Should be quite touch-sensitive. The latter two transistors might as well be a Darlington transistor. It needs no gain recovery stage as the output is already hotter than the input even at minimum gain. The Framus mid control has been slightly modified as to roll off treble up to mid-point, then actually recover some of that treble while the mids are cut (but there is a treble shelf cut before the mid control to soften the highs a bit).

U1 is rev log, U2 is linear.

I'm going to breadboard it at the weekend.

Quote from: fryingpan on April 16, 2025, 02:48:25 PMThe latter two transistors might as well be a Darlington transistor.

Better a CFP (Sziklai) with a 10k on Q3 Collector (across Q4 B-E), R10 3k3 and R9 3M3..

As it is, output is biased at about 3.2V (actually, a bit lower) hence Q4 current about 570μA..
Output at 4.75V and about 1.5mA working current should be better, IMHO..

Quote from: antonis on April 16, 2025, 03:34:47 PM

Quote from: fryingpan on April 16, 2025, 02:48:25 PMThe latter two transistors might as well be a Darlington transistor.

Better a CFP (Sziklai) with a 10k on Q3 Collector (across Q4 B-E), R10 3k3 and R9 3M3..

As it is, output is biased at about 3.2V (actually, a bit lower) hence Q4 current about 570μA..
Output at 4.75V and about 1.5mA working current should be better, IMHO..

I wanted to keep current consumption low. As it is, it's below 1mA (not counting the LED). Could work as an onboard battery-operated fuzz. The low bias is not a problem because negative swing is limited by the diode.

Quote from: fryingpan on April 16, 2025, 03:38:06 PMI wanted to keep current consumption low.

May I ask what impedance is intended to drive..??

Whatever you'd plug a guitar into. I see your point though. With such a low bias, the minimum load is going to be highish. I could increase R9 though to 3.3M or something.

And use a Sziklai for ony one V_BE drop..

I know you'll do whatever you want  but in case of utilizing a Sziklai configuration for output buffer, don't omit to place a 10k, or so, resistor across Q4 B-E..

It should be considered Q3 Collector load, resulting into almost constant Q1 current with dual effect:
1. Q3 current much greater than Q4/h_FE2 hence more acceptable Q3 h_FE..
(convenient for both total current gain and bias configuration high impedance..)
2. Good Q3 V_BE stabilization..
(if Q3's current is X% of Q4's Base current at peak output, then Q3's Collector current varies only within a range of V_T*lnX.. - Q3 should live in an almost linear wonderful world.. )

Sziklai is fine, after all I guess most people get their transistors off a "500 transistor kit" with all sorts of BJTs, and you gotta use those PNPs somewhere... Of course, linearity and absolute stability are not a big concern for a fuzz, but they don't hurt either, and the first stage is capable of 36dB gain as it is (could be higher still).

My only doubt is with the tone circuit and the treble cut through the cap across Q2 C-E, which will need some tweaking. The fuzz as is will be better for bass than guitar, probably, it would be nice to have a mid peak instead of a mid cut but that would need at least another amplifier.

Quote from: fryingpan on April 17, 2025, 07:03:32 AMlinearity and absolute stability are not a big concern for a fuzz

I beg to differ, sir..

A self-respecting output buffer has to pass unaltered whatever it receives..

Quote from: antonis on April 17, 2025, 05:06:44 AMI know you'll do whatever you want  but in case of utilizing a Sziklai configuration for output buffer, don't omit to place a 10k, or so, resistor across Q4 B-E..

It should be considered Q3 Collector load, resulting into almost constant Q1 current with dual effect:
1. Q3 current much greater than Q4/h_FE2 hence more acceptable Q3 h_FE..
(convenient for both total current gain and bias configuration high impedance..)
2. Good Q3 V_BE stabilization..
(if Q3's current is X% of Q4's Base current at peak output, then Q3's Collector current varies only within a range of V_T*lnX.. - Q3 should live in an almost linear wonderful world.. )

With a 10k collector load on Q3, base current is more than 1/10th the current through the biasing voltage divider... 15k makes it better. Or does it not matter much in this case because the configuration is inherently stable?

Quote from: fryingpan on April 17, 2025, 09:16:03 AMOr does it not matter much in this case because the configuration is inherently stable?

  

Anyway, with 10k or 15k on the collector, there doesn't appear to be much difference anyway.

This is my current idea (tomorrow, hopefully, I'll be able to breadboard it):



A possible change could be a 250kohm volume pot so as to present a heavier load to the bass cut (switchable).

Noise analysis also says that the fuzz should be very "clean", noise-wise.

A lot of what you guys are discussing is going a little over my head, but one thing I'd like to say I really like about this circuit is that it's pretty much all everyday parts.  Nothing wrong with building some bespoke fuzz with some unobtanium transistors crafted by Tibetan monks on a frosty mountaintop.  But I always enjoy seeing circuits built with good ole bog-standard 2N3904/2N3906 transistors.


Did either of you guys ever breadboard any versions of this?  If so, how'd it turn out?

Quote from: phaeton on April 23, 2025, 10:17:25 PMA lot of what you guys are discussing is going a little over my head, but one thing I'd like to say I really like about this circuit is that it's pretty much all everyday parts.  Nothing wrong with building some bespoke fuzz with some unobtanium transistors crafted by Tibetan monks on a frosty mountaintop.  But I always enjoy seeing circuits built with good ole bog-standard 2N3904/2N3906 transistors.


Did either of you guys ever breadboard any versions of this?  If so, how'd it turn out?

I was going to do it last weekend but work got in the way 

The fuzz should work, but of course the component values will need tinkering (especially coupling caps C2-C12, feedback cap C5 and treble bleed cap C9). Input impedance is low-ish but that's good for a fuzz (makes it less harsh), it could go lower but I'd rather work on C5 so as to make the Darlington stage cut the treble lots at higher gains.

By the way, the circuit already lowers the impedance it presents to the guitar with increasing gain. There are two phenomena at play here in making the circuit behave nicely for a fuzz, the Miller effect of ever increasing shunt capacitance in relation to circuit gain (which rolls off treble) and the lower value of the emitter branch itself as you increase gain (the pot is a rev log part, wired so as to decrease its value as you increase gain).

One addition I'd make to the circuit is a series resistance, something between 30k and 100k, to the tone pot. It controls treble a bit better. Also, some cap values are overkill for guitar, keeping in mind that I always design with bass in mind. The input cap could go down to 10n or even lower, the coupling cap after the Darlington amplifier can probably go as low as 22n with no apparent change in sound. And certainly the emitter bypass cap can easily be 22u.