FUzz, as i understand it,..

[Mugshot]

..is square-wave clipping.


*pardon the crude definition (and my obvious ignorance  ), but somehow i feel i have to let out my basic "knowledge" about fuzz circuits, expose it to more knowledgeable people, and learn more in the process. coming from a purely language-inclined course, i have certainly no solid foundation on electricity and stuff like that so i am really at a loss how circuits work. but dont worry (that's me telling myself), i am struggling to learn everyday. in the past 3 months, i have already built some regulated PSU circuits, several Tonebender circuits (silicon = a breeze; germanium = painful!), si FFs and some tweaks, a simple octaver, LPBs, and a bunch of other fuzzes. fuzzes are easy builds though  that i am now quite confident when i build one, and my soldering skills are improving, yay! definitely one of the first things to learn*


so back to fuzzes. correct me on these points, but this is how i understand these babies:


in a two transistor fuzz circuit for instance, i understand that it is designed to "clip" at a certain threshold, that any more signal (sine wave?) that enters that circuit will be clipped accordingly; hence, a boosted signal will be clipped more squarely, while softer signals get clipped less, with the clipping threshold remaining unchanged whether there is more signal that enters it or not, and probably explains the compression and clean up magic inherent in these circuits. i think this applies too to most other gain-based circuits no?


am i right to assume that diode-based clipping also works on the same principle?


yes i know that i need to read and learn more, that's a given  and i have scoured the net for info (re: RG's FF article and some at HC and others) so the point here is the practical and personal approach of our esteemed members here.

ah, fuzz on?

[JKowalski]

Yes. Transistors clip with saturation, which means that the transistor is trying to amplify the signal, but it reaches a point where it can't go any higher.

Diodes clip when the signal is 0.7v (depends on type) higher then what is on the other side of it, by shunting current through it.

It ends up working the same way, but it's not the same thing.

[Mugshot]

Yes. Transistors clip with saturation, which means that the transistor is trying to amplify the signal, but it reaches a point where it can't go any higher.

Diodes clip when the signal is 0.7v (depends on type) higher then what is on the other side of it, by shunting current through it.

It ends up working the same way, but it's not the same thing.

so there. good to know i "somehow" understood the principle. poor transistors 


----------------------------

you mentioned that transistors clip with saturation. another question:

is it right to assume that higher gain trannies have high clipping thresholds while low gain trannies exhibit the inverse? because if that is so that'd mean low gain trannies hit the threshold early on. in theory im not liking it. idk. 

[km-r]

Fuzz: 1-2 clipping [very]hard clipping stage, with flat equalization before clipping.
Tube Preamp: multiple transistor stage, trebly pre equalization, and interstage equalization.

so there. good to know i "somehow" understood the principle. poor transistors 

you mentioned that transistors clip with saturation. another question:

is it right to assume that higher gain trannies have high clipping thresholds while low gain trannies exhibit the inverse? because if that is so that'd mean low gain trannies hit the threshold early on. in theory im not liking it. idk. 

AFAIK, higher gain trannies have the tendency to sound distorted-er because they can amplify the signal far beyond the clipping region than low-gain ones...

uhm we need a graphical presentation of this...

[JKowalski]

This is a simple view of saturation. When you have a 9V supply, you can't amplify anything above 9V, right? And you cant go below 0V. You don't have a voltage source above or below those points. That's where the saturation will occur. So the clipping mostly has to do with your power supply limitations (known as headroom).

High gain transistors arent any different with their saturation properties. BUT. They do have more current gain. That means in the same set up, they will try to amplify the signal (lets say a pure sine wave) MORE. So more of the signal is in the saturation range, and it is cut off. The more of the sine wave is cut off, the more square it will look at the output. As km-r elegantly put, they will sound more distorted-er!

So it's not differing saturation regions, its a difference of how much signal is in the saturation region.

[petemoore]

  A wave goes down a channel, and there's a wall that comes down from the ceiling, any water above the threshold [bottom of wall^] gets 'chopped off', the rest of the water flows and all the peaks that are below the 'ceiling wall' pass by unadulterated.
  You put in a higher gain transistor is like making larger waves, the peaks that are cut off by the same 'ceiling wall' are a greater portion of the entire wave, so the wave gets seriously altered, and still nothing above 'X' gets through.
  So you can vary the size of the wave, and control how much gets cut off that way, make it small enough and the whole thing passes under to the other side of the threshold in it's original shape and form.
  And you can vary the 'ceiling' at which they are clipped to allow larger peaks to pass before they are clipped.
  Just barely clipping the tips off the peaks adds some harmonics, the event of clipping start and clipping stop, which of course occur just before and after where the peak is [er was] = more events in x amount of time...or..more complex harmonic overtones.
  Smashing a huge signal into a low threshold clipping stage can mangle the waveform so much it doesn't resemble its original self at all, and can sound seriously distorted, to the point of becoming ''unintelligible noise''. 
   Boosting the wave so more peak is chopped off is about the same thing as lowering the threshold...so more wavepeak is chopped off.

