Build your own DIY guitar fuzz box - fun project

[pma]

Build your own DIY guitar fuzz box – fun project

In essence, a fuzz box is a small amplifier that is designed to be overdriven. Any amplifier has a certain amount of headroom: the output power that determines the level of signal that can be sent through the amp without the signal being clipped.

Distorting the shape of the waveform will change the sound: if we start with a smooth waveform and clip it, we wind up with a more complex, somewhat harsher sound. The clipping adds harmonics to the waveform, giving it a very different character from the original. If the clipping is severe enough, it will tend to flatten out the natural decay of the guitar’s note, producing the well-known sustain effect.

Clipping is usually not an all-or-nothing situation: there is hard clipping and soft clipping. In soft clipping the amplifier will begin to round out the edges of waveforms before going into hard clipping. Soft clipping produces a warmer, more pleasant sound—one of the reasons why many guitar amps are still made with vacuum tubes rather than transistors in the output stages. The tubes’ clipping characteristics produce a more desirable sound than the harsher sound of transistor amps. While only a few esoteric fuzz pedals use vacuum tubes, designers of fuzz boxes expend a great deal of effort to get a perfect soft-clipping sound.

These people would be appalled by what we are about to do. We’re going to use an op-amp (operational amplifier), rather than individual transistors or tubes for the amplification stage. The clipping characteristics of the op-amp are anything but soft, so we are going to use an old fuzz box trick: a pair of diodes on the output to produce a softer-clipping curve. It’s not going to sound like a tube amp, but it won’t sound completely awful, and building it will cost not much.

The circuit

The booster circuit (fuzz box circuit) is built inside a shielded metal box, to prevent EMI fields from penetrating to the booster circuit that has considerable gain. We shall use good connectors to prevent unwanted loose contacts of the guitar cable that would result in infamous clicks, pops and buzz. The input connector (X1) is Neutrik NCJ6FI-S combo type. It contains a TRS connector and a XLR female in one common housing. The contact quality is excellent, with very low contact resistances.

This schematics was modified in post #5 as per recommendation by @KSTR

Right at the input is a Hitano VCR07D470K varistor (30Vac), R1, to protect the input from ESD discharge spikes, and also its capacitance of about 1nF is good to catch RFI. Then there is an op-amp, IC1, TL071P type. This cheap op-amp was selected during functional sound tests, based on a resulting sound and circuit stability and interference immunity. The gain is set by (R4+R3)/R3 feedback resistor divider at about 28.8dB (28x). The output of the IC1 would go up to some +/-13V swing, however the R5 resistor and D1, D2 diodes create a limiter that does not allow the output signal (behind the R5) to get over some +/-0.8Vpeak. This is the “fuzz” function of the circuit. At very low input amplitudes, up to some 10mV, the input signal is just amplified and undistorted, up to some 0.25Vrms output voltage. Then, when increasing the input level, the circuit starts to distort and the maximum clipped output rms voltage is about 0.55Vrms, or 0.8Vpeak. It will be shown in a distortion plot.
Behind the R5 we have an output buffer IC2, based on a monolithic buffer BUF634T. This buffer has high capability to drive long output cables and is able to give +/-250mA output current. It is necessary to use it to prevent overloading of the IC1 circuit behind the R5 resistor, as IC1 does not have enough output current to drive both the limiter circuit (R5, D1, D2) and the long cable + possibly low impedance of the next amplifier.
Output connector is Neutrik NC3MDL connected as a pseudo-balanced (R9, R10), which again helps to reduce possible hum/buzz due to long output cable. The amplifier used behind the booster is best to have a balanced XLR input.

The sample built

Galaxy metal case https://hifi2000.shop/site/en/catalog/galaxy was used for the sample, as I have had it here momentaly unused.

We may use another case, of course, in case it is a metal one. Input X1 connector must have “S” and “1” pins immediately connected to the metal box, to assure the shielding effect.
Power supply used is a regulated 2x15Vdc supply, my own design, can be replaced by any regulated 2x15V PSU.

