Three transistor OD/Heavy Fuzz

[Jack Deville]

While playing around with some garden variety OD topographies, I got frustrated and sidetracked and threw together a quick fuzz circuit.

I thought I'd share it with everybody as its cheap, easy and heavy heavy fun fun.

Slap it together on the breadboard. It'll surprise you!

EDIT: The switchable buffer at the front was added to bring up the input impedance. it adds quite a bit of noise with little audible difference to the overall tone of the circuit. I'm working on optimizing the buffer to reduce noise and allow for better interfacing with other pedals.... it is on the schematic as an "extra" not necessarily intended for use in this circuit as designed.

-jack



edited by culturejam: Original schematic removed at the request of the poster.[/i]

[jrod]

Hey, this looks pretty cool. I have been looking at a transistor book, that is mostly way over my head, that shows those circuit segments. Just regering to the first stage, I think R1 & R2 form a voltage divider and R3 forms voltage feedback. R4 & C2 form a bias stabilizer. Is this correct? Is VR1 a "fuzz" knob?

[Jack Deville]

Hey, this looks pretty cool. I have been looking at a transistor book, that is mostly way over my head, that shows those circuit segments. Just regering to the first stage, I think R1 & R2 form a voltage divider and R3 forms voltage feedback. R4 & C2 form a bias stabilizer. Is this correct? Is VR1 a "fuzz" knob?

kinda...

R1 & R2 form a voltage divider used as a bias network to set the operating point of the stage. at 9V the base of Q1 should sit at about 0.82VDC. Emitter will then sit at about 0.22VDC, collector at 4.2VDC. pretty close to linear operation.

R4 sets the collector current and C2 acts to increase gain at all frequencies that can pass through the network created by R4 & C2 (should be around 15hz).

VR1 is a fuzz control, simultaneously adjusting the amount of signal from stage 1 to stage 2 and from stage 2 to stage 3.
NOTE: this control also changes the bias point of Q2 and Q3 (although not by that much. Q2 and Q3 are still operating at or near their linear regions at each extreme of the control)!

correct me if my math/analysis is wrong.

[jrod]

Thank you for the explanation, Mr. Deville. This info is very helpful! I'm looking forward to seeing how you tweak the input buffer for less noise.

[Liquids]

Making the buffer less noisy, in theory can be done via a few subtle things...

Change transistors to something with a better noise figure--A good Darlington, an MPSA18, 2N5089 or the like. 2N3904 are more likely to add their own noise.

Avoiding the theory aspects, you can use a 'noiseless biasing' technique to get voltage the base of said transistor with a simple voltage reference (Vref)/divider arrangement using small value resistors; this will reduce 'thermal noise' etc of base biasing, compared to the 150k/220k arrangement at the base.

Run a 4.7k resistor from +V supply to your new 'Vref point;' then run 15k from that Vref to ground. Put a largish cap (10uf, 22uF etc) from the junction of those two transistors, to ground.

Now you have a voltage reference for the base bias. Run a 1M from that junction to the base. You should have your base biased right about where you want it (.7V above 1/2 supply voltage is ideal here), and with a standard input impedance value. Only two more parts used, and that stage, at least, will be less noisy.

But wait--maybe you like the base having a lower input impedance, as it currently does, which is like a 82k/100k resistor, no problem. You can run 100K from Vref to the base which will alter the tone possibly.

Maybe you like that resistor it even smaller, and it's effect on tone...47k, 22k, etc. Note that since the load of 100k (or less) resister is smaller than the 1M, and the Vref feeding that resistor to the bass is not a 'stiff' voltage supply, you'll make want to adjust the Vref resistors to compensate for changing this resistor. Measure the voltage, and you'll see it gets higher the smaller that resistor gets.

To compensate, for example, make the 15K in the simple voltage divider above a bit smaller, like 12K or 10K. Whatever pair works for the values used, the concept is to keep the base of the transistor (which is not the same voltage at the Vref junction) at about .7v above 1/2 the supply reference with a low-noise voltage divider network. For example, if you have a perfect 9v battery, 1/2 x 9v = 4.5v; 4.5v + .7v = 5.2v.

