CAKE voice project (let's design a Kosmo voice module together)

Okay, that looks like a challenge.

How about it? I propose this: Collaboratively design a Kosmo voice module. A single module offering the basics of a full synth voice, so at least one oscillator along with means to vary its sound (filter, waveshaper…) and amplitude (VCA, LPG) with modulation source(s) (LFO, EG, slope generator…). The idea is to have something not as powerful and flexible as a set of individual modules, but offering a good subset of that power and flexibility in a compact and relatively low cost package. It could be the basis of a small monosynth, or an addition to a full modular, or several could be used to create a polyphonic synthesizer.

“Collaboratively” implies contributions from multiple people — and they don’t all have to be circuit designs. Anyone with any general ideas, specific requests, block diagrams, panel art concepts, whatever, feel free to jump in. I recommend starting with general functionality, moving on to user interface, and only then designing circuitry.

To focus creativity and keep the project of manageable scope, some constraints (open to negotiation for now):

  • Conform to Kosmo spec
  • Design to allow building as stripboard or PCB (self etched or commercially fabbed)
  • Maximum panel width 20 cm
  • Bonus points for keeping PCB(s) small enough (under 11 cm vertically) to allow building it as Eurorack too
  • Use only readily (commercially) available parts
  • Limited maximum component cost — I’m thinking something like $75 US (including commercial fabrication of PCB and front panel — let’s say 1/10 of cost to make 10 at JLCPCB) (excluding postage) but let’s see. I’m a cheapskate and may be indulging in fantasy.

For inspiration here are a few Eurorack voice modules, find more via modulargrid:

So, anyone in? :smiley:


Very well thought out limitations, Im down to theorise.

A first thought is a drone made of 3 supersimple osc’ (to make chords) that could be CV controlled for 4 repeating steps and a Key pad of notes to jam over the drone?
(Scratch that, SS OSC cant be 1/v oct, leaving the idea up though)

Ill have a look at the inspiration and see what comes up =]
-Dual ADSR/AR/AD which ever, with a switch to add some rhythm expression. (thinking about hihats or chuggy guitar riffs)
-Bitcrushing? I havent seen many and love the sound.


I like the initiative!

Stable v/oct in DIY is a really hard problem though. Does the AS3340 violate the spirit of the project?

I haven’t seen much built around the VCDO wavetable chip, could provide an alternative to yet another 3340.


I’d say synth chips are fine, so long as they’re readily available (AS3340 qualifies as that). Though they do tend to be pricey, a voice made entirely of specialized chips might end up costing too much. And “yet another 3340” would be less interesting than a novel op amp design. But I don’t see a reason to rule it out.

Op amp V/oct is challenging but certainly doable. Some of us have built e.g. the MFOS or YuSynth designs, and @jkb’s Polykit VCO might also be an interesting starting point. Or a DCO design, maybe.


I was going to say, nobody has messed with the electric Druid VCDO, which already has a built in expo converter, subosc, and some effects. Could go a lot of different ways with it


It’s good to think about these things, but I think it’s better to figure out what we want the module to do before getting too hung up on what technology to do it with.


I dunno man, have you seen pin 9. =D


Two voices, FM, loads of sounds, glide and distortion out of the box really makes this thing worth the price in my opinion, and I haven’t seen any solid DIY’s built around it yet.

In terms of the other features? A HP/LP filter, a VCA and an ADSR/AD? I’m honestly not sure if it’s better to just build a great VCO or to put the whole kit and kaboodle into a single module.

Another route would be to go full “west-coast” Kosmo - VCDO+LPG+A/R, then maybe some kind of weird input source on the panel, like a contact mic or a ribbon controller.


Let’s use the Electric Druid envgen8 envelope chip. Advantages

  • Low External Parts Count
  • Duration function is cool, scales A,D,S, & R from 1ms to 10 sec
  • Can also function as an LFO or gated LFO saving space in our design
  • Open source firmware
1 Like

I dunno man, have you seen pin 9. =D

I’ve been waiting a month for my VCDO pcbs to arrive from JLC. Incredibly stoked.


VCDO seems like a good candidate, Still open to other IC’s/options.
What good could come from throwing a Teensy or Aruino in the mix? could give complex gates at a button press


I general can we try to avoid designs that require a lot of hand matched parts or CA-series chips that you can only maybe get from Rochester? It’d be cool if we can design something you’ll still be able to build 10 years from now.

I’d advocate for a teensy over an arduino simply because it already has a built-in DAC, and also it can enumerate as a usb midi device, so that opens up other options for the module with little effort as well.

well do you want this to be obtainable for the cash strapped and easier to assemble for our beginners ?
our do we make this the dream module with proprietary chips , etc. . I would vote for the simpler version myself . and the Kosmotron is born unto the on suspecting world , little did they know the chaos to come …


Hmm is that a name already?

It has to embody the spirit of Kosmo. So i think it needs to be mostly attainable parts or have an equal variable part (ie CEM/AS 3340)

1 Like

Definitely a good idea to think about keeping the module as accessible as possible for various skill levels and folks of different SES. Thank you for that devicex!

One of the cool things about the Electric Druid chips is that they are all open-source, and the PIC chips themselves only cost about $2, so if you can get ahold of a PICKIT programmer, you can make as many of them for yourself as you need.

Given the number of discreet components that a druid chip saves, i think in many cases they actually are the cheaper and simpler option, and in fact the accessible choice.

Thanks again for keeping everyone in mind. :+1: :+1: :+1: :+1:

1 Like

not against using the druid chips at all like you said they do cut down on many other components . I like the idea of a 20 cm all in one sound module . but done as simply as possible .

Even op amp VCOs generally don’t require “a lot” of matched parts. One pair of transistors, and there are monolithic options for those.

Good point about future buildability, though that’s hard to predict. Still, TL074s are likely to be around forever. PICs? Maybe…

1 Like

Zooming out from specific technologies, what do we want our module to do?

Minimum Features (imho):


Yeah, the “lots of matched parts” seem to come in the filters
A bunch of matched transistors for transistor ladders…
A bunch of matched diodes for minikorg 700 style filters…
Not sure about Steiner filters, I haven’t gotten to building one of those yet.
MS20-style might take less?