they just usually happen to have a solid state sound. Ideally, we'd want a more smoothed and random clip to better emulate a tube clip.

So here's my thought.. instead of simply sending the signal through a pair of clipping diodes, send it through a pair of oscillating circuits that oscillate between two clipping diodes for each circuit. Each oscillator would ideally oscillate at the input frequency to prevent "sweet spot" frequencies.

·Input signal gets split.
·One signal is used as audio signal
·Other signal is used to run two oscillating circuits.
·In place of each clipping diode is the oscillating circuit that switches back and forth between two diodes each.
·Each signal crest would control the oscillator to switch the signal path to either of two diodes, back and forth, every other crest.

Does this make sense? It theoretically could create an asymmetrical distortion that uses up to four diodes (two each path) instead of just one for each path. I suppose Forrester stages could be added to further randomize the clipping.

Possible? And if so, ideas on how this can be achieved? Or... maybe even not worth it? Or is this something that already exists?

[Blitz Krieg]

i don't know if it has been done...

You might look at the MXR Envelope Follower and CMOS switches.

good luck.

[robthequiet]

 I think it has, in a way, by having a circuit that takes the wave, clips the top in one stage, then clips the bottom in the next stage. If you wanted to split the signal to have the wave clipped by different diode pairs, it should not be too hard with a phase splitter with parallel distortion circuits, then a mixer/buffer at the output. In my opinion, the use of three stage JFET or MOSFET overdrive boxes come closer to a tube sound than diode clipping pairs, but a lot of the sound shape comes from the filtering capacitors and the amount of gain between stages, as well as in the art of biasing the transistors. One project to look at might be Run Off Groove's Thunderbird or the Brown Sound in a Box.

[Sirsnitty]

Thanks much my friends. If I get a few hours free this week, I may take a shot at it. Worst case, I lose $4 in parts

[Transmogrifox]

What you are describing doesn't sound like something to emulate a tube amp distortion but it does sound like it could have interesting results and well worth experimenting with it.

I have done circuit simulations of ideas where I put the signal on opposite sides of the head-toe diode.  One path has significant gain and the other does not.  As the path with gain starts to clip it clips against the same shape of the signal so you don't lose the original harmonic content on the signal peak.

As I thought about that I realized that's exactly what the Tubescreamer type of circuit does in a much more simple way.

However, the more brute-force method I described gives more flexibility to perform variations on the theme -- different filtering for the two paths, the possibility to invert one of the paths to get more of a fuzz sound, being able to dial in a gain relationship from path1/path2 so that the amount of "clean" compressed signal on top is adjustable.  Also could change the bias between the two ends to adjust the symmetry or 1/2 wave rectify the low gain signal to get the "clean" compression on one polarity and hard clipping on the other side.

Just that I have never breadboarded this to see what it sounds like.

[Sirsnitty]

Yeah I guess "tube emulation" wasn't quite the term I should have used. I think I would have been better to say "less diodey"?

At this point, I'd like to try it but I can't quite think of how I'd make the oscillating section

[PRR]

> oscillator would ideally oscillate at the input frequency

This is difficult.

> every other crest

This is much less difficult.

> I can't quite think of how I'd make the oscillating section

Nor can I.

Maybe draw the wave you expect out.

[Sirsnitty]

Ok yeah I'll draw it out in a bit, after I get this transmission pulled out of my vehicle.

If there's a way to send every other wave peak to a different diode, I guess that would create the same effect. But I'm talking every other positive wave as well as every other negative wave.

Let's say "up down up down" would be one cycle. Let's call four different diodes A, B, C, and D. A and B would be positive voltage diodes, C and D would be negative voltage diodes. Up down up down (one cycle) would be sent to A C B D respectively.

[samhay]

If you are going to drive this circuit with the output from a guitar, you will quickly learn that guitars do not output nice clean sine waves - they get close to this if you play harmonics on a single string, but a cowboy chord strummed with a big fat pick is going to give you something else entirely.

[Sirsnitty]

Haha in aware. I've done a bit of digital editing.

