thoughts on the design of the Schumann Two Face

[12XU2A3X3]

Schumann site states:

TWO FACE FUZZ

Two germanian transistor fuzz-face style distortions linked together at the ground rail.

One distortion controls the positive half of the wave. The other distortion controls the negative half of the wave. This lets you create very odd asymmetrical wave forms.

The TWO FACE FUZZ has two outputs which lets you have one fuzz come out of each output or you can create a mix.

A boost foot switch lets you adjust the positive and negative side to adjust how much you are going to boost.

Includes a +12 -12 bipolar power supply.

I think this thing sounds great and the basic premise has some real potential to be expanded upon. while i'm pretty certain you could achieve what's described with two fuzz face circuits and a phase inverter, what throws me is the description of "linked together at the ground rail." i fail to see how a common ground is germane (pun) to operation of the circuit. can someone enlighten me? the controls for master input gain, and individual output gain and boosts are pretty clear. are the sensitivity controls transistor bias? are the 1/n/2 switches transistor flips?

also, the split/link switch is for combining both side of the waveform into one output, not a control for series/parallel operation.

[multi_s]

'linked together at ground' may jsut mean that it is one negative ground and one positive ground circuit. both using teh same ground -> linked together. although thats basically liek saying all your pedals off a 1 spot are 'linked together' through ground so maybe it means some more magical.

[eatyourguitar]

I know what it means. "linked together at the ground rail" is a silly way of saying "two opposing half wave rectified fuzz signals merged back into one fuzzy signal"

there are two main styles of doing it. do you use comparators and VCA's? or do you use two precision half wave rectifiers with opamps and a summing stage at the end.

for a real-world implementation, check the datasheet for the LM393 comparator and the cool audio V2162 VCA.

I actually designed something for synthesizers that does the splitting of signals based on being above or below ground using TL072 precision half wave rectifiers. it processes the signals separately through two identical waveshapers and then combines the signals back together.

page 2 of this thread the tri to sin shaper is what I'm talking about.

but lets take a step back before we dive in head first. the fuzz face is a low impedance, low power audio amplifier that works sorta ok as a not so dirty audio amplifier until you actually plug in a guitar. the fuzz face loads down the guitar pickups in a non-linear way. that is where the fuzz comes from. how do you put two of them in parallel without changing the sound? you don't. impedance matching of the two fuzz face inputs would also be desirable but since we already know that fuzz face impedance is extremely non-linear, you would have a hard time getting that impedance curve right.

your better off starting with a fuzz that will fuzz buffered signals just fine and throw a TL071 buffer on the input. another consideration is that the fuzz face adds more harmonics on the top side of the waveform always. you would want to flip the input and output phase of fuzz #2 180 degrees before you start mixing half wave rectified signals back together. this will give you the maximum fuzz. more than a single fuzz face. if you don't, it will just sound exactly like a regular fuzz face and that would defeat the purpose of all this extra stuff.


a little bedtime reading http://en.wikipedia.org/wiki/Precision_rectifier

[12XU2A3X3]

dude, where to begin...

THANKS! great information. great to see another wiggler on here! will dig in and have more questions i'm sure.

[eatyourguitar]

had this idea a few days after this thread but I finally got around to posting my thoughts. maybe someone wants to work on what I started. the obvious challenge is getting the voltage drop across the diode string to match what you would like to have as a virtual ground. you are kinda working backwards from that virtual ground to have a correctly biased Q1 base with the right amount of current AND the right voltage drop while still having your Q2 collector where it should be at ~4.5v = 9v/2. in this example I assumed NPN + PNP silicon with a B E junction forward voltage 0.7 but biased at 0.6v forward. both Q1 B and Q3 B are tied together at 9v which is half way between the +18v and 0v rails. input cap is shared so impedance does not change. in a perfect world, only one transistor will be low impedance for half of the wave form at any time while the other transistor is immediately starting to get closer to B E reverse bias where the impedance becomes significantly higher. out of all the diodes, 1N4148 is probably the lower current needed to turn the diode on. there are diodes that require less but lets just say 1N4148 is better than an LED even though LED could reduce parts count with a larger voltage drop. if the diode does not turn on, we have no power supply. if we have enough current through Q1 + Q2 emitter to turn the first diode string on, we would have a virtual ground for the NPN fuzz face that sits at a voltage that will never change. same is also true for the second fuzz face. this means each fuzz face is running at a virtual supply voltage of 9.6v. they do not share virtual power supplies. all the grounds are virtual. the DC blocking caps on the input and output make this possible with referencing audio to 0v.