Voila, that should make things a bit quieter and functioning optimally, especially if you implement both ideas. Good luck!

[RnFR]

looks like a fun dual ganged fuzz blaster you got there. looks kind of like a higher impedance version of the gain stage in the roland bee baa. not sur if i've got the double 1M, but i can give it a try with 2 pots. are you getting any crackle on the pot? looks like it might be seeing some DC. i just ask, because i've done this same sort of thing before and didn't get any crackle, while someone else told me that i should be. i enjoy the sound of 2n222's getting the crap blasted out of them,too. it's a fun transistor.

oh, and i'm movin this guy to ready to build.

[Emanuele]

I'll add a treble bleed cap over VR1-2 something like 470p..it should sound very nice when the gain is rolled back then since 1M pot is a very big value..

[jrod]

So, is the idea here that each cascaded stage amplifies the signal and slams it into the next one, yielding fuzz?

[Jack Deville]

Making the buffer less noisy, in theory can be done via a few subtle things...

Change transistors to something with a better noise figure--A good Darlington, an MPSA18, 2N5089 or the like. 2N3904 are more likely to add their own noise.

Avoiding the theory aspects, you can use a 'noiseless biasing' technique to get voltage the base of said transistor with a simple voltage reference (Vref)/divider arrangement using small value resistors; this will reduce 'thermal noise' etc of base biasing, compared to the 150k/220k arrangement at the base.

Run a 4.7k resistor from +V supply to your new 'Vref point;' then run 15k from that Vref to ground. Put a largish cap (10uf, 22uF etc) from the junction of those two transistors, to ground.

Now you have a voltage reference for the base bias. Run a 1M from that junction to the base. You should have your base biased right about where you want it (.7V above 1/2 supply voltage is ideal here), and with a standard input impedance value. Only two more parts used, and that stage, at least, will be less noisy.

But wait--maybe you like the base having a lower input impedance, as it currently does, which is like a 82k/100k resistor, no problem. You can run 100K from Vref to the base which will alter the tone possibly.

Maybe you like that resistor it even smaller, and it's effect on tone...47k, 22k, etc. Note that since the load of 100k (or less) resister is smaller than the 1M, and the Vref feeding that resistor to the bass is not a 'stiff' voltage supply, you'll make want to adjust the Vref resistors to compensate for changing this resistor. Measure the voltage, and you'll see it gets higher the smaller that resistor gets.

To compensate, for example, make the 15K in the simple voltage divider above a bit smaller, like 12K or 10K. Whatever pair works for the values used, the concept is to keep the base of the transistor (which is not the same voltage at the Vref junction) at about .7v above 1/2 the supply reference with a low-noise voltage divider network. For example, if you have a perfect 9v battery, 1/2 x 9v = 4.5v; 4.5v + .7v = 5.2v.

Voila, that should make things a bit quieter and functioning optimally, especially if you implement both ideas. Good luck!

Fantastic notions. Looks like I've got my work cut out for me today. Thank you for the ideas. I'll post my findings.

jrod:

short answer: yes. raise the signal level till it hits the rails = hard clipping/square wave fuzz. sounds good.

[Jack Deville]

looks like a fun dual ganged fuzz blaster you got there. looks kind of like a higher impedance version of the gain stage in the roland bee baa. not sur if i've got the double 1M, but i can give it a try with 2 pots. are you getting any crackle on the pot? looks like it might be seeing some DC. i just ask, because i've done this same sort of thing before and didn't get any crackle, while someone else told me that i should be. i enjoy the sound of 2n222's getting the crap blasted out of them,too. it's a fun transistor.

oh, and i'm movin this guy to ready to build.

