smooth sounding octave?

[Bill Mountain]

At the risk of confusing everyone even further I was looking into octave up sounds to exploit a psycho acoustic issue I run into with bass reproduction.  You can't hear most of the fundamental on the lower strings (especially if down-tuned).  You hear an octave or so up and your brain fills in the rest.  I think solos and melodies would sound cool with the octave up sound but I'm really interested in hearing what happens when the octave up sound is added to a fundamental you can't hear (probably nothing).  The issue with the octave pedals I've tried is that it is nothing but a bag of farts in the lower registers and the pedals shine in the higher registers when an octave up makes a discernible difference.

This is another in a long line of though experiments before I hit the breadboard (with 2 kids this becomes harder and harder to schedule).

Additionally, I was looking for ways to do the full wave rectification method as easy as possible.  That's why I was looking at using the single opamp method in the app note.  I figured I could replace the opamp is the app note with my as-yet-to-be-determined-overdrive circuit.

[ashcat_lt]

Ring modulating the signal against itself is exactly squaring the signal, and will create the octave up, but it brings with it all the problems associated with ring modulation.

The problem with the full wave rectifier isn't really in the diodes, it's in the difference in slope in the wave at the zero crossing as opposed to at the top of the swing.  A triangle wave will full wave rectify perfectly (given an ideal diode), but a sine wave will have nice curvy positive peaks and really sharp spikey negative peaks.  Those spikes are all the high frequency nasty distortion that you don't like about these things. 

Seems like if you was to full wave rectify, then strip the DC component and gain it way up and then clip it down with diodes to ground, you should be able to come to a point where you're getting a pretty symmetrical square wave.  Would that be close enough?

[Mark Hammer]

The difficulty with any analog octave up or octave down effect is that we like our strings too damn long and too damn loose.  If we played heavy gauge strings on 20" scale guitars, it would all work out fine.

But seriously, anyone who has used any sort of analog octave up/down box will recognize that they tend not to behave particularly well until you get up to the 6th or 7th fret.  One of the reasons for this is that, as the string is shortened, going up the fretboard, it is essentially stiffened - longer length and thinner gauge associated with greater string compliance and wiggly-ness.  Thicker shorter strings have fewer artifacts and less harmonic content to confuse the octave-up/down extraction.  Keep in mind that when a string is long, applying your pick to it temporarily applies more pressure, tensioning it, and then when you lift the pick that tension is released, "re-pitching" the string.  All of that confuses the circuitry.

The higher up one picks, and the stiffer the string, the less of an effect that picking will have.  Most folks will likely note the "best" octaving coming from their B and G strings, rather than from their high E.

The problem with the full wave rectifier isn't really in the diodes, it's in the difference in slope in the wave at the zero crossing as opposed to at the top of the swing.  A triangle wave will full wave rectify perfectly (given an ideal diode), but a sine wave will have nice curvy positive peaks and really sharp spikey negative peaks.  Those spikes are all the high frequency nasty distortion that you don't like about these things. 

Actually, because of the gating action of series diodes, the "folded over" wave, will have a triangle/sine on one side, and a square on the other...an even worse distribution of harmonics.  That can be somewhat abated by use of the lowest possible forward-voltage diodes you can muster (e.g., selected Schottky types) to do the rectification, but will still be there in some amount.

I still want someone to provide an explanation of how the Foxx Tone Machine's network between the emitter of Q2 and base of Q1 helps to provide such a robust octaving effect.  Perhaps it can be improved upon if we understood its inner workings better?

[Bill Mountain]

The difficulty with any analog octave up or octave down effect is that we like our strings too damn long and too damn loose.  If we played heavy gauge strings on 20" scale guitars, it would all work out fine.

But seriously, anyone who has used any sort of analog octave up/down box will recognize that they tend not to behave particularly well until you get up to the 6th or 7th fret.  One of the reasons for this is that, as the string is shortened, going up the fretboard, it is essentially stiffened - longer length and thinner gauge associated with greater string compliance and wiggly-ness.  Thicker shorter strings have fewer artifacts and less harmonic content to confuse the octave-up/down extraction.  Keep in mind that when a string is long, applying your pick to it temporarily applies more pressure, tensioning it, and then when you lift the pick that tension is released, "re-pitching" the string.  All of that confuses the circuitry.

The higher up one picks, and the stiffer the string, the less of an effect that picking will have.  Most folks will likely note the "best" octaving coming from their B and G strings, rather than from their high E.

The problem with the full wave rectifier isn't really in the diodes, it's in the difference in slope in the wave at the zero crossing as opposed to at the top of the swing.  A triangle wave will full wave rectify perfectly (given an ideal diode), but a sine wave will have nice curvy positive peaks and really sharp spikey negative peaks.  Those spikes are all the high frequency nasty distortion that you don't like about these things.  

