Octaves and diodes

[Alex C]

Hello.  I'm interested in a clean octave pedal, either up or down, clean meaning just the octave, not the octavia-type fuzz and pseudo-ring-modulation.  I'd like to do some experimenting/breadboarding to try to understand how octaves work.  

I've read about a pedal or two that gives the Octavia fuzz without the octave, and it said something like "just leave out the diodes,"  or there was a switch to select between the diodes in the circuit (+ octave) or diodes out of the circuit (no octave).   I'm probably misunderstanding something or many things, but I have to wonder- it's not that simple, is it?

At GEO there is an article about the Rocktave divider, and simplifying it I believe, but I didn't quite grasp the concepts involved.

Could someone explain to me the basic principles of octave circuits, or point me to a reference that could clarify some things for me?  I'd really appreciate it.

Alex

[R.G.]

To start off with the answer, there are very few ways to make a clean octave, almost all of them are digital, and all of them are quite complicated. There are no simple *and* clean octave setups. Sorry.

Here's something about octave effects

First let's start with what an octave is. The human ear recognizes any exact 2:1 relationship between two signal frequencies as special. So to make an "octave", we have to do something - almost anything - to make a frequency twice or half of whatever signal we got to make the ear recognize the octave sound.

The human ear is also a pretty good spectrum analyzer. For it to recognize an octave sound, you only have to have the 2:1 signals in there somewhere. The ear will pick it out of even a slew of other harmonics and even non-harmonic junk (like from a ring modulator) or even raw noise.

So - some ways to generate octaves UP
- doubling/ring modulators and multipliers generate octaves by taking the signal and multiplying it by itself. This effectively converts a pure sine wave to the pure sine wave squared, which, if you do the math, contains the sine of twice the original frequency. On anything with two pure sine waves, you get twice each subcomponent, and also the sum and difference of the two components. So for a signal of F1+F2 frequencies, you get 2F1, 2F2 and F1+F2, and F1-F2.  For more than two, you get the sums and differences of all of them.

Guitars produce fundamentals, strong second, third, fourth, fifth, and many more harmonics. This is why you get the ring modulator sQuArK when you ring/double it.

- full wave rectifiers, either transformer/diode (Tycho Octavia) or all signal based (mayer Octavia, SuperFuzz, Blender, Foxx Tone Machine) all do this by picking out whichever half of an AC signal is (for instance) positive going, and inverting it so it's negative going.  This is a crude, brute force method that makes for a slew of other harmonics of the base signal if the signal happens to be a sine wave. If the signal is not a pure sine wave (no guitar signal is) then there is also a slew of sum-and-difference stuff that's not harmonically related and sounds either bell like, metallic, or hideous, depending on how you feel that day. It actually makes a multiplier/ring modulator sound smooth. It's cheap and easy to do, which is why that's what is in almost all octave up effects.

- Second order distortion circuits; I only know of one, it's my MOS/JFET Doubler scheme. What you do is set up a circuit with known second order distortion and nothing else, and arrange to cancel everything except the distortion.  It's about as good as a multiplier/ring modulator, but much, much cheaper.

- Phase locked loops; PLL's generate a new signal that is a true new signal, that can be made to be one octave up, and can be made to be an almost pure sine wave, or with more difficulty can be made to be other waveforms. What it can't be is an octave up version of the real signal.

- Frequency shifters; there are some complicated analog circuits that can make pure frequency shifts. They are complicated enough that they don't play much in the guitar world.

- Digital. The masterpiece. Maybe. You digitize the signal, and in real time play out each half cycle twice as fast as it went in, repeating the half cycles as needed to fill in properly. Works great, almost. It's a true frequency conversion up by 2:1, mathematically perfect. Except that no real physical process does that. The tailoff of harmonics in any real instrument comes from the acoustics of the real instrument. If you play an octave up on a real guitar, the harmonic tailoff spectrum remains imprinted with the physics of the real instrument. Digitally doubling it makes it sound as if it's coming from a toy guitar one acoustic octave up if not done cleverly.

