Wave Multipliers (VCM)

The Wave Multipliers function block is made up of three sections that create harmonically rich timbres from simple input source material. Each section accomplishes this in its own way, and one section (the legendary middle section) does what it does so well that its circuit has been studied, licensed, and cloned by other synth designers.

Briefly, the top section can act as a VCA or provide tube-like distortion depending on its switch setting, and is also capable of producing ring modulator-like sounds; the middle section generates a startling array of rich timbral material by emphasizing odd harmonics to input sounds through "wave folding"; and the bottom section changes the character of input sounds by introducing even harmonics through full-wave rectification.

I should mention that the knob settings and observations of resulting wave shapes and behavior described below depend quite a lot on the amplitude of the source signals. I was generally operating with 5V PP source signals. But if your source has larger voltage swings (cf. TGO vs PCO signal levels), your results will vary. Of course, that applies to any module, not just this one.

Top section — “The Ring Mod-ish Distortion Box”:

The top section is comprised of one audio input (IN), one voltage-controlled input (VC) with attenuator knob, and one (inverted) output (OUT). There's also a switch labeled LO/HI.

IN: With audio inserted here, and with the switch set to LO, only amplitude is affected when the VC knob is turned up, with an incoming 4V PP signal reaching line level at around 11 o’clock (8V PP reaches unity at about 10 o'clock). Larger amplitude source signals (>5V PP) result in subtle rounded distortion for all gain levels. This section is capable of 2x amplification or more, depending on the amplitude of the source. 

When the switch is set to HI and the knob is turned up (or control voltage is applied to VC), the signal is amplified and the waveform’s peaks and troughs become more and more rounded (aka distortion): a triangle becomes sine-ish; a sawtooth is still sawtooth but with rounded teeth; a sine is squished. The distortion isn’t a crunchy, nasty distortion, but more like a subtle change of character as harmonics are introduced.

VC: Unlike the other sections, audio inserted in the VC input will only work if audio is also fed to IN. Read on…

Ring Mod-ishness — Not hidden, but a non-obvious top section feature

The early Serge catalogs only discussed the top section in terms of a single input source that was acted upon in one of two ways dependent on the switch setting. But this section becomes even more interesting when audio-rate signals are used at IN and VC at the same time — producing a sound akin to ring modulation, which is when two input signals are multiplied, and their sum and difference frequencies (called heterodynes) are output, with the original input signals removed from the output. The heterodynes are rich in partials, and depending on whether those partials are harmonically or inharmonically related, the result will sound related to the source(s) but with new spectral characteristics, or may sound metallic or bell-like.

So when two audio signals are fed into the top section, I say they produce a ring modulator-ish effect because the overtones (a generic term encompassing harmonic and inharmonic partials) sound like the sum and difference result of a ring modulator, and yet the original source inputs can still be heard.

Switch LO: The RM-like effect is produced with a pair of simple wave forms, the effect becoming more pronounced as those oscillators’ frequencies (or other VC parameters) are modulated. The combined levels of the two sources sound evenly matched with the knob at 12 o’clock. Otherwise audio at IN responds linearly over the entire knob range, while audio at VC responds exponentially up to about 12 o’clock and then flattens out.

Switch HI: The RM-ish effect is even more pronounced. With simple wave forms as sources, knob changes are very subtle over the entire range, so the more interesting timbral changes are evident through changes at the source (e.g. frequency, wave shape, etc.). With more complex material, knob changes become more pronounced, with one or the other source responding more or less than the other. Experimentation is key.

Practical use notes — Top section 

With one audio source (switch LO): Acts as a simple VCA: For a 4V PP source (e.g. a triangle from a PCO/NTO), turn the knob down from 11 o'clock to attenuate the audio; turn the knob past 11 o'clock to increase the volume (generally up to 2x, but maybe more). Increase the source amplitude beyond 5V PP to introduce subtle rounded distortion.

