(Trying to) Kill Hiss, Oscillation and RF in Vintage Fuzz Circuits (MK II, Supa, Fuzz Face, etc.)...

[Big Monk]

Conscious of not hijacking the MK I Tone Vendor thread, I thought it might be a good idea to write a quick post about my methods for killing sonic artifacts in Fuzz pedals.


Firstly, let me copy over my MK II/Supa comments over:

For the MK II, I’ll usually start with a rough draft set of 100 pf, 82 pf and 82 pf for Q1, Q2 and Q3, respectively. It’s important to listen and mentally log the difference between this initial set and the sound with no caps installed.

With the Attack pot all the way up and volume at unity, increase the caps until the pedal is pretty quiet. Don’t try to find balance between noise and treble content. Kill the noise completely first.

When the noise has been addressed, compare the tone with the sound without caps installed. Now use a combination of the treble bleed cap (cap to ground on input), the input cap and emitter cap to tweak the circuit for the proper frequency response. “Proper” in this case is driven by personal taste.

I like to start with the bog standard cap set for the MK II at the outset: 0.01 uf, 4.7 uf and 4.7 uf for Treble Bleed, Input and Emitter caps, respectively. I find the treble bleed and Emitter caps most effective here. If the pedal is quiet but a bit muffled or muted after base to collector caps have been installed, tweak the treble bleed between 0.0047-0.0082 uf if need be and the emitter cap between 1 uf-3.3 uf.

Keep in mind, you may hit a point where the tone cuts well but is a little ragged. Also, lowering the treble bleed too much can bring on uncontrolled, high pitched, i.e. non-musical, feedback.

Conversely, you can go the Supafuzz route and ditch the treble bleed cap altogether while raising the input cap.

Many, many options. This is a circuit, much like the Fuzz Face, that IMHO requires breadboard tuning. It’s almost criminal to sell someone a cookie cutter kit or anything similar for these pedals.

I understand that the values differ between MKI and MKII, but would the principle be worth applying to MKI or is it so inherent already with MKI build that trying to silence it is a somewhat wasted effort? I mean I have some good tips on how to bring the noise down, bit perhaps I should not hope on getting it as quiet as with your MKIIs?

You could do the same thing. I’m just not as well versed in the circuit architecture of the MK I so I can’t say how changing frequency shaping components will affect the overall tone.

You can absolutely use what I described to clamp down on noise though. That is circuit independent.

The Fuzz Face can use roughly this template as well, but requires a few initial tweaks before proceeding to the base to collector caps. There are 2 issues that the Fuzz Face suffers from that the MK II/Supa (and the MK I, by extension) does not: RF Interference and High Frequency Oscillation.

The oscillation is really the first order of business in a Fuzz Face. I've seen many, many solutions proposed to handle this but the caveat is always that, "It may or may not work for you...".

One method I have found that works like gangbusters is to audition caps that bypass the feedback resistor. You'll want to start with something low like 220 pf and work your way up. Do not be alarmed if the caps get relatively HUGE. My recent Silicon Fuzz Face used a 4700 pF! Find a cap that gets rid of the majority of the oscillation without affecting the tone of the circuit too much.

With respect to RF, and also the last bit of the oscillation, resistance on the input can help. However, even a small reduction in signal from full up can start to neuter the Fuzz Face tone.

I use a "Pre-Gain" pot on the input of all my MK/Supa and Fuzz Face units. Besides letting me reduce the gain of the circuit for different shades of distortion, turning it just ever so slightly back from full often exposes a "notch" that effectively kills RF and the last of the oscillation. Of course, what is happening here is that on full, the "Pre-Gain" offers no impedance to the front of the circuit, so reducing it slightly is like putting a very small fixed resistance in series with the input. In lieu of a "Pre-Gain" control, simply audition very small resistances, starting with say 47 ohm and work your way up until RF and Oscillation disappears but the tone of the circuit is not affected greatly.

With that said, RF may be impossible to completely eliminate. I've tried all my methods and on a bad day i still get the local country station through my speakers. It's faint, but there.

Now you can attack hiss with the same method for the MK II/Supa, although start with lower values for the Fuzz Face than you did with the MK II/Supa. I would start with 82 pF for Q1 and 51 pf for Q2. You'll find that by bypassing the feedback resistor in the Fuzz Face, you need less capacitance across the base and collector of each transistor as opposed to the MK/II/Supa.

