guitar sine wave synthy pedal!..????????

[deadastronaut]

this has very likely been thought of already, or flogged to death ... but anyway..

is there a circuit/ic maybe, that will take a guitar signal in, and put out a cleanish sine wave?....(like a pure tone-ish)

is this possible...without the need for synths,pic's,hexpickups etc....

i messed around with a fuzz that put out a strange trumpety almost synth sound..(unbiased)
and i quite liked it..this got me thinking...

any thoughts on this?

rob.

[Taylor]

It's certainly possible, and EHX had a real guitar synth in the 70s that put out a sine wave. The schematics are available for that, if not on Mark Hammer's site, then I'm sure Stephen Giles posted it all.

But, it will always be very complicated. If you take one of the pitch-to-CV converters, then use it to control a sine wave oscillator, you have it. There's a pitch to CV converter around here by Harry Bissell I think. Tracking will never be great with this technology without increasing the circuit size to mammoth proportions.

I've done it digitally in the Spin FV1. Tracking is still iffy, but the size of the circuit is a lot smaller. This is one of those things that must be either huge or digital.

[deadastronaut]

cheers taylor, yeah i thought tracking would be a major issue with this..

so a square wave unbiased fuzz, say, cant be turned/converted into a sine?..without massive engineering involved then!..

[Taylor]

Not without a lot of processing, no.

You can use an integrator (fairly small opamp circuit) to turn a square into a triangle. That's close enough to a sine that it might be useful to you. It's also possible to use diodes to clip a triangle into something more like a sine. One thing that's a bummer about integrators is that the volume varies with frequency - low notes will always have more amplitude than higher notes.

You can of course filter a square or triangle  to be more sine-ish, but then spectral content and volume both vary with frequency, and you can't ever get a real sine wave with that technique except on one note - all others will be louder and have more harmonics, or quieter.

There are all kinds of waveshapers in the synth world for taking one waveform and turning it into another. None of them are simple.

[R.G.]

I've been looking for that for a good many years. Never found one what does it without being enormously complicated and full of special cases on what comes in. Fundamental extraction is what you're talking about, and it's a recurring problem in electronic music.

Closest I ever came was a filter bank which split the entire fundamental-tone spectrum of the guitar into frequency bands which were narrower than the distance between a fundamental and its second harmonic. About 1/3 octave does it OK-ish if you can get really steep filter skirts and no hole in the middle between them. After that, you can sense amplitude/envelope on the bands and use logic to pick the lowest band with content.

A guitar has at least three octaves of fundamentals, so you need at least nine bandpass filters with some decent performance for each, plus envelope detection for each one and logic to pick the one you want to hear.

That's for the single-note case. Double stops and chords get progressively more complicated, because sometimes we play a second interval (a third is vastly more common) and an interval of a second is only sixth-root-of-12 apart, about 12.2% apart in frequency. A minor third is about 19% difference in frequency, and that's getting better, but doing this over three octaves with analog filters is a huge task.

Another approach is to lowpass-filter the guitar output so it only has one zero crossing per fundamental cycle, convert that to a square wave, then run the square wave into the reference input of a phase locked loop (PLL). If you arrange a PLL with a VCO that has a sine wave output, then you get your sine wave directly because the loop will lock only on fundamentals if you lowpass the input enough and design the PLL loop filter well. Obviously, this is a single-note approach too.

What we really want is to get a Fourier transform, select a set of spikes from the spectrum, and let only the selected spikes through, using logic that always picks the real notes and not the harmonics. That is a tall order, even with a good DSP.

Well, maybe it's more possible and practical today. DSPs are a lot better and faster than last time I dug through them.

Otherwise, the simple thing to do is run it through a lowpass filter to get rid of some harmonics and take pot luck on what comes out of the filter.

This is another of those problems that are simple in concept, but hard to actually get running. Frankly, I came down to hex pickups being vastly simpler than building the circuitry.

Maybe someone else has a better solution.

[StephenGiles]

Yeah, build the front end of the EH Guitar Synth - as I have probably said 100 times!!! Constructors should learn to build big circuits, after all they are just several building blocks connected together. Once the obsession with tiny boxes is passed over, then the problem goes away!!!

Fast tracking is the trick which the EH Guitar Synth had. There are no short cuts I'm afraid, unless somebody produces a chip which does it all - which isn't going to happen.

[deadastronaut]

thanks for the replies guys...hmmm i knew this was going into deeper territory than i'm used too..
thought there wouldnt be a simple way to go on this..

i'll try the lowpass filter options, which seems the easiest to a novice as myself in such matters.
lots of reading to do as well i think!...

one thing i did read and saw a circuit for was using a 555 timer as a schmitt trigger, it could take
a bad signal and produce a clean one...could this be relevant in the idea of what i want to do in some mad way!
or am i way off on this?..

thanks again guys..very informative as ususal.!. rob.

[StephenGiles]

one thing i did read and saw a circuit for was using a 555 timer as a schmitt trigger, it could take
a bad signal and produce a clean one...could this be relevant in the idea of what i want to do in some mad way!
or am i way off on this?..

Yes, I'm afraid so - no easy way to do this.

