WaveThing Multiply/Divide Wave Shaping Circuits

Scott Stites · Posted: Sun Mar 29, 2009 8:25 am Post subject:

Well, that track never made it. Laughing

I recorded a whole series of samples yesterday morning, and thought I'd upload them last night. I uploaded the wrong sample for the first, then the real one wouldn't upload - it finally did, and I set the last one up, hit "submit" and went to bed. It was still waiting this morning, so I hit "submit" again - the post made it, but the sample didn't. Snow, I s'pose....

Anyway, right Ian - Sawtooth Animator. I haven't posted any more schematics because I haven't drawn anything. Automatic mixing would be cool. It would have been neat to make only four octaves and use a quadrature oscillator (BTW a sine quadrature oscillilator would be very nice, because the animator responds very well to sine). Is there such thing as a quintature oscillator? Laughing

The animators haven't made it to this re-build, yet. This is my second breadboard cycle.

Each waveform circuit is not all that big, but when you put them together, it turns into a rather large circuit. The trick is getting waveforms as pure as possible, so there's a fair amount of calibration. The problem with both multiplying and dividing is that you have to be careful to make each stage as correct as possible. Any non-linearity in one stage gets passed to the next, and this results in audible sub-harmonics in that next stage and any stage that may follow. Moreover, if the waveform you start with has some sort of discontinuity, you have to deal with that - it could be an offset (doesn't have to be much), a pip, or whatever.

If, say, your starting waveform has a pip at the peak, then you multiply it, the process leaves a pip on every other peak. Then you mutliply it again, and the pip will be on every fourth peak. Those pips would create a harmonic starting at the fundamental, so you would still hear it as an unwelcome visitor in the form of a lower tone.

The best method I've found for calibrating is to get everything looking as good on a scope as possible, then actually tuning by ear - the ear picks up stuff you don't necessarily see on the scope.

I think these waveform circuits would work "OK" for external signals, perhaps, but would be best if they were either calibrated to a specific VCO, or were part of the waveshaping circuitry of the VCO. It depends on how consistent the output wavforms are on your VCOs, I suppose.

Loss, I'll get to explanation as soon as I draw schematics and try to finalize things a bit more.

Here's a different track than the one I was about to attempt - I'm loading it because at the very end you can hear the highest octave, which is the X4 triangle. Not the best triangle, but certainly servicible. I should mention, even if your waveforms do not sound perfect alone, they always blend, really, really well with the rest of the gang.

This one has the four lower octaves (/4/2/fundamental/X1 as PWM) and the X4 is triangle. It's rather repetitive.....

Edit: Aarrrghh...never upload before coffee - I uploaded the first #$%!! file again. Reloaded correct file.

Cheerios,
Scott

WaveThing_5_O_4P1T_3hq.mp3

Description:

4 Pulses, one high triangle

Download (listen)

Filename:

WaveThing_5_O_4P1T_3hq.mp3

Filesize:

9.69 MB

Downloaded:

1500 Time(s)

Scott Stites · Posted: Sun Mar 29, 2009 9:12 am Post subject:

Here's the one with the exaggerated PWM. Again, four lowest octaves are PWM, highest is triangle.

WaveThing_5_O_4P1T_2.mp3

Description:

Klee Sequence with lots of PWM in four octaves and one octave triangle

Download (listen)

Filename:

WaveThing_5_O_4P1T_2.mp3

Filesize:

7.9 MB

Downloaded:

1398 Time(s)

Scott Stites · Posted: Sun Mar 29, 2009 9:35 am Post subject:

This one has the two lowest octaves as sawtooths, the X1 and X2 are PWM, and the X4 octave is sawtooth.

