Yeah, but then we’re looking at either obsolete CA-series parts, or SMD parts that scare off beginners, or tin-can parts that are pricey.
One or two matched pairs isn’t the end of the world (although it does require a beginning builder to have a multimeter with transistor testing function, but i guess fortunately those are usually found on the really cheap meters beginners tend to buy…)
So I guess if we’re trying to emphasize accessibility then we might try to:
*Stick to matched discrete components and avoid the monolithic options (because they’re often either obsolete or smd)
*Try to keep the matched parts count low.
But just my 2p.
EDIT: I’m talking too much, sorry! I’m gonna lean-back a bit
I dont know how beginner friendly it needs to be, theres things like the SS Osc’ and the hex 40106.
Re: Matching Components - It might be good to make it a ‘stepping stone’ lesson for beginners (like me) to step further into filter tech’.
Still, keeping it reachable.
I’ve been looking extensively into the Druid chips(microchip free samples…) and am currently designing a noise module, vcadsr, vcdo, bounce bounce but with less features, and starting a vclfo.
Most of my stuff is on GitHub, search for sonosus.
Happy to help with PCB, Sam style silkscreen, druid chips, most things!
I have to agree with @analogoutput on this one, analog oscillators sound so much ‘nicer’ and have a rich depth to them that you lose with 8-bit voices. If I remember rightly the digital ones are going to draw more power to boot. Plus there are at least two easy to DIY all on a single chip VCO’s out there (The 3340’s with their clones and the SSI2130DAB), so it’s not like we don’t have decent, stable and resonably priced options.
For my two cents on feature set, in a full featured voice module I would like to see:
-At least one oscillator with mutiple waveforms, if two (or more) present they should include FM
-A mixer section for said waveforms - maybe an attenuating one like you find on the MN 0-Coast
-A simple LFO - Something like the simple System 100m ones you see on stripboard is enough here
-Sample and hold with a noise source - The Druid noise chips maybe? Although with an analog noise source we wouldn’t have to worry about converting 5V power on the board at all, I just don’t know how we would build it.
-At least one envelope - Preferably the CEM3310, because again, analog circuitry for the win
and either
-A filter and VCA combo with an extra envelope - Plenty of all analog designs available, hell we could go the whole hog and drop in the CEM filter/VCA chips easy as pie
OR
-A low pass gate - I posted a while back a working design for a buchla 292 clone that I am currently fiddling with the resonance mod on to get it to squeal harder (its still a little tame at +/-12V the moment as the original was for +/-15V, but I know it can work - I have seen it done ).
Personally I would rather see the LPG with two FM’ed ocillators, and dropping the ADSR envelopes for a pair of AD ones for a space saving making a nice west coast voice and the VCF/VCA with the one VCO option for a nice east coast voice.
Route the standard signal path behind the panel from the switched normals on the jacks so they don’t technically need patch cables along the audio path and envelope CV for ease of use, but can easilly be repatched however you like. Keep the power usage as low as possible and the panel width at or below 40cm, as surely we would want people to see this as an easy startng point for a modular, something they can build, put in a sensibly sized case running off a bog standard FC power solution and still have room to grow rather than something that takes the mick and eats all their rack space and power.
I certainly cannot contribute to the overall design of the function of the module…just please someone give me some bi-color or RGB LED’s somewhere in the design
However, when we get to that point down the road when we need a panel I can and am willing to contribute to designing or helping design the panel. Thanks to @d42kn355 and @CTorp I can also create the pcb panel in kiCad for us and get the gerbers exported for production.
The Druid noise chips are easy to use, low parts count, I don’t see any reason why not.
Edit: a simple L7805 regulator is cheap and takes up next to no space on the board.
The reason not to go down that route is precisely to avoid the overhead of running the 7805, its an unnecessary power draw, especially in a module that is - in essance - going to be anywhere between 5-9 modules on a single board. Also the CEM3340 and its clones tend to be a bit tweaky about sharing their power source with noisey (in a power sense) digital parts which would end up complicating the VCO section design.
The Druid chips are not bad, but there may be a better way of deriving white/pink noise. By finding an analog solution, rather than just going for circuit design convenience, we can keep the modules power draw as low as possible, which is going to be a significant design challenge for such a large number of functions in a single module. I listed it in my version of a voice module spec as a ‘if we can’t do it another way’
You are absolutely correct - I didn’t think about the power requirements of the whole module. In that case, it is much better to use analogue circuitry.
My inclination is to not have a noise source or a S&H at all. I don’t think of those as part of a synth voice. If, after all, you were to build a polysynth, you probably wouldn’t include multiple noise sources or multiple S&Hs. One each would be enough.
(I guess I’m drawing a distinction between a synth voice and a monosynth.)
But if there is a noise source, a simple white noise generator needs only a transistor, a TL072, and a few passive components.
