Modular synthesis intro, part 14: Multiples
This is Part 14 of a series that started with Part 1.
The basic operation of patching a modular synthesizer consists of connecting an output to an input. Perhaps you have a MIDI interface with a V/oct pitch control voltage output, and you want to control an oscillator with a V/oct pitch input, so you plug one end of a patch cable into the output and the other end into the input, and the interface will control the oscillator. Fine.
Almost immediately, you find yourself wanting to make more complicated connections. Maybe instead of just controlling the oscillator frequency, you want the same control voltage to also set the frequency of a tracking filter, like a Leapfrog. At that point, you need to somehow split or distribute the voltage from the MIDI interface to send it to both the oscillator and the filter. The general problem is to connect one output to more than one input; and in modular synthesis, this distribution is usually done by something called a multiple or "mult." Note that the term multiplier usually describes something quite different (and itself refers to a couple of different functions); don't confuse multipliers with multiples.
As soon as a system grows to more than trivial size, it becomes pretty necessary to include something that provides the function of a multiple. There are a few different options on exactly how to achieve it.
Passive multiple modules
The most traditional way to split an output to drive multiple inputs is with a module like the Doepfer A-180-2. It's electrically very simple: just eight jack sockets in a panel, with their signal connections linked into two groups of four. You connect the output you want to split into any of the four jacks in a group, and then the other three jacks become copies of that output, which may be linked to up to three inputs elsewhere in your patch.
Many manufacturers sell similar modules, and they are also popular as do-it-yourself projects, often recommended to beginners as a way to practice soldering. Passive multiples are some of the cheapest modules in Eurorack. Variations exist. The basic arrangement of eight jacks in two groups of four seems to be popular and a good match to what people usually want, but it's easy to make a different arrangement (for instance, a single group of eight) by soldering the jacks together in a different pattern. For additional flexibility, one can also add switches, to make the connection pattern adjustable from the front panel.
There's potentially an impedance issue with passive multiples, as with any "passive" module: since the multiple is really just a cable adapter mounted behind a panel, with no connection to a source of power, it has to split the power from the output that drives it among as many inputs as it is driving. In Eurorack especially, we already have impedances only vaguely defined, with the general idea that outputs should have low impedance and inputs should have high impedance. If you drive more than one input from an output, then the output has to work harder (provide more current) in order to supply the greater load (parallel-combined input impedance). Depending on the specific modules involved, this increased load on the output can result in a lower output voltage.
For audio signals, you might lose a few percent of the voltage, which is a fraction of a decibel and you probably won't notice. But for pitch control voltage, which unfortunately is one of the most common reasons for using multiples at all, losing a few percent of the voltage can mean being off by a semitone or more over a couple of octaves, screwing up the tuning and tracking of oscillators and filters. The effect is especially maddening because it depends very much on the specific modules. An output with very low impedance will be able to drive multiple inputs through a passive mult with no problem, and many inputs with very high impedances can be driven even by a poorly-behaved output, so some patches will work fine, but then you try a slight variation with different modules and the patch stops working because you exceeded the limits.
Here are some banana plugs, typically used for connections to electronic test equipment; these ones in particular are on the test leads for a multimeter.
As seen in the picture, each plug has a matching socket at the back, into which you can plug another similar plug, potentially stacking up an unlimited number of them. The stacking capability is not an essential or defining feature of banana plugs - many are made without it - but it's quite common. If your synth were made with banana sockets and you used cables like these to patch it, then you wouldn't need multiples; you could just plug in as many cables as you wanted on each socket.
Some modular-synthesizer formats, Buchla in particular, are made with banana sockets and allow this sort of thing as a basic feature of the format. Some people have made efforts to replace all the phone jack sockets on their Eurorack modules with banana sockets - a move which can be difficult because of differences in the physical design, as well as likely screwing up any switching or normalling features of the phone jacks and possibly introducing grounding problems.
It is possible to get Eurorack-style patch cables with a similar stacking feature. TipTop sells "Stackcables," and holds a patent (US Patent number 8,033,860 B2) on at least some forms of such things. Modular Addict also sells stackable Eurorack patch cables. Debate exists as to whether Modular Addict's offering infringes the TipTop patent, and (since stacking cables of various kinds existed for a long time previously) as to whether the TipTop patent is legitimate in the first place. Threats have been made, but at this point no side seems really willing to take the matter to court. From the point of view of the modular user: these exist, they can be quite convenient, they have some limitations in terms of fragility and some details of how shielding is handled, and whether you want to use them instead of or in addition to other forms of multiple, is your call.
