freestompboxes.org

Thank you all for your work to trace this megalithical monster and for sharing it! Thanks to IvIark for his tons of very good layouts!

I've just finished it and it works, but it didn't sound so "spreaded" as in the clips from Youtube. Moreover, it has a lot of noise.
I couldn't find BC338 and I used BC337 instead. Can this be a problem?
Also, i observed that in the schematic the LED has the anode to the ground, while in the layout it has the cathode to the ground.

Anyone else has the same problems?

Best regards!

Thanks for the heads up, I've altered the layout. I don't think it should make much difference, I assume that is just clipping one half of the waveform and so it would just be clipping the opposite half. Unless I've taken that the wrong way and it's performing another function, but it certainly seemed to work as intended for others who have built it so far.

IvIark wrote: I don't think it should make much difference, I assume that is just clipping one half of the waveform and so it would just be clipping the opposite half.

Unless you have a very non symmetrical waveform at that point... just saying

Thank you IvIark and Duckman for your answers!
I will try as soon as possible to reverse the LED and to test the new configuration. Moreover, I will test a germanium diode instead of it, to obtain much distorstion.

Anyway, the main problem is the noise. Maybe BC337 are the cause? What trannies can I use instead BC338?

Best regards!

Not sure whether the noise is layout related, but when I breadboarded this at the weekend, I used 2x 2N2222 for the fuzz portion and it was not hideously noisy. In fact, it had a slight gated quality as I recall - Not gated as in cutting off the decay of notes, but a bit like a big muff where there lots of gain, but not a huge amount of hiss or noise like a hugely cranked fuzzface would be. I suspect this is down to the filter parts of the circuit.

I also tried a few different LEDs, as well as a back-to-back pair of 1N4148 and a double pair, and could hear virtually no difference. Others may have a different take.

I used BC337's and had no problems. Also used a ultra bright blue LED. My layout looks a lot like schematic. I tried to keep the main signal flow going in one direction and mine is very quiet with no noticable gating or hiss. Also BC337-25 have a gain according to DCA55 to be about 280hfe.

Nocentelli wrote:Not sure whether the noise is layout related, but when I breadboarded this at the weekend, I used 2x 2N2222 for the fuzz portion and it was not hideously noisy. In fact, it had a slight gated quality as I recall - Not gated as in cutting off the decay of notes, but a bit like a big muff where there lots of gain, but not a huge amount of hiss or noise like a hugely cranked fuzzface would be. I suspect this is down to the filter parts of the circuit.

I also tried a few different LEDs, as well as a back-to-back pair of 1N4148 and a double pair, and could hear virtually no difference. Others may have a different take.

Thank you very much Nocentelli and onebuzz for your answers!
I think that my problem could be from the trannies. The BC337-25 used by onebuzz have less gain that the BC337-40 that I used (just I saw that). As Nocentelli said, the fuzz part is like a FF, so it's possible that the gains of the trannies to have some importance. I will socket them today and I will test more types, with different gains.

My orange clear LED lights very weak (I can see it only in the dark), but it lights in the rythm of my playing

Best regards!

Anyone have a working notion of gains in the original transistors?

I had the dreaded squeal that about half the people who built reported at tagboardeffects, even after upping the filtering and putting a resistor in line to the DC jack. I backed off the original bc337/2n5088 complement, all above 350 hfe, to all 2n4401, all about 270 hfe, but still had squealing. Backed off to all 2n5551, all around 170 hfe, and it behaves marvelously. Less noisy than any demo vid I've seen, very well behaved, and all working well. Definitely socket from the start with this guy, since that made troubleshooting a breeze.

But I wonder if there is more corpulence I should be shooting for if I can keep plugging until the right doomy magic arrives. Anyone done any of this legwork already? Magic gain ranges or critical spots to swap to tweak its dirt character?

