[NB: While the focus in this essay is on the Eurorack modular format, the build principles discussed here are equally valid for any modular format, from simpler ones such as Tangible Waves' AE to brobdignagian monsters in the MU or Buchla formats, and everything in between. The current state of the art in modular synthesizers appears to be more centered around compatibility these days, as opposed to the earlier period in electronic music instrument development where different standards of many not-so-compatible types were more common, often due to manufacturers electing to adopt a more exclusive stance toward their competitors' standards. Thankfully, this is far less common now than, say, up through the first half of the 1980s when MIDI finally got a firm foothold and made consumers ask why you couldn't hook these older instruments up to each other nicely. The fact is that this issue, possibly more than any other, was the cause of the period of decline of interest in pre-MIDI analog synths from the mid-80s up into the early 1990s, and it's good to see it swept aside for the most part.]
Modular synthesizer. The words conjure up visions of walls of arcane soundmaking implements, connected with mazes of patchcords, looking for all the world like some sort of hallucinatory telephone switchboard. Long the instrument of the elite, well-paid, moneyed musician or the domain of the academic lab, this perception has begun a gradual change in the past ten to fifteen years as a new type of modular synthesizer format known as Eurorack has taken the electronic music world by storm. With smaller, often portable cases and cheaper, easily-available modules appearing on a seemingly-daily basis, the once-rarified domain has gradually been brought down to street level, where anyone who wants some sort of modular synth need not shell out tens of thousands of dollars on some enormous, immovable, studio-bound device. Certainly, even with Eurorack, it's possible to do this...but not necessary, as the sound-production machinery that would be found in something massive such as a Moog 55 can now be had for a fraction of the cost and stuffed into a case about the size of an average carry-on bag.
But should YOU have one? That's an interesting question. Even with the price-slashing and size-shrinking, these are still rather spendy pieces of gear that inevitably pose a learning curve – a dual learning curve, even, since most present-day synth users can freely create their own personalized instruments and, once having done so, they then need to figure out what exactly to do with this thing they've concocted as well as how to do those things effectively.
Initially, there's something very appealing...in a mad-scientist sort of way...of controlling a huge panel of dials and lights and cords running every which way, creating utterly alien sounds to the amazement of all looking on. But at the same time, synthesists have long said that creating electronic music, especially with something large and planted in place, is pretty much about as interesting as watching someone at a desk job. So there's actually a bit of a dichotomy about the 'crowd appeal' of playing a modular synth...or most any other synthesizer, for that matter. Consider: do you think you're putting on the show? Well, then you need to face the audience to have that sense of performer interaction...but if you do, then the audience can't really see what you're up to or, worse, the synthesizer is so huge that no one gets to see you, period. So if the synthesizer is the show, then you're forced into a playing position in which the crowd can see the lights and knobs and so forth...but your back is to them, and you're disconnected from the audience. If you're Miles Davis, this might be workable. But you're probably not Miles Davis. Sideways? Not much better, really. Ahhhh...so, is there a solution? Well...no.
OK, so if you're into 'star appeal', this probably isn't going to be too thrilling for you. You're better off sticking to conventional keyboards, which you can bounce around behind and connect with the crowd while still looking like you're doing something amazing. If you're a musician who's actually into music, though...then this could work. And work in an interesting way, as I've found over the years that performing with a modular setup really forces you to bring your 'A' game when it comes to the music you're performing. Oh, sure, you can also add trippy visuals to the mix, but then we're starting to talk about a level of production values that most synthesists just aren't going to be dealing with unless there's some major bank and logistics making all of that happen. And after whipping out a sizable stack of bills for a modular...even in these times when they're cheaper...money for wild lights and lasers and projections and fog machines and such probably is going to be pretty lacking.
So, quite honestly, modular synthesizers require people dedicated to MUSIC as a primary objective. Anything else is going to get swamped by the device's requirements, either in performance or in the studio. And speaking of 'in the studio', at that point the modular synthesizer is going to force you into a position where you MUST know your device's capabilities inside and out, because the temptation to noodle off into a corner is very real with these devices, and more often than not, this won't wind up being productive when there's a specific task at hand. Instead, these devices can wind up doing something very beneficial: engendering musical discipline.
And this is why very close knowledge of one's modular system is important. If a musician using a modular synthesizer can exercise suitable musical discipline, it makes their interaction with the instrument quicker, more to the point, and much more effective. Results tend to happen fast, especially when you understand the proper method of production of a sound you've envisioned or been requested to do by someone else who's envisioning it. But this comes with practice, with experience, and with diligence – in other words, musical discipline. Also, economy: you get to a point where you can immediately envision the most effective methods of realizing sounds, and this effectiveness leads to using only the exact things required, instead of the beginner method of “let's patch everything to everything else and tweak as many dials as possible”. If you can create a sound with only eight patchcords, then good. Those who can use more, do use more, but they're trying for certain levels of nuance in their programming. A good modular synthesist should only use what's necessary to achieve the correct result; the KISS (Keep It Simple, Stupid) approach definitely applies here.
This is not to say that the “let's patch everything...” approach doesn't have its place. Far from it. That sort of experimentation is best left to times of practice and exploration however, and with a modular synthesizer, even a fairly simple one, there's going to be a lot of that! This is how you gain knowledge and understanding of your instrument, how you get your 'voice', so to speak. And it ALSO takes time and discipline. Some people obsess over notating everything they do with these instruments, and I will note that if a performance requires a pretty complex patch, that's not a bad idea. But try as best you can to get used to coming up with the basics on the fly. Over time, you start to understand that certain waveforms work for this thing, that filter setting only gives me these sorts of sounds, this envelope generator has to be set just so for quick snaps and this other way for washes, etc. Time and discipline.
