Starting a little CMOS synth build

It’s unsurprising DIYLC doesn’t support what you’re doing, since what you’re doing is unconventional. To me it’s not at all clear what advantages there are to your putting components and wires on both sides. I don’t see that you really gain any space — you take up the same pads either way. In your photos I see three resistors mounted on the underside, evidently because if they were on top it would make it harder to connect their leads to pads on the other side of the capacitors. But not at all impossible. As for the wires on the underside, I guess the top side looks less cluttered that way, but I see no reason they could not go on top.

Of course components and traces on both sides is conventional with PCBs, and software like KiCad definitely supports that. There, having traces on both sides is definitely a benefit since traces on the same side can’t cross without connecting; wires can, so that benefit doesn’t apply in your case. Through hole components on both sides is a less useful thing, except for instance putting headers on the back side of a board and pots on the front side. I don’t think that really applies to your case either.

With either PCBs or proto boards, having wires (traces) and components on both sides does have the disadvantage of making it harder to understand the circuit by looking at it. But it’s hard in any case. I get cranky when people say “my project doesn’t work, is there a mistake, here’s the stripboard layout” — it’s 10 times easier to understand and debug a circuit from a schematic. Stripboard or proto board or PCB layouts are to describe implementation; use schematics to describe function.

3 Likes

All right, my first module based on my custom PCB is complete!

It’s very similar to the one immediately to the right of it, just done better :slight_smile:

A 40106 and a 4069UB providing 2 LFO and 4 audio rate oscillators, square and saw output. With some space left, I also added some LDR and pushbuttons for performance, and a pair of homemade vactrol LPG.

And here’s a video demo:

The wiring is a huge mess :smiley:

All in all, a very satisfactory result. The machine is almost already musically useful! Could use having a precise clock, something I’d like to add soon, probably a good way to learn MCU programming.

I’ve seen a project using KiCad to suggest a perfboard layout, but I still found it hard to read (especially as the end result is pretty much a 3D sculpture beyond my current build skills, pic on the project page).

Having wires on top on the 4040 board I made two weeks ago indeed make me realize that during patching, they get in the way of readability, as they are hard to tell apart from dupont cables. With the expected finished product lacking a proper panel or labels, the aesthetics of the top side are important.

I’m working 90% off existing schematics here, so getting the wiring correct wasn’t a problem, conveying a physical layout of it in an understandable fashion is the issue. If this physical layout were worth documenting, I think the best format would be step by step build instructions with simplified layers. And given the amount of wiring this board required, I think I’d have been better served designing a custom PCB.

3 Likes

Before continuing the build I wanted to fix the docs, since it bothers me how messy they were.

I really need to stress this once again: the schematic is NOT the pain point here. I didn’t make one beforehand because I was working off existing ones with very minor alterations. The difficulty I have is conveying a physical implementation in a way that would be useful for someone else to reproduce this build.
Even if I will probably regard this project as little more than a flawed learning adventure and just chuck the files on a git repo provided as-is with big disclaimers, I’d like to have clear diagrams.

Anyway, here’s my revised docs. Still using DIYLC (and KiCad for the schematic) and this level of complexity wrecks its performance. I’m trying to use a two-layer view for greater clarity, but I’m still not satisfied with the readability of it.

4 Likes

Experimenting with the 4017 Decade Counter, I managed to fry two chips, and I’m not sure why. My best guess is that it drove too much current. The article on this chip in the Logic Noise series suggests using two chips for this reason: one to drive LEDs, one to actually drive the signal.

But getting low on my stock of 4017 chips, I tried out something with transistors. It seemed to work on the breadboard, so I built it:

And well, the general idea works! At least, it works as a LED chaser (they probably sell kits that do no more than this for $29,90 on the official Hacker Stores for Makers).

But there is a serious voltage drop (not enough to flip logic gates), and not enough current to use it to trigger my DIY vactrols either.

Definitely a design mistake somewhere, but don’t bother trying to decipher that partial schematic in the photo lol, I will go back to this circuit another day to see if I can figure it out myself first. And also come up with a kludge to rescue this board, maybe.

Given that I’m running out of chips fast, I’ve taken a gamble on a Aliexpress seller, selling generic 4XXX chips for 1/5th the cost. Might be penny-pinching to try to save €0.35 per chip, but once you buy 100 chips, it kinda adds up. No idea what to expect, but I do have a few authentic chips I trust to be from TI, so I can compare their behavior.

3 Likes

Try to find something like the TL866Ⅱ Plus USB Programmer (that’s what I have, but it’s somewhat outdated, there are more recent programmers.)

It not only program ROMs and MCUs, but it also can check almost every CD4xxxx 74xxx chip.
Just drop it in the ZIF socket, select the (supposed) chip type, click “Check” and watch the green or red “light” !

2 Likes

I don’t see the TL866II for less than about $60, though, and if the goal is to save money, you’d need to test a lot of chips before breaking even.

