Complex Oscillators

Continuing the discussion from Instruo Cs-L:

Hello, I thought I’d start a topic dedicated to complex oscillators, their makers, differences, and most importantly the sounds they create.

My naive assumption is they are all derived from the buchla 259 [contains 2 oscillators, a modulation bus, a wave shaping section, and a wave folding (timbre) section]. Edit: great overview videos here: https://vimeo.com/137696291

I don’t own any distinct complex oscillators currently but have been patching them up from elemental parts.

Would love to know how you patch them, innovations and other complex-like patches. (see also: 2 x Mangrove vs complex oscillator)

So I don’t come empty handed, here is my attempt at one I created on the ER-301. I modified a preset created by @Starthief over at the Orthogonal Devices forum to use phase modulation for linear FM (https://forum.orthogonaldevices.com/t/starthiefs-custom-units-feedback-keeper-complex-osc-wavefolder-etc/2851).

complex osc 004.unit
complex osc 010.unit (73.3 KB)

  • fixed RM/AM section
  • added wave shapable modulator
  • added wave folding final section
12 Likes

Oh, what perfect timing. Thanks. I was planning just this weekend to plug a buncha slightly out of tune (with each other) triangle waves into an ER-301 unit and see what I could make of it.

I’ve always been wondering what exactly makes a complex oscillator and how it is different from just using separate components. Is it just the same? Is it more than the sum of it’s parts? Is there some magic that makes it “work”?

Did you notice any difference with phase modulation vs linear FM? It should be the same in theory, as long as the modulator is an audio rate sine.

Edit: I guess one advantage is, if you’re also using expo FM and have the ratio and level tuned for a nice harmonic result, you can also use dynamic PM without throwing off the tuning. Linear FM will mess up the tuning if you combine it with expo FM.

1 Like

The difference from using separate components is usually just that a complex oscillator will have some normalled connections to cut down on the patching you’ll need to do to get different sounds going. They also tend to have more fixed pathing than you’d have available if you’re using separate components, so that convenience comes at the cost of some degree of flexibility.

2 Likes

A complex oscillator invites/suggests particular ways to patch. It’s convenient if you happen to really like those methods :slight_smile:

Many of them are open enough you can use them separately, cross-modulate, sync either direction, etc. So I’d argue you don’t usually lose too much flexibility vs. separate modules.

(An exception, usually, is the wavefolder. There are a lot of tricks you can do with a wavefolder patched post-LPG, or with two oscillators mixed. Something I liked about the Pittsburgh Double Helix was the ability to patch anything through the folder & LPG, not just the primary osc.)

3 Likes

The Verbos has vactrol in the mod bus and I think the DPO has a bunch too, so you can’t just patch that up always and get quite the same slewed effect. Also the timbre/symmetry circuits can be very distinctive in personality indeed.

The Verbos is exponential FM and unipolar mod osc, whereas the Furthrrrr has a switch on the back and bipolar plus external input to the mod bus itself.

The FG does not have vactrols in order to have fast audio range modulation capability


but I highly recommend it for general full package melodic and percussion use, as the extensive timbre range and noise and ring modulation and simultaneous AM/FM can take you to great places while controllable linear FM, the dual tuner, crossmod CV, and external input amenities are great for melodic control, the tone is not full bandwidth modern.

I do like all complex oscillator designs in Eurorack for different reasons, and like I said in the other thread, would consider the Cs-L a good choice for sound design.

I also really like the Rubicon system but consider it a little different philosophy than a CO in terms of all the waveform outputs, FM options, and the through-zero metallic/filtery sweep power and with the 2 prefer a separate wavefolder, and not the rather vanilla ufold. To me it sounds “modern” in a retro-futuristic way.

2 Likes

I’d definitely be interested to hear a couple of those wavefolder tricks!

The key to one of those wavefolder tricks is this:

  • imagine a sine oscillator at 0Hz – or a sine lookup table.

  • feed a triangle VCO (at the right level) into the phase of that, and you get a sine out at the same frequency.

  • now mix in a second triangle VCO with the first. Effectively, one of them is “driving” the sine while the other is phase modulating it. (Really they’re both phase modulators and there’s not strictly a carrier, but whatever).

  • a typical wavefolder doesn’t have a perfect sine transfer function, but it’s got a vaguely similar shape – so it’ll work for this trick.

You can use sines or other “soft” waves instead of triangles. Sawtooth doesn’t work well because phase modulation in a continuous direction is just a pitch shift, and squares don’t work well because you’ll just get a square for the output.