If that 'sounds a lot like work'...well, perhaps you need to go do something else with your time and money. But if you understand that this...like any other musical pursuit...REQUIRES work, then keep right on going with it. Music is actually serious. Sure, it's also fun...but it's serious fun. Always approach the creation of music, irrespective of whatever instruments are involved, with the mindset that “I am going to create something that creates great change”. What that change might be, well, that's up to you. But always step up to what you play with that mindset from the start, and you will find, especially with a complex, nuanced, individualized instrument such as a modular synthesizer, that your footsteps on that creative path are firm and always forward.
Now, let's get into what you need to consider when committing to building your modular synthesizer.
First of all, you need a box to put the stuff in. Always remember that whatever size of box you start with, if you're fairly new to modular synthesizer design, that box is too small. DO NOT use the size of the box as a critical limitation. It is actually far better to set an initial monetary amount that you don't want to cross and then, having done so, work backwards to see how you can make a nearly-similar setup work within your budget constraint. To help with that, I've found a very useful rule of thumb that can help on more basic considerations:
price ÷ width = cost factor
By using this, you figure out that, where basic functions are concerned and various modules tend to be interchangeable, you can factor out a cost-per-hp number that actually shows how much it costs to fill up a given space in a row. The lower this can go overall, the cheaper your system gradually gets. It also has another interesting side-effect, I've noted: it tends to force one into builds in which there is more open panel space, and this can be useful in order to keep space clear to ease interaction with the instrument.
OK, let's say you've figured out that you've got $5,000 to spend. Next, put together something WAY too big for that dollar figure, either on paper or using various computer tools like Modulargrid, etc. This isn't what you'll eventually wind up with, but a 'sketch-space'. You will want ample room to move things in, out, around, assemble possible subsections, and so forth. Having done that, let's move on to SOURCES:
Really, when I talk about 'sources', I'm talking about 'things that make noise'. Oscillators, noise generators, external inputs – all of these are the initial sources from which the voicing section of any synthesizer, modular or otherwise, is made. Even the variety of digital generators that exist...samplers, wavetable oscillators, etc...still fit into the basic category of 'oscillator'. As a rule, you want to group all of these in their various forms in the same general area.
Which brings up an important point: before constructing ANY modular system of your own, look at how others have done this, especially in 'classic' instruments such as Moog systems, patchables such as the ARP 2600 or EML 101 and 200, or in modern builds either by experienced electronic musicians or firms such as Doepfer, Synthesizers.com, et al who have experience in creating factory builds which work well in their off-the-shelf configurations. You should notice that there is a certain type of grouping and work-flow that these various instruments tend to have in common. Note carefully, then continue...
To return to voicing: now, let's divide up your main sound generation devices in the oscillator section. You should have a few more-complex modules, and several very basic ones. How these get used is that much of your actual sound production happens with the complex devices, or with combining the simpler ones. The simpler oscillators generally (but not always!) are more useful in mixed groups, or individually as audio-frequency modulation sources for the complex generators and other processing modules and subsystems. Of course, if you're trying to emulate something simplistic in structure such as a Roland TB-303 or a Micromoog, using just one simple oscillator as the voicing source is just fine, too. But make sure to leave yourself lots of options.
But there's other options than just basic oscillators out there these days. Many digital devices exist in the modular world today that offer analog control over things that, even just several years ago, required full-on microprocessing power to generate. And they still DO...but the microprocessors are now cheap and tiny and tucked behind a modular synthesizer panel, hooked up to real knobs and jacks and such. So now, we can cram in a very functional sampler, various physical modeling devices, PPG-style wavetable scanning (with user-definable wavetables, even!), algorithmic FM synthesis with analog controls, audio loopers/choppers/scramblers...hell, if it generates an audio signal of some sort, you're likely to find it in a voltage-controllable synth module. Even bizzaro random things such as multiple FM radio tuners or even shortwave tuners can be had and hooked in right alongside everything else! So the sky's really the limit in terms of the number of ways you can generate audio signals; if you had a wild hair to try it, you could dispense with proper oscillators altogether and use only physical modelers to generate your initial audio, for example.
At any rate, contiguous to all of this, consider a few initial timbral modifiers such as distorters, wavefolders, audio dividers and harmonic multipliers, and the like. These are very useful at adding harmonic complexity to your initial oscillator outputs and beefing up these sounds into something richer which, when you get these sounds down to the subtractive section of the audio chain, will yield amazing results. And there's a lot of stuff that can do these functions; for example, you can use a comparator to 'square off' most any waveform, creating a variably-shaped clipped pulse wave from pretty much anything, with your pulse duty cycle determined by how you've set the comparator's threshold. And the more complex the waveform going in, the crazier the result coming out the other side will tend to be! Ring-modulators also fit here, since the general rule with them is that you'll be using an oscillator (usually with some simple waveform) in at least one input of it, if not both.
Right by this, or pretty close to it, is where you're going to want some basic mixing capabilities. Even if you're working in a more 'West Coast' direction where intermodulated oscillators create your basic sound, Eurorack still requires the mixing of signals to create a composite result. Even the Buchla designs, the epitome of 'West Coast', still require this for their audio paths. Why not just patch 'em all into a multiple, then? Ohhhhh...you DON'T want to do THAT. Why? Well, many modules use diode protection on their outputs to let you connect multiple outputs together for summing in that way...but many other modules DO NOT contain diode protection for their outputs. Don't get into the habit of doing this, therefore; the cardinal rule of patching is to NEVER, EVER connect an output to an output, no matter whether the manufacturer says it's safe or not! It only takes one misadventure with a reverse voltage coming back into an output, thereby wrecking one or both (or more!) modules, to show you first-hand why you NEVER, EVER do this! This is why mixers exist, folks!