There’s this

at $46 which is a little lower (though it doesn’t do programming, only logic chip testing).

And there’s this

which should be much cheaper, but not as versatile or easy to use.

4 Likes

Nope, the goal is to not pull out whatever hair you have left because you are using a bad chip, hair implants are much more expensive than $60 :stuck_out_tongue:
But it’s a valid point if it’s only use will be testing aliexpress chips…

I got a TL866 a long time ago for maybe 40€, but I got it mainly for programming ROMs for retro computers, the chip testing facility just came as a bonus for me (there are many in a retro computer).

4 Likes

Haha yeah, I don’t plan to use the 4xxx series enough to justify that sort of purchase. I’m just focusing on those right now since they seem very suitable to my current goals—cheap, relatively forgiving of poor design and experimentation, similar behavior across the series, and end up with something I can use.

But they’re just a stepping stone! Once I get a decent mini-system out of CMOS chips, I’m no longer using them as my go-to first choice.
It’s really hard to bootstrap a proper DIY modular synth system if you have neither existing modular hardware, electronics skills, or the willingness to buy expensive kits using circuits you don’t understand… so I’m working on having the second thing.

I’d say it feels more efficient to learn this way, but really I’m just focusing on what’s caught my interest right now lol.

2 Likes

When I started I had a Moog Mother-32 semimodular already, and in fact my initial goal was to augment that with a few modules.

But Moogs aren’t cheap (though the Mother-32 is a lot cheaper than most Moogs); a Behringer Crave, if you can bring yourself to give money to such a shitty company, would be an equally capable starting point at a lower price. Not necessarily for you, Aria, but maybe for others in a similar situation.

MFOS circuits are pretty well explained on their site; other good resources include the Kassutronics blog and Barton’s writeups.

4 Likes

I can’t even bring myself anymore to use a Behringer mic they made that I had for many years. Immediate mood wrecker. Anything that enters it will sound bad.

If you’ll allow me to get super serious for a moment: you know about the Tsukumogami?

The Japanese folklore idea that a tool that been loved for long acquires a divine spirit. Not in the metaphorical sense, I mean. A biwa or a koto that lives to be 100 years old will straight up grow a face and limbs and start playing riffs with its own body at parties with other yokai. I believe it’s 100% true and a literal account of a thing that happens frequently in real life. Instruments carry with them the circumstances of their birth.

When I made software modules, one of my instruments was directly inspired by a module a friend of mine created. What I created, in turn, inspired someone to port it to Csound—and also become a friend of ours.
Can’t do that on a Behringer.

3 Likes

No new electronics module, but a new module nonetheless:


Using a 18mmØ × 400mm tall pine rod from the hardware store, which is held with friction alone. Given that I can’t manage to remove the rod at all anymore, I will say it’s secure enough for its purpose.

Unlike other modules, the base can’t be printed upside-down, it needs supports, so the results look nasty… but it’s not like you’re supposed to see it.


Besides this, just a lot of experimentation on the breadboard, and with the already existing modules.

It’s becoming clear that my strategy of pumping mondo voltage into the cheapest USB powered speaker I could find isn’t gonna work out in the long term. Somehow, I have yet to blow them, but they react weirdly in my circuit: changing their volume changes the timbre. I can never trust that what I hear on those speakers is what I will eventually be able to record and play on good speakers.

The need for a proper output module with correct voltages and an op-amp becomes pressing.
That’s the part where my sourcing everything I can from Aliexpress hurts. I saw a capacitor not filtering DC voltage properly (whereas one with similar characterstics from a reputed vendor would), and a circuit with a TL072 I tried gave me trouble, but I know my chips are clones.
Still, it’s more likely I’m making basic mistakes than it is that I have bad components.

Another experiment with a Logic Noise circuit, this Twin-T filter is meant for drums, but also provides interesting filtering of audio signals. I could only get it to work partially (it calls for an additional buffer with a 4069UB, which I just can’t get to work). Getting good sounds out of it also called for a bit of experimentation with the 4 capacitors comprising the circuits, as the voltage influences the timbre, and this circuit is designed for 9V, while I use 12V.

The circuit also requires a stereo potentiometer, so I quickly made myself a little breakout board on a raft:


Yup, I still don’t like perfboard.

Now the question is, how to build a permanent version? Only perfboard would work with what I have (due to clearances and pin configuration), and I’d rather not continue using it for anything more complex than the raft above. I’m thinking of just using two separate pots, as it seems to work.

Another little board completed.

Taken almost verbatim from the Castle Rocktronics 002 Output Mixer (PDF), minus the mixer part, it’s doing some op-amp magic with a TL072, with some diode distortion. It’s definitely outputting more reasonable levels now! They’re low enough I have to use my Zoom recorder as an amplifer to hear it loud enough on my cheap USB speakers.