[Mugshot]

hold it there people! *head explodes* 

i appreciate the replies guys, the descriptions kinda give me a picture of what happens in a typical fuzz circuit. correct me if im wrong but to me it means that low gain trannies can barely amplify the signal such that it barely touches the "ceiling" as Pete put it, right? so low gain trannies offer more harmonics in the sense that only the peaks are chopped unlike high gain ones wherein more of the signal gets amplified, and - in a similar circuit - therefore looks more square. pardon my slow comprehension guys, but this whole thing is really fun!

question though:

1) if low gain transistors arent available (provided we are using the same circuit), why dont we just raise the "ceiling" to simulate (not sure about that term) clipping of the peaks of the signal generated by higher gain transistors?

2) how do we determine the "ceiling"? im assuming it is based on the circuit design, no?

3) can starving the circuit (i.e. lowering the power supplied) take me into lower gain territory since the transistors can no longer amplify much of the signal? or does starving the circuit affects the clipping threshold too?

4) when higher gain trannies are used, you say that the signal looks more "square". ah, octaved signals anyone?

pardon me for asking too many questions guys. im really frustrated because there is absolutely no one within or area that i can talk to and discuss about these things (sure there are techs here, but they are interested in other things *sigh*). you guys rock! 

[JKowalski]

You got some harmonic ideas wrong.

Every waveform can be created with sine waves. This is called a Fourier transform.

For example - we will look at a square wave. You can actually make a square wave out of sine wives - how you do this is you get your "fundamental" sine wave frequency, and then add an infinite amount of it's odd harmonics to it (sin(3x), sin(5x), so on) with diminishing amplitudes. Try it on a calculator, or just look at this animation:



It shows you how it becomes more and more like a square wave as the number of odd harmonics added to it increases.

Now - when we create a square wave in the electronic sense (such as with clipping), obviously we arent adding a bunch of sine waves together. But you can do a Fast Fourier Transform (FFT) on your signal, and it will show you a graph of all the signal strengths of the different harmonics required to make that signal. (theoretically, if you did it that way)


What this all means has to do with how your ear percieves it. A pure sine wave, without any added harmonics, will sound very boring to the ear. IF you add the second harmonic (sin(2x)) to it, or any even harmonic, it will have a more pleasing sound. If you add a odd harmonic to it, it will sound harsher and more grating.


With a fuzz, we are creating square waves out of our guitar signal. A FFT analysis comparing the input to the fuzz and the output will show that all of the third harmonics in the signal have been greatly amplified (because the shape is now more square!). Thus, the sound becomes MUCH more harsher to the ear, and we have our distortion.

so low gain trannies offer more harmonics in the sense that only the peaks are chopped unlike high gain ones wherein more of the signal gets amplified

OKay. When you only clip a little bit off a signal, the number of odd harmonics you are getting will only be a little more, because it's not affecting it that much. The more square it gets as you clip it more and more, you keep adding strength to those odd harmonics, and you keep getting more fuzz.

Since the guitar signal will mostly be composed of even harmonics (primarily the fundamental frequency, and the second harmonic (sin(x)+ a diminished sin(2x)), its just how the strings vibrate), clipping barely anything off will mean you have most of the pleasing guitar sound intact with a slight bit of added odd harmonic distortion. If you clip a ton off, you mostly have odd harmonic distortion.

4) when higher gain trannies are used, you say that the signal looks more "square". ah, octaved signals anyone?

Square does not mean octaving. To get an octave, you need to generate a signal with twice the frequency in hertz of the guitar signal. If you recall what I said earlier, generating a square wave creates ODD harmonics, not even. Octaves in musical terms are only even harmonics. An octave up is always twice the frequency.

How octaving is usually done is by taking a signal (lets say a perfect sine wave again) and full-wave rectifiying it. If you recall, that signal will be above ground for a period, then below ground for a period. What a full wave rectifier does is it makes that part giong below ground inverted, so it also goes above ground instead. Therefore your wave makes it's middle-up-middle-down-middle wave cycle at TWICE THE FREQUENCY. I hope you can see what I mean here.As you notice, the signal is not around ground anymore. It is around the middle of the new rectified signal.



Transistors can also be set up to do this full wave rectifying thing too, though it is more dependent on the strength of the input signal and generally less pronounced.








If there's anything here that you didn't get, id be happy to try and expand upon it.

[BAARON]

1) if low gain transistors arent available (provided we are using the same circuit), why dont we just raise the "ceiling" to simulate (not sure about that term) clipping of the peaks of the signal generated by higher gain transistors?

2) how do we determine the "ceiling"? im assuming it is based on the circuit design, no?

2) The ceiling, in this analogy, is the limits of the power supply.

1) I won't speak to what effect increasing the power supply would have on the sound of the circuit, because that'll end up raising all sorts of questions (I think)... but rather will point out that from a practical standpoint, there are lots of cheap, low gain, easy to find silicon transistors that you can use if you're trying to build a Fuzz circuit with a low-gain sound, which simplifies the matter and allows you to use a standard circuit and power supply.

PM me if you want more info on building cheap/easy/smooth-sweet-sounding Silicon fuzzes and I'll send you the info I've collected for my own builds.

[Mugshot]

wow! 

i AM so bookmarking this page! 

responses like these keep me up on my toes, and i must say JK's animated graphics really really helped - vivid (although i must admit i must do some required reading still  ) i am certain anyone can build fuzz pedals, but i guess it is necessary too to understand what actually goes on inside a fuzz box right?

im really grateful for the responses guys, thanks a big lot!