I've made a couple of octave pedals now and i want to understand how they work, so does anyone know of any good sites for info
thanx
octavia-related octave pedals use full-wave rectification as a means to double the frequency and thereby raise the frequency by an octave. it's interesting in that it's very similar to the first stage in converting AC power into DC power in old-fashioned power supplies. i'll try to explain.
diodes (also called rectifiers) are devices that only allow current to flow in one direction. they are extremely similar to one-way valves in plumbing. you put them into a signal path and only positive-going current can go through, unless they are wired in backwards... in which case only negative-going current can pass.
transformers are devices that take in alternating current (any audio signal, essentially) and transform it to a higher or lower voltage (or the same, if that's how they are designed). they do something else as well... they completely isolate the input side from the output side electrically so you can do anything you want with the ground potential on the output side.
octave pedals often use a combination of these two devices, one transformer driving a set of diodes (either two or four) to bend the waveshape in the following manner.
once the signal is isolated by the transformer, the diodes are used to "steer" the signal so that positive-going humps are all sent one place, and negative-going humps are "magically" transformed into positive humps as well. the resulting waveform as it exits these devices has spikey-looking lower humps that result from the waveform "folding" and doubled-up upper humps. the total number hills is doubled because of the original upper humps being combined with the flipped lower ones, and the total number of valleys is also doubled (although the waveshape on that side looks pretty pointy.) because diodes lose a fraction of a volt each and the overall amplitude is cut in half by the reflection trickery, the signal can end up rather small and have poor sustain unless the support circuitry and the transformer input/output ratio is selected correctly.
Thanks, Z - that was a great explanation. I never quite understood it myself until now.
"-)
I'll add some "gravy" to it (or nuts and a cherry, whatever your preferred food analogy).
One of the things you will often see in octave-up circuits is something called a "phase splitter". There are different ways to do this, but the most commonly observed one in these pedals will be a bipolar transistor with identical value resistors between collector and V+, and between emitter and ground (e.g., 10k and 10k). This will yield equal amplitude signals at the emitter and collector that are in opposite phase - whatever goes positive at the emitter goes negative at the collector.
Okay, suppose, looking at a scope, we were able to "shave off" one half of the waveform (and the obvious way would be, as Zach mentioned, using a diode, since it would only pass signal going in a particular direction). If this were a sine wave, what you'd see on the scope would be bump-gap-bump-gap-bump-gap-etc.
If we shaved off the opposite half-cycle, we'd see gap-bump-gap-bump-gap-bump-etc. View these on a dual-trace scope and you'd see the bumps of one lined up where the gaps of the other one are, only "pointing downward" (i Is the other half-cycle after all). Mix them together, and you end up with the original signal.
Okay, now suppose you were able to *invert* one of those before you combined them. That is, the "bumps" now pointed in the other direction. When combined, you would now have bump-bump-bump-bump-bump-bump, or double the number of bumps in the same "space". You have now officially doubled the frequency of the bumps.
Okay, back to the phase-splitter. Since the emitter and collector outputs are equal and opposite phase, if I take those outputs and run each of them through a diode pointing in the same direction, I end up chopping complementary half-cycles. If I combine these two altered signals, I will now have bump-bump-bump where each half might have had gap-bump-gap or bump-gap-bump.
In some instances, you will see the "signal shaving" via discrete diodes, where in other cases you will see it accomplished through transistors effectively used as diodes. Pretty much the same deal across different designs, though, whether you're talking the Green Ringer, Fender Blender, Superfuzz, assorted Shin-Ei Superfuzz cousins or the Foxx Tone Machine. Certainly one of the things that is important to achieveing a more obvious octave up is the relative balance of the two "halves". This is partly accomplished by the use of the equal-outputs phase splitter, but it is also achieved by things like matching the resistors and diodes coming off the emitter and collector (see JC Maillet's "nulling" trimpot mod for the Green Ringer) and also "balance trimmers" in some of the later model Superfuzzes. I had the pleasure of restoring a Royal/Shin-Ei Fuzz-wah this past weekend. This uses what is essentially the Superfuzz circuit. It lacked a balance trimmer and I have to say the octave tone was pretty weak compared to others I've heard.
There are certainly other methods for separating half-cycles so that they can be recombined to yield bump-bump-bump instead of bump-gap-bump/gap-bump-gap. Tim Escobedo's "Octup" circuit is one example. It does not use the discrete transistor phase-splitter, but it *does* use diodes to separate half-cycles early in the circuit, in preparation for recombining.
Another method you'll find in the Anderton "Octave Fuzz" and some of the magazine octave fuzz projects you might run across (like the Elektor Octave Fuzz), use a full-wave rectifier circuit, based around a single op-amp, a pair of diodes and some resistors, which does all the "folding over" for you. I've tried both of these, and I have yet to hear anythng with an octave signal as strong as what I get in my Foxx clone. That may be weaknesses inherent in the circuit design, but those weaknesses may also be the absence of compensating components that are already present in the discrete phase-splitter designs (hence open to improvement).
Finally, when I mention full-wave rectification (as does ZV), this may ring a bell. One of the things that compander chips do is derive a FWR signal for the purposes of driving their internal gain cell for either expansion or compression purposes. Several published magazine project/articles have exploited this signal for the purposes of "squeezing" both compression and fuzz from a single NE570/571 chip. The AMS-100 trigger-gate-envelope module found in the first issue of DEVICE (http://hammer.ampage.org ...scroll wayyyy down) also taps a "fuzz" signal from the NE570.
Hope this helps to augment your heads-up on octave-up.
This link should bring back some memories.
http://www.univ-valenciennes.fr/GDR-MACS/local/Montpellier/www.lirmm.fr/ftp/fxfaq.htm
Was surprised to see it still around. Here you will see an ascii (digicams were too expensive back then) drawing of how the rectified waveform from an octave up pedal should look. Lots more intresting stuff to checkout also.
Note the creator and contributers listed.
thanx for the info everyone
All excellent posts; thank you!
This is also required reading and was written ages ago!
Read up on Distortion 101.
http://www.geofex.com/effxfaq/fxfaq.htm