WaveThing_5_O_4s2s1p2p4s.mp3

Description:

/4 Sawtooth, /2 Sawtooth, X1 PWM, X2 PWM, X4 Sawtooth

Download (listen)

Filename:

WaveThing_5_O_4s2s1p2p4s.mp3

Filesize:

8.43 MB

Downloaded:

1430 Time(s)

Scott Stites · Posted: Sun Mar 29, 2009 9:46 am Post subject:

Again, four octaves PWM, one octave tri, fast Klee sequence, no delay or reverb.

WaveThing_5_O_4P1T_1.mp3

Description:

Four Octaves PWM, One Tri, moderate PWM

Download (listen)

Filename:

WaveThing_5_O_4P1T_1.mp3

Filesize:

4.32 MB

Downloaded:

1401 Time(s)

seraph · Posted: Sun Mar 29, 2009 9:56 am Post subject:

Scott
this theme sounds a little repetitive but the sounds are awesome Cool

Scott Stites · Posted: Sun Mar 29, 2009 10:07 am Post subject:

Yeah, Carlo, the repetion is to demonstrate the application of different waves to the same conditions. I felt that changing the pattern or the scaling would obfuscate what the WaveThing was doing.

Cheerios,
Scott

seraph · Posted: Sun Mar 29, 2009 10:14 am Post subject:

I'm with you Wink

it makes sense Very Happy

Scott Stites · Posted: Sun Mar 29, 2009 10:58 am Post subject:

Scott Stites · Posted: Sun Mar 29, 2009 11:19 am Post subject:

BTW, Ian, when I saw that SH1000/2000 VCO implementation, the first thing I thought of was the Double-Dekka. Like I mentioned before, your implementation is different - you use an ultrasonic core to drive dividers to form a single continously variable waveform. By logical extension, the Double-Dekka could be used to form various footages of different waveforms. The hard part (the core) is already done. If I understand it correctly, a version of the Double-Dekka, I think, could create octaves of fixed waveforms. It could even have an identical panel, only what the sliders do would be different - each slider could mix in an octave of a particular waveform.

The SH1000/2000 forms square and what I can only imagine to be approximations of pulse and sawtooth waveforms, ranging from 32 foot to 2 foot (there's no 2 foot sawtooth because, at that point, there aren't enough divided copies of the signal to form a sawtooth). It's all done with resistors tapping off the divide lines to mix the harmonics in proportion to what the selected waveform would put out. I think Roland used something like four or five different resistor values, just more or less of them, depending on the octave. The pulse waves are derived by directly mixing the divided outputs (on the 2000, those are done with diodes). The saws use ratioed resistors. I believe the square waves are a single copy of the divided signal for each octave.

It would be better than a 1000 or 2000, because the mix level for each waveform wouldn't be fixed by tabs. It would be more versatile than an SH-3A, because there would be more waveforms to mix. Only thing missing from the SH-3A and 2000 would be the chorus (1000 didn't have that, if I'm not mistaken). If you did that, I would call it the "Magic Bus". Laughing

Cheerios,
Scott

Scott Stites · Posted: Sun Mar 29, 2009 11:31 am Post subject:

Here's the video of the SH-1000 that sent me down this road. The section on the various footages/waveforms (which is sadly short) is what made me sit up in my seat....

Scott Stites · Posted: Sun Mar 29, 2009 3:59 pm Post subject:

And now for something totally different - I just accomplished what I set out to do since I saw the SH1000 video - I've created the WaveThing BOG:

Bottom Octave Generator Very Happy

Basically I multiplied the X4 Sawtooth up to X8, took the square wave from that, which is the top octave. That same signal drives a CD4516, which produces four octaves down from that. So, I've got:

X8
X4
X2
X1
/2

So, I've got five octaves, only in square wave form. Obviously one could go on dividing down from there.

I was hoping to get a decent enough X16 sawtooth, but that hasn't panned out - things just get too iffy. It's pretty good at X8, though.

I've got all five outputs going into my mixer - straight off the 40106 which is the X8 signal, and straight off of the pins of the CD4516. My mixer module allows me to center at zero. It sounds pretty good. This would be a pretty cool module in its own right, I think. Next thing to do is shape some other waveforms out of it, like-a-da-Roland.....

