Modular synthesis intro, part 3: How to get started
This is Part 3 in a series that started with Part 1.
I will assume that you've already read Part 1, which describes what a modular synthesizer is; and part 2, which says that you shouldn't buy one. Your spouse isn't allowed to kill me because I did warn you, and your cat has already formed an opinion on your intelligence anyway. In this third installment I'm going to offer some suggestions on how to get started with modular synthesis. Very much depends on your own reasons, which are ultimately known only to you, for why you've decided to do this. I can only offer some insights that may be helpful for what I think are typical newcomers.
This isn't the end of the "intro" series. When I first posted the articles on which these were based, in my personal Web log, I had a lengthy section in the third and final part on pitfalls for the unwary. Since that grew, and I found some other stuff to include after finishing the series, I'm going to split it off into at least one additional part.
I'm writing this assuming the Eurorack modular format, which is really the only one I'd recommend to a newcomer today. By "format" I mean the standardized physical dimensions of the modules, their electrical connections, sometimes conventions on panel layout and visual design, and so on. Devices built for a given format should all work together. Other formats beyond Eurorack exist, including the formats of the original Moog and Buchla systems from the 1960s, but Eurorack is by far the most popular one today. It has the widest range of modules available; because of its popularity, prices are sometimes lower and it has a stronger community; and it's physically compact, making the equipment more portable, all compared to other formats. Eurorack originated with the Doepfer A-100 system in the 1990s; it is based on, but differs in some important ways from, the Eurocard standard used for some industrial electronics.
You should expect that getting started in this hobby is going to cost you between about $1000 and $2000. Those are US dollars - I'm in Canada but I think readers of this will be all over the world, and probably more familiar with the conversion from their local currency to US dollars than to Canadian. There are some ways to economize, which will be described in this and subsequent articles, but using those techniques will help you keep the price tag in the $1000 to $2000 range, not bring the low end any further down. Don't expect that because I said $1000 you're going to be able to pinch some more pennies or save money by building your own or buying used, and get an almost-as-good system for $500 instead; my estimate is assuming you are already doing those things to the extent they're possible, and if you have $500 for synthesis gear then you should be looking at something other than a modular. In the other direction, if you make modular synthesis a big part of your life you may well end up wanting to spend much more money in the future. But it's also possible to buy a starter system and just keep using it indefinitely without spending more money.
Modular synthesis is a social hobby; a big part of the experience is connecting with others who are doing the same sort of thing you are, sharing ideas, getting into occasional flame wars, and so on. Even if you hate interacting with people, it's worth making the effort at least occasionally to exploit them as information sources, sell your old unwanted modules, buy fresh modules cheap, and so on.
There are a number of Web fora for synthesizer users, include Electro-Music, the Synth-DIY mailing list, and the Mutable Instruments online forum. There are often in-person events for modular synth users in large cities. Lurking on Web fora is probably the best way to find out about these.
Modular Grid is an online "planning" tool for modular synthesis. You can drag and drop pictures of modules onto a picture of a rack and build up an imaginary synthesizer. In principle this is useful for deciding what modules you want to actually buy, or for deciding how to lay out your collection of modules in your case, but really, I think most people just use it as fantasy material, building on the screen a synthesizer they imagine owning. Anyway, it's coupled to a database of available modules, with remarkably good data quality considering that it's maintained on the Wiki principle, so it's a good resource for looking up what modules are available in a given category and so on. The rack planner also includes automatic estimates of the total price and power requirements for a setup.
When you're considering buying a module, you should look for "demo" videos of it on YouTube. For instance, here's one of the North Coast Synthesis MSK 010. People make such videos for nearly all modules, and they can give you a good idea of what the module sounds like and how it's used. It's a bit unfortunate that YouTube has become the standard venue for such things, because its sound quality is atrocious even in the best case where people recorded the signal directly from the synthesizer; even worse when they used the microphone built into a Webcam. But that's the world we've created for ourselves.
Some wigglers claim to have a lot of fun with their synthesizers just by plugging in cables at random without knowing anything about how the sounds are made. I think a disciplined approach of actually knowing what you're doing is likely to give better results. For that, you're going to have to learn a bit about sound and how synthesizers in general work. You can pick up some ideas just by talking to other modular users, but there are also plenty of things to read and play with.
