maybe another way to get a good tracking octaver

[birt]

I'm no EE and this is just an idea. but it might be usefull if someone can help me to go from an idea to a schematic.

i was thinking about ways to get a good stable square wave from a guitar. the obvious way i guess is to amplify and clip it. but there's allways filtering needed and that seems to be the hard part.

but here's the idea:
when the signal is clipped why would you use the signal itself? the frequency of the signal is how many complete wavelengths you have in one second right? that means there are twice as many zero crossings as there are wavelenths. so if i could somehow use the zero crossings to trigger a circuit that outputs perfect clean squarewaves, the output would be exactly twice the input frequency. and that could be divided to get lower octaves.

this idea has probably been used before but i have no clue on how to start designing the circuit that is triggered by the zero crossings. a 555 comes to my mind because i know that can produce decent square waves. but i guess it's not going to be that easy...

[Gus]

First thing you should do is look at guitar and bass waveforms on a scope.  This square wave stuff you sometimes read on the web makes me wonder if anyone ever looked at a guitar and bass waveform on a scope.  None of it looks like a sine wave that could be easily made to a square wave.  How do you determine when and where to detect an edge to make a square wave?

This is a hard thing to do first you need to find the fundamental in the wave form.  Say 22 frets and the open string for one string.

Maybe a bandpass filter(s) with steep ends set to fundamental range(s) then to a edge sensing/etc  circuit

Look at synth stuff for guitar and bass often one pickup per string, placed in a certain spots under the string to limit the processing needed.

That all posted look at the EH  microsynth  schematic.

Boss

Also try a EBS bass octave one of the best tracking I have heard.  I did not get a look at the circuit

[birt]

i will look at that schematic now but my point was that whatever kind of shape the waves have, the zero crossings should have the right frequency. a sin or square or triangle look completely different but if they are all 5KHz the zero crossings are on the same place, whatever the wave looks like right?

[Gus]

Look at the waveforms from a guitar or bass on a scope then tell me how you determine the zero crossing.   Simple in theory for something like a sine wave.

Now using DSP is a different thing then using op amps and comparators and flip flops.

[R.G.]

The problem you run into is that harmonics can be of a size comparable to the full fundamental. In that case, there is more than one zero crossing per fundamental cycle. That's why there's nearly always filtering on the input signal - to reduce harmonic size compared to the fundamental.

A comparator with its reference voltage set to the zero-volt average of the incoming waveform IS picking off zero crossings. It changes output polarity each time the input moves across the 0V reference.

[Gus]

IIRC There is also the midi system Peavey midibass? that used frets cut into four to electrically sense the string to fret contact for each string to fret note.  IIRC the frets were wired to a circuit and sensed with the strings touching the frets.

R.G. and me are kind of posting the same thing.  You need to extract the fundamental to do a good detection.

What do we know

the range of the finger board in notes and upper fundamental frequency
so maybe a very steep roll HP off at the lowest note and a very steep LP roll off at the high end.
The lower strings MIGHT have a higher amplitude(fatter magnetic core of the wire) so the same note/frequency but on different strings might be somewhat detectable via amplitude.

[birt]

The problem you run into is that harmonics can be of a size comparable to the full fundamental. In that case, there is more than one zero crossing per fundamental cycle. That's why there's nearly always filtering on the input signal - to reduce harmonic size compared to the fundamental.

can we solve this with a HP and LP at the lowest and highest desired note as Gus sais?

A comparator with its reference voltage set to the zero-volt average of the incoming waveform IS picking off zero crossings. It changes output polarity each time the input moves across the 0V reference.

do you have a schematic example? it will be easier to understand for me i guess.

Look at the waveforms from a guitar or bass on a scope then tell me how you determine the zero crossing.   Simple in theory for something like a sine wave.

i know what those waveforms look like. but with a push pull gainstage for example you do set a point where there is a zero crossing right?

also if there are more zero crossings than the fundamental frequency.. they are there in the original signal so they will also be divided right? does this mean the octaver can't work or does it mean the octave signal (which would be a square wave, maybe added to the original signal or filtered to some kind of sine wave) would not be musical because the frequency is not even?

[DDD]

birt,
your idea seems to be good enough. As far as I know pickup signal zero crossings are "less sensitive" to the harmonic content relatively to the non-zero voltages. I don't want to say they are ideal, but take a look at the scope...

[Gus]

http://files.muziq.be/schematics/eh_microsynth.pdf

[gez]

As Gus said, take a look at some real signals through a scope.  Triggering a comparator at the 'zero crossing' is useless as there will be dips above and below the crossing; mainly on the wound strings due to the higher harmonic content.  If anything, you need to do the reverse: sample the very peaks during one cycle and use these to set the thresholds of a Schmidt trigger, which is triggered on the subsequent cycle.  Constantly monitor the amplitude and reset the thresholds accordingly.  At least that way, you have some possibility of avoiding false triggering and isolating the fundamental.  You can then use standard doubling techniques to get an octave up.

If you do this using analogue there soon comes a point where you start thinking, "hmm, maybe I should have paid more attention to learing how to program PICs".  Analogue requires a ridiculous amount of circuitry.  PIC requires hardly anything.  I sketched out some rough code to do this, but it needs a lot of work and these days I just don't have the time.  One day...

Then again, I might just buy a POG and have done with it!

[gez]

At least that way, you have some possibility of avoiding false triggering and isolating the fundamental.  You can then use standard doubling techniques to get an octave up.

That should have been dividing techniques to get an octave down.

For octave up, extracting the fundamental isn't so critical.  The zero crossing can be used for that:

http://hometown.aol.co.uk/Gezpaton/Fuzz+Saw+Mk+I.JPG

http://www.diystompboxes.com/smfforum/index.php?topic=39254.0

No idea how long that schematic will stay there as we changed provider a couple of months ago (they haven't deleted my files yet).  The schemo is convoluted.  All you need is two parallel op-amp comparators that are anti-phase, instead of all the 40106 stuff.

[Gus]

birt

  Did you look at the linked schematic?  Look at the parts used for octave down.

Also look for older transistor schematics of octave downs some good stuff to look at.

[birt]

i am trying to understand that schematic yes :p