 

[Big Monk]

A quick real world example:

I recently built a Silicon Fuzz Face on my Universal board. My benchmark was the "Time" solo from DSotM and the raging solos from "Echoes" off of Live at Pompeii.

I started out with a set of Amperex A115/A104 transistors with gains of 120/285, respectively. This had a nice sound but lacked the overall girth and sizzle that I was aiming for. From the outset, this set had major oscillation issues so i decided to get my transistors right before proceeding. I ended up using a BC109B/Amperex A104 combo with gains of 286/405, respectively.

First order of business was killing the majority of the oscillation so I could actually hear what it sou8nded like!

I started with a 470 pf ceramic across the 100k feedback resistor. I ended up tweaking that up to 4700 pf. Oscillation was mostly gone but the high end had been reduced. I still had some minor hiss so I put an 82 pf across B-C of Q1 and a 47 pf across B-C of Q2. That killed the rest of the hiss.

I have a 500kA volume pot on this unit. In order to restore the right amount of high end, I immediately dropped the emitter cap from 22 uf to 10 uf. This helped greatly, although I did not want the pedal to fart out or get flabby when playing power chords so I also dropped the input cap from 2.2 uf to 1 uf. That did the trick.

Now here is where you have to use your judgement. I use a 100kA "Pre-Gain" on the input of this pedal. With the "Pre-Gain" all the way up, there is a slight trace of whistling oscillation and RF interference is more active. I will typically turn the "Pre-Gain" down so slightly it's almost not perceptible and listen for when this last bit goes away. For those using fixed resistors, tack in a 100k pot and find this "sweet spot" on the dial. Remove the pot and measure the small resistance that killed the rest of the oscillation/RF and place that value in series with the input.

 

[BuddytheReow]

Dude, you should write a book on this!

 

[Big Monk]

Dude, you should write a book on this!

I'm just repeating things others have told me and passing it on!

 

[Dan0h]

This is why I spend so many hours on this forum. There are so many awesome nuggets of wisdom. Thank you for sharing.

 

[Feral Feline]

Thanks Big Monk for sharing the info aggregation — such amassed knowledge will surely reduce aggravation.

 

[andare]

I'll have to experiment with these ideas on a Fuzzrite. Those things are ridiculously noisy with both transistors on and they get radio with the volume rolled off.

A 2n2 cap from input to ground got rid of the radio but not the noise. Bigger caps eliminated the noise, I went up to 100n, however they reduced the gain and high end too much and the circuit lost its trademark sound.

 

[MBFX]

A quick real world example:

I recently built a Silicon Fuzz Face on my Universal board. My benchmark was the "Time" solo from DSotM and the raging solos from "Echoes" off of Live at Pompeii.

I started out with a set of Amperex A115/A104 transistors with gains of 120/285, respectively. This had a nice sound but lacked the overall girth and sizzle that I was aiming for. From the outset, this set had major oscillation issues so i decided to get my transistors right before proceeding. I ended up using a BC109B/Amperex A104 combo with gains of 286/405, respectively.

First order of business was killing the majority of the oscillation so I could actually hear what it sou8nded like!

I started with a 470 pf ceramic across the 100k feedback resistor. I ended up tweaking that up to 4700 pf. Oscillation was mostly gone but the high end had been reduced. I still had some minor hiss so I put an 82 pf across B-C of Q1 and a 47 pf across B-C of Q2. That killed the rest of the hiss.

I have a 500kA volume pot on this unit. In order to restore the right amount of high end, I immediately dropped the emitter cap from 22 uf to 10 uf. This helped greatly, although I did not want the pedal to fart out or get flabby when playing power chords so I also dropped the input cap from 2.2 uf to 1 uf. That did the trick.

Now here is where you have to use your judgement. I use a 100kA "Pre-Gain" on the input of this pedal. With the "Pre-Gain" all the way up, there is a slight trace of whistling oscillation and RF interference is more active. I will typically turn the "Pre-Gain" down so slightly it's almost not perceptible and listen for when this last bit goes away. For those using fixed resistors, tack in a 100k pot and find this "sweet spot" on the dial. Remove the pot and measure the small resistance that killed the rest of the oscillation/RF and place that value in series with the input.