Octave downs:
Octave down can only be done four ways - flipflop divider, PLL, frequency shifter, and digital.
- flipflop divider; this is what everybody does for effects. You preprocess the signal to find out where the circuit thinks it crosses the 0V (or signal midpoint) and uses that to trigger a flipflop that flips up on one signal, down on the next. This makes a square wave of exactly 1/2 the frequency of the original signal if you did everything right. It can be shaped, filtered, etc, but will always sound a bit buzzy. It's cheap, fits in a pedal, and that's what the Roctave Divider, Blue Box, and all the commercial stompboxes that do octave down do.

- PLL; the PLL can do better. It can be set up to phase lock to the original signal, but to also generate square waves of several octaves up from the original as well as an octave down from the original. These artificially generated harmonics can then be mixed into a more pleasing signal than the raw flipflop. Organs do the harmonic mixing part of this, and I really think that a pedal could be done, as I've had to sketch out most of the circuit in my heat to type this.  It has the potential to make an octave up, or down, or even musically related intervals to the original signal. Very interesting. But not clean.

- Frequency shifters can work. Too complicated.

- Digital again; read in the signal, space out the samples twice as far apart, filling in the holes before playing them out. Good, but complicated.

Yell where it's been confusing.
Copyright 2003 R.G. Keen Permission for viewing on diystompboxes forum granted. All other rights reserved.

[Marek]

Hi!

There used to be some website which dealed with this subject. Unfortunately I don't have the URL anymore. :-(

On the website, there was a filter circuit to make the guitar signal almost pure sine signal, so that when you double the freq., it doesn't sound 'rough'. (as far as I understood ) :-)

Never tried it. Dunno if it works.

Greetings,
MM

[Marcos - Munky]

I have a schematic here of a filter that if you use two and one LP and HP filters you'll get a clean octave up. I didn't trieds this. Tell me your e-mail, I send to you the schematic and the soundclip.

[B Tremblay]

There used to be some website which dealed with this subject. Unfortunately I don't have the URL anymore.

Is this it?
http://www.geocities.com/SunsetStrip/Studio/2987/octave.html

[ExpAnonColin]

Unfortunately, it's pretty hard to get very nice sounding perfectly clean octave without going digital-in which case I'd personally just buy your average Boss or Digitech  to save yourself a headache.

Something that I find isn't all to hard is converting the input to a different waveform (square, sine, triangle, etc) and halving the frequency, which is quite a bit easier than something that pitch shifts a complex waveform such as the guitar signal.... But you wanted clean.

R.G. really already got it all , go figure

I actually had a VERY long discussion with MorePaul on the HCforums about building octave.

-Colin

[Alex C]

Hey, thanks guys, (and especially R.G.), that's great info!
I've got another question.

Last night after posting this question I was looking at Tim Escebedo's Circuit Snippits and I saw the Rambler, a simple octave-fuzz.  It uses an LM386 with two diodes and a few capacitors to make a crazy octavia-like sound.  I quickly breadboarded it (very low parts count), and it sounds decent, I'll have to tweak some things and probably add a booster stage to the input, but I like it.   I have built the Smash Drive, which is almost identical to this (because of the IC used) in layout, the only difference is that pin 2 on the rambler is connected with a diode to the input (variable with "octave" control), which also goes through a diode to the input (pin 3), whereas on the Smash drive and Little Gem etc., Pin 2 goes to ground.  

I don't have a datasheet for the LM386, so I don't know what the function of pin 2 is.  Is it an inverter?  I understand that what is causing the octave effect is from adding whatever is coming from pin 2 to the input signal.  Does this sound correct?

Are the diodes there creating a FWR like R.G. said and on the site linked above?  I don't quite see why the diodes are necessary otherwise- maybe the one on pin 2 to keep the input from directly connecting to pin 2, but why the other one?
I would once again appreciate any help.  Thanks!

Alex

[hg81]

Thanks, RG, for the excellent explanation. I kind of just dived into this thing for a school project. It took me a couple of weeks of experimentation and calculation to figure out the ideas that you just posted. Someone named Gus posted this octave up schematic on another thread ( http://www.diystompboxes.com/pedals/ousb.gif ). I've tried building it, but the signal just does not go through (even after passing it through a pre-amp). Would you mind looking at it, and telling us briefly what the circuit actually does (it looks like a FWR scheme, but I'm not sure), and whether there is anything I need to add to it to make it work. I'm pretty new at this stuff, so I don't necessarily know what the authors of these schematics assume I know. Thanks again.
-HG

[R.G.]