Going one step further, you might consider a voltage-controlled mixer. Now, these are only partly mixers; the rest of a voltage-controlled mixer consists of voltage-controlled amplifiers, or VCAs. VCAs are the 'Rodney Dangerfields' of synthesis: they don't get no respect. People often ignore these boring modules...and they shouldn't! VCAs in an audio path, especially in a voltage-controlled mixer, allow you to control changes in oscillators, filters, etc etc as a unified part of a patch. If you, for example, want a gradual change of timbre by going from some complex FM subpatch to a pure tone like a single sinewave, VCAs are the things that allow voodoo like that to happen. Just patch an envelope to one VCA and set it one way, then another, different envelope to another and set it to rise in volume as the other falls. FM subpatch to one, sine to the other, mix...and voilá! It's magic! Well...no, it's just VCAs at work.
OK, back to our build. Now, let's move back over to below the oscillator section. Oh, I didn't mention the noise or external input? You're right...and that's because noise actually has a different intensive use which we'll get to in a bit, and the placement of the input preamp (and, potentially, an envelope follower) can actually go most anywhere in your audio chain sections, depending on what you think you'll want it for. If it's another 'source', then up by the oscillators is fine, but if you just want to mangle sound, then it's OK to locate it closer to filters, effect processors, and so forth. But back to that next section...in here, you'll likely want to put modulation sources, since they'll be used on both the oscillators above and filters, which you want to locate directly across from them.
Modulation sources that operate below the audio-frequency range really get divided into two types: running and triggered. Running sources are things such as LFOs, function generators, looping envelopes...anything that does its thing as long its active or by itself. Triggered sources, however, are where we find 'one-shot' envelope generators and their ilk, since these devices tend to only operate as programmed when they've received a gate or trigger signal from elsewhere in the synthesizer. And there's plenty of stuff that fits into an 'in-between' zone where certain modules can be configured to operate under either condition.
Why have these? Well, a good modulator compliment has loads of uses. You can use their subsonic control voltage signals to affect the pitches of audio sources, alter the behavior of filters, control VCAs, creating all sorts of sonic variation in this way. A hefty modulator compliment allows the synthesist to program sounds that evolve over time, that possess lots of internal complexity, or 'humanize' the sounds being generated. And this doesn't go for just 'East Coast' systems; Don Buchla's synthesizers relied heavily on modulation sources, both sonic and subsonic, to create their particular sound and programming methods. Plus, these signals can be mixed just like audio, by using DC-coupled mixers and/or VCAs that allow the subaudio frequency and/or DC voltages to pass through and be combined. As a result, it's possible with just a couple of basic mixers and VCAs to create insanely-complex modulation signals with just a handful of basic modulation sources. Also, without envelopes or LFOs or the like, the low-pass gates of these (a core part of their distinctive sound) would be totally useless in the way we're used to utilizing them, as one example. And there's lots of other examples where, with a little imagination, you can see that not having modulation sources would equal some pretty boring sounds.
Anyway, one other thing you want in this area, close by the oscillators, is processing for the oscillator control voltages. And this gets us to multiples, another unfortunately-neglected family of devices.
Multiples are pretty much what they sound like: one signal goes in, several come out, preferably identical. They're a backbone of modular programming in that they let you send a single output to lots of different input locations at the same time. Any decent system, even of the smallish variety, needs these. But when we're talking about more precise signals, such as a control voltage that drives several oscillator pitches, then a problem can arise: voltage droop. You don't want this if all of your audio sources have to stay in tune with each other, and it's caused by some complicated electronic mumbo-jumbo wherein different signal loading factors across several multiple-split inputs cause a precisely-scaled single voltage to...well, droop. Not...good. So there's another sort of multiple, the buffered multiple, that fixes this. Buffered multiples are actually active devices; they take an incoming signal, regenerate it so that all of the buffered outputs are identical, and increase the overall strength of the signal so that a single control voltage can remain identical at each CV input. In theory, you can drive 3-4 or a couple more VCOs off of each buffered output, whereas if you tried this with a more typical passive multiple, you'd start running into intonation problems after hooking several VCO pitch inputs up to a passive mult. So, for larger, more oscillator-dense builds, using a buffered multiple to split up a basic VCO pitch control voltage is essential. But on smaller systems, not so much, and a regular ol' passive multiple works fine to split signals with only a slight loss due to loading.
Sooooo...up in that spot I mentioned, you put the appropriate mult, and then you also put slew generators right by this, since their function is to modulate the oscillator control voltage in various an interesting ways. A lot of the time, these function as basic portamento (or 'glide') modifiers which can slide control voltages between discrete voltage level steps. But there are lots of variations, such as slew gens that can vary their behavior depending on whether voltage levels are ascending or descending, slew limiters that allow you to voltage control their behavior across intervals of time, and then there's the special case of variable slope generators. Now, these arcane thingys are definitely weird, and definitely neat. Technically, yes, you can use these as basic slew generators. But they actually fit more into the realm of function generators, devices that either continually or upon triggering output certain user-determined voltage curves (or, as they're termed in mathematics, functions, as the behavior of a function generator actually depends on mathematically-determined processes imposed upon a static voltage source). These things kick ass! You can use them as two-stage envelope generators, user-defined-waveform LFOs, even as oscillators...and the waveforms can either be altered by changing their rise and fall times manually or...get this...with OTHER modulators. Nuts! They originally appeared in Serge Tcherepnin's famous West Coast design, the self-named Serge synthesizer, but have seen an ever-increasing adaptation to Eurorack application as well. Why would you want some of these? Hell...why WOULDN'T you want some of these! In fact, many savvy builders make absolutely sure that space is available for them, with the Make Noise MATHS being the most-employed version of these modules. So, not just for portamento anymore!
