Octave Up

[Fancy Lime]

Hi there,

I put together a little circuit snippet that some may find useful, I hope. Its an octave up effect using an op-amp precision rectifier. This has been suggested on this forum several times and I'm sure there are similar thing out there but I've never seen or heard it implemented so I thought I'd give it a try. This was originally meant to be the oct-up part for an octave up and down pedal I was working on but I could not get the oct-down to work right, so I thought I'd share this part in the mean time.



This is not a clean octave, more of a "dynamic fuzz octave" sort of thing. That's because the wave shape is closer to an impulse than a sign or square wave. So for me it is not terribly useful on its own unless you are hunting for a unique sounding fuzz that does not squish the dynamics. It gets a bit less fuzzy and more octavy when replacing the 1N4148 silicon diodes with Schottkys (I tried BAT41's). But if one were to add some overdrive/distortion/fuzz and tone controls to both audio paths just before the mixer, this can probably be developed into a great effect for bass. My suggestion would be to use 3 inverters of a 4049 or 4069 hex inverter in each path, the first to wired as overdrive stages (like in the Red Llama, Tube Sound Fuzz or Onkotherium) and the third as a tone control. Should be good for that layered wall of sound kind of thing. You can probably leave out the fundamental extractor, I just had it on the breadboard already and thought it may allow me to better judge if the octave up is working as expected.

Have fun,
Andy

[Gus]

Good circuit to study
http://www.experimentalistsanonymous.com/diy/Schematics/Guitar%20Synth%20and%20Misc%20Signal%20Shapers/Electro%20Harmonix%20Micro%20Synthesizer.pdf

[Mark Hammer]

Actually, I thought the 3-pole lowpass "fundamental extractor" was one of the smarter parts of the circuit.  It is common lore about octave-up units that best results are obtained by using the neck pickup, rolling down the guitar tone, and picking above the 7th fret or so.  All of that is directed at giving the rectifier sub-circuit mostly fundamental to double.  So inclusion of suitable LP filtering allows for whatever follows to double content without having to do quite so much ahead of the input.  I don't know about others, but I always like to design/mod stuff such that I can have a clean bright sound, hit the stomp, and get the optimal sound of the effect without having to change anything on the guitar.

I've made a couple of FWR octave units in past (one in Elektor, Anderton'scircuit) and was never as thrilled with them as I was with the more traditional stompbox approach of half-wave rectifying complementary outputs of a phase splitter.  To be fair, I built both of those a while back, and may have simply not known how to tweak them right, so there may be more promise in them than I experienced. 

In recent years, it has become apparent that compressing the signal (something often accomplished with a simple 2-diode clipping circuit near the output) is part of what yields a more audible octave.  That, too, is part of showcasing the fundamental.  Plucked strings start out much more complex (inaccurate or unstable pitch, most harmonic content), and settle down to mostly fundamental after a few moments, such that the octave sort of "emerges from the bushes" after a few hundred milliseconds.  Allowing the fundamental to linger, "purified" for a while longer, makes the octave more audible.

Finally, my growing sense is that Schottkys are underutilized in octave-up units, and as you experienced, may yield a more pleasing (and reliable) octave.  Part of the reason is that, by having a much lower forward voltage, they do not clip the "sides"  of the wave quite so much.  They still clip, and contribute additional harmonic content, just not quite so much as Si diodes that square off the sides of the wave more.

[Fancy Lime]

Hi Mark,

my thinking for using the low pass for this (other than already having it) was pretty much to chisel out the octave effect some more. And it does that. My suggestion to leave it out was meant to mean: you can leave it out without affecting the workings of the rectifier but it will of course change the sound. If the whole thing is supposed to be used with guitar, the filter needs adapting, as it is now, the cutoff is at 330Hz; for Guitar I would suggest moving it up to 1000Hz. Maybe switchable?



Brilliant idea wit the clipping diodes, thanks. I tried a few different arrangements and found that asymmetric clipping seems to bring out the octave best. It also takes away most of the annoying/charming 8-bit or arcade fuzz character so I made them switchable. I also added a simple low pass at 720Hz for the same reason. Now it actually sounds good enough to box it. This is kind of nice: easy, versatile and somewhat unconventional sounding but very punchy. Should stand out well in the mix. I think I'm going to change the input stage to a discrete mosfet stage with gain control, though. I don't like having unused op-amp halves sitting around. Doesn't feel right.