The OUSB

The first opamp is a variable gain noninverting stage, gain of one to 4000 or so. The second opamp is inverting, gain of one. Both are biased to the midpoint of the 9V power supply by resistors from the + input to the bias supply V 1//2.

The two 10uF caps couple the two opposite phase signals into the 33K/germanium diodes. This is indeed a full wave rectifier. Both diodes are turned on by the 100K to +9V trickling current through the Ge diodes to the 33k's to ground. The current is really close to 9V / (100K + 33K||33K) or about 77uA total. Of this, about half goes through each diode, and it is turned slightly on.

When an incoming signal appears, one diode's cathode is pulled down, the other up. This turns on the "down" diode harder, and the signal appears at the anode side of that diode, as the only place the extra current can come from is that 100K resistor and the signal out capacitor. The other diode's cathode is pulled up by the reverse-phase of the signal, and turned further off.  So a negative-going half cycle of the signal appears at the output 0.47uF capacitor.  On the opposited half cycle, exactly the same thing happens, but to the other diode(s) and so *both* half cycles appear negative going at the 0.47uF cap.

From there the 1N4148's clip off anything bigger than +/- 1.2 to 1.4V peak, and the signal level can be controlled by the output pot.

If yours didn't pass signal at all, it's highly likely that something is mis wired. There's nothing inherently wrong with the circuit as shown.

For instance - both opamps should have about 4.5V on all pins except 4 (0V) and 8 (+9v).  If that's not true, either the opamp is bad, or the components are hooked up wrong, or there's otherwise a flaw in that circuit. The voltage at the cathodes of the two Ge diodes should both be about 1.27V. The joined anodes should be about 0.1 to 0.3V higher than the cathodes. The output voltage at the top of the output pot should be 0.00V. If the DC voltages are not like that, there's a hookup or component bug, or a short/open/etc.

[hg81]

I think that's going to be very useful. I really appreciate your help. Can I ask you one more question ... what is the role of the cap at the input?

[C Bradley]

I've done some thinking along the lines of a clean octave too, and think I've found a fairly simple method.

A full-wave rectifier takes a waveform and doubles the signal and then filters it. remove the filtering. To get the two opposite phase wave forms to feed the two diodes, use a transistor phase inverter and then use two germanium diodes (for the low forward voltage characteristics) to rectify the signal, giving you double the frequency. To compensate for the loss of signal in the diodes, use an opamp or FET for a clean boost and the added benefit of high input impedance.

It's a fairly simple circuit. If you can find any flaws, please point them out so I can give it further consideration.

Chris B

[AllyP]

Have you looked at the Shin EI?  It seems to be a clean octave down pedal....

http://www.fortunecity.com/tinpan/humperdinck/223/effect.htm

I'm not sure what method it uses to produce the octave down(R.G.?) but the sample sounds great.

You might want to think again about the switching and amp section tho......

I was gonna use a single dpdt to switch in the octave and add a "blend" pot for the amount of wet and dry signal using one of the panning circuits from RGs article "Panning for fun!"

Anyhoo....I'll stop rambling,


Ally

[brett]

Octave - I feel like I've been playing with then FOREVER.  After biulding several and dreaming up a few of my own, I've come to these conclusions;
enjoy the dirty octaves
you'll never get clean octave unless your guitar puts out a sine-wave signal
:? ring-modulator and multiplier circuits will always have non-harmonic "junk"/"crazy tones" no matter how much you filter and tame them
to my ears, the rectifier circuits that use transistor-based phase-splitting sound better than the transformer-based circuits (this may be  due to the assymetric phase separation - one phase dominates the other, often by a ratio of 2:1 or even 3:1)   This means there's a fair bit of fundamental in there, but as RG pointed out, the human ear is quite sensitive to the presence of some octave sound.

I like the various Roger Mayer circuits and related octaves.  Some are now more complex than the older circuits without sounding any better (in my opinion).

If you're interested, there's a transformer-based ring modulator (the Dalekator) and a transistor-based octave (Octavia) at my website.

Have fun!