Octave Down and Five Octaves Up - and Sawtooth Waves for Smoooooothness

[R.G.]

The Rocktave excepting the 4013 divider is the top batch of stuff. The three yellow rectangles are two CD4024s and one CD4046. One of the 4024s is used to sub in for half of the original 4013. This chip provides the divide-by-two signal from the top signal processing stuff to the 4046 PLL. The 4046's oscillator is set up to run between 2.2kHz and 22kHz, and then run into the second 4024, which obligingly produces seven divided-by-two-more outputs. The fifth of these is sent back to the phase comparator of the PLL, which locks this frequency to the incoming frequency from the first 4024.

So we get one and two octaves down from the input frequency, and unison as well as up to four octaves up. More to the point, that mess of resistors takes all those frequencies and electric-organ-style makes stairstepped sawtoothed waves out of them, so that the output frequencies are not only octaves, but they're NOT square waves, and sound much more natural and smooth. From here you could do the whole mess of organ filtering for voicing, or just make the filtering/voicing you like for guitar.

The original square waves in the same range of octaves are of course available at the second 4024 - plus a third octave down! - if you lean that way. I decided not to bring that out as it really ought to be resistor-divided down to the same volume as the sawtooth outputs, and that's another two resistors per note, or another 14 resistors on an already crowded board. But it's possible, and left as an exercise for the student...  

NOTE - THIS SCHEMO WAS REPLACED TO CORRECT ERRORS, SEE CORRECTED ONE FURTHER DOWN THE THREAD. I'M BEING MY OWN STUDENT.   

There is a lot more to do to this to make it into a full pedal, as you'd need to set up some way to select how much of what notes and so on that you hear. And again, it's a single-note thing. Like the Roctave whose shoulders it stands on, it would get very confused if you fed it chords.

[midwayfair]

 

What's the Compander helping with? Keeping the signal a somewhat constant level so the octaves don't go wurWURwurWUR?

[R.G.]

The compander is doing two things. One section is sensing the envelope of the input signal, and the second section is then impressing that envelope on the output signal.

This is important, and one reason the Roctave is a more sophisticated octave down than something like the Blue Box. The output of any digital-divider device is a constant-level square wave. It has very ugly sounds when the input disappears, a lot of which comes out as "spluttery tracking". The idea of impressing an envelope over digital outputs is clever in that it makes the digital-square-wave stuff sound more like a real instrument which has dynamics, but also suppresses the sound entirely when the controlling instrument/input goes to nearly zero, and shuts off the grack and sputtering.

The E&MM Melody Generator does something of the same thing by a different way. It generates a DC-ish octave, then uses the square waves to chop this between full and ground, with similar if not so sophisticated results.

[Mark Hammer]

The trick that Craig Anderton used was to set the decay time of the expander half juuuuuusssstttt a little bit faster than the compressor half.  The net effect was to produce a bit of downward expansion and suppress the sputtery sounds as the input signal hovers around the triggering point of the flip-flop.  Keep in mind that the 4013 flip-flop (or indeed, whatever chip is used to divide by going positive after X number of positive going input peaks) requires that those peaks be at or above some minimum amplitude. 

When the input signal is well above, or consistently below, that threshold, you either get consistent division, OR you get no division (silence, in our case).  The compressor half of the companding maintains the average level high enough that one gets consistent triggering of the flip-flop.  Of course, reducing the dynamic range of the signal in a way that provides consistent triggering for a while, post-pluck, also means that the signal will be hanging around in the almost-but-not-quite zone for a longer period than normal, too.  This required more aggressive use of the expander half to not only restore original dynamics, but to exaggerate those dynamics a teensy bit to make the period, when the signal just might trigger the flip-flop every once in a while, seem to go away faster or not even show up at all.

I guess a useful way to frame it is to note that the divider chips are essentially digital chips, co-opted for analog purposes, that expect digital levels.  Our strings can be coaxed into providing digital levels, but eventually fall back to analog ones; at which point the digital chips refuse to co-operate.

Incidentally, that maybe-and-maybe-not portion of the signal when the string's vibration occasionally produces a very brief peak that exceeds the flip-flop's trigger threshold is also the very same thing that delivers "envelope ripple" in things like compressors and autowahs.  Maybe our brains apply a kind of "tachometer algorithm" to the whole signal and treat it as if there is a smooth decay after the string is plucked/strummed, but if one scopes the string, you'll see plenty of small brief perturbations n that overall decay.  And that's what causes the problem for both envelope-controlled circuits and triggered/divider circuits. 

