Fender Blender Octave Generator Circuit

[mimmotronics]

Hi all,
I'm wondering what the two diodes are doing in this Fender Blender circuit:
http://www.geofex.com/FX_images/blender.gif

Not the hard clipping diodes, but the ones feeding back into the collector and emitter of the 3rd transistor, respectively. I know they have something to do with the Octave generation, but I'm stumped as to exactly how they are functioning....anyone wanna shed some light/take a guess at how the circuit is operating there?

[antonis]

3rd transistor is a phase splitter..
Collector gain is unity, 'cause Collector/Emitter resistors ratio is unity..
Emitter "gain" is also unity (for the sake of simplicity)..
Collector and Emitter outputs are 180^o out of phase..
So, for signal positive waveform, Emitter output is "blocked" by diode facing to Emitter while Collector output is "permited" by respective diode..
For signal negative waveform, Emitter output is "permited" while Collector output is "blocked"..

In other words, a full-wave rectification occurs, doubling signal's frequency, hence resulting into octave (up) effect....

[imJonWain]

Hopefully this makes sense.  Q3 "splits" the signal into 2 versions that are of opposite phase with each other (same as superfuzz, etc).   The diodes into Q4 act to combine these 2 signals and double the frequency (The superfuzz uses a differential amplifier instead).

Antonis beat me to it!

[ElectricDruid]

Yep, +1 what they said.

You've got out of phase signals coming from collector and emitter of Q3, so by half-wave rectifying them and then mixing them back together, you get a full-wave rectifier. It's worth comparing this circuit with the op-amp "precision fullwave rectifier", which does the same thing in a slightly different way, cancelling the input signal with the half-wave rectified signal, instead of combining two half-wave signals (which would also work):



From https://sound-au.com/appnotes/an001.htm

Once you start to understand how these full wave rectifiers work, there are only a few ways to do it. Now you've seen two of them!

[mimmotronics]

Amazing explanations guys. Been combing through a lot of octave fuzz schematics the past couple hours and it seems like this is one of the most common ways to get that sound. Thanks a ton, ya'll are great!

[antonis]

Once you start to understand how these full wave rectifiers work, there are only a few ways to do it. Now you've seen two of them!

Let's see a 3rd (all time classic) one..
Also implemented to some vintage pedals..

[ElectricDruid]

Thanks Antonis!! That is indeed a classic, could be considered "the original" since it comes from the Octavia, which must (??) be the earliest example of an octave fuzz, or any kind of octave pedal, come to that.

[Mark Hammer]

The Foxx Tone Machine, and several other octave-up fuzzes use the same approach.  The two complementary outputs of that third transistor produce a positive half-cycle where the other has negative.  If you block the two negative half cycles (which is what the diodes do), and then combine the result, you get two positive half-cycles where there used to be one.  That's how you get a frequency-doubling, in theory.

As I learned from experimenting with the Tone Machine, and several other octave-up units, the circuit produces plenty of fuzz on its own, thanks to the front end.  The so-called "hard-clipping diodes" add a bit of distortion, but their primary purpose is to to be a simple limiter that maintains a relatively constant signal level for a while. 

The challenge with ALL octave-up fuzzes is that a plucked string starts out with gobs of harmonic content, all of which is doubled and tends to obscure/hide the octave.  Once the string starts to relax, the fundamental starts to stand out more, and so will the octave, as a result.  Some might describe the octave as "blooming".  But that string relaxation is accompanied by a rapid decline in amplitude, such that the emerging octave seems to be fading out just as you get to hear it.  The hard-clipping diodes make that emerging octave seem to stick around by reducing the dynamic range of the doubled/rectified signal.  The "normal" Green Ringer lacks those diodes.  I added some to mine, and they improved the audibility of the octave noticeably.

One of the other things to note is that the series diodes in the Blender, Tone Machine, and others produce what is termed "crossover distortion".  They don't slice the waveform surgically at the midpoint to yield the seamless"folding" of the wave we see in the diagrams shown here.  Rather, they simply prevent any portion of the wave from being conducted UNTIL that half-cycle exceeds the forward voltage of the diode in question.  The result is a small gap of nothingness between each successive "hump".  With silicon diodes, that threshold is going to be a little over half a volt, and about 1/3V with germanium.  I like to use Schottky diodes in their place, with forward voltages of >200mv, so that less of the rise and fall of the wave is blocked.  YMMV

Why mention this?  Well, in the Tone Machine, and Fulltone Ultimate Octave, and probably some others, the octave is eliminated when one of the diodes is lifted.  Do the same mod to a Fender Blender, and you'll have the choice of not only blendable octave-up but blendable at-pitch fuzz as well.  I carried out the mod on a Prescription Electronics COB clone recently, and it works like a charm.  HOWEVER, there is the matter of crossover distortion introduced by the one diode that remains in-circuit.  What I've done for Tone Machines I've built is to use a 3-way toggle.  One side position connects the 2nd diode for octave-up.  The middle position lifts that diode to eliminate the octave, but still leaves the other diode in place.  And the other side position leaves the r2nd diode disconnected but connects the two ends of the remaining diode, effectively taking it out of circuit such that the entire wave is allowed to pass.