It took me a minute to find this thread. Search Function to the rescue. I'm not familiar with the fuzz blaster, there is one called the "master blaster" listed on this forum.... is that what your were referencing?

sadly, the dual-ganged 1M is what makes the design. two 1M could be substituted i suppose. what i thought was neat was the way that the two worked with each other.

no crackle on the pots heard, the voltage is pretty low there, i tried decoupling the pots from the bias networks, and could not detect a significant change in tone or noise and deemed the additional parts un-necessary. ymmv.

i also tried subbing 2N3904s for the 2N2222As (in an effort to reduce cost of the unit), found that the 2N3904s did not have the same smooth characteristics that the 2N2222A displayed. Ran the circuit through three amps:

Valve Jr. head -> 212 V30 cab : good frequency response across the board. thick saturated fuzz.

later 70s master-vol pull-boost twin: aggressive scooped tone. stoner hell.

traynor studio mate 48 (freshly capped w/ 2 new jensens): bottom, heavy. did i say HEAVY?

all setups tested yeilded very heavy tones. guitars used: 2007 gibson MMRI, 80s ibanez roadstar w/ dumb hot blade humbucker. i preferred single coils for the increase in treble response. the humbucker seemed a bit more wooly that i like, but that could probably be overcome by tuning the first stage for higher frequency response.

sustain was approaching 9 seconds.

[Jack Deville]

modified the input buffer as follows:

voltage divider setup as: +9VDC 7.5k -> 10k -> GND

470k from voltage divider to base of 2N3904. 15k emitter resistor used.

no notable decrease in noise....

more results to come

[RnFR]

looks like a fun dual ganged fuzz blaster you got there. looks kind of like a higher impedance version of the gain stage in the roland bee baa. not sur if i've got the double 1M, but i can give it a try with 2 pots. are you getting any crackle on the pot? looks like it might be seeing some DC. i just ask, because i've done this same sort of thing before and didn't get any crackle, while someone else told me that i should be. i enjoy the sound of 2n222's getting the crap blasted out of them,too. it's a fun transistor.

oh, and i'm movin this guy to ready to build.

It took me a minute to find this thread. Search Function to the rescue. I'm not familiar with the fuzz blaster, there is one called the "master blaster" listed on this forum.... is that what your were referencing?

no, i was referencing YOUR circuit. just my lame attempt to be witty.

[Jack Deville]

no, i was referencing YOUR circuit. just my lame attempt to be witty.

ha. i'm stupid.

[Jack Deville]

and!

updates with a successful first build:

first noteworthy update:
i omitted the switchable buffer and raised the input impedance of the first stage to around 42k.
the other stages were modified to reduce noise and current consumption. these changes worked stellarly (is that even a word?), but i recommend breadboarding both versions as there are minor tonal changes.

from the schematic posted above:
R3 changes to 220k
R4 to 10k
C2 to 100uf
R6 to 100k
R9 to 100k
R10 to 4.7k
C4 to 100uf (if ya feel like it)
R12 to 100k
R15 to 100k
R14 to 4.7k

ADDED: 220pf bypass cap from signal-in to wiper on VR1B. THANK YOU Emanuele!

These revisions radically reduced noise at max gain, and also allowed the circuit to clean up better at lower gain settings as well as with the guitar volume control.

scope report:

at min gain light clipping occurs with guitar at 10. reducing guitar volume cleans signal up entirely.

at max gain: full squarewave fuzz with all harmonics visible. totally dude.

i built a proto on perfboard to test with different guitars and amps. more to come, with an updated schematic. breadboard or build this one. its quick, easy, and makes you play loud stoner rock.

[mictester]

modified the input buffer as follows:

voltage divider setup as: +9VDC 7.5k -> 10k -> GND

470k from voltage divider to base of 2N3904. 15k emitter resistor used.

no notable decrease in noise....

more results to come

That should help, but if you increase both the emitter and collector resistors (in equal proportion) to reduce the collector current, you'll get a major reduction in buffer noise.

[Jack Deville]

Updated schematic...

[squrl]

is that recent schem correct? or should one follow this:
R3 changes to 220k
R4 to 10k
C2 to 100uf
R6 to 100k
R9 to 100k
R10 to 4.7k
C4 to 100uf (if ya feel like it)
R12 to 100k
R15 to 100k
R14 to 4.7k

just curious as i wanted to give it a whirl.

[Jack Deville]

Sorry. I sent in the wrong version schematic. Changes noted have been tested with "improved" results, the schematic above does not reflect those changes.

Hopefully, I can manage to breadboard this baby again, and get back to work. I've got a few things that are higher priority right now, but that may change.

Apologies.