With one audio source (switch HI): As the voltage at VC increases, the additional harmonic content introduced results in pleasing mid-range distortion.

With two audio sources (switch HI): With a simpler source at IN and a richer (preferably rhythmic) source at VC, the rhythms sound like a viscous acid bath splashing tantrum (with bleeding). Or rather, the rhythmic source (VC) sounds filtered through and distorted by the simpler source (IN), with some amount of the IN signal bleeding through to fill in the spaces in the rhythm. An envelope follower tracking the rhythmic source applied to the amplitude of the simpler source before the latter hits IN would tighten that up.

Middle section — “The Wave Folder”:

The middle section is comprised of two audio inputs (IN 1 and IN 2), one voltage-controlled input (VC) with attenuator knob, and one output (OUT).

Middle section inputs

IN 1: With the knob all the way down, audio is inverted and attenuated (equivalent to that of the top section with its knob at 9:30, switch HI). An increase at the knob immediately begins folding the waveform back onto itself over a very wide range. Control voltage applied to VC animates this effect and sounds brighter than control voltage applied to IN 2.

IN 2 (with audio inserted): Audio here begins to fold with the knob around 8 o'clock. This is more obvious with waveforms richer in harmonic content: a triangle wave has normal looking peaks, but the troughs are rounded; a sawtooth appears inverted and squished. An increase at the knob sounds similar to the folding effect using IN 1; it just sounds like it has a different starting point. Contrary to expectations, more harmonic content in the starting waveform does not result in more interesting sounds — so a triangle wave results in richer sounds than a sawtooth as the knob is increased. And like IN 1, the output from IN 2 is inverted.

IN 2 (with control voltage inserted): With audio at IN 1 and control voltage at IN 2 only, the effect is similar to control voltage applied to VC, but with inverse effect: as voltage increases, the effect decreases. Also, as observed with audio, it sounds like the starting point for control voltage applied to IN 2 is different than that of VC (e.g. if IN 2 starts high at 1 o’clock, then VC starts low at 8 o’clock). The output has less brightness (but is no less interesting) than when using VC.

Things get really interesting when applying audio-rate waveforms at both inputs. Two different waveforms and two different pitches droning together already sounds interesting, but a slow sweep through the knob’s range reveals a huge variety of juicy timbral material. And of course applying a slow-moving LFO or envelope animates an already rich sonic tapestry.

Similarly, applying two control voltage signals to one audio source is more interesting than one modifier. One slow control voltage at VC and one slow control voltage at IN 2 (keeping in mind that the action at IN 2 is inverted) results in the two control voltages competing as you’d expect any time two CVs veer near each other or collide.

More on the middle section VC input

VC: Surprise! This (non-inverting) input (which could have been called IN 3) also accepts and responds well to audio. With the knob full left, a triangle wave is already folded. Turning the knob clockwise traverses through the folding effect as the other inputs do, in this case rounding the triangle wave to a sine wave at 2 o’clock **. With the knob turned full-right the triangle wave looks nearly like a triangle again (with rounded peaks), but is doubled in frequency.

** Incidentally, the same source signal applied to the top section input, with its knob at 2 o'clock and switch set to HI, results in an inverse sine wave of the same frequency.

With audio inserted in VC, and with control voltage applied to IN 2: The behavior by this point is what you’d expect (i.e. as the control voltage changes, the effect becomes animated), except that the action at IN 2 is not inverted this time. In other words, when applying control voltage at IN 2 with audio at IN 1, the effect decreases as the CV goes high; but when applying control voltage at IN 2 with audio at “IN 3” (VC), the effect increases as the CV goes high.

With audio inserted in VC, and with control voltage applied to IN 1: But wait, IN 1 is a black jack indicating that it’s intended for audio, right? Well, with Serge, you may as well try it because it just might be interesting. In this case, an LFO with a gradually changing voltage (e.g. sine or triangle) results in the effect sweeping up and down through a small range for half of the LFO cycle, and then the movement repeats. I believe what's happening is that control voltage applied to IN 1 is rectified, so the result has a doubled effect where the action occurs twice per LFO cycle. An LFO with a more abrupt voltage change (e.g. square or saw) sounds similar to ringing a filter with pulses, including a pronounced click associated with the abrupt voltage change of the LFO — which may be desirable for a rhythmic effect.