Some measurements

As already said, the booster yields a clean signal if the input level is very low, and starts to distort quite gradually above certain input level.

THD vs. output level, to see the distortion rising with output amplitude:

Sound sample

How does it sound? Attached is a short sample, unfortunately Eric Clapton has been unavailable at the moment, so I had to play it myself .

 

[KSTR]

@pma,
Pavel, are you sure about the right side of R4 connection? Once the diodes start to conduct, the opamp goes open-loop because the feedback source is clamped, slamming into to the rails, counteracting the soft-clipping idea quite a bit. Shouldn't it go to the output of the opamp (pin6)?

It won't make too much of a difference, though. Except at the point where it matters most, at the onset of distortion.
Which is exactly what I hear in your bass guitar sample, doesn't sound like gentle soft-clipping at all, and doesn't look like that either, very visible in the note decay phase:

And, 100k input impedance is a bit low for passive guitars, making the volume control of the guitar (often a 500k pot) behave compromised. And it's the only volume control we have, in this case, to adjust the clipping threshold.

 

[KSTR]

Zooming further, at strong overdrive when the input was loud, I see always the exact same waveshapes after the clipping (per polarity, that is), while onset of clipping varies. This indicates severe rail-sticking with a long recovery from the open-loop state:

Note the two kinks shortly after the level drops below the clipping threshold.

The apparently different clipping levels are an artifact of the AC-coupling of the sound-card used for recording. Clipping asymmetric waveforms produces DC offsets, and the AC coupling tries to restore DC balance, visible in the envelope of the track:

 

[pma]

Pavel, are you sure about the right side of R4 connection? Once the diodes start to conduct, the opamp goes open-loop because the feedback source is clamped, slamming into to the rails, counteracting the soft-clipping idea quite a bit. Shouldn't it go to the output of the opamp (pin6)?

Hi Klaus,

thanks, it seems to be o good hint at least in the simulation.

As of now (original unmodified circuit), the clipping looks like this:


Your suggested modification:



Looks good, I will try it!

 

[pma]

@KSTR

works well, thanks again, Klaus!

 

[DonH56]

Nice to see DIY here! I made money back in HS and college making various fuzz boxes based on various "cookbooks" of the day. Some used simple diode clippers, while others overdrove a discrete transistor. Diodes, BJT, and FET boxes all had a slight different sound. Most of them added a straight path around the clipping circuit and control on the "clipped" path to control the amount of added "fuzz", which could be another mod. I also played with a pot in series with the diodes (etc.) to both soften the clipping and control the onset of clipping.

One problem I had was that some guitar amps and many guitarists only used the "guitar" input, meaning added gain would overload their (pre)amp. The parallel paths allowed me to adjust overall gain in the middle or at the output with a little less noise than a single pot at the input.

 

[pma]

I made money back in HS and college making various fuzz boxes based on various "cookbooks" of the day.

Same here, but just for myself those ancient days (but I made some money making amplifiers during tech university years). My "old days" boosters suffered from capturing all kinds of interferences from the environment, it took me years to learn how to wire it and shield it properly.

 

[pma]

I changed the TL071 op-amp for the fast OPA637 in the modified schematics (post #5). The result is much better behaviour at high audio frequencies. Below is the wideband measurement of 10kHz clipped sine, up to 6MHz. Distortion profile is now independent of frequency up to 20kHz. With the TL071, there was quite fast distortion rise with increased frequency. BAW62 clipping diodes are very fast and keep the signal shape even at high frequencies.



Sound clip sample is attached as well.

 

[JohnnyAudio]

One germanium diode served Jimi well.

 

[EngineerNate]

What’s the SINAD score?

 

[bmc0]

Possibly silly idea: Replace R5 with a potentiometer, taking the feedback from the wiper. Should work as a "clip hardness" control.