Guess since this thread has moved out of the sub-octave region, I'll re-name it....

Cheerios,
Scott

loss1234 · Posted: Sun Mar 29, 2009 3:59 pm Post subject:

scott

these samples sound GREAT!!

damn

also...are you running the output of the Klee through a quantizer before it hits the CV in of the VCO?

this really makes me want to build my Klee!!!

thanks

Scott Stites · Posted: Sun Mar 29, 2009 4:01 pm Post subject:

Nope, Loss - I don't have a Quantizer. I imagine I'll have to do some business with Fonik....

Take care,
Scott

cbm · Joined: Oct 25, 2005 Posts: 381 Location: San Francisco

Scott Stites · Posted: Sun Mar 29, 2009 6:00 pm Post subject:

That's an interesting module - sort of what I had in mind, but different yet the same. I notice this one takes a triangle input - and that frequency tripler is interesting. The mention of adjusting the triangle input level to get the right multiple suggests that perhaps it uses the Lockhart design(?).

Here's a sample of the BOG - the inputs are taken directly (and I mean directly) off of the CD4516 and CD40106. I've got the breadboard hooked up to the mixer with alligator clips. Don't try this at home. Very Happy

I'm driving it with the same VCO, controlled by the Klee. I vary this one considerably. Basically I'm changing the mix ratio of the five outputs, then the freq of the VCO, then the rate of the Klee, the scale of the Klee, etc.

If anything, you can get an ear-full of what my wife puts up with on a routine basis....

Cheerios,
Scott

WaveThing_BOG_5_O_1.mp3

Description:

WaveThing Bog (5 Octave) sample 1

Download (listen)

Filename:

WaveThing_BOG_5_O_1.mp3

Filesize:

12.5 MB

Downloaded:

1243 Time(s)

frijitz · Posted: Sun Mar 29, 2009 8:44 pm Post subject:

aerogramma · Posted: Sun Mar 29, 2009 11:01 pm Post subject:

sound clip 2 also is pretty amazing ian Wink

aero

Uncle Krunkus · Posted: Mon Mar 30, 2009 2:26 am Post subject:

There's some majorly cool stuff you're doin here Scott.
Puts my little experiments to shame. Laughing

loss1234 · Posted: Mon Mar 30, 2009 5:32 am Post subject:

keep it coming!

loss1234 · Posted: Mon Mar 30, 2009 5:34 am Post subject:

oh and the reason i asked if it was quantized scott was really because the notes sounded so IN TUNE.

as i have never used a klee before i wasnt sure what level of tuning accuracy it had. many step sequencers i have used have been hard to get sounding that pretty!

frijitz · Posted: Mon Mar 30, 2009 6:06 am Post subject:

numbertalk · Posted: Mon Mar 30, 2009 6:58 am Post subject:

Any chance of seeing a stripboard layout from someone (for someone who cut their teeth on PCBs and has made nothing but a mess out of the stripboard layouts he's tried to do, with circuits much simpler than this even Smile

)?

loss1234 · Posted: Mon Mar 30, 2009 7:08 am Post subject:

so i got this up on the breadboard!

super simple to build, especially if you start with just the TRi and Saw sections.

now my goal is to try out the possibilities of this for Audio, and as a pesudo ring mod.

thanks scott

aerogramma · Posted: Mon Mar 30, 2009 7:23 am Post subject:

loss1234 · Posted: Mon Mar 30, 2009 7:36 am Post subject:

the tri output does seem to work with a few different waveform inputs...maybe not perfectly but using my Ray Wilson VCO...i got a an octave down tri out inputting the "Sine" wave from my vco.

The tri out on this IS very heavy duty. this will be a welcome way to add nice Bass to leads. or hell, it would be nice on its own too as it actually sounds a lot better than the TRI out on my VCO.