Cheers,
Andy

[Mark Hammer]

Dino/digi2t suggested a good idea.  A simple diode pair on the output does produce the desired compression, but at the cost of introducing more harmonic content, via clipping.  Dino uses the Si diode pair but sticks a small-value resistor (1k-ish) between the diodes and ground to soften the clipping while retaining the compression aspect.  I suspect your 2+1 diode arrangement, by lifting the clipping threshold for one half of the signal, does a similar thing.

[Fancy Lime]

A variable resistor between the diodes and ground is what the zvex machine does for this very purpose. Come to think if it, its not so different an effect, just uses two frequency triplers instead of one doubler for that extra bit of mayhem. That may be a better choice for an extra control instead of a gain control, ctually. The asymmetric clipping might also benefit from extra softening, I suspect.

Andy

[amptramp]

If you like the idea of a lowpass on the output, you may want to try making the output into a current source feeding a capacitor in order to get a sawtooth waveform.  A square wave has the fundamental plus a third of the 3rd harmonic plus a fifth of the 5th harmonic plus a seventh of the 7th harmonic etc.  A sawtooth is integrated mathematically so you get the fundamental plus a ninth of the 3rd harmonic plus a twenty-fifth of the 5th harmonic plus a forty-ninth of the 7th harmonic.  You are much closer to a sine wave and you have better clipping options.

[Fancy Lime]

Hi Ron,

well, for one thing we don't have a square wave here. A full-wave rectified guitar signal resembles more of an impulse wave. If we integrate that we get something that looks a little more like a square wave. So to get reasonably close to a sawtooth, we need to integrate at least twice. For that to work, our integrator corner frequency must be close to or below the lowest fundamental, so for an octave up we want to use with bass around 80Hz. Everything above that would be attenuated with 12bd/octave. An integrator is a lowpass filter, two integrators are a second order lowpass. So this method would only work in a very small window of fundamental frequencies unless we have a very effective adaptive volume recovery stage after it (aka compressor) or we move the filter around so that its corner frequency is always at the fundamental. The latter is even more complicated than the former because it needs a frequency analyzer, frequency to voltage converter and voltage controlled filter. Possible but waaaaay beyond the scope of this little circuit.
So we need to make some trade offs. I added a passive low pass and clipping diodes. The clippers reign in the high impulses, making the fundamental come out petter because it moves the waveform towards something a bit more square-wavy (still an impulse wave but with the impulse tops cut off) but is not frequency dependent. The passive low pass has a corner frequency around 720Hz, meaning that for high notes it is an integrator, for low notes it at least takes the fizzing off.
Designing static filters to work as wave form shapers is a tempting concept. Unfortunately the wide range of input signals we are dealing with in Guitar/Bass applications forces a lot of compromise. That is much easier to do when you have a synthesizer that gives you one defined frequency per key that you can use as the audio signal and as control signal for moving around voltage controlled filters.

Cheers,
Andy

[thehallofshields]

Finally, my growing sense is that Schottkys are underutilized in octave-up units, and as you experienced, may yield a more pleasing (and reliable) octave.  Part of the reason is that, by having a much lower forward voltage, they do not clip the "sides"  of the wave quite so much.  They still clip, and contribute additional harmonic content, just not quite so much as Si diodes that square off the sides of the wave more.

I always enjoy reading your posts on Octave Up circuits.
I've probably read 50 of them over the past few years, and even though you usually make the same points, it's always interesting to think about.

In the case of the Precision Rectifier circuit, doesn't the Opamp nullify the Diode Forward Voltage?

I believe I read that while experimenting with this: http://www.lynx.net/~jc/superFullWave.html

[ElectricDruid]

If you like the idea of a lowpass on the output, you may want to try making the output into a current source feeding a capacitor in order to get a sawtooth waveform.  A square wave has the fundamental plus a third of the 3rd harmonic plus a fifth of the 5th harmonic plus a seventh of the 7th harmonic etc.  A sawtooth is integrated mathematically so you get the fundamental plus a ninth of the 3rd harmonic plus a twenty-fifth of the 5th harmonic plus a forty-ninth of the 7th harmonic.  You are much closer to a sine wave and you have better clipping options.

That's a triangle waveform you're describing, not a sawtooth. Unless saws have triangular teeth around your neck of the woods, I suppose!

Your point is still the same though - integrating to get a triangle reduces the harmonics. As you said, much closer to a sine.

Tom