Two points:

1) The input of this unit is strange if there is nothing before it which has not been posted. The input impedance will be virtually the resistance of the first control. The gain stage after it has a very low impedance input due to the feedback employed. It isn't 47k as some people imagine it to be, that has to be divided by the gain which makes it much lower. Imagine the current when you put 1V on one end of a grounded 1k resistor, it's 1mA. Now do the same but with the other end dropped down to -9V. Your signal is still 1V but the current is now 10mA because there is 10V across it. even though the resistor value is still 1k, the input 1V is "seeing" a 100R resistor. Now imagine the same thing but with your guitar signal. That is what you have, the input on one end and the amplified signal at the other. The current drawn from the input must increase many times and the apparent resistor value drops by the same factor. It seems odd to me that people will gasp and talk about "tone suck" because a modern well designed pedal doesn't use "true bypass" then do this sort of thing which is the definition of tone suck at anything but full input settings. Driving a low impedance from your high impedance guitar sucks tone, driving a high enough or, better, a matched impedance does not even if it is FET switched.

2) If this unit oscillates for a lot of people the answer is to try to stabilise the oscillations, not search for a mythical transistor which will not do so! People seem to be effectively saying, "we have to find a transistor which is low enough gain so it cannot amplify enough to oscillate". Is it then doing its job in inducing overload as well? My car wheels wobble at high speed so I've put in a smaller carb so it can't go so fast? I don't have a wobble now!

If it oscillates with some transistor types it will be too close to the verge of instability for most others. A good circuit is one which does not change when you change the transistors, that's what feedback is for, and feedback is used in almost ALL circuits in one way or another. For example, Q2 has 100% feedback around it to do its job as a follower! Instead try putting small value capacitors, up to a couple of hundred pF, across EITHER the 47k feedback resistor from Q1 to Q2 OR across the collector/base of Q1 or the same for Q3. Use as little of that as you need to stop oscillations at full gain settings. The only factors which should matter for all non-specialist transistor choice should be Polarity (Duuhh!), Max Voltage (easy), Noise (for inputs/very high gain) and perhaps Bandwidth (largely irrelevant for guitar frequencies if you know how to stabilise a circuit). hfe or ß (beta) should be irrelevant if the transistors you use are undamaged and reasonably modern general purpose or better. If you design the standard parts of a pedal so they require a specific gain to work then you have designed a bad pedal badly!

This does not hold for JFETs which are a different beast.

bordonbert wrote:Two points:

1) [...] It isn't 47k as some people imagine it to be, that has to be divided by the gain which makes it much lower. Imagine the current when you put 1V on one end of a grounded 1k resistor, it's 1mA. Now do the same but with the other end dropped down to -9V. Your signal is still 1V but the current is now 10mA because there is 10V across it. even though the resistor value is still 1k, the input 1V is "seeing" a 100R resistor. [...]

Node 47k-10k(emitter resitor Q2) is AC-grounded. Minimum input resitance is 47k.

[...]
If it oscillates with some transistor types it will be too close to the verge of instability for most others. A good circuit is one which does not change when you change the transistors, that's what feedback is for, and feedback is used in almost ALL circuits in one way or another.

This is a Fuzz, not a 'good' circuit! Neg. feedback is detrimental to the sound of this type of circuit.

For example, Q2 has 100% feedback around it to do its job as a follower!

This is not an emitter follower.

[...] hfe or ß (beta) should be irrelevant if the transistors you use are undamaged and reasonably modern general purpose or better. If you design the standard parts of a pedal so they require a specific gain to work then you have designed a bad pedal badly![...]

And how do you do to get rid of the excess gain? Emitter resistor reduces gain but introduces neg. FB and rises input impedance and it doesn't behave nicely' when overdriven. Suddenly you have a lot of additional resistors, diodes and capacitators in your circuit to be 'well designed'.

electrip

Did you ever have one of those days where nothing seemed to go right? Apologies to all, Electrip is spot on in his comments. I have been looking at three open schematics and confused them. Some of my criticisms do not apply to this unit. Hey it was late and I was really frustrated with this whole issue. You raise a few good points which must be acknowledged Electrip, and a couple which are interesting and deserve a comment or question.

You're absolutely right about the AC grounding of Q2 emitter, my comments about low impedance are wrong, pure and simple. I was looking at and had commented on a feedback pair in, I think, something like a Pharaoh circuit and another multistage design at the time. I should have been concentrating on what I was doing.

