tube fuzz

  How's that for a good-looking schematic.  I know it looks like I did it with a sophisticated CAD program, but believe it or not, it's actually hand-drawn.

Yes, I think the Zener diode level shifters may be an important piece to this puzzle.  Without the Zeners, i.e.  with the simple resistor dividers as level shifters, I'm up against a no-win situation as regards gain.  The tubes must be operated near voltage midpoint, i.e., with large cathode resistors.  This implies gain of 1.  In this configuration, all three resistances, i.e., the plate resistor, the effective resistance of the tube plate-cathode path, and the cathode resistor, are approximately equal.  To shift the plate voltage down to the proper grid voltage, requires at least a 1:1 resistor ratio, i.e., gain of 0.5. There seems to be an inevitable no-win voltage-scaling situation, in the absence of Zeners or other non-resistor level shifting.  Operating the tubes so that the target voltage at the grids is below midpoint to increase the DC gain above 1, means the plate output voltage must then be scaled down with a more-aggressive resistor ratio, ...

Thinking more about this DC gain issue.  If, indeed, what is needed to create the magical fuzz effect is relatively high DC gain, well, I sure don't have it. The innate DC gain of each of my tube stages is right around 1.  But then, I run through resistor dividers: one between the two sections, and the other (effectively) in my variable-bias circuit.  In each case, the divider is acting as a level-shifter, to bring the high DC level of the plate outputs down to the bias voltage required on the grids, which is considered to be roughly 4 volts below the cathode voltage.  That's why I'm running the tubes with large cathode resistors, to raise the target grid voltage up to closer to the plate voltage.  But nonetheless, the target voltage will always be well below the plate voltage, since the plate-cathode resistance of the tube is effectively somewhere around 60k.  So I need to shift the levels downwards, but the gain-reduction of the resistor ladders is an unf...

So as I continue to wait for the replacement of this digital oscilloscope to arrive (maybe by this Saturday, they're now saying), I'm running through thoughts and ideas in my mind about how to make the circuit work better. I've decided not to do any further work on the actual physical prototype, until I have a scope and can thus see what I'm doing.  This is something I've learned, in my experience of designing new circuits: it's very important to manage my own energy and emotions, and not let myself get overwhelmed by too much disappointment or failure at any one time.  Building circuits designed by someone else is a totally different endeavour.  In that case, there's a known standard of performance that I'm seeking to replicate.  If it doesn't work at first, it's always possible to keep trying and searching for the mistakes, and with enough effort, it's guaranteed to work eventually.  With a novel design, there are no guarantees, and so one ...

https://youtu.be/2Xx6cmFttiA

 Just sharing some of my research for those who might be interested. So my best candidate appears to be the Hantek 6022BE.  This is a 2-channel, 20MHz USB scope (so no built-in display or controls, just two BNC jacks).  Hantek themselves only provide Windows software, but there's a free software project called OpenHantek which specifically supports this model, and sounds like it's actively maintained.  The 6022BE costs about $65 on the Internet. For those who want to pay more and get more, there's also the BitScope BS-10, at 100MHz, with 8 logic channels as well as the 2 analogue channels, signal generator, and considerable other nice features.  They explicitly support Linux, and also Raspi, etc..  These cost about $245.  Worth it, I'm sure, but I'll be trying the cheap one first! ---- (Later...) So far, nothing but failage. I bought the Hantek 6022BE from Amazon. Hooked it up to my Debian Linux PC. Executed (as root): snap install openhantek snap conn...

 I thought maybe I should try reducing the input capacitor to a "normal" value: maybe too much bass response was making the sound gritty and less-optimal, or whatever; maybe there would be little to no change, in which case, no need to keep the anomalously large cap value which I had simply copied from the (nutty) transistor circuit.  So I swapped in a 22n.  And functionality completely ceased!  Putting the big 10u electrolytic back in, the sound returned.  Strange. So I take this as an indication that the circuit is very unstable, operating on the edge of ultrasonic oscillation probably.  I really need a scope at this point, but unfortunately, my scope is 1400 miles away in San Diego.  And wow, digital scopes aren't that cheap!  Didn't I recently see little digital scope modules from the Internet in the under-$50 range?  Not finding any such now, everything is well over $100, even for the plasticky toys.  Darn.  Still looking... Wi...

Here, thanks to electrosmash.com, is a nicely-annotated schematic of the famous Big Muff distortion pedal.  This is often called a "fuzz pedal", but as you can see, at least according to my newly-developed definition of fuzz, this circuit is not it.  There is no adaptive biasing feedback path.  It's just a pair of cascaded diode clipping stages, plus buffers and tone control.  This comports with the fact that Big Muffs never seem to produce what I call the fuzz effect, although they definitely serve up plenty of thrashy distortion. Same goes for the ProCo Rat pedal: Again, no active feedback-biasing, just a diode clipper under the hood, driven by a fixed-EQ mid-boost stage.  I've heard people call the Rat a "fuzz pedal", but according to my definition, it is not any such thing.  I think these pedals are called "fuzz" just based on what I consider to be a misunderstanding, the notion that there is a single one-dimensional axis of "amount of dis...

So I got my dual-gang pots in, and hooked up the full two-tube circuit.  And... it's better...  But still no "home run", I think.  There's a good amount of gain, for sure, and I do hear the time-varying effects that I associate with "true fuzz", as opposed to mere overdrive or simple distortion.  However, it seems like there's a lot of tweaking and adjustment to be done, to really dial this circuit in. For one thing, the dual-gang pots are dual 250k, whereas I was using 500k and 1M pots before because that's what I had, just to get started.  But now, the one-tube mode seems to have a lot less gain than it did before.  Given the nature of the circuit, I don't quite understand how decreasing the pot values can have reduced the gain, but that's the main change I've made.  So clearly, I don't really understand how this circuit is working, vis a vis how it's *supposed* to work. Also, as I turn the bias pot, there is a time-lag effect d...

Well, I wired up basically what is seen in the schematic diagrams I've posted, plus-or-minus some different component values depending on what I had available.  I only ran one tube stage, because I'm still waiting for shipment of my dual-gang pots.  And initial results are, well, promising -- but not yet "home run" territory. With one stage, into my Fender Champ, well, it definitely distorted.  Did it do anything different or more than just a regular tube overdrive stage?  It's hard to say, but I think maybe so.  Almost any kind of tube overdrive or boost into the Champ tends to sound pretty awesome.  It definitely produced some tasty distortion which was well beyond what a Champ can do on its own, but this was to be expected.  At the very max limits of the controls, it seemed like I could begin to detect a slight edge of that "fuzz thing", that shifting vocal quality that seems to ride the envelope of the incoming signal.  But I'll have to see ho...

 

  So here you can see the two identical fuzz stages and the switch wiring which interconnects them. The large 10u electrolytic is only on the input of the first stage; I'm not sure this large value is needed or wanted, I'm just blindly following the general design of the transistor version.  It seems like there are some slow, long-time-constant behaviours to the classic fuzz tone, and my supposition is that maybe this is due to some of the large capacitor values: both on the input, and in the gain control circuit. The other coupling capacitors are standard tube amp values, 22n (or could also be 20n).  In 1-tube mode, the first stage is loaded by the 1M output pot.  In 2-tube mode, the first stage is loaded by the feedback network of the second stage, 33k into something like 500k equivalent to ground; and the second stage is then loaded by the 1M output pot.  Thus, there are no explicit load resistors.