Xaoc Devices (pronounced "chaos devices") is among our favorite designers in the Eurorack modular synthesizer space. Xaoc was formed in 2011 in Poland, originally conceptualized by graphic designers Tomek Mirt and Marcin Łojek. Today, their lineup is vast—with various themes including everything from fresh takes on analog synthesis all the way to complex digital sound generators and wild modulation sources. They're among the small number of makers to have created what we feel is a full complement of modules with a unique and distinctive unifying sonic and visual aesthetic...which is certainly no small feat. While their work certainly plays well with others, it's safe to say that an all-Xaoc system could make for a rich and well-considered instrument in its own right.
Aside from cult favorites like the Batumi quad LFO, Odessa additive synthesis oscillator, and Belgrad dual-peak filter, Xaoc has also developed an entire collection of modules for creatively exploiting the processes of analog-to-digital and digital-to-analog conversion: the Leibniz subsystem. That might sound like a mouthful—and indeed, we're not aware of an exact precedent to these modules' approach to signal manipulation. The idea is relatively simple: you can convert an analog signal to a stream of digital "bits" (gates/triggers), manipulate those bits, and then recombine them back into an analog signal. Or, you could use those "bits" as gate/trigger/modulation/audio sources themselves; or, you could combine a number of gates/triggers directly into an analog signal; etc. The Leibniz subsystem includes modules for handling the analog-to-digital and digital-to-analog conversion, as well as modules for manipulating bits, performing lookup functions, and much, much more. Musical applications can range from peculiar sequencing tricks to wavetable-like distortion effects, creation of unruly control voltages, and much more.
That still sounds like a mouthful, eh? Well, you're in luck—we were able to talk directly with Mirt and Łojek about the Leibniz subsystem: its origins, core concepts, ideal starting points, patching tips, and much more. Along the way, we learned a lot about the Xaoc design philosophy and approach to music-making; find the full interview below.
An Interview with Xaoc Devices
Eldar Tagi: Hi Tomek & Xaoc Devices team! Thanks for chatting with us. Let's begin with a brief introduction: when and how did you get involved in designing synthesizers, and what do you do at Xaoc Devices?
Tomek Mirt: More than decade ago, as modular enthusiasts Marcin and I had this crazy idea to start Xaoc Devices. I might sound like an old fart, but everything was a bit different then, and it wasn't as mad an idea as one could think now. Not much later Maciek, our main engineer, came on board and everything started to look serious for us. And this was how we got involved in designing synthesizers. I did some DIY things earlier, and I studied industrial design—I once designed an interface for a MIDI controller, but it wasn't anything serious.
ET: If you were to briefly summarize the overarching philosophy behind the Xaoc Devices approach to synthesizer design, what would it be?
TM: I think it is something hard to grasp, as there are three completely different personalities behind our projects, and probably each of us sees it a bit differently. I think we all agree that we are always trying to bring something new to the table, even our more "normal" modules need to have some unique features. Even though we are able to build a whole system with our modules only, we are focused on how they work in various systems.
I think we have more than one path we follow: modules like Zadar or Odessa are digital and conceptually completely new to Eurorack or even hardware synths; Sofia, Berlin or Sarajewo are quite old-school analog circuits, but greatly expanded and with modernized; finally, we have the whole Leibniz Subsystem, which is entering completely new and experimental territories, even though it uses old 8-bit technology.
ET: Where do you tend to draw inspiration from for your work?
TM: There are many sources where I look for inspiration, the most obvious one is constantly dealing with the modular synth, making my own music, but also observing what others are doing and what may be missing in the ecosystem—which is exceptionally crowded now, so there is just a slight chance that you create something really unique.
I constantly try to look at even the simplest issues from a new perspective, sometimes it is worth turning something upside down. Even if it seems absurd, it often unlocks a new way of thinking. It's a bit like collecting small parts that you never know when or what they'll be useful for. I try to read as much as possible about the history of synthesis, many academic ideas from years ago have never been implemented, because it wasn't possible then. Now it is. Together with Maciek we also talk a lot about techniques used in electronics, sometimes outside of audio, wondering if there is a place for them in the modular realm.