Actually, because of the gating action of series diodes, the "folded over" wave, will have a triangle/sine on one side, and a square on the other...an even worse distribution of harmonics.  That can be somewhat abated by use of the lowest possible forward-voltage diodes you can muster (e.g., selected Schottky types) to do the rectification, but will still be there in some amount.

I still want someone to provide an explanation of how the Foxx Tone Machine's network between the emitter of Q2 and base of Q1 helps to provide such a robust octaving effect.  Perhaps it can be improved upon if we understood its inner workings better?

Your explanation is quite interesting.  I assumed the octave worked better on higher strings/notes because it's easier to hear the difference between 1000Hz and 2000Hz than it is to hear the difference between 30Hz and 60Hz.

[StephenGiles]

The front end of the EH Guitar Rack Synth provided an excellent octave up - arguably needs a lot of circuitry to get there for many, but well worth the ride in my view.

[Quackzed]

if your looking at a dual opamp type fwr check out tim escobedo circuit snippets page, theres a dual opamp 'octup blender' that is both simple and effective as a full wave rectifier.
you can omit the 3rd opamp stage if you dont need clean blend...
here it is...

[R.G.]

Not a terrible trade secret. Yes, the overdrive and octave up are separate paths internally. This is one case where the distortion of the full wave rectification is hidden inside the harmonics generated by the overdrive.

Given what you're trying to do, Bill, I think you're going to be disappointed.

There are techniques for deriving a pretty smooth octave up from single notes. If that's what you're after, I can offer some advice.

The PLL-follower technique works pretty well for generating square waves at multiples of the frequency it's fed. The Rocktave Divider has a pretty good "shell" for picking out a note and feeding it to a PLL. This was the basis of my PLL thing I mentioned earlier.

Square waves (the extreme case of what you get from "overdrives") have only the odd harmonics of the waveform. The PLL multiplier generates the even harmonics the square wave is missing.

The only issue from there is how much "smoothness" to add in terms of even harmonics and filtering.

It is *possible* to use a 32x (fifth octave up) or 64x (sixth octave up) output from a PLL to outright synthesize a *sine wave* at the fundamental or second harmonic by using a five or six stage shift register plus weighting resistors on the output. That's about as smooth and clean as it gets.

This sounds complicated, I know. But a PLL is one chip, the divider to let you get out 1..7 octaves up is one chip; that gets you sawtooths with a resistor network to add them in the right proportions. This is how electronic organs used to work before LSI chips. The shift register to make pseudo-sines is one more chip.

So the question becomes - how hard do you want to push for your experiment? Is 2-3 CMOS chips and some resistors on top of a Rocktave divider pedal enough?

[PRR]

> a frequency 2F and a DC component that gets blocked

DC only if the tone started infinitly long ago and has not varied.

The "DC" changes on the first half-cycle, and with any level change. For a non-sine input, the level changes if overtone phases aid and null. So on a plucked string there's a lot of thumping going on.

Which may be the side-view way to say R.G.'s "intermodulate badly".

[Bill Mountain]

Thanks guys.  The PLL doesn't sound too scary.

I guess what I really want to do is get a good overdrive tone then run it through a clean octave up like the PLL or something else.

[R.G.]

I've had a few email messages and the PLL setup in the posting/schematic I noted seems to work pretty well. It might make a good starting point for you.

The craft in getting a clean-ish output at an octave up would then consist of how you used the PLL output to make the signal you want. The organ and synth guys refer to making the signal you want from other pre-existing signals as "additive synthesis", as opposed to making some waveform with lots of harmonics and then filtering out what you don't want, which is termed "subtractive synthesis" (well, duuh!    ).

If you want, I'll try to sketch out a sine wave maker from the output of the PLL multiplier. It's kind of on my interest list anyway.

[Bill Mountain]

I've had a few email messages and the PLL setup in the posting/schematic I noted seems to work pretty well. It might make a good starting point for you.

The craft in getting a clean-ish output at an octave up would then consist of how you used the PLL output to make the signal you want. The organ and synth guys refer to making the signal you want from other pre-existing signals as "additive synthesis", as opposed to making some waveform with lots of harmonics and then filtering out what you don't want, which is termed "subtractive synthesis" (well, duuh!    ).

If you want, I'll try to sketch out a sine wave maker from the output of the PLL multiplier. It's kind of on my interest list anyway.

That would be cool.  And maybe a recommended chip as well.

Thanks!

[R.G.]

Looks like the CD4015 will do an eight-step approximation to a pure sine, and a simple R-C filter will make it sound pretty good - I think...

I'm messing with this now. I'm trying to avoid using a "jellybean" inverter gate chip for some housekeeping on the 4015. Actually, using a CD40106 or CD4584 would be cleaner, as that would make getting a clean reset pulse easier. But it costs another $0.50 for the chip.

The divider needs to be reconnected so it gives out 32X the input frequency, as the shift-register sine approximation needs a 16X clock and you want an octave up. Just a trace moved.

Looks like one or two chips, both in stock at Mouser, and seven 1% resistors to do the sine approximation on top of the stuff from the Roctave Up design.