However, I don't agree with you entirely that 'fuzz' is "not a good circuit"! I can see why you may say that but I maintain it doesn't have to be, nor should it be. The circuit we have is basically a pair made of 2 cascaded common emitters, with DC stabilisation via emitter degeneration in both and overall feedback, but with AC feedback removed and gain increased by the 2 x 47uF emitter caps to the point where it overloads the second transistor. Such a raw approach, as you pointed out, will of course depend on the hfe of the transistors. I assume the idea was to produce as much gain as possible in a simple stage, although it may be that there is a sweet spot in the gain where it sounds just right. I would argue that to design a circuit where the gain is dependent on device parameters like this is still poor design and unnecessary no matter what it may sound like.

If you want more gain, another general purpose transistor will cost about 10p retail and a couple of resistors and a cap another maybe 30p. Say 50p at the outside. You could even double that to another two transistors without cost being an issue. You can then have a more predictable first stage pair with overall feedback defining a closer tolerance highish gain from any transistors. This does not necessarily need to overload, just produce a much amplified signal for the following new stage. That could be an overload single common emitter stage almost identical to the existing Q2 but which can have a lower well defined gain by emitter degeneration again independent of transistor type. With its higher input signal, the single transistor stage will still overload without any alteration by overall feedback around it, so bottoming and saturation characteristics will not be altered by the feedback, but you have an independent "good" circuit which is repeatable and reliable. It should be more stable too. Is that not a better way to approach this and achieve the same effect?

And as a separate point, I would also maintain that the number of resistors and capacitors needed to get a design to work is largely irrelevant within reason. A few additional feedback components is nothing.

I may still be missing something of course, (I have just demonstrated I am very capable of doing so, and incidentally I'm genuinely very grateful for you pointing it out, we only learn by making mistakes and being corrected), but in principle this is what good design is about. It should include producing the same result as the "design brief" reliably, rather than as a chance thing which you occasionally achieve with a magic component. That is a difference between what the big boys do who produce the great classic pedals that sell in droves which the "forum designers" do not. You may get a one off great sound but with a design process like that no one else will. Witness the number of people who report they cannot achieve the same sound as the YouTube demo shows and who report stability problems. They may of course select components for tolerance but that makes for much increased expense and is once again moving away from a good engineered approach. We need to make our designs such that others with no experience or understanding of the workings can build them and achieve the same results reliably with whatever components they can get their hands on.

Does that make any sense?

To clarify, the whine I experienced occurred at any setting when not playing guitar at that moment, and I believe is a power-supply only issue to see how others brought it up. (I avoid batteries, so didn't wire up a snap to test.) The current noise behavior with lower gain devices is fantastic for such a primitive circuit topology and it still gets pretty loud and big rock screaming and growly. I'll try some things including some base-to-collector caps, to see if I can push it further without the whine, but I could definitely live with it as-is.

Staring at the schematic more and noting the family resemblance to the Fuzz Face and similar ZVex (Wooly Mammoth and Mastotron), where certain ranges tend to play nicer, I'm guessing the original was built with some specific gain ranges in mind, and thus the mix of NPNs that would generally be close enough to each other in a feedback-stabilized circuit. The gain ranges for many types in my transistor stash from Tayda tends to be much higher than most report for the same types from other sources, so I always have a good chance of pushing the stability of a non-stabilized circuit if I just go by component designation. Chasing low-tech fuzz, I'm willing to deal with these things.

I backed off the original bc337/2n5088 complement, all above 350 hfe, to all 2n4401, all about 270 hfe, but still had squealing. Backed off to all 2n5551, all around 170 hfe, and it behaves marvelously.

I rest my case . What would be really interesting here would be to source the magic transistor type and find 'lowest in the range', 'centre of the range' and 'top of the range' hfe examples. Then put them in and see if it sounds any different for the same amount of generated fuzz. Same type, extremes of gain, same degree of effect, any difference in quality of sound?