ET: With the recent introduction of the Berlin oscillator, we wanted to focus this interview on the ever-expanding family of modules in Xaoc's roster, the Leibniz Binary Subsystem. What are the origins of the idea, and how would you describe what LBZ is?
TM: I don't remember exactly how it happened, but at some point Maciek was explaining to me how analog to digital converters work and it was a moment of enlightenment for me—an analog signal comes in and you have multiple gates at the output, it's practically a complete module already!
When we started thinking about what it could actually be used for, a hundred ideas came up. It was hard for us to predict at the start which way it would all go. I know that it might be a bit intimidating for many, as it seems to be aimed at nerds with a penchant for sound experiments. But the real strength lies in its shape-shifting nature: on the one hand it can operate as a large ecosystem, but on the other hand it can act as multiple specialized microsystems comprising two to three modules. If you want to build a glitchy, crackling generative behemoth that will live its own life, that's possible, but just as well you can limit yourself to a simple, interesting pattern generator; add a pinch of digital dirt; create a wavetable oscillator with a twist, or a unique voltage processor.
As I said, even with a small set of two or three modules you can awaken the dormant potential. We encourage users to "misuse" the Leibniz modules, e.g., create a pattern generator to generate an audio signal, or use an oscillator to generate a CV sequence. The Leibniz Subsystem is deeply modular in its nature—I see it as a shapeshifting beast.
ET: What do you find to be exciting about working with 8-bit signals in the context of modular synthesis? What creative opportunities does this offer to artists?
Maciek Bartkowiak: What first comes to mind is the multitude of gate signals available, which makes this system perfect for experimenting with rhythmic patterns, switching, and everything that involves managing a patch. Eight bits means that most of the generated events can be extremely complex, but they are never actually random. For me, this works great in combination with more traditional ways of generating sequences. It is easy to create something unpredictable, but still repetitive. By balancing between these two methods of creating sequences, you have a wide palette of patterns at your disposal: from tight and traditional with just a hint of craziness to totally unexpected clusters of throbbing beats.
Another Leibniz strength is the high sampling rate, which allows for the generation of extremely detailed, harmonically rich audio signals, perfectly suited for further processing with filters. When we showed Berlin in tandem with Jena at Superbooth 24, many people pointed out this feature. Another exciting thing is processing analog signals in the digital domain. Messing with bits, instead of audio, has its own unique character. It's easy to go extreme, but with a little restraint you can learn to control it to achieve subtle and unique results. The possibility of generating dozens of dependent events that will not be as obvious as when using simple Boolean logic or comparators is fascinating. I sure could go on listing Leibniz's uses for a long time.
Leibniz is a fascinating tool, especially if you want to look at music creation in a slightly different way, but not necessarily through the eyes of an engineer skilled in advanced methods of digital signal processing. This is the moment when, accustomed to straightforward 4/4 rhythms, you hear something polyrhythmic or a different division for the first time and it turns out that there are more possibilities than you thought.

The idea?
Obviously, we didn't design and even didn't conceptualize all the Leibniz modules right at the start. At first, it was just the idea of doing the analog/digital conversion and messing with the binary data. The initial inspiration came from observing how the activity of individual bits of data from the converter depends on their weight (bit number or position)—the lower the bit weight, the higher its activity and its seeming randomness. The conversion process is quite similar to multi-stage wavefolding, which is known to make even simple signals increasingly complex.
So the individual bits changing over time are interesting signals derived from the original signal. They are related in a complex way. On the other hand, since the conversion is a reversible process, modifying the bits may have a surprising impact on the signal reconstructed from them. Thus, a pair of converter chips offers unusual ways of manipulating voltages, adding structure and texture without the need for numerous nonlinear circuits that often require trimming and are sensitive to temperature fluctuations.