Alongside those, LFOs make the most sense, particularly if you want to push some into the audio range where you might want some CV scaling that gets patched in tandem with the audio oscillators. Also note that you can send that same regenerated, scaled voltage to a lot of places where pitch (or a specific voltage level for some other use) is important, such as filter cutoff frequencies. Below or beside your LFOs should go envelope generators, and these can be either looping (which actually makes them more akin to function generators, especially if you can voltage control their envelope parameters) or one-shot ones that only work when triggered by a gate or trigger signal. You never want just one of these devices. By having a few different envelopes, all triggered simultaneously but set differently, you can create sonic behavior that changes across an entire sonic event, by sending one envelope to your VCAs for your oscillator mixing, another to a filter's envelope cutoff frequency, a third to that filter's resonance or Q, and maybe a fourth to reshape the final signal going into the output mixer. BTW, these devices are called 'envelope generators' because a very old term for the amplitude contour of a sound event is that event's 'envelope'. But like other things in a modular synth, you don't have to use them just for shaping that amplitude event. Always consider other purposes for things; your results may amaze you!
Back across the rack again, to the right of all of these modulators and under/by the VCO/waveshaping section, we're going to want some filters. Yes, even in a West Coast setup, because the subtractive processing that filters do have lots of uses in timbrally altering those complex modulated audio signals there. And speaking of West Coast, the most ubiquitous 'filter' you find in those systems is the aforementioned low pass gate.
Low pass gates are actually a combination of devices: a VCA and a low-pass filter. They work in tandem, as a rule, and almost always see use on audio signals. The 'classic' low pass gate also makes use of one other arcane element: an optical isolator element. Now, this thing as originally designed by Don Buchla used that optoisolator to keep control-voltage-level signals in his synthesizers separated from direct connection to the audio chain, since the control and audio systems in Buchlas are at different voltage levels and voltage leakage from one could cause harm to the other. But the neat thing about the optoisolated low pass gate is that the optoisolator doesn't 'perfectly' control things. Instead, there's a certain aspect to the decay, caused by the dimming of the light source as the control voltage drops off (instead of simply stopping cold, which would happen if a direct connection occurred here) that gives the LPG a certain character that is sort of unique. Now, since those days, there's also been the development of LPGs in which the control voltage DOES directly control the VCA and low-pass filter in the same way as the optocoupled signal, but these tend to sound smoother and, as a result, don't have that 'woody plunk' (as it's been described more than once) of the original opto versions.
But those are weird (and fun!) and rather unlike the vast majority of filters you find in synthesizers, which are designed for a dual purpose: 1) they remove parts of the frequency domain of the raw audio signal coming out of an oscillator (or several) to reshape the signal's timbre, or tone color, and 2) they accentuate other parts, usually around what's referred to as the filter's 'cutoff frequency' by using a feedback circuit to emulate physical resonance of a cavity or other sound-producing object. The amount of this control is either referred to as, simply, 'resonance' but you also see it called 'Q', which is an engineering abbreviation for that. While there's a lot of other, more exotic resonant filters, the vast majority come in four distinct flavors:
1) Low-pass filters. These are the 'bread and butter' of subtractive synthesis. They remove signals above their cutoff frequency, and when resonant, accentuate frequencies just below the cutoff frequency. When you think of what a basic synthesizer sound is like, I can pretty much guarantee you that you'll be thinking of a sound processed through a low-pass VCF.
2) High-pass filters. Take a low-pass filter and turn it on its head. Same idea, except now all of the frequencies below the cutoff go away and ones just above the cutoff get accentuated. If you tandem these, you get either...
3) Bandpass filters. Low and highpass in series give us the classic bandpass response. In this case, the 'cutoff' is better termed the 'center' frequency, and changing the resonance of this filter tightens the span of frequencies being passes while peaking the ones right by the center frequency. Or...
4) Band-reject (or 'notch') filters. Pretty much the opposite of the above, which you get with low and highpass filters in parallel. The center frequency is then where everything drops, and the more you increase the resonance, the more that frequencies around the center frequency disappear. However, this can be kind of weird, because the span narrows as you increase Q and more signals around the center start to appear instead of vanish.
Multimode filters are simply a couple of low and highpass filters behind the same panel and some sort of switching and/or mixing capability to combine their responses to create bandpass and/or band-reject behavior. And that's pretty much it for your basic resonant filters.
But modular synthesizers often make use of NON-resonant filters. These often take the form of filterbanks, a paralleled array of fixed-frequency cut-only filters that allow the user to manually shape the timbral curve of a signal by altering the levels of these different filters, sometimes by changing an input level to the individual filters, or changing the respective output level. A few even have 'break-outs', which channel the fixed bandpass filter to individual outputs; this arrangement is necessary in a vocoder, where one filterbank would divide up a 'control' signal (such as a voice) into multiple discrete bands, sending this to envelope followers which then control the bandpass levels of yet another filterbank dividing up the frequency domains of a 'carrier' signal. A 'graphic equalizer' is another form of filterbank, although these can also (non-resonantly) boost signals as a rule as well as cutting the levels of their discrete bands. Using a filterbank as one typically sees in a modular synthesizer is akin to using it as a cut-only graphic equalizer.
