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

[toneman]

Wouldn't there need to be a "attack" knob to mute the 4046 until it locked onto the fundamental?

(and, of course, a VCA to do the actual un-muting)

I found an easy way to to do octave up and 2x octave up was to run my guitar into 2 Behringer PS600's in series.

And it works with chords!!!! 

[Keppy]

I'm having some trouble with the resolution of the graph, and to me the whole top row of resistors off the second 4024 (leading to the +4oct pad) appears to be shorted. Am I reading it wrong?

[slacker]

No you're reading it right, looks like the connection between the right hand side of each resistor and the horizontal trace above them shouldn't be there.

[R.G.]

Wouldn't there need to be a "attack" knob to mute the 4046 until it locked onto the fundamental?
(and, of course, a VCA to do the actual un-muting)

That's what the compander does - it mutes the 4046 stuff when there is no signal.

The 4046 is used in a way that will only lock to fundamentals, not harmonics.

I thought about putting in a lock detector to additionally mute the digital outs, but then it would be additional, and hearing a glide up to pitch for a few cycles at the start of a note might be interesting if one wanted to mis-tune the loop filter.

I'm having some trouble with the resolution of the graph, and to me the whole top row of resistors off the second 4024 (leading to the +4oct pad) appears to be shorted. Am I reading it wrong?

No you're reading it right, looks like the connection between the right hand side of each resistor and the horizontal trace above them shouldn't be there.

You're reading it right, it's drawn wrong (I told you that there were some mistakes in it as drawn, right?  ).  Slacker is correct - the right hand side connection of the top row of resistors should not be there.

The horizontal lines in the schematic from the second 4024 are the square wave octave signals. The top one is the highest frequency, five octaves (i.e. two-to-the-fifth, 32x) up from the signal fed to the 4046. The signal fed to the 4046 is one octave down from the input fundamental, so the highest frequency is four octaves up from the input. The highest signal is formed from just two resistors, the 22K and 2.2K at the upper right. It is just a square wave.

The next lower octave is formed from the third octave up square wave plus a smaller part of the fourth octave up, giving a crude sawtooth, but with most of the even harmonics of the third octave supplied by the addition of the fourth octave signal. This continues down the octaves, with the lowest note being a 22K/2.2K pair, and harmonics added from every higher note by connecting in ever-larger resistances to add in the harmonics.

This resistor matrix is not my invention. I copied it wholesale from the Schober organ schematics in "Electronic Musical Instruments" by Richard Dorff. I highly recommend that book to anyone wanting to know about musical electronics. It covers a whole lot that's useful for effects ideas. It's where I learned about the auto-magic tremolo idea that I adapted to the EA tremolo.

I'll go correct the schemo. Thanks for catching that!

[R.G.]

Updated schemo, correcting (some!) errors and adding in mix/key matrix for PLL'd tones.

[electrosonic]

This resistor matrix is not my invention. I copied it wholesale from the Schober organ schematics in "Electronic Musical Instruments" by Richard Dorff. I highly recommend that book to anyone wanting to know about musical electronics. It covers a whole lot that's useful for effects ideas. It's where I learned about the auto-magic tremolo idea that I adapted to the EA tremolo.

As a side note - I found that book online here...  http://babel.hathitrust.org/cgi/pt?id=mdp.39015002089400;view=1up;seq=15

Andrew.

[R.G.]

Good tip. I highly recommend anyone who is really interested in good design of musical gear to read that, as well as Carl Seashore's "Psychology of Music".

[maartendh]

Just completed this and it works in my Rocktave. Octaves up are kind of thin, which is to be expected (clean though, and 4 up is very high), but coming down to unisono you get a full sound.

I settled for 2n2 for the 4046 cap (could not read the proposed value) after some experimenting. 

Suboctaves on the low E-string, up to about the 5th fret may not follow your notes; in some instances they will not go lower than some certain frequency, giving the same tone whatever you play - this also depends upon the capacitor selected. Mixing in or out some of the fuzz might help. Higher up on the neck there are no problems. Different capacitors may improve the low end, but then you will find the highest notes (high E, 10th or 12th fret and up) will get into difficulties. In general the high E-string can have difficulties, the B-string might experience problems up from the 5th or 7th fret. These are my preliminary observations. Different cap values can also have a more pronounced effect on the sound itself: spluttering sound (if that is the right word) or a sound reminding of the synthesizers of ELP in Peter Gunn, or some kinds of oriental music, a kind of portamento, I guess.

But then again: there is enough spread in bandwith to have some fun with this - and maybe can be improved further.  Concerning the mixing stage: I ended up connecting the 22K/2K2 points of the octaves straight to the 2u2 cap on pin 14 of the 570 (no pots and no 10 K's) in order to get the volume level more in line with the clean signal (from the first stage of the Rocktave). Mixing in Fuzz and/or changing the tone control can have some serious impact.

Maybe the resistormatrix can be improved also: the higher octaves could profit from some more gain (graduall diminishing values for resitors when going to higher octaves?) - but that is also a matter of taste, and of your personal Robinson-Dadson curve, of course - and yes, I am that old.

The real fun will be in doing some serious filtering of the signal we now have obtained (thanks to RG!), so we can create some guitar-unlike sounds...

Maarten

[R.G.]

Just completed this and it works in my Rocktave.