The Synth Secrets articles from Sound on Sound magazine are a good place to start. This is a 63-part series covering a wide range of synthesis techniques. Early parts claim it'll just be about "subtractive" synthesis, but in fact the series also went to many other places in its five-year run. If you read through all of that, you'll probably be ahead of most of your peers in terms of technical knowledge.
It's worth getting Csound or another software modular synth and playing with it. I talked about these in more detail last time. They're a good way to try out different synthesis techniques, and they have some unique advantages over hardware. Using one for a while before you start buying modules will help you get a clearer idea of what modules you want.
The user manual for every North Coast Synthesis Ltd. module (linked from the shop pages for the modules, and all together on the DIY page) includes patch suggestions and technical discussion of the circuitry. Even if you don't buy the module, you may want to download and read the manual as a source of ideas. Some other module makers also offer patch suggestions.
If your main focus is on making music, then you may be better served by some other instrument than a modular synthesizer. Modular synthesizers appeal most to people who want to spend a lot of time tinkering and building up a customized system from individual pieces, possibly to the exclusion of actually making music with it.
Nonetheless, if you want to skip the part that most modular users find fun and just get started with a usable system with a minimum of fuss, it is possible to buy a complete modular system as a package deal and not have to think about it further. You may get a slightly better price this way compared to buying the pieces one at a time; and you'll certainly get up and running making music faster.
Doepfer makes a couple of different standardized complete systems, with prices ranging from around US$1200 to US$2400. Any of those would be quite reasonable ways to start off with a complete usable system. I favour the "Basic System 2" (which comes with a USB and MIDI interface, so you can drive it easily from a computer or keyboard synth) and the "Mini-System," which is a well-balanced small system. Any system built completely from Doepfer modules like these will probably be influenced by the Doepfer philosophy of splitting the synthesizer into many small, simple functions. This is more or less what people mean by "East Coast" synthesis.
If you buy one of the Doepfer ready-made systems, then for an extra charge you can have them install it in a bigger case, with an empty row for adding other modules in the future. It's a good idea to do that. Otherwise, you'll be facing another case purchase as soon as you want to expand.
MakeNoise offers the Shared System, a packaged collection of their own modules that form a complete instrument for US$4500. These are bigger, more complicated modules, and fewer of them, partaking more of the "West Coast" synthesis tradition. Most are also digital, with microcontrollers at their cores, should that make a difference to you. The Shared System also includes a small amount of empty rack space, which you can use for adding one or two extra modules of your choice (maybe from some other manufacturer).
The MakeNoise panel graphics are polarizing. Some people love the look, and some hate it. I'm in the latter camp. However, I must admit that when it's an entire system those graphics look a lot better, especially in the black-and-gold edition. My visual objection is to a single MakeNoise module in the middle of a system built out of other manufacturers' modules, where it tends to clash. It's easy to argue both sides of that question. One side: if you're making music, shouldn't you care more about how it sounds than how it looks? The other side: if you're spending thousands of dollars on a thing for your own amusement, why should you buy it if you don't love it?
If you get a ready-made system, you're basically done. The rest of this article is about the more complicated, and more modular, approach of building up a system from parts obtained separately.
Eurorack modules require ±12V DC power and sometimes +5V as well, so you need a source of such power. Most people buy power supplies designed specifically for Eurorack, which will usually come with either a rigid "bus board" that has male header connectors on it, or a "flying bus," which is a ribbon cable with such connectors attached directly to the cable. Then each module comes with a female-female cable of its own for connecting the module to the power bus.
Bus boards are preferable. Flying busses introduce enough resistance in the power distribution that they can cause crosstalk between modules. However, that's not a dealbreaker; you can still get very good results with a flying bus, especially in a small system. They're just maybe not the absolute best. Flying busses are often a fair bit cheaper, which is why they remain popular. I'd say, go ahead and get one if that's what comes with the power supply you want to use, but all other things being equal, it's nice to have a bus board.