The caveats: 
1) Using the entire wave and not just the positive-going half-cycle will be nicely fuzzy, but also noticeably louder than the octave-up setting, so that toggle side-position is something that will require a volume adjustment, and is not the sort of thing you want to make a footswitch. 
2) Use of the entire waveform will sound pleasingly fuzzy but a bit less harsh, with the tonal change depending on the diodes used.  Some may like it, and others not.  If one uses Schottky or germanium diodes, the change in tone between diode and no diode will be modest, and maybe not even audible to some  (it DOES fuzz a lot).  If the diodes used for rectification are 1N914 silicon types, there will b enough crossover distortion introduced that the difference between diode and no-diode WILL be audible.  That's not a caution, but rather a statement of what yields a viable tonal choice and what won't.

Many different effects are produced by splitting the input into two different "copies" of the signal, manipulating one copy, and then combining them again.  Although the manipulation is not quite as sophisticated or complex as what one gets in a flanger, chorus, delay, Uni-vibe or phaser, the basic logic is the same.  That 3rd transistor ("phase splitter") yield two opposite copies of the input, and the diodes select what portion of each is to be combined to generate the octave-doubling effect.

[antonis]

Just to add the fact that rectifier diodes orientation doesn't matter, as long as both of them face towards the same direction..

[Keppy]

From https://sound-au.com/appnotes/an001.htm

I love that article. I actually designed an octave fuzz based on the image below the one you posted:

[ElectricDruid]

From https://sound-au.com/appnotes/an001.htm

I love that article. I actually designed an octave fuzz based on the image below the one you posted

Me too! It's a fantastic article and taught me a lot, and I've built several octave fuzzes based on precision rectifier designs^*.

*Some better than others!

[iainpunk]

One of the other things to note is that the series diodes in the Blender, Tone Machine, and others produce what is termed "crossover distortion".  They don't slice the waveform surgically at the midpoint to yield the seamless"folding" of the wave we see in the diagrams shown here.  Rather, they simply prevent any portion of the wave from being conducted UNTIL that half-cycle exceeds the forward voltage of the diode in question.  The result is a small gap of nothingness between each successive "hump".  With silicon diodes, that threshold is going to be a little over half a volt, and about 1/3V with germanium.  I like to use Schottky diodes in their place, with forward voltages of >200mv, so that less of the rise and fall of the wave is blocked.  YMMV

don't the circuits have DC bias resistors (R9, R10 and R13 in that Tone Machine schematic) to prevent this crossover distortion from happening?
in my experience, such octave generators with bias resistors work even with no pre-fold gain, even with un-matched diodes, without gating much at all.

hmm, now yall made me want to build an octave fuzz. bet a single opamp and diode can do the deed.

cheers

[antonis]

bet a single opamp and diode can do the deed.

Make diodes two, to be sure..
http://generalguitargadgets.com/wp-content/uploads/simple_oct_up.jpg

[GibsonGM]

Oh great, that uses the same Radio Shack transformer I've had in my junk bin for 20 years.  Think I'll toss one of those together

[radio]

David Morrin has also a summary of octave up circuits with a bit of history https://sites.google.com/site/davidmorrinoldsite/home/trouble/troubleeffects/octave-up

[iainpunk]

hmm, now yall made me want to build an octave fuzz. bet a single opamp and diode can do the deed.

owkay, i lost that bet to myself 
diodes and emitter followers are less ideal than i expected

best i can do is 2 opamps and a single diode, and it loses half its amplitude.
opamps are LM358, as they operate quite well close to ground

it works IRL as well, as long as there is no input capacitor, the diode arrangement moves the bias over time

cheers

[Mark Hammer]

One of the secrets to pleasing octaving is to rid the input signal of as much harmonic content as one can.  All of that harmonic hash tends to obscure the doubled note, like having too much dry ice and smoke makes it hard to see the band.  So, for me a filter stage of some kind at the front end is crucial to octaving, regardless of what doubling technique one uses.

[m4268588]

https://www.falstad.com/circuit/circuitjs.html?

[iainpunk]

https://www.falstad.com/circuit/circuitjs.html?

im in awe, i hadn't seen this topology yet, but its amazing.

cheers

[amptramp]

There are other ways of getting a doubled signal but there are a few issues using them.  There are trig identities where the angle can be replaced with angle = 2 * pi * frequency that go like this:

cos ( 2 * Frequency) = 2 * cos² ( Frequency) -1

If you have an analog multiplier, you can multiply a signal by itself and get the square of the signal minus a constant which is a DC value you can block with a coupling cap.  The problem with this approach is you get a lot of intermodulation that is decidedly unmusical between harmonics signal and partials which are non-harmonic but nearly harmonic.  For example, in an acoustic guitar, the second, third and fourth partials are slightly flat and this mixes with the perfect harmonics of the multiplied signal to provide a hashy mess of an output.

Fullwave rectifier circuits provide a much better output in a musical sense because the intermodulation is not as obnoxious.