Differential Fuzz Design

[mdcmdcmdc]

Antonis - with regard to proper biasing of the buffer stage, here's what I'm reading at the moment:

Source: 9.40

Q1 (NPN):
c - 8.77
b - 5.60
e - 5.17

Q2 (PNP):
c - 5.18
b - 8.77
e - 9.40

You're talking about the emitter of Q1, correct? It's slightly above 1/2 supply voltage, should it ideally be higher? If so, what's a good target voltage?

Apologies if I'm misunderstanding your comment!

[iainpunk]

You're talking about the emitter of Q1, correct? It's slightly above 1/2 supply voltage, should it ideally be higher? If so, what's a good target voltage?

Apologies if I'm misunderstanding your comment!

yes, he is talking about the emitter of Q1 (which is also the emitter of the Sziklai pair, which can be calculated as one transistor with Higher Hfe.) and i think the higher the current the better. a reasonably strong guitar signal is max 1.5v peak to peak, if you want that to be transferred cleanly you need .75 v headroom on the positive side, and due to the double forward voltage from the Sziklai pair, a good starting point would be 2v under Vcc.

HOWEVER in your case, you need not have a perfect buffer, you use it as part of a filter (C6 22nF) and asymmetry is not a bad thing since you are going to distort the h#ck out of the wave anyways, so what Antonis and I were talking about is basically a bunch of theory for ''proper'' buffer design mainly applicable in control systems and Hi Fi equipment...

cheers

[antonis]

what Antonis and I were talking about is basically a bunch of theory for ''proper'' buffer design mainly applicable in control systems and Hi Fi equipment...

@mdcmdcmdc: The above in an inside joke 'cause I love to theoretically tease Iain when he speaks about practical solutions and, of course, also practically tease him when he speaks about theoretical issues.. 

P.S.
I do insist on splitting Emitter resistor to get a bit of gain together with impedance (mis)matching..!!

[mdcmdcmdc]

Lovely, thank you. This is all kind of theoretical to me, so it's primarily just a good exercise for me to go from "emitter voltage on Q1 needs to come up to ~7.5v" and then figure out how to make that happen. Whether or not it's an absolutely necessary thing or a can't-hurt-might-help situation isn't a big deal.

That said, changing R2/R3 from 100K/150K to 39K/220K gives me 7.5V on Q1-e. I don't fully understand the interrelation of everything happening in the bootstrap loop/etc, so "properly" biasing this circuit might be more complicated than just adjusting the ratio of those resistors...

[antonis]

I should say 7.5V is overkill a bit but it strongly depends on voltage dividing effect during current sinking (signal negative waveform)..
(Q1 Emitter can't go all the way down to 0V due to 3k3/Load effect..)

Considering 100k Pot + 4k7 stopper resistor as "load" (omiting next stage input impedance), minimum Q1 Emitter voltage during signal's maximum negative amplitude is about 3% (45mV for 1.5V_peak) so Q1 Emitter should be ideally biased at (power supply minus reverse polarity diode forward voltage drop) / 2 plus 45mV..
(we intentionally omit BJTs V_CEsat..)

So, get back to 100k/150k bias configuration both for the above said reason and for not dominating buffer's current driving ability 'cause (as said before) Emitter total Load is considered R6//(100k pot + 4k7)//R3//R2..

[mdcmdcmdc]

@antonis - if I bridge Q1 emitter to Q2 collector with a 2.2K resistor and change the Q1 emitter resistor to a 1K:



I get no output from the circuit... doesn't seem to work! A jumper between Q1-E and Q2-C w/ a 3.3K emitter resistor works fine.

[antonis]

Weird enough..
(I can ensure you it DOES work..) 

[mdcmdcmdc]

Ahhh, I didn't have the load resistor (R2 in your schem) in the circuit when I tried it and I was taking the output direct from your C1. I bet that's it... I'll try it again this afternoon.

[mdcmdcmdc]

Hmm, very weird - I've pulled that section apart and rebuilt it on the breadboard 3 or 4 times now and still nothing per your schematic. It still works fine with a jumper and a 3.3K (although it has suddenly started motorboating when there's no instrument attached, which is new). I noticed you switched to a 22uF cap instead of the 10uF between the between bias resistors and Q1-E, so I switched that on mine as well and still no output with the 2.2/1K setup.

