My Ultra Fuzz's output looks good, but sounds like steaming garbage... why?

[deadlyshart]

Hi guys, I thought I'd make a nice simple fuzz from Craig Anderton's EPfM book, since I had all the parts lying around.

I assembled it on Veroboard and tried it out. It... makes sound, but just sounds so crappy. The thing is, he baaaaaasically says in the book that it sounds crappy, so much question is just, is it supposed to sound this bad?

I mostly understand the operation of it, I think. It essentially sets a threshold (the "sensitivity" knob), and when the signal is above the threshold it gives the high value, and when it's below, it gives the low value.

Providing a sin wave to the input and probing the output with my $15 scope seems to reflect that it's working. When the sensitivity is set to the center, it turns a sin wave to a square wave, because only the parts above 0V are above the threshold.

If I turn the sensitivity up, now only the peaks are above the threshold, so it basically gives an asymmetric square wave (more like a pulse wave), where the "high" part of the square wave has a smaller duty cycle. This is behaving as expected. It does the opposite if I lower the threshold (asymmetric in the opposite direction).

I tested it for 100Hz - 20kHz. The higher the frequency, the smaller a range of sensitivities you can use before it cuts off those high frequencies, but that's fine. At 20kHz there's still a range it's clearly working.

So why does it sound so bad? Is it actually just supposed to sound this shitty? I listened to some youtube video samples of other fuzz pedals that were even simpler (just using a single transistor, no IC) and they sounded better than this. I don't think I've ever heard fuzz like this being used in music, mostly because it does sound so terrible.

What's a nice, simple fuzz pedal that's relatively easy to make, but doesn't sound like satan getting a prolonged papercut?

[chuckd666]

Fuzz Face. Ever heard of that one?

Alternatively there's Craig Anderton's Tube Sound Fuzz (also known as the Red Llama) which is closer to an overdrive/distortion, or what about.. the BAZZ FUSS. That's a super simple and low parts count fuzz.

[robthequiet]

-- Have you tried increasing C1? It might round out the tone a bit. If all else fails, there are enough parts in the unit that could be repurposed into something like a Rat clone.

edit: I second the motion on red llama, that's a good one too.

[GibsonGM]

Could your fuzz be mis-biased?  I've built all sorts of fuzzes, and many sound BLAH as-is...non-impressive.  Then, if the bias is tweaked, some COME TO LIFE!    Just putting that out there...

[deadlyshart]

Fuzz Face. Ever heard of that one?

Alternatively there's Craig Anderton's Tube Sound Fuzz (also known as the Red Llama) which is closer to an overdrive/distortion, or what about.. the BAZZ FUSS. That's a super simple and low parts count fuzz.

Oooh, the bazz fuss looks reaaaaally simple. I might trying getting it in one of those slim 1590A chassis..es.

[deadlyshart]

Could your fuzz be mis-biased?  I've built all sorts of fuzzes, and many sound BLAH as-is...non-impressive.  Then, if the bias is tweaked, some COME TO LIFE!    Just putting that out there...

Sorry, what do you mean by mis-biased? Polarity? I'm sure the polarity is right if that's what you mean. I've probed around the circuit and the potentials seem to be right at the places I tested (op amp rails, diode ref points for the comparator).

[slacker]

If it sounds something like this, then it's working fine. It will sound pretty horrible, it's turning your guitar into a square wave with no dynamics at all.

[GibsonGM]

Could your fuzz be mis-biased?  I've built all sorts of fuzzes, and many sound BLAH as-is...non-impressive.  Then, if the bias is tweaked, some COME TO LIFE!    Just putting that out there...

Sorry, what do you mean by mis-biased? Polarity? I'm sure the polarity is right if that's what you mean. I've probed around the circuit and the potentials seem to be right at the places I tested (op amp rails, diode ref points for the comparator).

Oh, ok - that's one of those "Atari type things", ha ha.  Calling it a "Fuzz" is...yeah....not very descriptive of them!

Build a fuzz face or similiar (start with Bazz, that's cool!) and then we can talk about biasing!

[Mark Hammer]

Pity that Craig published EPFM before he ever discovered compander chips.

In the Rocktave circuit, Craig used a trick whereby he had the expander half of the 570 chip cut out before any sputtering around the trigger threshold occurred.  Sadly the Ultra-Fuzz lacks any means of either maintaining the input signal somewhere near or above the trigger threshold, or any means of identifying when the signal gets dangerously too close to the trigger threshold and tending towards sputter gaps in triggering.

Really and truly the Ultra-Fuzz needs to be preceded by a compressor and a gate.

[Mark Hammer]

I should add that the UF "decides" whether to trigger and produce a square wave on a cycle by cycle basis.  The guitar signal itself is inconsistent, such that one can expect that there will be "pits" in the signal that don't result in triggering.

By contrast, the envelope detector that guides a noise gate averages out over an extended series of peaks and valleys in the signal, and slight perturbations in the signal are forgiven.  The gate passes or blocks passages or notes.  The UF "gates" on a millisecond by millisecond basis.

There are a few ways that the UF looks like it could be made more predictable.

1. Make the input resistor 10k instead of 100k.  That will hike the gain up to 100x.
2. Make the feedback cap 68pf instead of 47pf to shave off some more of the harmonics and make things more about the fundamentals.
3.  Stick a back to back diode pair in parallel with the 1M feedback resistor and 68pf. That will reduce the dynamic range of the signal in terms of peaks.  Silicon will do.
4.  Stick a back-to-back pair between the 1uf output cap on stage 1 and the 10k after it.  Here, use a schottky or germanium pair.  This will restrict the dynamic range at the low end and essentially gate out anything that isn't a least as high an amplitude as the forward voltage of the diodes..

Both steps 3 and 4 will increase the likelihood that the threshold setting will be valid for most ofwhat you play.  If it passes the series diodes, it  probably won't sputter.