Shmøergh A1 is a fully analog semi-modular monosynth. It has a default hardwired patch inside and a couple of modulation sources and targets on the top so we can consider it semi-modular. Everything is handmade by me at home. Most of the module circuits are based on very well-known schematics and a few of the circuits are original. I list references for each module and I either provide links to circuit explanations or I try to explain them myself. Check this video to hear how it sounds.
So what can you do with this? First of all, the sound of an pure analog synth is another dimension. I could rarely if ever get this rich tone from a computer or a digital synth. Of course I'm biased but I honestly haven't had this much fun with Ableton and MIDI controllers as I had with this simple monosynth.
I used my Korg NTS-1 for some plate reverb but everything else is coming straight from the synth. Pardon my playing.
I don't know whether I would recommend building this as a first synth DIY project. There are quite a lot of points where things may go unexpected which can be a great source of frustration—as well as joy, when the bug is found. The individual modules are however not complicated at all and could be useful even for small projects (e.g. for a DIY Eurorack modular) or for beginners. So maybe yes, I would recommend as a beginner project 😄. Shematics below!
- Fat AF analog sound, much fun
- Two voices with 3 waveshapes (saw, triangle, pulse) and additional noise and square suboscillator sound sources
- 24dB/oct diode low pass filter with self-resonance and dedicated envelope generator
- Triangle and square waveform LFO
- Mod matrix for overriding the default patch
- Follows Eurorack standards: +/-12V power supply, 10Vpp audio and modulation signal, 0-8V envelope levels
- All modules are designed and built on separate PCBs for reusability and hacking
The A1 follows a pretty standard analog synth architecture.
Everything here is pretty self-explanatory. One thing to highlight tho' is that the LFO output amount runs through a dedicated VCA so it's possible to control it via voltage, e.g. with the modwheel output of a Keystep—similar functionality as the Minimoog modwheel.
On the top of the synth are some modulation sources and inputs which allows overriding the default patch. Modulation inputs break the original modulation signal path through switched jack sockets. This is nothing new, for example Korg does the same in the ARP 2600, one of the most popular semi-modular synths ever.
I built the modules on separate PCBs because I make boards by hand at home and the PCB size is limited. Plus this way I could reduce the risk of a single mistake in one of the circuits ruining the whole synth circuitry. Additionally the modules are interchangeable and I can keep experimenting on the system (it's a prototype). It's pretty much like a hardwired modular synth in a box.
I used thick traces and large spacings on the PCB to make it less error prone for DIY builds. There's a Eurorack power supply on all modules to make them reusable as standard Eurorack modules without modifications.
KiCad projects including schematics and PCBs are available on the project's Github page.
I'd suggest to download all schematics from the link above and open them one-by-one as you read along the details of each module. I only highlight certain parts of the schematics in the text below. Please use the link above for the full schematics of each module.
- 1x dual-rail power supply (+/- 12V / 830mA) with 12-15V DC input
- 2x identical VCOs with three basic waveshapes, with pulse width modulation and linear FM input. One of them has an extra noise circuit, the other has a suboscillator circuit (2x2 PCBs)
- 1x mixer, 1x LFO, glide and output stage circuits on a single board (1 PCB)
- 2x VCA
- 2x ADSR
- 1x 24dB/oct diode ladder low pass filter
- 1x modulation source mixer
- A couple of extra stripboards for small stuff, like controlling an RBG led or signals multipliers etc.
The most common power supply unit in synth DIY circles is the famous "Wallwart PSU", popularised by Ray Wilson. It is a simple, cheap and easy-to-build circuit. The only problem with it is that it requires an AC/AC wall adapter which is practically impossible to get in Europe. So I went with a switching power supply even though it's significantly more expensive and less DIY.
The synth is powered by a Meanwell DKMW20F–12 switching power supply that outputs +/-12V voltage and 830mA current. It draws about 240mA so there's plenty of headroom if one wants to add more modules to it.
Unfortunately since then this power supply seems to be unavailable in most suppliers (everything for synth DIY is getting impossible to to get). An alternative could be NSD10-12D12, which is half the current but it still should be more than enough. Note that I haven't tested this power supply and if you plan to use it then the PCB would have to be updated. Based on the datasheets, it has pretty similar noise figures to the DKMW so I assume it should be a fine replacement.
The VCOs follow a classic design from the 60s-70s. If you want to understand how it works I recommend reading Hal Chamberlain's excellent book, Musical Application of Microcontrollers and watching Aaron Lanterman's amazing series about analog music synthesis on YouTube (here's the part about this particular VCO).
A somewhat interesting detail is that I'm using a BC847BS dual transistor for the exponential converter for temperature stability. If you plan to solder it by hand, prepare your good eye and steady hand because this thing is extremly tiny! Alternatively you can use a matched pair of standard 2N3904 or BC547 transistors.
Another part to highlight is the octave switch for which I needed a precise 5V source (provided by the LM336 regulator) and 0.1% resistors to make sure the synth stays in tune no matter which octave is selected. The circuit for that is on the waveshaper board and it's connected to the core board with external wires.
Everyone who ever tried to build an analog VCO knows that the biggest challenge with it is temperature compensation. There are three very common ways to deal with it:
- Using an ~3300PPM/°C tempco resistor somewhere before the exponential converter (typically in the feedback loop of the CV input mixer). This should be the ideal and simplest thing to do but of course it is impossible to get 3300PPM/°C tempco resistors.