There are a few different ways to get the power onto the bus. Some cases (see next section) come with built-in power supplies, which might plug with a cord directly into the wall (like most of the Doepfer cases), or have a low-voltage input supplied by a "brick" or "wall wart." The outboard component in turn might be a switching supply producing DC at a relatively high voltage to be regulated down by components on the bus board, or it could be AC at less than mains voltage, with the AC-to-DC conversion done inside the case. The connection to whatever cord there is, could be on the side of the case or through a panel that mounts like a module into regular rack space (like the 4ms Row Power or Tiptop μZeus).
Part of the reason for separating part of the power supply into an outboard device is that in most places, anything that plugs into the wall is required to go through an expensive "certification" process of safety testing before it can legally be sold. Manufacturers of these power systems make an argument (which may or may not be legally correct) that if the system consists of a wall-wart which plugs into a circuit board with a low-voltage connection between the two, then only the wall-wart has to be certified (ideally it can be a commercial product bought from some other manufacturer who already paid for the certification) and the circuit board doesn't need to be certified. Another reason for the two-part design is that the voltages and plug shapes needed for dealing with mains electricity are different in different parts of the world, so with some care it may be possible to design the system so the circuit board is usable everywhere and you just swap in different versions of the outboard transformer according to the local voltage.
Some people make a big deal of wanting "linear" power supplies for audio applications and turning up their noses at "switching" power supplies. People who say this usually are unable to explain what those terms actually mean. The difference has to do with how the voltage regulators work - linear regulators continuously intercept part of the current flow, dissipating it as heat, whereas switching regulators turn the power fully on and off for different periods of time, causing less to be lost in the switch itself but then requiring other components downstream to smooth out the flow. The different kinds of regulators are discussed in more detail in my green modular series. Linears are inefficient and less able to handle variable input voltages, so by audiophile logic they must be better. My perspective is that a properly designed present-day switching regulator is perfectly acceptable in audio applications, and a badly designed regulator of any kind will cause trouble, so you should be looking for quality rather than looking for the word "linear." Many present-day designs actually use both, with a switching regulator in an outboard box feeding power into a linear regulator built into the case. This has the advantage that if you like the product you can say it's "linear" and if you don't like it you can say it's "switching" and either way you'll be right!
There's similar fear, uncertainty, and doubt going around regarding the idea of using a "wall-wart" or outboard power converter at all. In Eurorack, it's not realistically possible to avoid them. Advice to never ever buy a wall-wart is, for most of us, advice to never use Eurorack. Very few manufacturers offer cases with built in power supplies that plug directly into the wall without an intermediate device (see above for the reasons why not), and even if you can find or build such a case, there is no automatic electrical advantage to putting the mains-to-low-voltage conversion inside the box. On the contrary, having a large transformer inside the same box as your modules may increase the risk of introducing mains hum. All other things being equal it may be better to have it outside - but either way you still need to know it has the right power-handling and regulation capabilities. Discussion of outboard or built-into-the-box power supplies often becomes a proxy for the "linear" versus "switching" discussion, with people who don't really know what they're talking about overstating genuine electrical engineering issues that they heard as rumours and didn't fully understand.
Occasionally someone floats the idea of trying to use a PC power supply to power a modular synthesizer. That is usually not a good thing to do. PC power supplies are appealing because they are produced in huge quantities and therefore cheap, but they are not well-optimized for modular synth use. The average PC power supply is meant to supply massive quantities of power - often hundreds of watts - at a low voltage like 5V or even 3.3V. It will have +12V and -12V outputs too, but those will have much less current available, and the ±12V outputs may be poorly regulated. Those outputs are usually only meant for driving relatively undemanding mechanical things like fans and hard drive motors. On the other hand, a Eurorack modular synthesizer will usually draw most of its power on the ±12V busses; a few computerized modules may also want significant amounts of +5V, but not the huge amounts available from a PC power supply. And modular synthesizers are relatively sensitive to the cleanliness of the power, especially on the -12V bus because of the way op amps (a basic analog building block) are typically designed. So with a PC power supply hooked up to a modular synthesizer, the power supply's strengths are not being used, and its weaknesses are being sorely tested. The stuff you'll have to do to compensate for this situation is likely to cost you as much time and money as you wanted to save by using a PC supply.
