"Uncanny Valley" - analog bitwise wave manglor

[earthtonesaudio]

Just breadboarded the one with the 4051 and a bunch of pots.  On the scope it looks pretty much like the simulated scope shots Taylor put up earlier.  Fun.

[Taylor]

The first thing I was thinking would be interesting, beyond multiplexing dc signals, would be to have some oscillators going into the channel inputs. A 40106 wired as 6 oscillators would do nicely.

The result might be just a racket, but in principle it's kind of like frequency shift keying, which I would wager has not been yet exploited as a guitar effect. This might actually work better on bass than guitar, since the lower the input frequency, the more time spent on each channel, so might sound less like pure noise and more like an intentional sound. I think we're onto some pretty sweet stuff here.

[artifus]

how about a 4040? fuzz to 4040 to 4051?

fun with sea moss
cmos synths

[earthtonesaudio]

@Taylor:  You're making me want to get a bass.  I would imagine 6 squarewave oscillators would be a horrendous racket, even for my tastes.  In a completely opposite direction, you could use the 4051's common as an input for the raw audio, and the 6 outputs could be 6 separate voltage dividers.  Kind of like a compressor or tremolo but with six discrete, and arbitrary, gain settings.
@artifus:  Complete madness.

[Taylor]

But the thing is that you would only hear one oscillator at a time, so if the frequencies of the oscillators were in the same range as the instrument input, they would create new harmonics which might only be heard every few cycles, or at even longer periods. With low audio range oscillators or lower, it would be like a bit crushed fuzz but with slowly evolving harmonics. I think.

If the oscillator frequencies were around 2-5 times the instrument frequency, it would create some FSK-ish sounds, which I think would be in the realm of Speak N Spell-style synthesis.

Lots of potential with these ideas.

[earthtonesaudio]

Ah... I was thinking the oscillators would be higher than the audio.  Near the fundamental frequency or lower would indeed be interesting.

[Taylor]

I mocked up a bunch of these ideas in a modular computer program (Plogue Bidule). Some very interesting and weird sounds.

I first tried a bunch of free running square wave oscillators into the multiplexor inputs. All kinds of nasty weirdness. This was definitely bordering on harsh noise. But as I thought, lower frequencies than the fundamental made for more usable sounds. I think this is worth building. You can have 6 square waves using a 40106, 6 caps and 6 pots, so it won't be a tough build.

Then I tried something emulating a cd4040 into the inputs - in other words, as per artifus, I divided the instrument's frequency by 2, plugged that into the first channel, then divided that by 2, plugged it to the second channel, etc. So the multiplexor switched between differently divided versions of the input. This was pretty cool. Sounded much like some FM electronic bass sounds.

I also tried something where the square waves were all the same freq as the input, but with different phase offset. That created some really cool tonewheel organ sounds - not straight sine-y octaves, but real organ sounds which are kind of "grunting" and nasal, or almost like a bassoon or oboe.

If I get time to really build some of these circuits, I think we'll have some pretty fresh stuff on our hands. More unique than a hundred boutiques combined. 

[a soBer Newt]

I cant wait till you get these ideas ironed out I will definitely will be buying pcbs if/when they are available.

Sabata Forever!

[defaced]

I mocked up a bunch of these ideas in a modular computer program (Plogue Bidule). Some very interesting and weird sounds.

Just what I did not need to find out existed.  That is beyond cool.  Their chip sounds software sounds really cool too.  There goes my Saturday...

[Taylor]

Yeah, Bidule's awesome. I used to use it a lot for my audio experiments before I got into building pedals. It's a million times more flexible and easy to build stuff in there than with hardware, so when I started working with pedals, I was actually really disappointed with how much you have to fight with physics.

I actually haven't done much with the program in years, but I pulled it out to do those tests and remembered how cool it is. It's much of the flexibility of Max/MSP or Puredata, but a lot more intuitive to me.

[Strategy]

I use Audiomulch, has been around for years, is similar to Plogue and extremely fun.
www.audiomulch.com

Yeah, Bidule's awesome. I used to use it a lot for my audio experiments before I got into building pedals. It's a million times more flexible and easy to build stuff in there than with hardware, so when I started working with pedals, I was actually really disappointed with how much you have to fight with physics.

I actually haven't done much with the program in years, but I pulled it out to do those tests and remembered how cool it is. It's much of the flexibility of Max/MSP or Puredata, but a lot more intuitive to me.