With the conversion covered by Drezno, we introduced subsequent Leibniz modules corresponding to the classic building blocks of digital control systems and computers (Erfurt, Poczdam, Rostock). Various combinations of these modules can exhibit complex behavior patterns that can be musically exploited. Building sensible setups from these basic blocks does require some understanding of binary logic, but really nothing more difficult than the concept of waveforms, filtering, and modulation in a purely analog modular system.
To create interesting patches, you need to go beyond simply connecting the modules with ribbon data cable at the back, just like you need to go way beyond the VCO-VCF-VCA scheme in your classic modular. Experimentation with individual bit signals is easy and safe when delicate CMOS circuits are equipped with some rudimentary UX and protected with buffers. You really cannot break anything by patching (however, you can damage things by plugging dangerous voltages to the interface pins at the back, beware!).
What makes it difficult to grasp is that the Leibniz data may represent very different things: these are either 8-bit numbers, or just parallel gate signals. At any given moment, the combination of all eight binary signal states (ON or OFF meaning a logic 0 or 1) altogether defines a value between 0 and 255, which may be converted to and from a certain voltage. However, the same binary signals considered individually and changing over time are just streams of gates or triggers with diverse applications. They can be rhythms, frequencies, event triggers, etc. The choice between one interpretation and the other depends on the context, but you are free to decide.
Revolution or just plain old circuit bending?
Some enthusiastic users claim this is a revolution in modular synthesis. Well, it is undeniably different from most modular approaches. It breaks the traditional paradigm, forcing users to change their mindset. But once you understand it, it's just another set of fancy tools. Patching with the Leibniz modules is indeed very much like the circuit bending technique popular in the 1980's, which involved people randomly modifying simple electronic toys that used early digital circuits with many signals exposed. Modifying and purposely breaking the circuits forced them to run erratically, yielding surprising sonic results. Here, you don't have to do it blindly—everything is documented, and it is quite robust.
Note that we often stress that these modules are implemented in hardware, using discrete logic chips, instead of just a fast microcontroller running some code (except Jena, which is CPU-based). The reason is that only hardware can offer nearly instantaneous response (like analog circuits do), hence it opens all the crazy feedback possibilities. Also, hardware can run at a super wide range of data rates. For example, with Berlin you get waveforms with variable sampling rates and no aliasing because it doesn't reduce the number of samples at high pitches like almost every other wavetable oscillator.
Complexity vs. Chaos
Let's face it, electronic sounds generated from basic waveforms and processing tools often do not evoke the same complex emotions as they did in the past, especially since they are so ubiquitous in modern music. Also, running a patch from a simple step sequencer can be boring unless you put a lot of creativity into breaking away from repetitiveness. Reaching a perfect balance between the familiar and the unexpected in modular music is quite elusive, both on the level of composition and sound palette. Complexity that keeps listeners engaged can be achieved through complicated modulations programmed in the patch. However, the common practice, especially in compact systems, is to fake complexity by adding elements of randomness, sampled noise, etc. This is the lowest hanging fruit. When you create an ambitious complex modular patch with plenty of cross-modulations, your whole setup becomes a chaos generator, something you created, but still there may be some blind chance involved.
Random and chaos are not the same thing: chaotic behavior is deterministic, it is just sufficiently complex that it escapes our ability to see patterns. Chaotic analog patches are often over-sensitive to initial conditions, where a slight turn of one knob may yield completely different results, which may be fascinating. Still, it also means they are sensitive to even minuscule amounts of noise in analog circuits. So they are to a degree still driven by randomness.
Leibniz subsystem patches make it easy to introduce various degrees of complexity, even with not very elaborate patches. Note that a purely digital system is never random. It may behave pseudo-chaotically, but is in fact purely deterministic (until you introduce some analog randomness, like, for example, a quantization error in an A/D converter). Chaotic or pseudo-chaotic behavior may have many forms and uses: super complex and impossible-to-follow rhythms, twisted and glitchy waveforms, ultra-rich signal spectra or bursts of unpredictable events.
So is it all about chaos and glitches?