Another (usually) non-resonant filter typically seen is the 'formant' filter, a set of passband filters which have rather wide passbands that center on typical harmonics found in vocal sounds. This was a key device in an early electronic instrument called the 'Trautonium' and was used to filter the audio generated by a florescent-tube relaxation oscillator (exotic, hm?) that was very harmonic-rich, in order to create different timbral combinations. The Trautonium, invented by Friedrich Trautwein, actually saw a good bit of use in compositions for the instrument by Paul Hindemith and Ernst Krenek and others, particularly in the 1930s. However, the formant filter also comes in resonant varieties...and these are weeeeeird! You can sort of imagine how this works in a signal path: consider how so-called 'overtone singing' works, where an upper partial of a vocal sound is pushed via various methods by a singer into sounding out clearly, as if the singer was producing a second note, or potentially even a third! There was a very famous analog monosynth some years back, now quite rare, called the Synton Syrinx that was based around this very sort of resonant formant filter, and its qualities were very prized among synthesists because of how its resonant formant filter could create a seemingly-endless variety of timbres and textures. Sort of a Trautonium on 'roids!
Technically, a phase shifter is sort of a filter device as well. In its case, a very odd creature called an 'all-pass' filter is slowly modulated and, sometimes, very slightly time-shifted against the original input signal to create the classic 'sweeping' timbral changes we know as 'phasing'. Note, though, that this is actually NOT like the original 'phase-shifting' technique one began to see in the 1960s. It sounds sort of like it, but it's a 'cheap' method to recreate the real thing, which was actually caused by taking two identical tape recordings and slightly 'dragging' one out of 'phase' with the other by physically slowing one or the other tape machine with a finger against the supply reel flange. Hence the real time-domain effect's name: 'flanging', which is actually capable of more extreme sonic transformations than a typical phase shifter. This doesn't mean, though, that phase shifters are just some 'cheap copy' effect device; because of how they work, especially the modular variety, you can use a modulation signal very different from the typical sine or triangle wave LFO in a phaser stompbox. The results can be quite interesting, especially if some very discontinuous jumps in the modulation signal get fed into the phase shifter's mod input.
One filter that's actually not a filter, by the way, is the sort-of-not-well-named 'comb filter'. Don Buchla's version aside (his is just a misnamed filterbank, actually), this is actually better described as a 'tuned delay line' which rings at a specific frequency and its overtone nodes when energized by an incoming signal. It gets the 'filter' confusion because the ringing behavior is more accentuated when sounds run through it have a wavelength similar to the harmonic resonant cavity the delay is set to emulate. But in reality, the 'comb filter' actually belongs more to the wonderful and weird world of physical modeling, and that gets complicated. If you want to know more on what to do with tuned delays, look into a few articles on Karplus-Strong synthesis, which has these devices as the core of a whole different method of sound synthesis.
...anyway, there's all sorts of other weird, fantastic, freaky variations on filters in the modular synthesizer world. But let's move on to randomness. Literally.
Random signals are very necessary in pretty much ANY type of synthesis. At their most basic, noise, they form the basis of a vast array of percussion sounds in the natural sound world and the electronic one as well. But there's more to noise than just noise. The term actually applies to any signal that fluctuates randomly; what we call 'white noise' is actually an incredibly-fast voltage fluctuation that can appear at any given frequency at the same level at any time. It sounds the way it does because our hearing processes experience it in that way. White noise is far from a single sound, however, and by using a ubiquitous module called a 'sample and hold', we can see this noise has all sorts of too-fast-to-perceive discrete states.
Sample and hold circuits are part of a huge gamut of devices (as are noise generators) which fall into the class of 'random signal' modules. S&Hs are able to slice up random signals, such as different varieties of noise, when they receive a trigger pulse. When that happens, a 'voltage buffer' then locks into the voltage level that was present at the S&H's input at the instant the trigger was received. And lo and behold, you have a discrete voltage taken right out of the noise's instant state at the point in time the S&H was triggered! Sample and holds, however, are sort of the interface between the 'determinate' parts of the synthesizer (audio signals, etc) and the 'indeterminate' world of signals which we don't parse as something specific in frequency, timbre, and so on. You can control filters, oscillators, and so on with the voltages sample and holds derive from all sorts of sources; a very nifty trick is to use an LFO's fluctuating voltage as a sample source with a trigger pulsed at a rate that is faster than the LFO's frequency, then controlling a VCO with that while sending the trigger pulse on to an envelope generator, etc etc. It's neat and spacey. Try it sometime if you haven't before.
But other randomnesses exist beside noise. There are all sorts of 'chaotic' circuit behaviors that have been harnessed in synthesizers, ranging from pretty straightforward (cascaded and/or paralleled LFOs) to very exotic (modules that use complex equations which result in semi-deterministic outputs, such as Euclidian sequencers, Sources of Uncertainty, etc). But you really want your noise devices in the vicinity of these critters to make it easy to sub-patch all sorts of chaotic activity and send that all over the place, or just as well, have randomness devices that incorporate their necessary noise circuits.