[R.G.]

+1

The Roctave is one of my favorites because whether through intent or luck, it solves this particular issue well in actual playing.

I thought about putting in a lock detector to sense when the PLL was locked and the notes out were accurate, but fell back on leaving is as is.

As a side note, there are some serious issues with fitting controls for this thing. The output notes bank really needs some way to mix in X amount of each note. That's missing from the schemo entirely. Four octaves up is probably not all that useful, so maybe it could be cut back to five mix controls as against the three of the original Roctave.

[kingswayguitar]

thanks for sharing
why do these crazy elaborate circuits interest me? very cool!

[mth5044]

What the whaaat. That's intense. If only we had a source for those organ drawbars...

[midwayfair]

thanks for sharing
why do these crazy elaborate circuits interest me? very cool!

This one's actually shockingly simple for how much it does. I think it's actually less circuitry than the Pearl Octaver.

[kingswayguitar]

i may just try again since some of the building blocks look familiar. i was experimenting with compandors, flip flops, and square wave comparators of late. with some success too!
cheers

[thelonious]

If only we had a source for those organ drawbars...

It's not too tough to find terrible 70's transistor organs that people are practically giving away. Someone gave me one back a while, and I ganked all sorts of interesting stuff from it, including drawbars, a reverb pan, loads of PCBs ready for part-scavenging (unfortunately no Ge stuff in mine), and hundreds of feet of wire. It's a pain in the rear to gut it, but it's also kind of fun. The wire alone makes it worth it, IMO.

I'm not sure exactly how the old organ drawbars work, though. I haven't used them yet, and they might not work for this application.

[Mark Hammer]

Some of them will also have the smaller "cheesewheel" rotating speaker, suitable for making your own Vibra-tone.

[Eddododo]

...        ill see y'all in a few years..  =bookmark=

[R.G.]

Once again I'm back at the complexity/flexibility conundrum.

This is most flexible if you are able to select which octaves and how much of each is combined into the output. But the obvious simple way to do it, with a pot and maybe a switch for each octave, makes for a very busy front panel. Maybe that's OK - if you ever once get it running the way you like, then you can punch it in and out for a given sound. But I think that changing the octave mix is going to be tough, not something you do on stage.

If that's the way it is anyway, maybe it's OK to make the octave mix pots be almost-hidden, accessed by self-knobbed pots.

[midwayfair]

Once again I'm back at the complexity/flexibility conundrum.

This is most flexible if you are able to select which octaves and how much of each is combined into the output. But the obvious simple way to do it, with a pot and maybe a switch for each octave, makes for a very busy front panel. Maybe that's OK - if you ever once get it running the way you like, then you can punch it in and out for a given sound. But I think that changing the octave mix is going to be tough, not something you do on stage.

If that's the way it is anyway, maybe it's OK to make the octave mix pots be almost-hidden, accessed by self-knobbed pots.

What about these?
http://www.smallbearelec.com/servlet/Detail?no=1277

A row of those on top of a 1590BB doesn't look busy at all (especially because I think someone would be thinking of this thing in terms of synths or keyboards, in which case it looks positively simple). You could maybe have a master blend for the upper octaves as a whole, and use the external trims to set how much of each upper octave is there.

[Mark Hammer]

Sometimes, the optimal solution is to use a couple of well-chosen rotary switches (4P3T, 3P4T, 2P6T), pick your preferred sounds, bite your lip, forfeit a couple of theoretically attainable possibiltiies, and live with an easily usable and repeatable cluster of presets that can be gotten with a flick of two rotaries, and maybe fine-tuned with a single pot.  Fits in a box better, too.

Incidentally, you can fit a LOT of those little PCB mount DPDT push-button switches like these ones ( http://www.dipmicro.com/store/SWITCH8X8B ) in a small space.  Less wiring than with toggles, and you don't need as much space between switches to get your stubby little fingers in there.  I don't know why we don't use them more often.  The downsides are that a) they only come in up/down, not on-off-on or on-on-on, as toggles do, b) switch state can be harder to detect visually than with toggles ("up" isn't much higher than "down", no matter what button style), and c) I've only seen DPDT, with more than 2 poles being a bit of a rarity. (although I see Small Bear carries these: http://www.smallbearelec.com/servlet/Detail?no=1095 ).  The good side is that they are pretty dang cheap these days, and the button caps come in a wide array of colours that can provide "intrinsic/implied legending".