With audio inserted in VC, and with control voltage applied to both IN 1 and IN 2: As with using audio at IN 1 and control voltage at the other two jacks, competing CVs can create some interesting effects, rhythms and sense of movement. The same applies here (using audio at VC and control voltage at IN 1 and IN 2), except that as above, there are more clicks and “bumps” when abrupt control voltage changes occur.

Finally, what happens when you feed audio into all three inputs? Well, mayhem, really. And of course feeding (say) three oscillators into the three inputs takes away CV control on the middle section itself. But those oscillators do have VC inputs of their own and could be modified independently from the middle section, and this could result in some interesting timbres or rhythms or textures. But I found it took way more work to get three audio sources to sound good together and to animate well together compared to what I got from just plugging in two sources and turning the VC knob or using one control voltage modifier to animate things. Generally speaking, when it comes to audio and CV with this section, it seems 2 is better than 1, and 2 is also probably better than 3.

Practical use notes — Middle section

As I found from running through the various processes above, there is a point of diminishing returns when it comes to sources fed into this section: more harmonic content is not necessarily more interesting than less harmonic content (i.e. three audio sources is often less interesting and more chaotic). This section excels at a few really useful cases:

  • Creating lush, juicy textures and razor-like timbres from the humblest of sources.
  • Using short percussive sounds (like pulses or drum machines or even dry real drum sounds) to ring it like one would ring a filter, while playing with the frequency and/or resonance at the source.
  • Adding some dirt to otherwise clean (and not too harmonically rich) content.

Bottom section — “The Rectifier”:

The bottom section is comprised of two identical audio inputs (IN 1 and IN 2), one voltage-controlled input (VC) with attenuator knob, and two outputs (OUT 1 and OUT 2).

Bottom section inputs

IN 1 and IN 2: With the VC knob turned full-left, an audio-rate 5V PP triangle wave presented here will have slightly rounded peaks and very rounded troughs. Amplitude is attenuated (equivalent to that of the top section with its knob at noon, switch LO). As the knob is turned clockwise, the amplitude is increased and at 9 o’clock the rounded troughs begin to flatten. Just beyond this point new peaks rise up from the troughs while the existing peaks shrink down until the amplitudes of the new/old peaks match. Thus the rounded triangles return, but at twice the frequency of the original.

So, full-wave rectification. Turning the knob back to full-left for comparison reveals that the doubled frequency (2nd order harmonic) can be heard very quietly due to the distortion of the triangle wave. As the knob is turned clockwise, the fundamental fades to the background while the harmonic comes to the fore until about 10:30 where the first doubling is achieved. Continue turning the knob and the 4th order harmonic comes forward with the new doubled waveform presenting at about 1:30. One more doubling (8th order harmonic) occurs by about 3:30, above which there is little to no change.

While turning the VC knob back and forth, another interesting thing is revealed: If the knob is turned quickly, the sonic character at the new position is not heard instantly as it normally is. Instead, the new character takes a few seconds to settle into its new knob position (especially in the full-left to noon range). This soupy response to changes in voltage makes me think of the behavior of a vactrol in a low-pass gate. In any case, the delayed response will be good to keep in mind in practical use.

VC: As with the middle section, audio inserted in VC (a DC-coupled input) works fine, and the amplitude is about the same as IN 1 and IN 2. With the knob turned all the way down, it’s not readily apparent what’s being done to audio applied here. It’s easiest to see with an oscilloscope using sine or triangle waves as sources. The best I can describe is that as the knob is turned up, the troughs rise up and grow Batman ears. As the knob is turned up further, the Batman heads move aside until new troughs appear, and then those rise up and grow ears. With a sine wave (which appears non-inverted, but with only the positive polarity part of the signal visible since it's a DC-coupled input), as the knob reaches full-right, the wave shape simplifies and ends up looking like the positive portion of the sine with a bite taken out of the peaks — similar to the full-left position, but with greater amplitude. At full-right with a triangle or sawtooth, the signal simply disappears.