To make the clearances work with the 3D print, I had to sculpt a bit the supports. To do this I’m using a heat gun from my soldering station at 200°C, apply it point blank to the print for 15 seconds or so, then mold things using tools, or by hand (it doesn’t get hot enough to burn on my setup, but yours might be different, be careful, I’m not your mom etc.)

And here’s a pic of the full system so far:

5 Likes

With tails to scale of course.

2 Likes

A while ago, I purchased a bunch of 10nF capacitors without realizing they were rated for 300 whole damn volts, making them absolutely huge.

Finally found a way to use a few of them! Here’s what’s concealed beneath them:

12 fixed audio rate oscillators, using the simplest 40106 oscillator there is:

Before building, I tried out various resistor values, to cover a wide range of pitches, and for my 12V system, settled on the following: 1M, 330K, 68K, 33K, 12K, and 3K3. Two of each: I think it might be useful later to have closely but imperfectly matched pairs of square waves.

Using every oscillator of a 40106 generates a lot of cross-talk, it seems people who build serious VCO with them recommend using only 2 out of 6 outputs of the chip. But I mostly expect to use them to build percussion sounds, so it should be fine.

This simple circuit was entirely YOLOed without any draft, the layout existing only in my mind while building, and somehow it worked on the first try :smiley:

(Although, looking back on it, I misremembered advice of putting 1K resistors on outputs and 100K on inputs, and used 100K for outputs… but it seems to work just fine with my current circuits).

3 Likes

I don’t know if you’ve seen them before, if not I think you might be interested.
the very instructive videos of Moritz Klein on a VCO with CD40106 (part 1, part 2, 3…) with V/octave & waveshaper :slight_smile:

3 Likes

That Shapes VCO is neat! It’s meant for more traditional modulars with a -12V rail so I don’t plan to build his stuff yet, but I’ll probably add an exponential converter to a 40106 and other features when I can.

(BTW, another problem with following Moritz Klein’s stuff is that it’s video-only. I am dependent on subtitles to properly understand most videos due to auditory processing disorders, and Youtube auto-subs interpret many of his videos as being in the wrong language. It’s worth noting, if it’s useful to anyone else, that most of his videos were transcribed closely in the corresponding manuals of Erica kits, the PDF can be downloaded for free).

More designs I was looking at for the 40106 when I feel ready to try out something more complex:

Also, I have not tried to stick to the “pure” Lunetta style, where everything is a square wave, but I’m trying to keep it the main focus of the system.

I don’t work very fast and am not too concerned since I’m mostly trying to understand those circuits better and make them mine through small customizations, I think most of the boards I’ve made so far are drafts that won’t make it into the final system.

Right now the system is very noisy despite my using bypass caps everywhere (you can literally listen to the power supply by leaving the input of the final output module disconnected), and the most common problem I run into is that I don’t have the current under control—making chips output too much of it in places, and not providing enough to drive other parts of the circuit.

3 Likes

Now I know what to expect:

Yup. And that’s despite multiple reviews for the item showing them shipped in antistatic foam, and a 97.6% rated seller (albeit, only 1 year old). I’d name and shame but it’s pointless, those kinda shops close down and remake the second they tank their quality and their ratings follow suit.

Not every Aliexpress deal is a winner.

3 Likes

Some circuits I have been experimenting with have been giving me a ton of trouble, so it’s time to empty the breadboards of stuff that kinda works, even if I don’t fully understand why they work… Always time to revisit them later.

This is what I plan to build next time, I’m trying to customize at least a little every circuit I take from another source.

Feel free to let me know if I’m committing horrible breaches of schematic etiquette. It’s for my own documentation but eventually I want to be able to share stuff that’s easy to read and understand.

3 Likes

(Rather used to the stripboard), if I had to build this I would be lost. Not being conductive continuous lines (like the stripboard), but studs, I don’t really understand the connections between the components.
I have the impression that he will miss the underside of the plate to understand, did I miss something ?

2 Likes

And they are built!

Unfortunately, only two out of three circuits worked on the first try, the rightmost one, based on Ken Stone’s CGS24 gate to trigger converter, doesn’t work, but I had it working just fine on breadboard, so hopefully it will be easy to diagnose later.

For this hi-hat envelope vca (which I have no clue why it works), instead of potentiometers, I settled for something more fun: the performer places manually a capacitor in the circuit.

Still some space left for more stuff on the board. I wanted to build the Twin-T filter described in a Logic Noise post, but that circuit gave me a ton of trouble getting to work. Might build something else in that space, of the percussive persuasion.

I’m definitely building on both sides of the boards!


My custom PCB is very similar to a breadboard layout, which I use a lot for experimenting, so I feel pretty comfortable with it.

However, the layouts I made in DIYLC are probably impossible to follow for anyone except me, and I need to cross-reference with the schematic a lot… I would need to come up with a better solution if I were to share designs like these in the future.

1 Like