[R.G.]

If that's the way it is anyway, maybe it's OK to make the octave mix pots be almost-hidden, accessed by self-knobbed pots.

What about these?
http://www.smallbearelec.com/servlet/Detail?no=1277

That is what I had in mind.

Sometimes, the optimal solution is to use a couple of well-chosen rotary switches (4P3T, 3P4T, 2P6T), pick your preferred sounds, bite your lip, forfeit a couple of theoretically attainable possibiltiies, and live with an easily usable and repeatable cluster of presets that can be gotten with a flick of two rotaries, and maybe fine-tuned with a single pot.  Fits in a box better, too.

Yeah.  But... how do I make that work without using a uC and sending everyone running screaming for the hills? The matrix of switching possibilities gets out of hand.

Incidentally, you can fit a LOT of those little PCB mount DPDT push-button switches like these ones ( http://www.dipmicro.com/store/SWITCH8X8B ) in a small space.  Less wiring than with toggles, and you don't need as much space between switches to get your stubby little fingers in there.  I don't know why we don't use them more often.  The downsides are that a) they only come in up/down, not on-off-on or on-on-on, as toggles do, b) switch state can be harder to detect visually than with toggles ("up" isn't much higher than "down", no matter what button style), and c) I've only seen DPDT, with more than 2 poles being a bit of a rarity.

That's a good idea. The DPDT configuration actually helps, as this will let you insert an LED per switch easily without going to yet more circuitry.

One switch per octave lets you use the switch like a drawbar/rank tab on an organ.

Shoot, the only difference between this and an organ is that it's (1) one note at a time and  (2) doesn't have the selectable voicing filters - yet.

[Mark Hammer]

Don't know how many remember this, but the old Elite model Strats used a trio of push-buttons for pickup selection instead of the more traditional 3 or 5-position knife switch.  There may well have been some budget models (Bullet?) that used this as well.

In any event, the 3 push-buttons allowed one to direct as many fingertips as you wished in the direction of the buttons, and hit all the ones you wanted at once, such that a person could go from, say middle-only, directly to N+M+B, in one move.

In a sort of analogous way, the drawbars on an organ let a person grab a fistful of of them, and either push them in or pull them out.

I recently made a Jen Fuzz III clone in a 1590B with a trio of those little push-buttons to add/remove a treble-cut cap, add/lift the clipping diodes, and increase/decrease the Q2-emitter-to-Q1-base feedback resistor.  Having them all in a little line lets a person hit whatever combination they want of the three, and get an instant tone change.

I really need to make myself a little PCB layout for those switches that I can cut to whatever length, for however many switches, I want.

[mth5044]

Perhaps a potentiometer with a switch would be useful? Pull the knob out to activate the octave then twist to adjust the blend of the activated octave. Pulling could change the potentiometer turning which might be a nusance if everytime you pulled the knob you had to adjust a bit.

There are MIA Fender Telecasters that come with a push button on top of the knob that sets the picks into series. I can't seem to find a source for them at the moment, but they could be very useful in this situation.

[Quackzed]

What about these?
http://www.smallbearelec.com/servlet/Detail?no=1277

A row of those on top of a 1590BB doesn't look busy at all (especially because I think someone would be thinking of this thing in terms of synths or keyboards, in which case it looks positively simple). You could maybe have a master blend for the upper octaves as a whole, and use the external trims to set how much of each upper octave is there.

not to dampen the ideas from flowing but i agree, this is probably the cleanest/easiest solution.
sliders would be slick, but a pita to implement...
4th octave up may be 'over the top' but i'd keep it... this thing IS over the top, thats what separates it from your run 'o the mill octavers...
i'd keep it as streamlined as possible ,with optional mods or addons kept separate. ie. waveshape distortion tonestacks etc..

8)this thing is shaping up to be pretty slick at least in concept!

[R.G.]

Yep. It desperately needs someone to breadboard it.      

By the way, the schemo is clearly non-functional as shown - the reset pins on the 4024s are shown not pulled to ground.

Someone with a real Rocktave or one of the now dearly departed GGG boards could probably modify theirs with a three-chip-and-resistors board.