Hey, thanks for verifying that, Maarten.  My bench time is *very* limited, so it might take me a long time to get to the actual breadboard.

Suboctaves on the low E-string, up to about the 5th fret may not follow your notes; in some instances they will not go lower than some certain frequency, giving the same tone whatever you play - this also depends upon the capacitor selected. Mixing in or out some of the fuzz might help.

Ah. That's a design mistake. For quickest tracking, I limited how low the VCO in the PLL could go. Making the value of R26 bigger than 470k would help this. R26 can be open, and that lets the VCO go as low as DC.

Higher up on the neck there are no problems. Different capacitors may improve the low end, but then you will find the highest notes (high E, 10th or 12th fret and up) will get into difficulties. In general the high E-string can have difficulties, the B-string might experience problems up from the 5th or 7th fret.

That's probably also my choice of limiting the VCO for quicker tracking as opposed to wider range. It's tunable.

These are my preliminary observations. Different cap values can also have a more pronounced effect on the sound itself: spluttering sound (if that is the right word) or a sound reminding of the synthesizers of ELP in Peter Gunn, or some kinds of oriental music, a kind of portamento, I guess.

Tuning this involves messing with the loop filter on the PLL. It can be purposely misadjusted to cause slow tracking and glides, or burbling overshoots and wobbles as well.

But then again: there is enough spread in bandwith to have some fun with this - and maybe can be improved further.  Concerning the mixing stage: I ended up connecting the 22K/2K2 points of the octaves straight to the 2u2 cap on pin 14 of the 570 (no pots and no 10 K's) in order to get the volume level more in line with the clean signal (from the first stage of the Rocktave). Mixing in Fuzz and/or changing the tone control can have some serious impact.

Ah. Good point. The resistor mixing for giving non-square waves at the octave outputs drops the level. Probably the simplest thing is to adjust the gain on the opamp after the second compander section I'll have a look at that. It needs to add some gain for the synth notes, and drop the gain from the dry signal a bit.

Maybe the resistormatrix can be improved also: the higher octaves could profit from some more gain (graduall diminishing values for resitors when going to higher octaves?) - but that is also a matter of taste, and of your personal Robinson-Dadson curve, of course - and yes, I am that old.

I would probably also do this in the mixing into the compander opamp.

Thanks for the verification. I'll go look at some changes along those lines. The big, huge, critical thing is getting it to synch in and track with the notes. The other stuff is tuning.

[maartendh]

Posted some raw sound samples in dropbox for you to hear.  Straight from Rocktave+ into soundcard, same notes on the same strings every time, only changing the selected octave, going from -2 octaves up to +2.  Second sound sample is with some fuzz from the rocktave added in, and maybe I changed the tone to bright on the Rocktave.
With any octave selected, you can also mix in other octaves, but this I did not record. It is useful though, as the higher octaves sound more harsh (they are very imperfectly shaped as sawtooth  - that is why I stopped at +2), but these higher octaves do sound good when mixed in with the lower octaves.

I also included some photographs and a drawing of the layout I used, which may help you save some time in making a compact pcb. I used 3 wirebridges, which I probably did not include very clearly in my drawing.


https://www.dropbox.com/sh/qk9t2n74klg1cim/xKBtahWfky


Maarten

[R.G.]

Very nice. I probably do need to do an add-on for a real Rocktave.

I'm somewhat disappointed in the roughness of the sound. I expected better. I think I may have made some mistake in picking resistor values for the adder networks. Maybe it just needs some additional filtering to take the rough edges off.

But boy, is the original Rocktave a great base for a PLL note manipulator. The tracking and immediacy of the notes is great. And a very nice, compact layout indeed.

Here's what I did:

This is the whole Roctave plus the additional octaves in a form factor to fit a 1590BB. The dark purple outline is the size of the 1590 BB box outside, the pink rectangle is the board outline.

There are 8 onboard pots, the four from the original Rocktave and four more for the additional octaves.  There's room at the bottom for a footswitch, battery, LED, etc. It's very much a work in progress, as I only had time to complete it to the point where I was sure it WAS complete-able. (there are known issues in there, folks)

When I got through with that, I thought, why not do another one, with just the original Roctave and not the additional four octaves. This was largely a matter of subtracting the top four pots and their associated resistor matrix, so I did one of those as well, and moved the four additional Rocktave pots up to the top row.

I'm hoping I get some time to finish this up to a state where a paint-by-numbers builder can deal with it. You're clearly way past that level! 

[maartendh]

Well, the roughness of the sound also depends on many variables, which I do not control as of yet.... that was why I referred to "raw" sound samples, I think I heard nicer sounds as well in my experiments, but I did just some recordings to give an impression of the possibities, not of the nicest sounds as there still are a lot of things to be improved;  you will get a better sound by mixing different octaves, e.g.; probably some filtering will also help. But the basis is there, and I agree that it is a very good starting point. The tone control will make a marked difference, it goes from very dark to very harsh, and I don't recall to which position it was set at the time.

In these recordings the 4046 cap was 1n. I changed the resistor on pin 11 to 10 K, and the one on pin 12 to 2M7. This works ok for the moment.

As to decreasing the straight sound volume / increasing the volume of the octaves I will try and see if the problems will be solved by lowering R22 to 33K or maybe 22K (increase C5 to 470n?) , increasing R18 to 33K and increasing R17 a bit more, to 39K or 47K.

Maarten