Here are two triangles an octave apart in the ER-301, mixed into a wavefolder, with Maths driving their levels and clocking a Stages sequence.

20190503T223129.flac (3.7 MB)

From what I understand, the Cs-L can do this by routing one oscillator into the symmetry input of the opposite wavefolder, which is equivalent to mixing in front of the wavefolder :slight_smile: Or it can be manually patched using triangles instead of sines.

Double Helix has two inputs to its wavefolder (which is immediately followed by an LPG), but it doesn’t offer triangle outputs.

3 Likes

Another Candidate would be the Furthrrrr Generator-yet another Sound

2 Likes

I have the Sputnik Dual in my main case which is rather nice. Sounds huge but is a little less complex (!) than the Cs-L so I’m not sure it offers the same diversity of sound.

In my small case I don’t have space left for a large complex oscillator so took a more modular route based around the Ts-L. That packs a lot into 6hp.

I pair it with an old v1 Disting acting mostly as a simple modulation oscillator although it also stands in a balanced modulator when using other sources as the MO. Along with an attenuverter module and a 2hp unity mixer its not as streamlined as a single complex oscillator module can be, but it covers an awful lot of the same ground (and is more flexible about patching on the days when you don’t want/need all that a complex oscillator offers) and does it all in 18 hp.

I also prefer the sound of the DPO because it has a defined character, finding the CS-L too full bandwidth for my taste just as crazeebo said, and like you implied, the Make Noise interface is very intuitive and not cramped. The CS-L would probably be best for sound design though.

I think the FG and Verbos most of all are the most inviting-sounding, although the former is set up to do a lot more than the latter (such as linear FM, noise and timbre modulation, and I find the FG has better pitch stability), but again that’s more about my personal taste, which tends toward the less pointy or aggressively toned side of things.

2 Likes

I’m still on the fence to buy a Complex Oscillator (I have the Verbos Harmonic Oscillator). I watched lots of videos and listened to a lot of demos, and if I would have to decide right now, it would be the FG exactly for the reasons you mentioned. I like the sound of Verbos CO and FG best, but the FG offers linear FM (with the upgrade) – and probably a better quality control and service (lots of issues with Verbos).

The linear works analog core as well, it’s just a switch on the back to set exponential or linear. Verbos mod FM is exponential and unipolar mod osc. I think you mean through-zero upgrade, which is different.

Personally I prefer the sound of the analog version to the through-zero version. For through-zero I like the Rubicon.

I highly recommend complex oscillators in general and the FG for melodic work. It has a lot of conveniences and doesn’t sound especially modern like some of the other options, although the noise modulation and normalized cross mod (the red fm knob on modulation half) and ring mod can get you quite a lot of interesting percussion sounds. The Furthrrrr/Timbre section has a little more range than the Verbos, although the Verbos has the most standout “vintage” tone which I love.

The FG is the more complete package, in other words.

Should this thread topic be split or renamed into a broader Complex Oscillator discussion? I’m personally interested in a CO myself (or maybe CO-like with a rubicon2).

4 Likes

Yes, you’re right, I meant through-zero. I think the upgrade makes only one of the OSCs digital, the other one stays analog, but I didn’t have a chance to compare both versions yet.

Yes the upgrade is the carrier osc, but according to manual can be done to mod osc as well with loss of LFO function. I assume that the FM and pitch would be more stable with change to mod osc as well, but don’t know as manual doesn’t say.

I’m basing the sound determination from the Learning Modular through-zero FG video. I know the analog core sound because I use the module all the time and I just don’t think the trade off is worth it for me. Although the FM can go deeper obviously with the strong digital core, it just doesn’t sound as good to me over analog for linear FM melodic purposes, which is why I use the FG.

Interesting! I’m not 100% following the “feed into the phase” of the sine. I’m not sure any of my oscillators have phase manipulation (dixie2+, mangrovex2, sto, just friends), and plaits though i dont count that in this exercise.

In math terms, I"m just talking about sin(x) where x is the output of the triangle VCO. But like I said, substituting a wavefolder for that step works pretty well.

One way to achieve PM if you don’t have a VCO with a phase input, is with a very short, clean delay with no feedback. Modulating the delay time can act as phase modulation :slight_smile: Some delays (like Chronoblob) smooth out delay time changes and can’t do it, but the delay modes in Disting will do it, though. (This only works when the signal is audio rate though, not my hypothetical 0Hz sine wave.)

Ah ok I’m with you. DLD will definitely handle that