Battery power is possible, but likely to be inconvenient. The voltages may be a problem. You need a total of 24V, which would be 16 standard dry cells at 1.5V each, and for anything but a very small system the current requirements will drain your batteries quickly. Synthesizer modules usually don't use huge amounts of power, but they're not designed with power economy in mind as let's say tablet computers today might be. Then you may also run into issues of controlling the voltage as the batteries drain, since a stack of 16 dry cells is really only approximately 24V when fully charged, and some modules may start to misbehave as the voltage decreases when the batteries drain. All in all, you're probably better off using some other kind of synthesizer for portable use (like maybe a keyboard synth designed to run on batteries); or else, let's say for an outdoor music festival, you could go the whole way with an AC generator and plug in your regular power supply.
Racks and cases
You're probably familiar with 19-inch equipment racks, sometimes called "telephone racks" or "server racks." There are two vertical rails, and equipment stacks vertically between them, screwed into the rails, usually consuming integer multiples of the "rack unit" (abbreviation U), 1.75 inches or 44.45mm. You might buy an industrial PC in a 1U or 2U case and install it in a co-location facility to use as a server.
For smaller pieces of equipment, you can have a "cage" that mounts inside the rack, with a couple of horizontal rails to which things mount in a horizontal stack. Sometimes there's a backplane built into the case, into which the smaller devices fit. Blade servers mount this way; so do 500-series modules used in pro audio. Eurorack synthesizers use a similar system. One row of Eurorack is 3U high; there are rails at the top and bottom of that space, and the modules screw into the rails. The standard rails used by Doepfer and most but not all other manufacturers have a lip at the top and bottom into which the module faceplates fit, so the module faceplates are a little less than 3U high. See the Doepfer mechanical details page for some pictures.
The horizontal width of each module is measured in units called HP, which I always want to read as "hit points" but it's actually meant to stand for "horizontal pitch"; 1HP is 5.08mm or 1/5 inche (exactly). Most but not all modules are an even number of HP. A standard 19-inch rack frame holds 84HP of modules.
But since Eurorack has become popular, it's become a thing that manufacturers have created cases specifically for Eurorack synthesizer modules with the mounting rails directly attached to the case, no 19-inch racking system in between. And in such a case, there's no requirement for the width to be 84HP. Eurorack-specific cases are built in many other widths, both wider and narrower than the standard 84HP.
So if you're starting out to build a Eurorack system, you have a number of choices. Probably the cheapest thing to do, and perfectly adequate for starting out, is to buy or build a wooden box with about 110mm vertical space inside and whatever width you want. Storage units intended for CD jewel cases may be about right. Then you can screw your modules into the wood at the top and bottom with wood screws. You'll need to add a power supply, maybe the kind with a wall wart plugging into the front panel and a flying bus behind; or you could use a bus board and screw it into the back of the box.
One 84HP row of Eurorack is enough to contain a decent starter system. I'd strongly suggest buying more case space than you need at the start, though, so as to have room to grow. If you're starting with one row, get a 6U case (two rows). People are always complaining about not having enough HP, or wishing that a given module consumed fewer HP, even without thinking through how there'd be no room to get their fingers onto the knobs if the panel were shrunk any smaller.
I think a depth of about 100mm from the front panel to the back of the box is probably ideal. That's about what's available in the Doepfer cases. Almost all modules will fit within 60mm, but it's nice to have some extra space for running cables and so on. If you use a shallower box, you may have to pay close attention to how deep your modules actually are and whether they will fit. ModularGrid's database can be helpful with that. Shallow cases are often called "skiffs," especially when you put one on the table in front of a larger synthesizer and use it to hold your control and sequencing modules.
A wooden box without rails saves some money, but the wood will get chewed up if you remove and replace modules a lot. A fancier option is to get a pair of rails and some "ears" that connect to the ends of the rails and make a simple frame, which you can then mount in a 19-inch rack cabinet (which in turn could be a do-it-yourself makeshift or plain wooden box, if you like). There's a thing called a Happy Ending Kit which is quite popular: it's a set of rails with rack ears that can rest on a tabletop if you don't screw them into a 19-inch rack, bundled with a flying-bus Eurorack power supply, for about US$160.
The next level up from rails and ears in a 19-inch rack is to use a complete case built with metal rails specifically as a Eurorack case. Doepfer offers several of these; so does Pittsburgh; Tiptop recently introduced the moulded-plastic Mantis case; and there are many others. At the high end, outfits like Lamond Design make bespoke wooden cases to whatever specification you can afford.