- Using an NTC (great article about it here)
- Using a KTY81 type resistive temperature sensor. The idea is that a combination of a standard 50PPM/°C resistor and a KTY81 resistive temperature sensor would be an equivalent of a similar value 3300PPM/°C resistor which is required for good temperature compensation in a VCO.
For my past VCOs I always went with NTCs but this time I tried the KTY81 method and it worked pretty well. I made a small calculator for equivalent resistance values. At the end I used KTY81-120 (which is about 1kΩ) and a ~1.4kΩ resistor in series in the feedback loop of the CV input mixer that results in a pretty close temperature coefficient equivalent to the required value.
Befaco Even VCO
Leapfrog VCF – temperature compensation (p. 61)
Hal Chamberlain – Musical application of microprocessors (p. 181)
Aaron Lanterman – VCO: sawtooth core
The greatest explanation of exponential converters ever
Rene Schmitz: Exponential converters
KTY81 tempco equivalent calculations
Mixer + output stage
The mixer is an original design but it's basically nothing else than a couple of inverting op-amps with multiple inputs. Very simple and basic stuff.
The filter is heavily based on Moritz Klein's excellent diode ladder video. But TBH, while the circuit sounds amazing it has some issues which I'll try to fix in the future: the CV input is a passive mixer which makes it a really hard resistor-balancing act to extend it. The other issue (which I couldn't really fix yet) is that the resonance path can pick up a lot of hum from the environment if a longer cable is used to connect the resonance pot to the filter PCB. My fix right now is to use a shielded cable and ground the resonance pot itself but that still feels quite temporary.
While this filter sound awesome I'm still planning to try various other filter designs in the future and compare their sounds to choose the one I love the most.
Probably the simplest circuit in the whole shabang.
R28 is just a pulldown resistor so that there's no unwanted noise on the input when there's nothing plugged in. The control voltage input is buffered by
U4A and then it charges a fairly big capacitor through the
If the pot is turned all the way down, the capacitor charges instantly, if it's turned up it takes some time. The bigger the capacitor, the longer glides you can make with it. Then the output is buffered again and there's a small resistor for small output impedance. I used a logarithmic potmeter to have more control on the lower end of the spectrum.
The LFO circuit is based on the other most popular VCO design in the world, which I already used for the classic VCO in my modular. It's great for LFOs because it generates a true triangle and squarewave without any extra waveshaping. The disadvantage is that there's no pulse width modulation. If you're interested in how it works in details, check out this great video by Aaron Lanterman.
Because of the size of the project, I wanted to go as simple as possible for all the modules. So I first built a couple of discrete (transistor based) VCAs based on this design by René Schmitz. Unfortunately there was some very minor bleeding and a little distortion which was annoying enough to try a different direction. I eventually went with the other classic circuit: using a soon to be dead LM13700 OTA chip and some control circuitry and built a dual VCA from a single IC. I kept both of the schematics just in case, but at the end I used the OTA version.
This is again a classic synth DIY circuit called "The Fastest Envelope In The West" by synth builder and designer guru René Schmitz. My version is pretty much using the exact same values, the only addition is a 1k trimmer at the output to be able to fine tune the peak voltage of the module.
Note that I tried to use a regular 555 timer chip instead of the CMOS 7555 but the decay and sustain part didn't work, plus – as I learned – the regular 555 draws significantly more current than the CMOS version. Just go with the 7555.
Rene Schmitz – Fastest envelope in the west (v3)
The modmixer is a custom part of the synth. It's inspired by the Minimoog's modulation mix knob which allows mixing two modulation sources to get interesting modulation source signals.
In my case though, since I already had a modmatrix that can be used with Eurorack patch cables, it made sense to build a generic version of it with inputs and outputs.
The circuit is a standard inverting amplifier that has a reverse fader on its input which allows to balance the amount of two modulation sources. Then there's a level knob and the signal is reversed back with a second inverting op-amp which also has a ~3x gain on it so it's possible to "overdrive" the modulation if needed.
The A1 proto case is a temporary, oldschool-looking wooden box which is just enough to hold these many modules with a million meter cables. It's been real fun to design and figure out how to build one but if there's going to be any future version of this synth, I'm 100% sure it's going to look very different.
I had a couple of variations before I got to the final thing. Here are some WIP pictures.
Aren't there enough monosynths in the world? Why make yet another one?
I hear you. There are enough analog monosynths for sure. I wanted a monosynth for quite some time and I could have bought an Arturia Microbrute or a Korg Monologue which would have been a million times simpler. But I'm a maker and I love working on DIY stuff both electronics and woodworking. I also love understanding the underlying physics and maths so I decided to just make one.
How long did it take to make this? How do you have time for this?
About a year, but I've already built quite a lot of modules for my modular synth for about 2 years so I had a bit of a headstart.
I have a family that I love, a job that I love, I love playing basketball so it's definitely challenging to find time for a project like this. I've been working typically about 6 hours almost every weekend for about a year - it's a weekend project in the truest sense of the word. If I sum up the time I spent on it (about 300 hours) it's not that much but it requires strong dedication and accepting that it's not going to be ready in a couple of days.
I love it! Can you make me one?
Nope, sorry. But you can download all the schematics and build one for yourself 😊
Why the pig?
Pig cool! 🐖