Absolutely not. Processing of Leibniz data (when you treat the 8 binary signals as numbers) may yield smooth and very musical results. It is all about mathematics, which is the base of all music. For example, a simple clock division with Erfurt by accumulating numbers often yields rhythms in the same ballpark as the popular Euclidean ones. Also, Jena with Berlin may generate smooth waves with deep through zero-FM that sound very crisp and clangorous. The very high sampling rate helps to prevent aliasing in many situations.
ET: That all certainly sounds fascinating. What would you say are the most challenging aspects of dealing with the 8-bit domain?
TM: I think that for many people the major obstacle is exactly what lies at the heart of the Leibniz Subsystem, i.e., the analog to digital conversion, understanding that we are just changing audio or CV into a set of 8 gates. At the start, you don't need to know much more. You don't need to follow what is happening with each bit every second. Another problem is the idea that Leibniz is a monolithic whole where you have to connect all the modules into a long chain. This is possible, but often doesn't make sense. Leibniz works best as a set of a few small islands that interact, but can also operate completely independently.
ET: Now that Berlin is available, some see it as a module that opens up the LBZ to newcomers. Why do you think that is the case, and how do you personally see Berlin fitting into the existing ecosystem?
TM: You're right, Berlin is a more specialized module than Drezno, which in this case means it's much simpler to grasp. Jena is also a natural partner for Berlin, so that pairing is a good starting point for further explorations. If you add Gera or Lipsk, you get an extremely powerful oscillator and you don't even have to know what's going on with all these bits. It is just a rewarding playground for sound experiments with a lot of patchpoints.
This is a perfect and natural place to start for all those who just want to begin their adventure with the Leibniz Subsystem, but even more advanced users should be satisfied. For example, you gain a way of tuning a clock that runs the whole subsystem; also, Berlin offers much more precision compared to the Drezno/external oscillator combo. I believe there were many exceptionally attractive small sets of Leibniz modules already, but the introduction of Berlin increases that number significantly.
Berlin with Jena is fun, but how about Berlin and Rostock, or Berlin and Jena with Drezno used as an insert point between these two?
ET: At first glance, Berlin might be mistaken for a traditional oscillator, but it's much more than that. Can you outline what makes it unique, and what sonic/control possibilities it offers when used in combination with other LBZ modules?
TM: Berlin looks inconspicuous, but it is actually three modules in one—a tunable clock for the entire system, a perfect sawtooth generator and a DAC converter—the output from the system.
Its strength lies in what it offers when combined with other modules. On its own, Berlin is not so keen to show off its strengths, but together with a wavetable memory (i.e., Jena), it works similarly to oscillators found in the classic PPG or Fairlight synths. It means it does not lose samples when you change the frequency, so aliasing is not a problem. Additionally, it is able to generate very high frequencies. All this is already a certain advantage and the reason why it makes sense to buy this pair, instead of just the oscillator module.
But remember that you can (and should!) expand the setup. Together with Jena, apart from the wave available at Berlin's output, you have access to 8 bit outputs, which already expands the audio palette, but add a simple module such as Lipsk or Gera, and you're in for some advanced waveshaping by simple cross-patching. The results may be additionally enhanced with simple logic modules, dividers—stuff you have in your modular system anyway. These operations expand the number of harmonics, so the resulting waves constitute a great base for further processing.
I talk a lot about Berlin paired with Jena, but it can be an extremely original and interesting sound source together with Rostock. You can generate many unique waveforms, especially if you insert Lipsk between these modules. You can also use Poczdam to combine two Berlins running at audio rate—a patch that definitely requires more modules, but renders distinct and uncommon results.

ET: Where would you personally recommend someone start, if they decide to explore LBZ?
TM: Currently I see several interesting paths. We have already talked about a small system starting with Berlin, but a completely different good start is Lipsk connected to Erfurt. Together, they make a versatile gate pattern generator. If we add Gera, we can create feedback and generate more complicated patterns. With Rostock, we can loop different fragments of the patterns, so it is a system that is easy to expand.