I even measured every component and tested both transistors and everything seems to be working as it should. I'll give it another shot later on with fresh eyes, I feel like it has to be something pretty simple...

EDIT: switching the 22uF back to 10uF stops the motorboating.

[antonis]

Never mind..
(it was a suggestion for getting a bit of gain but you don't actually need it so let it be..)

[mdcmdcmdc]

Oh there's plenty of gain for sure—it's the frustration of knowing that something *SHOULD* work but not being able to figure out why it won't.

[antonis]

Oh there's plenty of gain for sure—it's the frustration of knowing that something *SHOULD* work but not being able to figure out why it won't.

Try it without bootstrap cap..
(just a thought..)

[mdcmdcmdc]

Try it without bootstrap cap..
(just a thought..)

Still didn't work, so I'm OK with just letting it go...

Spent some time balancing the gain and rolling off the high end here and there and I think it sounds... pretty good? It gets a hint of that kind of torn speaker crackle with the gain maxed, which I like. I think it's time to set it aside for a while.

If you feel like breadboarding it, I'd love to hear your thoughts!

[iainpunk]

what Antonis and I were talking about is basically a bunch of theory for ''proper'' buffer design mainly applicable in control systems and Hi Fi equipment...

@mdcmdcmdc: The above in an inside joke 'cause I love to theoretically tease Iain when he speaks about practical solutions and, of course, also practically tease him when he speaks about theoretical issues.. 

yes, but it has made me a better designer/engineer because of it, so thank you.

the progress of this design is/was very interesting to follow so far, and im sure you learned a lot from this process. can't wait to see it to completion, and maybe even build a perfboard of it myself

cheers, Iain

[mdcmdcmdc]

My significant other came in the room earlier and asked me what I was doing, and I explained that two strangers on a DIY messageboard were helping me tweak the buffer section of a fuzz pedal circuit, and she said "I wish my internet was more like that," wistfully.

So yes - thanks greatly for the help so far! I've learned a bunch and it looks like this might turn into a fairly usable fuzz pedal.

[antonis]

this might turn into a fairly usable fuzz pedal.

What..?? 

Without an output buffer..?? 

[mdcmdcmdc]

What..?? 

Without an output buffer..?? 

"Pete Cornish has entered the chat"

[mdcmdcmdc]

Pulled the breadboard for this off the shelf, ripped it up and started again...

Musical interlude first, schematic second...



OK, so here's where I'm at:

I rebuilt the original version of the buffer and A/B'd it against the one that iainpunk and antonis helped with; tbh the first version sounded better so I went back to that one.

Adjusted the attenuation and LPFs between differential stages.

I added a basic SWTC2 tone control, which seems to work quite nicely; I haven't messed with the values at all, but the suggested starting ones definitely sound a-ok.

I added 560pF caps between C and B of basically every stage to help with the HF noise and minimize string wiping. It seems to have helped a lot.

Per Antonis' ribbing I added an output buffer in the form of a ce-cc cascade stage (does this count as a buffer? maybe?) at the end to give a bit of gain recovery as well after the SWTC2.

There's still some filtering that needs to happen as there's some HF noise, but I'm also using a jazzmaster which is maybe the best/worst guitar to breadboard with.

Current big issue to suss out is when I roll the volume off on the guitar the pedal sounds like a vacuum cleaner; huge low frequency hum. Not sure where that's coming from... overall it's getting closer though, and it sounds pretty ripping.

[mdcmdcmdc]

Dug into it some more this afternoon and made some good progress chasing down the noise. Not sure if this is possible, but I think the buffer was maybe too efficient?? Pulling C5 out (the 47uF cap from Q2-E to 9V) helped a lot and then attenuating things with a 51K resistor before the distortion pot basically fixed everything. I was able to more than recover whatever was lost by letting  some gain and high end back in after the first differential amp by reducing the R15 from 51K to 33K.



I think it may be at the point where it's kinda hard to tell what's noise/interference related to the design and what's related to it being a sprawling mess of leads and wires on a breadboard...

[Steben]

Starting to look like an amp channel