I would have to say I still suspect that the character of the basic distortion this type of circuit produces is generally about the same, I believe the majority of the character is in the tone shaping and filtering before and after distortion is introduced. Like most of you I am a working guitarist myself, not metal or similar so not bent on fuzz, old school classic rock. So my personal choice would be more a gentle to heavy overdrive rather than fuzz. But taste aside, it stuns me to sit in music shops and hear guitarists trying out pedals and talking about how they sound different to what they expected when they are using the shop's choice of amp and a totally different guitar to the one they heard played in an online demo or they tried out last time. Do they think these things do not have any effect on the final sound?

How square can a virtual square wave get? The number of diode clipping pedal designs is enormous and a huge number use the basic 2x 1N914/1N4148 setup? Can these pedals ALL have a different sound based on what they do with these 2 diodes? I think not. I would ask whether it matters much if you use coarse hfe dependent gain, feedback controlled transistor gain, or opamp gain, all to the same extent, a lot of the sound character is in meticulous and painstaking filtering after the distortion and in the way the circuit blocks interact with each other. Like Electrip's point about the input impedance, points like that are what is important and to my mind would have more influence than the way we chop the top off the wave.

Yes, I know there is much more involved here. Transfer function bending is achieved across the whole waveform and gain control is a part of the sound. I've done some work investigating diode ladder mechanisms of up to 5 individual tuned series diodes in each direction to see what sound more sensitive handling can give. It's quite astounding how much the transfer function can be tweaked if you want. But no one who works seriously in this field seems to be trying to identify what are genuinely the important points which give pedals any sort of uniqueness. There will still be a stream of 2x back to back diodes of a 'magic type' which will continue to appear. (Yes I know germanium and LEDs are different, I still mean 1N914/1N4148 types).

In the light of my 'fakuck up' of yesterday, I felt guilty so I've done a little Spice simulation work. Yes, I know that isn't the same as listening and hearing but it's the best we can do in the time and it gives us something to talk about. I mocked up the exact distortion stage of your pedal. I mocked up my own suggested circuit from above with a feedback controlled pair giving controlled gain with no deliberate distortion followed by a single overloaded transistor stage. This second stage had emitter degeneration to set the gain but no form of feedback from the output point, much the same as your own second transistor. I fed both with the exact same signal and ran the simulation of the open gain version to find out what its gain was. I adjusted my own stage gains via the feedback to achieve the same overall figure. I then compared the output of each of these without any form of loading or filtering, remember this is only the distortion creation mechanism I am considering.

I got what looked to me to be incredibly similar results! Clipping was essentially the same. Distortion at the onset of clipping at each end was very similar. Effects like overshoot and slow recovery from saturation did not appear to be an issue either. And my circuit did not change much at all that I could detect for wildly different transistor types, (it must change a little, there is a limit as to the effectiveness of feedback and open loop gain does come into the equation but the factors involved suggest it won't be audible). I don't doubt there is a slight change to the distortion content with changes in every minor factor, but in the light of the amount of distortion we are producing would this be audible in any way? In a 0.002% distortion hifi system yes, in a 10%-20% distortion pedal? Hmmm!

Now I could easily be wrong on all of this, (it has been known ), I don't sit there designing pedal after pedal or building dozens in the hope of the Holy Grail appearing, I prefer to spend time playing using my own best design, (SRPP JFET), and I stick to that because it sounds just as I want. You guys seem to be more into the build process than I am and would have informed and valid, (for the most part), opinions it would be worth hearing. I hope someone out there who genuinely knows will offer their thoughts on this. I'm really not into pooh-poohing the 'established knowledge' for its own sake, but all 'known facts' must be held to account in a scientific way at some point in time. Without calling it to task you will stay reliant on a 'Priesthood' who continue to push forward the beliefs in magic and the mystic arts. This is engineering and we should be able to sort out what really counts and, more importantly, why. I always remember the old maxim, "In the country of the blind the one eyed man is king"! (And the others can't see to know how restricted his own view is).

Yeah, I know, long posts, fanaticism, you've heard it all before, sorry. But I genuinely would like to just get to some kind of truth, independent of 'it's just known' or the placebo effect, (yes it does come into it).