So the VC input appears to do some combination of rectification and wave folding. Suffice it to say that simple waveforms applied here are transformed into new shapes rich in harmonics. It’s also worth noting that the delayed response to changes in knob position is still present when applying audio to the VC input, but the response is much faster than that observed when using the other inputs.

Bottom section outputs

The descriptions of this section’s inputs so far have all been done while using one of two available outputs: the non-inverting OUT 1 (black jack indicating AC-coupling, or bipolar -2.5v to +2.5v). What about OUT 2 (blue jack indicating DC-coupling, or unipolar 0v to +5v)? The Serge catalog description for this section says that one output is a squared-up version of the other. So which is which, and what does that mean in terms of how it sounds?

Viewing the two outputs on an oscilloscope answers the first question readily: the blue jack (OUT 2) is the squared-up and inverted output. Since the output is squared up, feeding it with a sine, triangle or sawtooth results in the same thing: a square wave. And the output sounds essentially the same regardless of the source, with a sharper, nasally character.

As the VC knob is turned clockwise with a sine wave source, the square waves divide and move aside until the 2nd order harmonic becomes dominant. With a triangle source, the 2nd order harmonic is already present at full-left. As the knob is turned, the square waves widen until they join, bringing the fundamental back to dominance. As the knob is turned further, the square waves join and the signal disappears by full-right. Conversely, with a sawtooth source, the fundamental is dominant at full-left, and as the knob is turned, the square waves narrow until they and the signal disappear.

What this translates to in practical terms is that with more complex source material, changes to voltage at VC will cause some harmonics within the program source material to come and go gradually while other sounds might wink out abruptly and then return just as abruptly, which could be really interesting.

As with the other two sections of this function block, things get even more interesting with two audio sources used with OUT 1. And though things get chaotic quickly with three audio sources (as observed with the middle section), it’s much easier to fine tune those sources to get interesting results.

Using only a pair of simple sources (e.g. triangle and sine waves), rich timbral material is produced. With turns of the VC knob or changes in frequency of the sources, the interplay of rectification, wave folding, and harmonics coming and going means there’s a lot going on here.

As interesting as these mixed and mutated signals can be, adding control voltage to animate the behavior can add an entirely new dimension, but may require some fiddling to get there. This section is capable of producing some dense and rich textures, but the result can be noisy and indistinct when attempting faster control voltage movements (recall the VC input's soupy response to voltage changes). On the other hand, slowly moving control voltage animation creates mesmerizingly evolving sounds.

Practical use notes — Bottom section

As with the middle section, sometimes less is more when it comes to feeding audio into it. Taking advantage of the harmonics resulting from rectifying the incoming signal is a satisfying process since those harmonics occur at octave intervals and are always in tune. This section is useful for changing the tonal character of audio fed into it in a way that the other sections don’t, with the added ability of adding brightness by mixing in the secondary output.

In conclusion

With all the incredible sounds possible with the Wave Multipliers, it’s tempting to send one section into another — and doing so can potentially create some great surprises with enough patient experimentation. But it’s worth keeping in mind that this function block isn’t like a subtractive filter, where you can keep piling things on and carve away at it with frequency cutoffs and different response slopes.

This function block is a wave multiplier: what you get out is more than you put in. In some cases a lot more. The Wave Multipliers block excels at creating new and interesting timbres from very simple sources, and it’s easy to go overboard and end up with a murky mess. But never say never, right? Regardless, when you generate some jaw-dropping sounds (and you will), hopefully you'll have a better sense of what's happening within each section. And for goodness’ sake, animate!

No comments:

Post a Comment