[slacker]

http://www.diystompboxes.com/smfforum/index.php?topic=84872.20 The Electrax circuit mentioned in this thread, and other places, works pretty well for single note stuff, still has the problem of the second harmonic creeping in as the note decays, but if you set the threshold at which the PLL samples and holds the note right you avoid this. The original circuit doesn't give you a sine wave but you can generate one from it, using additional circuitry.

I'll have to try the EH Guitar Synth circuit sometime as well.

[deadastronaut]

http://www.diystompboxes.com/smfforum/index.php?topic=84872.20 The Electrax circuit mentioned in this thread, and other places, works pretty well for single note stuff, still has the problem of the second harmonic creeping in as the note decays, but if you set the threshold at which the PLL samples and holds the note right you avoid this. The original circuit doesn't give you a sine wave but you can generate one from it, using additional circuitry.

I'll have to try the EH Guitar Synth circuit sometime as well.

cheers ian, thanks for the links, interesting, electrax sounds good, did you ever finish your slacktave to the point of where you
were happy with it...was it still monophonic?...sounds like a cool experiment!. rob.

[Nasse]

There was complex project in ETI magazine, used MF10 filter chip but was it filtering original guitar signal, tried google but no success

[Mark Hammer]

Doesn't the Boss DF-2 Superfeedbacker produce something that could be redeployed for this?  Or is the wave produced by the 4046 PLL unsuitable in some fashion?

[slacker]

I think you probably could use that as a starting point Mark, I'm not sure exactly what the final outputs from the PLL are, looks like they lowpass filter the squarewave and add some envelope shaping, but there could be more going on.

Just remembered this amazing thread from Markus W http://www.diystompboxes.com/smfforum/index.php?topic=55075.100 go to page 6 for breadboard pictures and sound clips.

[deadastronaut]

wow that is clean too...nice..

thats some breadboarding too...

thats exactly what i was talking about on page 3...nice and clean, wonder how it is fast though!...

[DDD]

Severe filtering - say 1-10 Hz pass, then compressor, then filtering and compression again, then noise reduction.
Totally 5 OpAmps and 5-10 trannies. Why not?

[EATyourGuitar]

a 24db/octave LPF with pitch tracking is all you need. or even a 48db/octave filter with a LM386 after. the compressor may also be useful at the end of the chain.

with ^this you wont ever have the problem of the osc latching on to the second harmonic, since the harmonic is mostly filtered. no osc needed. just filters with pitch tracking.

[StephenGiles]

a 24db/octave LPF with pitch tracking is all you need. or even a 48db/octave filter with a LM386 after. the compressor may also be useful at the end of the chain.

with ^this you wont ever have the problem of the osc latching on to the second harmonic, since the harmonic is mostly filtered. no osc needed. just filters with pitch tracking.

And for your next joke

[Gurner]

with ^this you wont ever have the problem of the osc latching on to the second harmonic, since the harmonic is mostly filtered.

This would be great plan .....*until* you then play a note that is twice the fundamental of the note you've just previously played - which even with a hex pickup will happen at least twice along the length of each string! The follow on higher note would be filtered out - it wouldn't exist for the fundamental extraction aspect of the circuit.

no osc needed. just filters with pitch tracking.

Also how can you track pitch, if you can't pull out the fundamental in the first place?!!! (which is what everyone is saying here - ie it's difficult to pull out a fundamental!)

[StephenGiles]

with ^this you wont ever have the problem of the osc latching on to the second harmonic, since the harmonic is mostly filtered.

This would be great plan .....*until* you then play a note that is twice the fundamental of the note you've just previously played - which even with a hex pickup will happen at least twice along the length of each string! The follow on higher note would be filtered out - it wouldn't exist for the fundamental extraction aspect of the circuit.

no osc needed. just filters with pitch tracking.

Also how can you track pitch, if you can't pull out the fundamental in the first place?!!! (which is what everyone is saying here - ie it's difficult to pull out a fundamental!)

It's easy to track pitch with the EH Guitar synth, what's harder is to pull folks out of the "can't build anything bigger than will fit in a small 1590 box" rut!!

[PRR]

Pure sine-wave is boring. It won't compete with other instruments. You don't want that.

If you did, guitars would be plucked with a large felt plectrum, plucked and picked-up in the center of the string speaking-length. And you'd never clip the amplifiers.

The most siney traditional instruments are flute and selected organ pipes. The flute is not very sine-shape and also has large wind-noise. The mellow organ pipes are mostly used ganged together to build selected non-sine tones.

When I worked with modular synths, the sine was mostly used as LFO modulation or into ring-mod patches. Most everything else was better starting from triangle or rectangle through a low-pass to mellow but not kill the overtones.

What you probably want is some attenuation of the 2nd and 3rd harmonics and large attenuation of the higher overtones. If you play in one octave, a 12dB-18dB low-pass at the top of that octave will give pretty mellow (maybe too mellow) sine-like roundy waves.

Tracking is major work. You may do better with a "wah" pedal controlling low-pass corner frequency: you know what you are about to play, and you can over- or under-shoot to give less-mellow or more-mellow flavors on different notes.