I myself started out with a dual-row Pittsburgh Modular Cell case, but they don't make those anymore, and that may be just as well. I wasn't thrilled with it. It's a little on the shallow side; I didn't like the sliding nuts (more on those later) with nonstandard screws; there were problems with the power supply; and I didn't like the factory finish on the wooden end cheeks and ended up sanding it all off and redoing it myself by hand (with results that I liked, but it was a lot of work and more than zero expense). Anyway, here's a picture of it.
Pittsburgh Modular makes several other kinds of cases now, which are probably worth a look - they seem to have learned from some of the mistakes of the Cell series.
I expanded into a rack case of the kind travelling acts use for mixing boards, with 10U of 19-inch rack on the top (into which I've put three rows of Eurorack rail-and-ears frames) and 4U at the bottom (once planned to fill with a Roland V-Synth XT, but now I'm thinking in other directions). This is a very bulky case, with a lot of empty space inside, and it wouldn't be very practical for travelling; it also wasn't cheap. But it solves most of the things I didn't like about the Cell.
In a case that uses the standard Eurorack rails, there's a question of how exactly to fasten the modules to the rails. Modules have mounting holes on them. Rails may contain a row of threaded holes 1HP apart to accept the screws; or they may have a slot, into which slides a metal strip with the threaded holes; or they may have so-called "sliding nuts," which just fit inside the rail and can slide to whatever horizontal position you want.
I think sliding nuts are really annoying, but they're the cheapest option, and a few people actually prefer them. Sliding nuts have the small advantage that you can slide them to any horizontal position, so it's not so necessary to mount your modules perfectly on the 1HP grid (useful for the occasional module with nonstandard mounting hole locations relative to the panel). Holes directly in the rail are less often seen; they're the sturdiest way to do it, but if you strip the threads on one, too bad, there's not much that can be done about it. The threaded strip inserts are what I prefer.
The threads for the mounting holes ought to be M3 (metric machine screw), but a few oddball cases are built with M2.5 or American #6 machine screw threads. Since your modules will come with M3 screws included, you can have fun searching for the right kind of screws if your case uses a different size.
People frequently want Eurorack cases that can be taken on commercial airliners as carry-on luggage. I think it might really be better to buy an appropriately-sized Pelican flight case and send your equipment as checked baggage, with as much insurance as it needs. That won't be cheap, but travelling safely with delicate electronic equipment is not going to be cheap in any event. Demanding that your case should be carry-on means it can't be as large as you want, to begin with. It's going to be built with the bare minimum of padding and protective structure in order to squeeze in the maximum number of HP. You'll have to fight with other passengers over whether you can put it in an overhead bin, and probably also with the airline staff over whether it really meets the requirements even though you thought you had that in writing from them before you started. Fat chance of your being able to take advantage of the rumoured "musical instruments are ALWAYS allowed as carry-on" policy. And so on.
Note that transporting a Eurorack synthesizer is really no harder than, and not much different from, transporting any other musical instrument. More or less the same considerations apply.
If you want to buy just one module
You need a case, even if it's as simple as a plain wooden box, and a power supply, or else you can't do modular synthesis at all. But if you're really pushing the budget, first stop and ask whether you can afford this hobby, and then if you want to buy just one module to start with, my current suggestion is my own product, the Middle Path VCO (about US$500).
The Middle Path is what's called an analog "complex" VCO (voltage-controlled oscillator), meaning that in addition to having two sound-generating cores that can be used separately or together, it also has special features for combining them to create more complex timbres. In particular, it has an unique shaping section that allows the VCO signals to modulate each other, with quadrature outputs that can create a stereo effect or can be used in frequency-shifting patches with other modules. Using this module alone without any others to control it or process the result means that you can't generate much except "drones" (long sustained sounds rather than individual notes), but there's enough front-panel controllability that you can still have some fun with it all by itself. There's also a sensible expansion path: many other modules you'd add in the future can be used well in combination with this as a starting point.
When I first wrote this series of postings, I didn't have a suitable first module for beginners in my own product line, and so my recommendation then was the Klavis Twin Waves. It's about US$250.