Right around this area as well, which I like to put below the modulators and near the left-bottom/front tiers, you also want a lot of devices that either generate or alter clocked or unclocked gate/trigger pulses. This dovetails well with the chaotic section, strangely, because a number of the 'chaotic' modules need the very 'deterministic' modules of this other sort to do what they do. It also puts them in a nice place where they can work with a sequencer or two, either onboard or outboard, and places them in reach of your main controller for easily altering time-based activity on the fly. Lots of things can go in here, ranging from comparators which can 'read' fluctuating voltages in various ways to output gates or triggers, quantizers which constrain fluctuating voltages to specific levels in step (or not) with a clocked device, logic arrays that combine and alter pulsed signal patterns to derive all sorts of rhythmic mayhem, and various clock modulators that divide, multiply, alter the patterns of, and count clock pulses for loads of different uses. Those of you reading this who like the notion of 'generative music' will want to take a rather close look at these types of modules, as they're key to creating a constant cascade of shifting, varying rhythmic structures depending on how they're patched up and what's present there for that patching.
To the right, you'd like to see your effect processors. But between all of this pulsing craziness and those aural scramblers, you're going to want to try and shoehorn in a few linear, DC-coupled VCAs (for working with control voltage levels) as well as some other boring devices, namely attenuators (or attenuverters, which allow you to polarize or invert control voltage signals in addition to changing their levels). These seem pointless, but the fact is that you can create a lot of control and sonic nuance by having these simple things there where they can branch their processed signals back up into the morass of patchcords forming above them as you fiddle with them. Never neglect things because they might seem 'boring', as they're only as boring as their user's imagination when within a modular synthesizer.
OK, effects. This one's up to you. A lot of what goes into this area depends on the sort of sound the music you create demands. DO make sure, however, that your effects processors are capable of interfacing with the various modulation sources you've put in. Don't just get a 'thing with a knob' and expect it to yield the same amazing variety as a similar 'thing', but one which has a few CV inputs that allow it to merge in with the activity in your patch. Best placement for all of this is above or beside the final mixer and below or after the filters; if you think about 'why there?' for a minute, you should realize that there's a pattern to all of this. It's NOT consisting of 'toss crap into case, hope for the best, then get really confused and angry when realizing how dumb that tactic was' as a build method.
The key to understanding how you're seeing me step through these different sections has to do with 'signal flow'. Anything that flows needs to flow as simply and directly as possible. Water in pipes doesn't deal very well with all sorts of convolutions and diameter changes to the pipe it's in. A story has a distinct start, middle, and end, and deviating from how that should work gives you an incoherent mess. And modular synthesizers...since they are musical instruments, after all...need to be thought out as rationally and as carefully developed as any other instrument. You wouldn't put the bridge on a guitar about two inches to the right of where the strings are strung, right? It has a place.
Likewise, in almost all cases when I design a build, I always follow a single rule of thumb as to how signals flow in a modular synthesizer. This pattern is what I call “up-left/down-right”. Control signals are derived and modified along the LEFT side, since right-handed people (like myself) tend to use our right hands in a more precise manner, and a little 'slop' is OK sometimes in working with control generators, modifiers, and modulators. But more precise and frequent actions, such as tweaking filters, adjusting mixers to achieve timbral/tonal balances, carefully tuning VCOs, and the like are best done with ones' dominant hand. So if you're a lefty, just flip this pattern around the other way, works just fine.
So, control signals mostly flow UPWARD on the left, increasing in automnity until the 'voicing' section is reached. At that point, control signals change to audio signals, and then flow mostly DOWNWARD on the right, from 'voicing' thru 'filtering' then 'fx' and to, yep, the output. Sure, there's also going to be some criss-crossing of the patch panel, but with everything grouped in this way, it actually makes it very easy to track the source and destination of these patchcords. These are also, quite often, sources and destinations you're going to want to pay more attention to while playing that patch configuration, so the fact that the 'sideways' patchcords act sort of like arrows pointing at these makes things even more intuitively interactive. And this flow method applies equally well for smaller builds as it does for huge studio systems.
And that brings us to the lower-right, where our output stage is. Remember how the effects are just above this? This makes it easy, if the system's mixer has an AUX send/return capability, to drop one or more of the effects processors into that loop, and just as easy to tweak it in conjunction with using the final mixer. Now, also in this area, you might want some submixers which, again, incorporate VCAs to control signal levels. But in this area specifically, ALWAYS use AC-coupled exponential VCAs. There are important reasons for this...
First of all, our hearing apparatus (ears and brain) responds to what's known as 'apparent loudness' as an exponential function. Not linear. As things increase in decibel level, they are actually increasing in kinetic energy in a 'law of squares' manner, and not in a straight line. So if you want something to seem as if it's increasing smoothly in volume, you're actually not increasing the level of the sound energy in a straight line, but more in something akin to an asymptotic curve. So to get actual instrument-like contours of volume imposed on our final signals, the last VCAs they go through have to respond to their control signals with exponential changes in dynamics.
The other critical reason for using this particular variety of VCA has to do with a nasty, lurky issue that crops up quite a bit: DC offset. Remember the bit a ways back about Don Buchla and his optoisolated low pass gates? This factors into that, and was part of what Don avoided with that curious bit of circuitry. DC offset is what happens when a DC (or subsonic) voltage gets imposed onto an audio signal. It's sort of like the desirable 'offset voltage' where a certain DC voltage gets added to another existing one to transpose, etc something like a filter cutoff or VCO pitch. But in this case, you absolutely do not, ever, EVER want to run a DC offset out with your audio and into any sort of amplification. The results are ugly. At the very least, you'll introduce a lot of undesired phasing issues into what should be a nice, clean output signal, which is especially nasty if your modular outputs in stereo and the DC offsets aren't the same in each channel. At the very worst, you can literally destroy speakers, amplifiers, and probably your synth out of your own reaction to seeing smoke coming out of your speaker cabs and bad charry smells emanating from the amp. Blame the messenger and take it out on the modular, right? Well...no. It's your dumbass mistake for not isolating the DC out of the outputs by using the right VCA in the right place! So...use the right device in the right place. That's why there's AC-coupled VCAs.