But that's actually beneficial to my idea. The back and forth clipping would be even more sporadic, creating an even more asymmetrical effect

[robthequiet]

The oscillator concept suggests something like a phase lock loop that tracks frequencies, but as mentioned above you would expect a guitar signal to be too chaotic. The idea of perhaps having a separate oscillator in the circuit that might modulate a phase splitter to swing between two paths is possible? As you say, you want to avoid sweet spot frequencies in case of resonant squeals or ending up with a ring modulator. Maybe one approach would be to say what happens with a single note as opposed to a chord.

[Quackzed]

> oscillator would ideally oscillate at the input frequency
This is difficult.
   > every other crest
This is much less difficult.
   > I can't quite think of how I'd make the oscillating section
Nor can I.

could you possibly use a flip flop? that seems like it could work,low passed input signal to drive it? ,  though i'd guess at some point  it'd find the frequency far faster then its flip could flop... 
sounds interesting . i'd wanna drive one low threshold diode pair against a higher threshold diode pair, see if you get the difference or what... buth more and less distortion at the same flippin time. cool

[Transmogrifox]

Actually I think some of the octave-down circuits derived from the input signal could be of benefit.  The basic strategy is this:
1)  Low-pass filter the  audio signal
2)  Amplify it into  something approximating a square wave.
3)  Feed it into the CLK input of a D flipflop configured as a toggle flip flop.  This gives you the frequency divider effect.

For your purposes you can modify this strategy slightly.  You use the audio signal to shift a logical "1" around a shift register in a circle.  For 4 diode pairs this can be done with a quad package of D flip-flops.  It's a bit of a logic design but the general idea is to use each rising edge of the audio signal to switch to a different clipping threshold.  The trick is to detect the zero-crossing and switch diodes before the signal is large enough to clip.  That way you aren't making the switch while a diode is conducting.

You could actually do this with a single diode pair.  You use the output of your shift register, or logic "thing" to change the voltage on the other side of the pair so the clipping is always changing symmetry.

Then to take this idea further, you use something like a 12-bit shift register and a couple of XOR gates to create an LFSR (https://en.wikipedia.org/wiki/Linear-feedback_shift_register) that is clocked by the audio signal.

If the LFSR is modulating the clipping threshold for each 1/2 cycle then you get a sequence that takes so long to repeat that there won't be any periodic modulation.

OTOH you might like it to be periodic every other cycle because that will give you a sub-harmonic element to the sound.

Just some ideas.  Main idea is using the sub-octave type of circuit for some ideas how to modulate the symmetry:
http://www.valvewizard.co.uk/uboat.html

[Rixen]

How about clipping against an envelope follower, so it always clips at, say, 80% of the peaks regardless of amplitude..

[Sirsnitty]

@Transmogrifox ...
It's like you are right inside my head.. except, you have way more understanding of how to do it. My experience with logic circuits dates way back to about four days ago. So, I'm not entirely sure how that all works, but I know enough to say that everything you said is exactly what I'm talking about.

@Rixen ...
Envelope follower... I'm not familiar, but I'm interested

[Keppy]

I once breadboarded a circuit that only clipped every other cycle. It sounded terrible.

It's been awhile, but I think what I did was similar to the ideas here:

1) Run the signal through a gain stage
2) Split the signal, one to diodes and one to a comparator
3) Run the comparator signal to a 4013 flip flop to divide by 2
4) Run the comparator output to a MOSFET gate

The clipping is done by diodes to Vb after the gain stage through the MOSFET (drain to diodes, source to Vb). Thus every other zero crossing in the signal turns clipping on/off.

Like I said, it sounded TERRIBLE 

[Sirsnitty]

Every other zero crossing, or every other cycle? Every other zero crossing (I think) would be just like only using a clipping diode in a single direction. I could be wrong here, though

[Keppy]

Every other crossing turned clipping on or off. So one whole cycle was clipped top and bottom, and the next was unclipped, top and bottom.

[Sirsnitty]

That's actually kinda half what I'm going for.. except the unclipped peaks would be routed to different diodes. And each diode pair would be non identical

[robthequiet]

@Transmogrifox, thx for the insight -- I don't mean to be noisy on this thread -- the idea of the LFSR is interesting if it can be tuned to pick certain harmonics to emphasize. The U-Boat sounds great, btw.