[Taylor]

Oh yeah. I think I got into Bidule when I was a Mac-only person, and AM is/was not available for Mac. AM is pretty cool, I think it was my first foray in computer music, way back before I had ever heard any of this kind of music. I remember playing around with that, and my friend asking, sincerely, whether there was anybody who would actually make this kind of "music" for people to actually listen to.   The "metasurface" is amazing, and that's the kind of thing that totally kills Bidule as a live tool.

To bring it back to hardware, I think these can be great tools for understanding circuits conceptually. They don't model electrical characteristics, but some ideas like the ones in this thread are so abstract that it's necessary to get an idea for what they will really do with audio before trying to build them.

[artifus]

@artifus:  Complete madness.

i'll take that as a compliment

[Top Top]

I mocked up a bunch of these ideas in a modular computer program (Plogue Bidule). Some very interesting and weird sounds.

I first tried a bunch of free running square wave oscillators into the multiplexor inputs. All kinds of nasty weirdness. This was definitely bordering on harsh noise. But as I thought, lower frequencies than the fundamental made for more usable sounds. I think this is worth building. You can have 6 square waves using a 40106, 6 caps and 6 pots, so it won't be a tough build.

Then I tried something emulating a cd4040 into the inputs - in other words, as per artifus, I divided the instrument's frequency by 2, plugged that into the first channel, then divided that by 2, plugged it to the second channel, etc. So the multiplexor switched between differently divided versions of the input. This was pretty cool. Sounded much like some FM electronic bass sounds.

I also tried something where the square waves were all the same freq as the input, but with different phase offset. That created some really cool tonewheel organ sounds - not straight sine-y octaves, but real organ sounds which are kind of "grunting" and nasal, or almost like a bassoon or oboe.

If I get time to really build some of these circuits, I think we'll have some pretty fresh stuff on our hands. More unique than a hundred boutiques combined. 

CMOS logic stuff seems to be a whole world of untouched guitar FX. I have messed around with it a bit and have a few things that I can throw down on the forum when I get a little time for it.

[artifus]

Chapter 24: Analog to Digital Conversion, Sort of

*edit* Tim Escobedo has some cool cmos stuff too.

[earthtonesaudio]

I discovered something somewhat interesting on the breadboard.  Using a TL074 for the "comparators" I am able to generate every word except for 101.  At certain input voltages the op-amps oscillate and/or are slewing so that it is possible to generate the word 010, even though ideally it should not. 

I feel like it should be as simple as placing one extra resistor somewhere to generate all 8 words (reliably), but so far I've been unsuccessful.  Using two comparators and the same general scheme it's trivial to get 00,01,10,11 in order, but for some reason 3 is proving difficult.

[Jazznoise]

Isn't 101 5? Those 4 comparators should never be able to generate 5 - there's four of them! You sound like you've changed the design but I'm not sure what you've done.

Was thinking of building this. The osccilator part sounds very interesting. I wonder how it'd sound if you used the guitar signal going through a D-Type flip flop as your osc source!

[earthtonesaudio]

Finally, I was able to generate all 8 words (in order) using this circuit:



It may use an entire TL074 but only 3 of them are wired as comparators.  I hooked the outputs up to LEDs to test and indeed when I put in a DC sweep at the input I get
ABC
000
100
010
110
001
101
011
111

The supply voltage is +9V and GND and AGND=4.5V (junction of two 1k resistors with a 10uF cap).  This actually matters because AGND must be low impedance relative to R3/R4 or amp C gets squirrely.  The only other caveat is that TL0** series op-amps will do the phase reversal thing if you bring their inputs too close to their negative rail.  Since I'm mostly concerned with small signal inputs this is no big deal.  I make R4=10*R3 and similarly R2>R1.  This puts all the voltage references near 4.5V/AGND.

So far, with a DC voltage as the source signal, it works perfectly.  I'll report back once I've put a guitar through.

Isn't 101 5? Those 4 comparators should never be able to generate 5 - there's four of them! You sound like you've changed the design but I'm not sure what you've done.

Was thinking of building this. The osccilator part sounds very interesting. I wonder how it'd sound if you used the guitar signal going through a D-Type flip flop as your osc source!

When I was mentioning generating word 101 and 010, I was referring to the schematic in this post, not specifically to Taylor's original.

[Paul Marossy]

Interesting idea Taylor. Got any sound clips of this mad circuit?

[earthtonesaudio]

I discovered an error in my circuit above.  As drawn it can't generate the words 001 and 110.  Adding a resistor R_c from the junction of R1/R2 to the output of IC1C solves this.  A starting point for values would be R_c=R2>R1.