Why does the B-C capacitor help reduce hiss?

 

[SillyOctpuss]

A quick real world example:

I recently built a Silicon Fuzz Face on my Universal board. My benchmark was the "Time" solo from DSotM and the raging solos from "Echoes" off of Live at Pompeii.

I started out with a set of Amperex A115/A104 transistors with gains of 120/285, respectively. This had a nice sound but lacked the overall girth and sizzle that I was aiming for. From the outset, this set had major oscillation issues so i decided to get my transistors right before proceeding. I ended up using a BC109B/Amperex A104 combo with gains of 286/405, respectively.

First order of business was killing the majority of the oscillation so I could actually hear what it sou8nded like!

I started with a 470 pf ceramic across the 100k feedback resistor. I ended up tweaking that up to 4700 pf. Oscillation was mostly gone but the high end had been reduced. I still had some minor hiss so I put an 82 pf across B-C of Q1 and a 47 pf across B-C of Q2. That killed the rest of the hiss.

I have a 500kA volume pot on this unit. In order to restore the right amount of high end, I immediately dropped the emitter cap from 22 uf to 10 uf. This helped greatly, although I did not want the pedal to fart out or get flabby when playing power chords so I also dropped the input cap from 2.2 uf to 1 uf. That did the trick.

Now here is where you have to use your judgement. I use a 100kA "Pre-Gain" on the input of this pedal. With the "Pre-Gain" all the way up, there is a slight trace of whistling oscillation and RF interference is more active. I will typically turn the "Pre-Gain" down so slightly it's almost not perceptible and listen for when this last bit goes away. For those using fixed resistors, tack in a 100k pot and find this "sweet spot" on the dial. Remove the pot and measure the small resistance that killed the rest of the oscillation/RF and place that value in series with the input.

I wish I'd found this thread earlier, I've been going through this exact thing with a few fuzz faces over the last few months. I've pretty much cracked it now and I'm getting ready to move onto tonebenders

 

[andare]

Why does the B-C capacitor help reduce hiss?

The high end is reduced. Common trick on silicon fuzzes.

 

[Big Monk]

Why does the B-C capacitor help reduce hiss?

The high end is reduced. Common trick on silicon fuzzes.

You have to always remember that noise isn’t created in a vacuum. Hiss and RF in particular are part of a specific frequency band.

You filter that band in some way and you remove the noise. You also dampen the high end but you can compensate for that in other ways.

 

[Ginsly]

I'm really, really glad I stumbled upon this thread. @Big Monk this is about as comprehensive an "order of operations" as I've seen for tackling various gremlins, and addresses both 3-transistor and 2-transistor designs. @andare it's funny you mention the Fuzzrite, as breadboarding a variant lead me here! And it is indeed pretty dang noisy. I also have a tweaked Bosstone version (T-Bone) that oscillates at Full-Attack, and I'm wondering where to start in addressing that.

Rather than start a new thread about noise, figured I'd bump this excellent primer. I'm curious what other noise-killing steps people take when modding/designing/building high-gain pedals. Living in a big city makes for a very noisy environment, chalk full of EMI and RFI. I'm wary of ground loops, but I can't say I know what that kind of noise sounds like in a pedal.

I'd especially love to know where all of you start when it comes to addressing noise. For example, I've found that before I start messing with other components it's a good idea to try some different transistors and see what happens. One of my first builds was a Modded Si Fuzz Face from BYOC, and although it sounded good its noise floor made it pretty unusable unless biased to gate. I tried a lower hfe transistor for Q1 and it had a drastic impact on overall noise. I was happy to give up a little gain for a less annoying experience. It's a delicate balance, though!

 

[Big Monk]

Modern transistors is a good place to begin.

I couldn’t believe how much quieter modern BC type transistors are than my old NOS stuff was.

 

[Ginsly]

Modern transistors is a good place to begin.

I couldn’t believe how much quieter modern BC type transistors are than my old NOS stuff was.

I've certainly found this to be true. Not an option with Ge of course, but I mostly stick with Si anyways. Even within that realm I had at one point considered sourcing 2N3565 buttons for some builds; @PedalBuilder tipped me off that it not only wasn't worth the cost, but pretty much any equivalent hfe, modern Si would have lower noise and sound the same.