(And, oh, sure, I know that lots of synths of the past didn't use AC-coupled outputs. The ARP 2600 comes to mind. But you learned to keep that DC 'thump' OUT of the outputs...sometimes, the hard way, although I never got to see anything go kaboom due to my own errors here.)
Then, your mixer. Do get something appropriate to the modular's mission. If this is a smaller rig, mono output is just fine since you'll likely be doing simpler, single patches, one at a time, and this is another great place for voltage-controlled mixers. Make sure to have an output stage, though, since synthesizers have operating voltage levels that are generally considerably higher than typical line levels that a mixer, amp input, etc would like to see. Note that 'generally', though; some things actually have sufficient headroom to handle a raw synth-level audio signal. The thing to keep an eye out for in this case are pieces of equipment that have a headroom that's quite high, either capable of handling +20 over 0 db or better, or which have some sort of input attenuation that can back the device's level sensitivity down to a point where it doesn't distort when presented with a synth-level signal. Be careful, however, as not all things that appear to attenuate the signal at input are actually doing that, and it's very possible to fry sensitive input-stage components with a typical +/- 8V AC signal directly out of a synthesizer mixer that has no output stage.
If you're going to be doing more elaborate work, though, stereo mixers are the better choice, because...well, let's say you're doing a live gig in which there's four different pieces that use four distinctly different sounds out of your modular, and each uses processing, etc that definitely results in a stereo image. The thing to do (and the reason some of these modular systems are SO FRICKIN' HUGE) is to patch each 'instrument' up on a separate channel of your stereo mixer, and then fade between these as needed. This sort of thing is also where submixers come in handy, so you can do your final submix on a separate module, then perhaps use a panner (or not) before the stereo mixer module's input. Also, the stereo performance mixers that are becoming more prevalent often offer VCA control over levels, stereo panning, and even AUX sends and returns, so having something that does more for you, rather than you doing a lot with it...well, that's always a 'win'.
Lastly, perhaps the most boring of all the boring stuff that can bore you with a modular synthesizer: power.
Having figured out what you want in this box after what's likely to be months (or more) of twiddling, searching, head-scratching, conniption fits, seizures, and your doctor asking you if you've been to regions where 'certain fungal infections' are common, and you're ready to toss down that plastic and make it SCREEEEEEEEAAAAAM...well, hold on. Behind all of these panels and cords and lights and knobs and jacks and so on is something very critical that also doesn't get a lot of thought, and that's your power system. It won't work at all without that, right? You'd feel pretty derpy and defective if, when it came time to turn the synth on, there was no...thing...which turned it on. Duh.
Modular synthesizers have, as a rule, three parts to their power systems. The first, obviously, is a power supply. Now, it is extremely possible, if you weren't paying attention to the cumulative current draw of all of these modules, to wind up putting a power supply into (or connecting it to) your modular synthesizer that is underspecced in amperage. And that won't be good. Quite often, too much draw on a power supply that's not rated for high enough amperage results in power supply go BOOM...and sometimes, parts of your synthesizer as well! So, keeping track of accurate current draws as they add up is critical, and will determine the final cost and current capacity of your supply. While there's probably going to be some readers who think this is excessive, I like to have a power supply for something of this sort that can supply at least 1/3rd more amperage than the system presumably requires. On power-up, some devices can actually draw more current than the rating that they have (tube devices are notorious for this, as was Roland's 'Aria' module series), and if sufficient current headroom isn't present, again, boom, etc. Also, a less-taxed power supply is a happy power supply, because it's a cooler power supply, and the thing that causes more device failures over time in electronic components is sustained component heating. So, if you get a result of 3000 mA as a total draw on your +12 rail, go with a power supply that can deliver 4 ½ amps to that rail.
One other thing about power supplies is that there's two different types. One is the linear variety; these tend to be extremely stable, generate no noise onto the DC outputs, but the tradeoff there is that they're rather expensive and definitely a lot heavier. For a portable rig, you probably wouldn't be happy with lugging a modular equipped with a linear power supply, although it would sound awesome. Then the other, and vast majority, are of the switching variety. These are cheap, lightweight, but produce their DC through extremely high frequency transformation methods, and so some of them can actually impart noise onto your DC rails if they're shoddily-made. There are methods of filtering this HF crud, however, ranging from filtering schemes on distro boards by several manufacturers to using ferrites to block high frequency AC garbage on your DC lines. But the best rule of thumb there is to obtain power supplies that are well-made and especially those that are recommended and/or used by module or system manufacturers, as these firms really don't want to deal with the problems that substandard power supplies can cause.
Part two of this important system is distribution. No, not the distribution boards...the wiring and power inlets, especially between the power supply itself and the distros. If you're thinking of doing this yourself, make sure to check amperage ratings for wire gauges, and whatever these say, go one gauge bigger. Again, the issue partly lies with keeping the power supply happy, because a less-resistive load before the distro boards will make for a more stable supply. But also, some of the issue lies with the wiring itself, in that not allowing for ample current flow will lead to both voltage drop to the distros as well as heating (and possibly damage) of the DC wiring itself. Again, when in doubt, always overspec. Power inlets, also, should offer some way of letting you know that not only is the system switched on, but that the power rails themselves are properly supplied. Indicators for both 12 volt rail polarities should be present if the panel size allows for it, and also +5 volts if you're making use of numerous modules that require it.