The Twin Waves is similarly a dual VCO, but it's implemented with digital techniques, basically a DSP chip running a bunch of different signal-generating algorithms that can be selected from the front panel or via control voltages. It's a lot of bang for the buck.
If you don't want a digital module, nor the analog Middle Path, then you might consider another manufacturer's analog complex oscillator module, like the MakeNoise DPO (US$600). Like the other two mentioned, the DPO is two oscillators in one with some features for making them interact, it can produce interesting sounds immediately without needing any other modules, and it won't be left behind when you expand to a larger system.
The DPO's manufacturer claims it is Vactrol-based, which if true means the module is probably illegal to sell in Europe. Vactrols are old-fashioned optocouplers made with cadmium sulfide photocells, and cadmium is banned by the EU Directive on Reduction of Hazardous Substances ("RoHS"), largely because of waste-disposal problems. Nonetheless, some European retailers have DPOs for sale. I don't know what to tell you.
If you add a CV controller, or want to buy only two modules and have one of them be a MIDI or USB to CV interface, then the North Coast Synthesis Ltd. MSK 007 Leapfrog VCF (about US$400) is worth a look. It can function as a very minimal synthesizer voice all by itself when given pitch and gate CV, as well as forming the core of a more complete analog subtractive voice once you add other modules. The Leapfrog VCF really needs a control voltage or signal input from some other module to do much of value, but given another module providing control it can almost function as a single-module noise maker; and it's a reasonable addition to almost any growing system. It works well with other North Coast modules too, of course.
Patch cables and multiples
As soon as you have more than one module, you'll need patch cables to connect them together. These should be male-to-male cables with 3.5mm mono unbalanced "phone" plugs (not "phono," that's a completely different style of connector); these are often called "1/8 inch" plugs even though they're really a tiny bit larger than that. Don't use stereo patch cables. There's some variation in where the contacts in the mono sockets on modules actually touch the plug body, so that it's possible for a stereo cable between two mono sockets to fail to make the connection. You may still want to have a couple of stereo cables on hand for doing things like connecting to a computer's stereo line input; just keep those separate from the mono cables used within the synth.
Places that sell modules usually sell patch cables, too. It's nice to have a selection of different lengths and different colours; if your cables are all the same, it can be hard to keep track of what goes where in a complicated patch.
Sometimes you want to connect a single output to more than one input. You can get a thing called a "passive multiple," which contains several sockets wired directly together behind a small panel that fits into the synth rack like a module. Many manufacturers sell these; a typical example is the Doepfer A-180-2, with two four-way multiples (reconfigurable to a single eight-way) in 2HP for about US$40. The way it's used is that you patch your output into one of the holes, and then patch from the other holes to your multiple inputs.
Passive multiples are often recommended as beginner do-it-yourself projects, to practice soldering and general module construction, and many places sell kits for building them.
You can also get passive multiples as separate devices that don't take up rack space, such as the Intellijel Hub (single four-way, US$10). And there are specialized cables, like the Tiptop Stackcables, that allow multiple plugs to share a socket, providing the effect of a passive multiple built into the cables.
If you drive several inputs from a single output through a passive multiple, that places correspondingly more load on the output. It's possible, though not really common, for the increased load to interfere with the output's ability to maintain a consistent voltage; the effect tends to be most apparent with a pitch control voltage driving several oscillators, filters, and so on all at once. Too much load and all the pitches go a little flat. To guard against such problems, you can get a "buffered multiple," which is essentially a distribution amplifier that boosts the available current while keeping the voltage fixed. Plug the CV into the buffered multiple's input, and then it has several outputs available for driving other things. These cost more than passive multiples, but it's nice to have at least one or two, depending on the size of your system.
My suggestion is that it's good to have about one multiple module (containing two four-way multiples) per full row of Eurorack, and at least one should be a buffered multiple. So, for a small system, once you're near filling your first row, you should probably plan to get a buffered multiple, and then add some passive ones when you grow larger. You can substitute a few Stackcables, Hubs, or similar devices for the passive multiples, if you like.