 

[Cbdb2]

Nice to see some one else here doing some DIY. Just a couple of comments.
There are many commercial opamp distortion pedals. Most opamp distortion circuits put the diodes in parallel with the feedback resistor (if you switch one in/out you can get asymetrical clipping). This creates soft clipping. You can do ESD protection with reverse diodes from input to the rails (cheaper). Guitar inputs are usually higher impedance, at least 500k (you can add a switched lower R for active guitars.) I would also include a RF bypass cap at the input and DC blockers at the in and out. And a design that can use a standard 9V pedal supply (or battery) would reduce the cost by quite a bit, but make the design a bit more complicated.

Heres a good referance: https://generalguitargadgets.com/how-to-build-it/technical-help/articles/design-distortion/

 

[pma]

Just a couple of comments.
There are many commercial opamp distortion pedals. Most opamp distortion circuits put the diodes in parallel with the feedback resistor (if you switch one in/out you can get asymetrical clipping). This creates soft clipping. You can do ESD protection with reverse diodes from input to the rails (cheaper). Guitar inputs are usually higher impedance, at least 500k (you can add a switched lower R for active guitars.) I would also include a RF bypass cap at the input and DC blockers at the in and out. And a design that can use a standard 9V pedal supply (or battery) would reduce the cost by quite a bit, but make the design a bit more complicated.

Heres a good referance: https://generalguitargadgets.com/how-to-build-it/technical-help/articles/design-distortion/

Thank you for your comments. As you will see, I am familiar with them and also with the fuzz box commercial solutions. That is really why I decided to make it different. As an example, you will see a comparison with cheap Behringer SF300 Super Fuzz. The plots are taken with a dual channel oscilloscope, the blue track is always from my circuit and the yellow track is from the SF300, which has 3 different switchable settings. Input signal is a 1kHz sine, for simplicity. SF300 is 9V battery operated, and its circuit is supposed to be this one:


1. SF300 BOOST mode

In this mode the signal is only amplified, there is a gain pot, the SF300 just adds a small amount of distortion due its primitive transistor circuitry. This is the only usable mode, regarding the resulting sound.

2. SF300 FUZZ1 mode

We can see that the base frequency has been doubled by SF300, with a bit asymmetrical clipping. However, the sound is unusable.

3. SF300 FUZZ2 mode

The frequency is doubled again, sound is extremely sharp, lacking any lower frequencies, as the shape of the signal indicates.

So that is the SF300, not mentioning its much worse immunity to ambient fields and corresponding unwanted interference sounds. It is so-so usable in the gain boost mode, the FUZZ1 and FUZZ2 are useless, from my point of view.

Now technically:

a) ESD protection. Yes clamping diodes conventionally connected to rails would work, however the varistor can absorb much higher energy without being destroyed.

b) it would be no problem to change the input impedance from 100k to usual 500k. JFET input op-amp makes it easily.

c) RF filter - I would certainly include it if it was necessary. The varistor capacitance seems to make the job well, measured up to several MHz decades.

d) 9V supply - exactly the thing I wanted to avoid

e) case etc. - this is a sample only. If I was to build a "serious" version, it would be mounted in a 19" 1U panel rack. It would have some gain and limiter level setting, but the basic circuit would be as described here.

I am sorry but the document you linked gave me no useful information, moreover, the diodes that are connected directly to the op-amp output, without the current limiting resistor, are used completely wrong. Similarly for FB diodes, this must be done in a more sophisticated way.

 

[pma]

In the previous post I have shown comparisons in the time domain. Now I am attaching the frequency spectrum comparisons. However, in the case like this, the human ear is a decisive tool .

Please note that in both cases the Behringer SF300 doubles the frequency with the highest amplitude, from 1kHz to 2kHz. In other words, H2 is higher than H1. Similarly the H6, H10, H14 are higher tan H1. The sound is totally diverted and is perceived as too harsh, too sharp, and disintegrated.