And last, the distribution boards themselves. Get filtering. I can't state this more clearly; filtering is essential. The sad fact is that lots of modules send signal leakages of all sorts back down their DC lines, and this can get into other modules if there's not any filtering present in there to get this under control. This undesirable signal leakage can do lots of irritating things, creep out into your patches, destabilize modules unexpectedly, and...just trust me, get filtered distros. You'll be a lot happier than if you didn't. As for the whole keyed/unkeyed, 16-pin versus 10-pin, and other arguments...meh. These are matters of taste, convenience, and what different manufacturers demand. Suffice to say, as far as the 16-versus-10 pin thing goes, just go with 16-pin distros, as there's plenty of 16-to-10 pin ribbon cables to deal with running across modules that insist on 10-pin connection. Keyed and shrouded versus unkeyed/unshrouded...just exercise way more caution that you think you should about connection position and rererereREread manufacturers' instructions on how their gear likes to be connected, as there is still some variation out there (but why? Shouldn't this all have been sorted out long ago?).
Then stick it in a real box. Not the virtual one, or a bunch of grid lines. A physical, actual thing.
Now, this is where things get strange. You're all ready to slap this monster together and then OHMYGAWD just look at how expensive these cases are! And yeah, they do get spendy, although in the past couple of years there's been some movement toward cases that cost less and which are still substantial. But even with that, if you want to put together a modular system of any size...large, small, or in-between...there's definitely a broad range of prices in all of these capacities. And while some prices, to me at least, are just outright damned unreasonable, you have to take some things into account while rushing for the bathroom to deal with that nosebleed that case-shopping has caused.
First of all, consider what the rest of the synth costs. Now, would you like to put that in some flimsy piece of crap? Probably not. And good-quality hardware simply costs real money. Manufacturers who build heavy-duty hardware such as pro-grade Pelican cases into Eurorack housings might charge a chunk of change, but you know exactly what you're getting there: security and smash-resistance. On the other hand, someone dropping rails into a vintage thrift-store suitcase might be making something that looks trendy, but...ahh, would you, say, stand on it? Pile a bunch of other heavy crap on top of it? I think I know the answer to that...and I KNOW I know the answer to what happens as a rule if you did!
Also, keep in mind that, at some point or another, you're going to want to take that big-ass cab out of the studio and stick it in front of an audience. Now, do you have health insurance for the hernia surgery? Bigger systems take some care in selection, and quite often it is better to build larger modular synthesizers in several smaller cases so that you can disassemble the whole mess and pack it up in a more convenient space for transport. Should you do this, I note, you'll likely wind up with separate power systems in each case, and so you'll want to calculate your current draw PER CASE in addition to figuring out what the entire setup should be. A very few large cabs are fairly transportable (ADDAC's Monster series, for example), but the basic rule is that if you go big, also go 'broken up', unless you never, ever intend on taking the instrument out from where you install it.
Ultimately, the watchword for case selection is 'practicality'. It shouldn't be all that difficult, given all of the case styles, manufacturers, and configurations out there these days, to figure out which case works in your...uh, case. Sure, some designs might look snazzy, but how will they look in five years time and usage? Or ten? Or longer, because modular synth cases are not the sort of thing that most users turn around and pitch as if some fashion trend dictated it. Are you OK with the rails? The hardware? The way your hands move over the patch panel's plane? Does it 'feel playable'? Again, be practical. After all of the other effort, you definitely can afford to!
And this brings up one more case-but-module point: the current (as of this writing) 'tile wars'. I like tiles...the little 1U minimodules that can add a lot of extra functionality to a system. They've gotten pretty capable, too; some tiles are the equivalent of basic, standard-sized Eurorack modules at this point, and I think that's great. It allows you more versatility and opens up more room in 3U-land. If you can fit 'em in, by all means do so. They're great. But not if tile A can't fit in case B. And right now, this is the situation as there are two different tile size standards, and while one size can be (sort of) made to fit the other, the reverse is not true. I don't think, frankly, that this is tenable. Consider the whole situation between Analogue Systems' 'Eurorack' and the actual form factor cooked up by Dieter Doepfer from the industrial standard. True, some manufacturers make panels which have the oval holes needed to fit either the AS racks or the standard-conforming ones. But not all. And putting AS modules into regular Eurorack cabs which have fixed fastener positions just...doesn't...work; the only way this works is with sliding nuts, and that gets into a whole 'nother debate altogether. So I'm going to take a second in this paragraph to get up on my soapbox and say to the firm doing this schismic thing, “Guys...please think this over again.” Everyone making tiles, irrespective of form factor, is doing awesome work, and this being Eurorack we should be able to put these modules anywhere, and not be forced into taking some proprietal side. After all, how big of a player IS Analogue Systems and their not-exactly-on-center module standard these days, hmm? Hmmmmmm???
All that aside, and as for the rest, once you've taken the plunge and either mortgaged your firstborn for that 21U system, or just cobbled up a little box of toys...it's just a matter of getting used to this strange device you've allowed into your home and life. Takes time, diligence, patience, and a good ear...but the rewards are worth it, and the whole thing is one of those matters that's more about the journey than the destination. Remember to have fun with the process, and then you're pretty assured that you'll have fun with the resulting modular synthesizer you've created.
GOOD LUCK!
DAC Crowell (aka Lugia @ Modulargrid)
http://daccrowell.bandcamp.com