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

$ 9.00 in SMT form at synthcube . [ ss12130 ]

1 Like

The DAB variant is the DIP adaptor version, which are being made in lower quantities than the QFN package.

Edit: Which I just realised is DIP Adaptor Board

still a proprietary IC though , but does save a lot of parts .

I worry about using a part as new as this whose future availability seems very much up in the air.


It’s far from new, its the re-issue (with improvements) of the chip used in the Prophet 5 (and this version is in the new prophet 5). And if JLC don’t Digi-Key have through hole mounted sockets for the QFN 32, then you just push the chip in and away you go.

EDIT: Twasn’t the P5, that was the 3340… but it was definitely in a Sequential synth… the new VCF from the same series as this VCO is in the new P5’s and its original was in the old P5’s

EDIT SOME MORE: It was in the P5, just the Rev 1 & 2 before they replaced it with the CEM3340


I’m coming in late here, but would love to help. I can only help with concept and perhaps visuals.

I suggest we keep this within 10cm wide or smaller for a few reasons:

  • ease of build
  • focused design
  • it’s the first collab here
  • it’s a common footprint for those of us with wooden rails

I don’t think it should be a standard oscillator. Something about it should be spicy. I also think that it shouldn’t be too hard to put together using existing circuits. At least not yet.

I think there is a good list of “expected” features already:

  • PWM
  • FM
  • fine tune
  • octave switch

With all of that in mind, I think it could include one of these spicy features:

  • Detune/Swarm/Unison
  • Wave folding
  • Sub octave
  • Distortion

Perhaps some combination of these could yield an interesting module. Here are a few super rough mocks leaving out some of the expected features for now:

SWARM - The Swarm module could have 2-3 oscillators that can detune with a single knob. I’m not clear what the wave shape should/would be. Perhaps only saw or square. I’ve never built anything like this.

FOOL’S FOLD - a single oscillator and wave folder combined

BIG SUB - Inspired by @CTorp’s Sub module. If we’re going heavy, it could be fun to add an overdrive/distortion circuit as well.

Ignore the silly names and the layout is just for getting ideas out quickly.

Edit: Mocks in figma. Happy to invite folks to collaborate. Just let me know.


Subosc is quick and dirty compared to a second osc, but it cannot be detuned because it is based on the first osc frequency. Pretty sure you’d need a distinct oscillator for detune

Also, 20cm width or less in my opinion


Totally. That’s why they’re separate. I think swarm might be a bit of a pain to build with 2-3 oscillators.

1 Like

LOL well we will be at this for a few months trying to refine all the ideas into one module that is fairly easy to build and not to expensive and is 20cm or less or maybe 30cm or … dam I hope we work better than my government works [ U.S. ]

1 Like

If it’s just that I think it could easily be fit into 20 cm. The VCA would need no controls other than a switch between ADSR and drone. Four knobs and two jacks for the ADSR, two knobs and 3 jacks (in, out, frequency CV) for the filter, and then there’s plenty of space left for the VCO. And probably space enough to add more usability, e.g. CV attenuators for the VCF and VCA. It wouldn’t be as spacious as I’d normally like, but again, for a supplemental voice I think that’d be okay. It’s not intended to replace full modules.

But it’d be maybe too limited with no LFO, only one EG, no noise.


yeah a nice simple voice to start with and build from there with future modules ?
but having said that there are a lot of those [ simple vco’s ] out there so maybe something unique would not be a bad idea ?

1 Like

This is why I have been pushing for a west coast design, by all means build a good ole fashioned voice module as well (it’s not like it would be all that hard, especially with the level of experience folks on here have with the EDA software), but surely we have all had enough of that simple VCO/VCF/VCA by now, I mean I have 2 complete three oscillator voices in Kosmo these days, and I am craving a different sound.


so i can see we are headed for a feature list that we give everyone a week or so to vote on ?

Having spent extensive time on open source projects, the trick here is to keep the goal very focused or it won’t move quickly. Creating a voice is pretty abstract so perhaps some parameters to narrow the focus could help or we will all be envisioning very different things.


kinda why I suggest a list , the top 3 or 4 things with most votes is the focus .

Do we want this to be educational for a newcomer to the synth DIY world or just a cheap way to have a voice?

If it’s the former than I’d shy away from using complete VCO or Filter ICs, they’re kind of black box and you probably won’t learn as much from them as having everything broken out separately.

If we’re trying to keep component count and cost down, then have at the single IC solutions. But if we’re going to go down the QFN route that’s making it a little more difficult for DIYers, and we may as well allow monolithic matched pairs in surface mount packages in a discrete VCO build.

Swings and round-abouts really.


1 Like

From what I gathered from @analogoutput’s first post, we are after a single module (so one panel) housing all the necessary ingredients in order to design a sound that can then be played via CV. So that means a sound source, a shaper of some description, and modulation sources in a single unit. As small as resonably possible, whilst keeping in mind the design considerations set out in the Kosmo standard - so a playable user interface and big jacks and knobs.

Personally, my thoughts immediately jumped to a well rounded ‘starter’ module, that doesn’t have much in common sonically with what we currently have available but builds on what do have and something I have been brewing myself in isolation - a west coast style voice complete with a complex FM oscillator, the basic modulation sources required to produce west coast sounds (LFO/Noise/AD Function Generators) and a Resonant Low Pass Gate. I didn’t have it as ‘educational’ in the sense of learning circuit design, but rather experemental in terms of usage and learning to patch creatively. The kind of thing that could be put together by a novice, from easilly sourced parts, which gives a lot of rewards when it comes to cross patching it with itself and wouldn’t be instantly replaced when the system was upgraded to include other modules. Similarly it would be something that we as current Kosmo users/builders would actually want in our systems by its very nature of being radically different from the voice architecture we usually get from Sam’s ‘official’ modules.

Also @jaradical those QFN chips we have been talking about can be mounted using through hole sockets, so it’s no different to soldering a DIP socket, it’s the SOIC ones we would need to avoid to prevent scaring folks off :wink:


sounds like a good basic idea .

Fair. I suppose I found most of the proposals well within reach of what already is available currently as PCBs/panels (outside of maybe a function generator and noise noise source). To me, the proposals seem like multiple modules on one panel which doesn’t sit well with me. However, my opinion shouldn’t stop you if that’s the end goal.

Similarly it would be something that we as current Kosmo users/builders would actually want in our systems by its very nature of being radically different from the voice architecture we usually get from Sam’s ‘official’ modules.

I’m not sure how much of the above differs just yet. The biggest deviation is the function generator which I would absolutely LOVE as a stand-alone so I can sell Maths. As for the FM, I found Sam’s oscillator modules already are an FM playground with FM, sync, and cv inputs.

1 Like

LOL so basic or unique . dam I like ease of build to be all inclusive of everyone but why build something that isn’t different from everything else out there ?

1 Like