Speaking as the owner of an LMNC-like 8 step sequencer, I can say 40 cm seems to me pretty huge as a starting point.
If you look at the Eurorack voices I mentioned above, 3 of the 4 are square or pretty close to square. I think if you were to try to repanel one of them as Kosmo, you could do it or come close in 20 cm width — maybe 25 cm, and yes, I know that’s not listed in the Kosmo spec as a standard width, but as a multiple of 5 cm I regard it as fine. Anyway, I advocate aiming for 20 cm max. It won’t be enough to include every fun feature of every submodule, but that’s part of the whole point, to keep it focused on the important stuff.
Yeah, 40cm is big, I know, but I was thinking about the ‘performance’ aspects of the design ethic Sam uses in his modules. The euro ones that are square tend to be cramped and unpleasant to use, whereas the more musically playable modules tend to be rectangular in about 40-60hp making them about half the width of an average eurorack case - think the Moog Mother 32, the Pittsburgh Modular SV-1 or the Make Noise 0-Coast; all synth voices but still workable as a standalone unit. Also, I did say at most 40cm
As for Noise/S&H they are an essential part of a West Coast voice, as they tend to do a lot of the heavy lifting with the sound shaping modulation, god only knows why the West Coast Mini Synth didn’t include it. I use my Wogglebug in pretty much every patch for this kind of thing, and as of yet its something we just don’t have in Kosmo as a PCB set. They are less important if you want a more Moog-y VCO-VCF-VCA voice like that cramped Cwejman module (and the absolutely awful Doepfer voice) , but that would be a simple case of stringing together a couple of Sam’s existing schematics and calling it a day. Also, you are right in the sense of only really needing one decent source of uncertanty-esque module in your rig, but I find two is my ideal number when it comes to experimental sounds, but that is going way beyond a simple S&H/Noise/LFO combo, whereas by including it in the functionality of a ‘starter’ module means you can unlock extra layers of functionality without ever having to build a second module. You could even build the voice as I envision it in a little skiff with its own PSU and have a lovely desktop Kosmo synth without needing a big case if you didn’t want to clutter your studio space, and use it to dip a toe ino modular without having to go out and buy a load of wood or dedicating a whole meter or so of table space to the effort. The features I have listed for a West Coast style voice would probably fit in a 30cm panel without too much in the way of spacing compromises.
I think for both the width and the noise/S&H the issue is that I’m, as I said, distinguishing between a voice and a monosynth. Certainly I regard a noise source as essential in the latter, but if I were building an 8-voice polysynth I wouldn’t require 8 noise sources. As for width the Mother-32, for instance, includes a noise source, a sequencer, and a MIDI interface, all of which are good parts of a monosynth but none of which I’d include as part of a voice. It also includes a CV mixer and a 32-point patch bay which are great but inessential. Remove all those and you could get the rest, Kosmo-ized, into something closer to 25 cm.
Yes, spacing is tight on those voice modules I listed. Not what I’d want most of the time in a primary voice. But it is what I might want if I wanted it just as a supplementary voice, or as a unit of a polyphonic system especially if it were built to allow a single external controller module to set parameters of all voices the same way. And for those who do want something roomier, why not a alternative, larger front panel as an option? One with more space and maybe more functionality, like a noise source, S&H, MIDI interface, and so on? In other words a monosynth that has the voice board as its core.
On the other hand, it could perhaps go the other way: One could take your monosynth idea and build it with the unwanted sections unpopulated and stuff it behind a more compact front panel, as long as the PCB isn’t too wide to allow it.
I’m not saying your approach is a bad one at all. Just that it differs from what I was thinking. Yours could be the better way to go. But I’d suggest the same core could work both ways, if planned that way.
Edit to add: You know, I may be changing my mind regarding noise. Thinking about it more I can see you might want separate noise sources for each voice.
On the original Illustrator document there is space at the bottom for the lower rail and mounting screws as well as a few mm on the two edges to allow it to breath next to other modules, it just didn’t export it as it was blank space and I couldn’t be bothered going back and doing it again, lol. Spaced for the 33mm, 23mm, and 19mm Boss style knobs made by SCI you can get from Rapid.co.uk and elsewhere. It would also accept Rogan RB knobs, but that would be… costly.
EDIT: Also, we could drop the mult section and expand the LFO and Noise sections to include independant clocks, as drawn there the stepped output on the noise would be clocked by the square wave output of the freerunning LFO. Without the mult you could then sync the LFO and the Noise to seperate divisions of a master clock from elsewhere in the overall system
no reason to stop at one module . so start with a basic voice module and add extra goodies in later modules . 20 cm max. remember these panels will have to be shipped all over the world and the bigger they are the more expensive it will be to ship , even at 20cm the shipping in a lot of cases cost more than the boards .
).
In that case, it is much better to use analogue circuitry.