Unexpected. Found the magic in CD4049 pedals

[brett]

I've stuck my nose into the hex inverter (CD4049UB) designs after trying to recreate a particularly nice one I built 15 years ago, then gave away. 
I'm not sure if it was a Tube Sound Fuzz (Craig Anderton) or a Forty Niner (Mark Hammer).  There's also the Red Llama (?), Whipped Llama and more.
I just couldn't get the soft, fuzzy warm tone of the old one.  It's only 2 inverting gain stages, with a gain-setting pot on the first, and a high-gain second stage.  Both have treble roll-off via resistor-bypassing small caps (10pF and 10pF in theTSF, 51pF and 100pF in the RL).
Output via a 10uF cap and 10kA pot.
So my input signal is a square wave.  Coz I'm interested in seeing them "knock off" and finding where the creamy tone comes from.
Here's the input with the stats on it (excuse my $35 'scope).



And with the gain control of the first stage at about 3 o'clock (gain of just 2 or 3) here's the signal being fed into the second gain stage.



Note the amplitude Maybe affected by the output impedance of Stage 1 and the 0.1uF coupling capacitor?

Next is the BIG gain stage.  How much is uncertain, but probably 10 to 20 (depends on the output impedance of the previous stage, about 60k? and the 1 M feedback resistor. 1,000/60 = 16).  Here's the output.



Whoa!! It's back to a square wave.  And it looks like it's "hitting up" against the supply rails (+5.5V and 0V in this case).

So where is the creamy magic?  Where's the softening due to those feedback caps?  No wonder the values used in different designs vary by an order of magnitude.  Although my square wave input would have affected the result a little, it's fairly obvious these gain stages are good at clipping signals into a square wave.  Presumably, sine waves as well.

So there was only one, seemingly trivial thing to do: 'scope the output to see if a big electro cap and a pot changed things a little.  Here's the output with the volume at about 11 o'clock on the pot.



Bingo! That's a whole lot more magic.  Firstly, there's a LOT of signal loss.  Arpund 90% of the amplitude.
On the top leading edge, at least. But also a whiff of a rounded corner on both lower corners.
My "Eureka" moment is realising that the current driving and current soaking capacity the CD4049 stages are Asymmetric AND quite Low. Just a mA or two. 

I've just thrown out almost everything I thought I knew about these designs.

My first guess at what happened last time, to get a super soft fuzz, is that I used a 1kA or 5kA pot instead of a 10kA.  Or maybe not?

I'm also thinking (from memory) that Mark Hammer's FortyNiner has a 1k resistor, 10k (B?) Tone pot, and then a 10kA volume pot.  A lot of extra resistance.

I'll come back and add notes as I find out more.  I'm hoping that adding clipping diodes to the second stage will pass a more useful waveform to the cap and potentiometer  and output jack.
(And yes, the output waveform changed shape as well as amplitude as I raised and lowered the volume pot.  From memory, high volume = square wave, low volume = rounded square). 

[amz-fx]

You might find some additional ideas in here:

https://www.diystompboxes.com/pedals/insanity.GIF

Also, you might try a 0.1u capacitor in series with the diodes in your schematic.

regards, Jack

[Fancy Lime]

Interesting. Did I understand correctly thet the second to last image, the 5.5V square wave was taken after the second stage while the volume pot was disconnected?

My understanding of your findings is that the wave form produced by the second stage depends on the load that this stage sees at its output. To get your chased after sound, independent of the volume setting and connected next circuit, I would suggest adding a simple BJT buffer (like those in Boss pedals) before the volume. You can the adjust the bias resistor of this buffer to taste. 10k seems a good starting point. I would try this myself, if I had time, but that may be a while. I have in the past had great results when putting a low input impedance unity gain cmos stage after the main cmos clipping stage. I never asked myself why, but it may have been for the reason you describe.

Andy

[Rob Strand]

I'm not 100% on your connection but are you achieving any more than adding an RC filter at the output, after the chip ?

[brett]

Hi again. 
Fancy Lime:  the big square wave is at the output pin of the IC, WITH the volume pot attached (no load on the wiper of the pot).
The last wave is taken from the wiper of the output pot. 

Rob Strand: Quite likely this IS some sort of combination of RC filters, possibly with one or more parts being internal to the IC.

For me, the key finding is that there wasn't much wave-shaping in the inverter stages. The wave-shaping was in the interactions between the output of the inverter and the output socket.

Will build a sine wave test oscillator and try clipping diodes later today...

[Rob Strand]

Quite likely this IS some sort of combination of RC filters, possibly with one or more parts being internal to the IC.

Yes indeed.   The output impedance of those CMOS gates is around 1k, depending on supply voltage.  So if you load it down with enough capacitance it will form an RC filter.

You can try playing around with the *chip* supply voltage.   The resistor in series with the supply can set the chip supply voltage.   Ideally the supply is bypassed with a large enough cap.

[Mark Hammer]

The late Charles R. Fischer had a project in Electronic MUsician, that used a 4049, controlled by an LM334 current regulator, as its distortion element.  JD Sleep had it on the GGG site for a while (may still be there, but not immediately evident).  Here's a scan of it.







The text accompanying the project notes that:
      " Like any other guitar-effects box, the Spectra Distortion Generator can be
      placed in the signal chain at the beginning (i.e., between the guitar and
      the input of the amplifier) or in an effects loop. However, line-level
      loops will overdrive this unit, and it won't sound right, so use it only
      with guitar-level signals.
      I invited some guitar players to evaluate the sonic qualities of the
      Spectra Distortion Box. A typical comment was "It has a fairly wide range
      of solid-state-type distortion textures, very reminiscent of 1970s styles.
      Depending on the settings, you can get a sound similar to the MXR
      Distortion+. Other settings make it sound like the old Sunn Beta
      amplifier." This makes a lot of sense, because Sunn also used a digital IC
      biased for linear operation in its overdrive circuits.
      The really cool departure from these traditional distortion sounds is the
      Spectra control. By setting just the right amount of Drive and Spectra
      distortion, you get that MXR Distortion+ sound with a funky compressor
      built in. It reminds me of the Neil Young tune "Hey Hey, My My," but
      Fischer's Spectra Distortion Generator has qualities not available in any
      of my stomp boxes.
      What's going on to produce this effect? When the digital chip is driven to
      draw enough current from the power supply, it causes the regulator chip to
      start limiting the current. In addition to changing the waveform in some
      unique ways, it also limits the amplitude, thus compressing the signal at
      the same time. Of course, as is usually the case, you can have too much of
      a good thing. We found some settings that are not very usable in our
      application, but the Fischer Spectra Distortion Generator is definitely
      worth experimenting with."

[Rob Strand]

The late Charles R. Fischer had a project in Electronic MUsician, that used a 4049, controlled by an LM334 current regulator, as its distortion element.  JD Sleep had it on the GGG site for a while (may still be there, but not immediately evident).  Here's a scan of it.

For a multistage 4049 you will probably need a cap across the 4049 supply.   Then it works a lot more like a resistor.