Drezno on its own, without any additional modules, can be used in so many different ways that it should feel at home in many systems. It is an excellent tool for processing control voltages. Just plug in the simplest triangle LFO, mix the bits in a random way with Drezno's patchbay and you get a complex shape.
To sum up, there are currently three modules you can start with: Berlin if you prefer to build an oscillator; Erfurt as a great start for a gate generator; and finally Drezno for signal processing, but also many other purposes (basically anything really). In most cases, Jena or Lipsk are perfect candidates for the second module.

ET: Do you have a favorite patch or setup within the Leibniz subsystem that you can share with us?
TM: I consider it more of an idea for further exploration, but I like to use Rostock with Drezno to create a very short delay line. If we add feedback outside, we can create interesting comb filtering effects. I've spent a lot of time just messing with this patch for fun. Here's just a few other ideas off the top of my head. If you want to use Drezno as a waveshaper, a VCA before the module is your best friend (together with careful attenuation). Drezno as an insert point for a VCF between Berlin and Jena is a joy to use. If you use a chain of Berlin, Lipsk, Drezno and Jena, you can additionally use the bit outputs on Jena to feed them back into Lipsk. I know it is easy for me to suggest various module combinations when I have all our Leibniz stuff available in spades in our office—sometimes we create rather absurd monstrous patches just because we can—but you really don't have to have everything to have fun.
ET: How do you envision the LBZ subsystem evolving in the future?
TM: It's hard for me to answer this question as we have many ideas, but I think it would be good to give the users a chance to master what we've already released. User feedback is also essential to envision new paths for the future. For now, we want to focus a bit on other projects. When you spend a lot of time with a large Leibniz Subsystem, you come up with ideas for many new modules, but then you realize that they might address the needs of only a handful of users (such as me for example). At some point you have to exercise a bit of common sense and responsibility. Personally, I'd love to have a binary summing unit.
MB: Are we done with releasing new Leibniz modules? Is the subsystem complete? I don't think so. However, what we have now is a very powerful set of modules. We certainly do not advise customers to jump in at the deep end, because it might feel overwhelming. Starting with just 2 or 3 modules is sufficient to achieve spectacular and inspiring results, and quite often it is a good place to stop as well. However, we do understand that some users may want to grow their subsystems.
Our advice is to abandon the idea of making a huge setup of Leibniz modules all connected into one monstrous patch using the ribbon data cables at the back. This often limits creative applications, as it is hard to comprehend what is happening, and it may be difficult to use some combinations of modules independently. It is better to divide your setup into smaller task-based combinations.
Users enquire about a flexible solution to freely route Leibniz data between many modules, which would hypothetically solve the problem of tedious re-configuring the setup with ribbon data cables. Unfortunately, such a device would have to be massive and very expensive. We know, we've tried. The combination of a huge number of wires and data speed is a real challenge. Leibniz signals run at MHz rates, and the connections must provide sufficient bandwidth. Switching them may be implemented in several ways. Mechanical multi-pole switches of the required capacity are super rare and prohibitively expensive. Electronic switching is possible, but a common microcontroller is simply far too slow. It would engage hundreds of middle-scale chips, or a giant FPGA chip that costs hundreds of USD per piece. We seriously doubt there is a market for $1K+ switchboard for Leibniz, and simply cannot take the risk.
ET: Would you like to share with our audience any other projects the XAOC team currently working on and/or passionate about?
TM: I can't reveal any details yet, but we have at least three pretty big projects in the pipeline. They should be out by the end of this year or at the beginning of the next year.
ET: How do you like to spend your time when you are not working on synthesizers?
TM: I still spend a lot of that time with synthesizers, making my own music and recording albums, besides that I am fascinated by field recordings, and I try to spend as much time as possible listening and recording my surroundings. I am crazy enough to release my own and other people's field recordings on my Saamleng label.
Thank you very much for your time!
(Ed.: Want to learn more about Xaoc Devices? Check out Waveform Magazine Interview with Xaoc Devices on Signal!)