You shouldn't patch two outputs into each other, neither directly nor through a passive multiple. The best case of what will happen is that (as seen by any inputs sharing the multiple) the resulting voltage will be the average of the voltages the two outputs are attempting to send, resulting in mixing of the signals; but that's optimistic. Both outputs are trying to have very low impedance and drive the signal to a specific voltage. If they aren't aiming for the same voltage, then they will fight over it, both drawing significantly more current than is normal. In the worst case, the weaker one could be damaged. That's rare, but what's fairly common is that you get an unpredictable unequal mixture, with maybe a lot of distortion. If you want to combine two outputs equally into a single signal, the right way to do it is with a unity mixer like a Doepfer A-138u (US$54), a voltage-processing module with unity mixing included like the North Coast MSK 008 (US$207), or a mixer with adjustable levels like the North Coast MSK 011 (US$160).
Buying more modules
First, some words on budget and timing. You should know how much you're willing to spend, but you shouldn't spend it all at the start. It's pretty much a certainty that after you make your first purchase and start playing with it, you'll discover some module that you just have to have and didn't think of when you were starting. Set aside some money so that when this must-have module shows up, you'll be able to get it. For those keeping score, mine was the Intellijel Dr. Octature II, because I found I needed a clean sine wave LFO. It's not bad for that, but since I really wanted many independent sine wave LFOs and couldn't afford to just keep buying more Octatures, that was also the genesis of the Fixed Sine Bank.
It's frequent advice to start with a small system and add modules one at a time, and if your budget is very constrained, you may have to do that whether it's what you want to do or not. I have some recommendations above for which module to buy if you're going to buy just one. But I think if your budget is a little less constrained, and you can afford to buy several modules right at the start (with some money reserved for your future lust module, as mentioned above), it's okay to buy a fairly complete system right at the start. There's not much point flailing around with a few modules that you can't really use properly yet, when you can already afford to expand. I wouldn't suggest buying more than about one 84HP row right at the start, though.
Play with different configurations in Modulargrid and see how appealing they seem to you. It's a commonly done thing - which some people hate, but it remains commonly done and generally accepted - for newbies to set up the systems they're thinking of buying in Modulargrid and then post the links in a Web forum asking for advice and suggestions. You'll get both good and bad advice that way, and both can be interesting. I maintain a Modulargrid rack for a nice one-row system featuring North Coast modules; note that rack is designed to be nice rather than cheap, but if you can afford to rush out and buy that all at once, you'd certainly be able to have a lot of fun with it.
Here's a rundown of some of the common types of modules that might be of most interest to someone starting out.
Oscillators: Also known as VCOs (Voltage Controlled Oscillators). Waveform generators controlled by a pitch control voltage (CV) which would normally come from a controller or MIDI interface, as well as front-panel knobs. You really need at least one of these; having more than one allows playing more than one note at once, or connecting them together for modulation effects. Both my suggestions for single-module systems are so-called "complex" oscillators (one digital, one analog), which produce more or less complete sounds by themselves. The more traditional kind of synth VCO produces just a simple waveform like a sawtooth. Those don't sound good by themselves and really need other modules to shape the spectrum, so they become appropriate when you're building a system with more than one module.
I bought two Tiptop Z3000 oscillators (US$255 each, no longer visible on the manufacturer Web site and possibly discontinued) when I was starting out, and I like them a lot, but not everyone likes the digital frequency displays on these analog oscillators. The Intellijel Dixie II (US$220) is also highly regarded, and takes up very little rack space. (Don't make the mistake of planning with Modulargrid to use the earlier and even narrower "Mark I" Dixie - those are no longer made and have become hard to find used.)
Controllers and interfaces: If you're not just going to twist the knobs on the oscillator and make drones, you'll need some way to play notes. I think for most beginners a MIDI interface makes the most sense; then you can plug in a keyboard synth or a computer. Many of these will accept both USB-MIDI and real DIN MIDI connections, like the Doepfer A-190-3, US$140. A fancier choice, with DIN input only but a bunch of special features, would be the Mutable Instruments Yarns, for about US$360. Mutable Instruments used to sell a do-it-youself kit for a very basic USB-MIDI interface called a CVpal, and those were really nice for anyone with a DIY bent, but unfortunately the kits have been out of production for a while. The design is open and in principle could be revived, but trying to build it without a commercial kit, including getting the chips programmed and the boards manufactured, is no longer a beginner project.
If you can afford just one module, I said the first should be an oscillator, but then if you can afford just two modules, for most people it would make sense for the second to be a MIDI interface. Be aware that most synthesizer USB-MIDI interfaces are USB devices, not hosts, and most USB-MIDI controllers (like the Akai MPK Mini I use) are also devices. That means the controller won't plug into the interface. You need a computer in between. If involving a separate computer is a problem, you should either look at controlling your rack through traditional DIN MIDI, or seek one of the rarer Eurorack USB-MIDI hosting modules, like the Expert Sleepers FH-1 (US$280).
Filters: Subtractive "East Coast" synthesis revolves around the voltage-controlled filter (VCF), which is used to remove higher harmonics from a harmonic-rich VCO signal. Much of the timbre of the sound in a subtractive synthesis patch comes from the filter. These are usually low-pass filters (LPFs) because that's the most useful type in making musical sounds, although other types are available and one common family of circuit topologies routinely offers multiple different filter types on separate outputs. There are many filters to choose from. North Coast's flagship product is the Leapfrog VCF, about US$400 assembled, and it includes a built-in VCA which can reduce the need for other modules in a small system. But it has its own unique sound, which you may or may not like. The North Coast Synthesis Coiler VCF is also worth a look when building a first system; it's a smaller module at 8HP and about US$200, and it has multiple inputs and outputs for a variety of different sounds from a single module.
The first filter I bought for myself was a Tiptop Z2040 (US$205), which is generally thought to sound similar to 1980s Roland analog synths - a classic, middle-of-the-road synthesizer sound. The Z2040 also has a couple of built-in VCAs, which can be very nice in a small system with few or no dedicated VCA modules. Doepfer makes a number of filters of which the A-124 "Wasp" (US$92) seems to be especially well-regarded, notwithstanding that it sometimes has reliability problems. Intellijel makes some fancier multi-function filters of note, including the Dr. Octature II (US$280) and Korgasmatron II (US$390).
Probably my favourite of the filters I currently own excluding North Coast products is the Erica Polivoks VCF (USD$113 fully assembled, but I built mine from a kit for a little less, and made some modifications). I like it because it has what I'd call a very expressive sound - much like an acoustic instrument and not a sterile "electronic" sound at all. However, it is electrically rather poorly behaved, cannot produce a "clean" sound under any adjustment, physically deeper than will fit in some Eurorack cases, and so on. I'm not sure I would recommend that as the only filter to someone who plans to buy only one.
Voltage-Controlled Amplifiers (VCAs): VCAs allow one voltage to control the level of another. That is necessary in shaping continuous signals into separate notes - the output from an oscillator usually runs all the time, so you might patch it and the gate signal from your MIDI interface into a VCA to allow the sound to start when you press a key and stop when you release the key. With the addition of an envelope generator, it can also follow a loudness profile (envelope), typically loudest right at the start of the note and then trailing off. But VCAs also have many other uses, in modifying and routing control signals through a large patch; and they are often overlooked because of not being as "sexy" as some of the complicated sound generation and effects modules marketed by up-and-coming manufacturers. That's why it has become a cliche to tell newbies their proposed initial setups "need more VCAs."
The Pittsburgh Dual VCA module (US$160) is a decent choice. Doepfer makes several variations of which the A-132-3 (US$113) seems like a good pick for a small starting-out system. At the higher end, the Intellijel Linix offers six VCA units in a single module for about US$330.
Other utilities: There are many small modules that do things with signals in ways that become more useful as a system grows: these include LFOs like the MSK 010 (Low Frequency Oscillators, which can be used for stuff like adding vibrato or creating a regular pattern of notes in a self-running patch); mixers (both for control voltages and audio signals) like the MSK 011; switches like the MSK 008 and clock-manipulation logic (again, useful for patches that run themselves); sources of random voltages (either to play notes, or as "noise" to add to a timbre, especially in percussion sounds); and many others. I think I'm going to leave detailed discussion of these for some other article, because they're less relevant to beginners just starting out.
This should have given you some idea of what's involved in getting started with modular synthesis, should you be foolish enough to try it. In the next installment, I'll discuss common pitfalls beginners encounter.
Continue to Part 4 of this series.