Although not exactly a special kind of cable, I'd also like to mention Intellijel's "hubs," which are basically just four-way passive multiples in their own little plastic cases instead of modules to mount inside the rack.
Although not in the form of a module, all of these stacking-cable and outside-the-box splitter methods are electrically identical to passive multiple modules. They all share the issue of placing increased load on the output as it tries to drive the combined lower impedance of several inputs in parallel. To distribute a precise voltage without loss, such as for pitch control, you may need to look further.
So, what if you do want to distribute a precise voltage to several inputs without worrying (much) about impedance issues and voltage loss? Then you need what in modular synthesis is usually called a buffered multiple. Other subfields of electronics would usually call it a distribution amplifier.
The buffered mult has groups of jack sockets, with one in each group marked as an input. You plug the output you want to split into that one, and then the other sockets in the group become outputs with identical copies of the input voltage. So far it sounds much like a passive mult.
The difference is that with a buffered multiple, the signal goes through an amplifier. This module draws very little current from the output that feeds it, and it attempts to provide each input it drives with as much current as that input requires, at the same voltage. Because it increases the current, it exhibits current gain and really is an amplifier despite there being no change in the voltage. It also has the effect of isolating (thus buffering) the different inputs it drives, from each other. The extra current has to come from somewhere, so the buffered multiple requires a connection to the power supply, unlike the passive multiple.
Although you can often get away with driving several pitch CV inputs from a single output through a passive multiple, if you want to really be sure of avoiding voltage-droop problems, you need a buffered multiple. My usual recommendation is that when building a system, it's appropriate to plan for at least one 2x4 multiple module (or equivalent in stacking cables) per row of Eurorack, and for at least one of those to be a buffered multiple module per rack of three or four rows.
The main disadvantages to using buffered multiples (i.e., why not just use them all the time?) are that they are more expensive, and they can't really be implemented in any form other than screwed-in modules because of the need for a power connection. There is also potentially a quality issue with some of them: sometimes there is a very small voltage gain or loss, or a voltage offset, introduced by the buffer because of inaccuracies in the components used. It's normally not very much, and it's usually less than the accuracy you'd lose to impedance issues in a passive multiple, but if you are terribly finicky about precise tuning, you may find that running your pitch control voltage through a buffered multiple will throw your oscillator tuning a few cents off compared to just connecting it directly.
Abusing multiples as mixers
What happens if you try to use a passive multiple in the opposite direction, to connect two or more outputs to a single input instead of a single output to two or more inputs? In such a case, the outputs will conflict with each other, each trying to drive the voltage to its own preferred output voltage, and the input will see whatever is the result of that conflict.
In the best case, the effective voltage of two or more outputs connected together would be just the average of them. If one output would drive the voltage to 2.0V and the other to 1.0V, then ideally the combination would be 1.5V. In practice, that is seldom reliable, and it's a good idea not to attempt it. Other things that can happen, collectively more likely than nice averaging, are that one output or another "wins" and overwhelms the others, driving the combined voltage to its own; that you get some kind of mixing, but with a lot of distortion; or that one or more of the modules overheats and maybe even suffers permanent damage. The actual likelihood of damage to the modules from output contention is remote. I've never seen it really happen in practice. But it's traditional to warn people against it, exclude it from warranty coverage, and so on.
It's easy to understand why patching outputs to other outputs, directly or through a multiple, would be a bad idea. Outputs, especially on Eurorack modules, are meant to have low impedance themselves and to drive high impedance. If you set them up to drive each other, then you have them driving low impedances for which they're not designed, and you can expect the results to be different from what the modules would normally do. There's also the problem that "what the modules would normally do" in this case is contradictory anyway. If one module is, by design, driving the signal to 1.0V and the other to 2.0V, they simply cannot both win.
So the basic advice on using multiples instead of mixers, is "don't." To mix signals you should instead use a proper mixer module, like the North Coast Synthesis Transistor Mixer. If you want a very simple one with no knobs and superficially looking like a buffered mult, but in the opposite direction, then do a search for a "unity mixer"; many manufacturers make them.
To muddy the waters further, some other synthesizer formats (not Eurorack) attempt to control impedances in such a way that you can safely patch outputs together to mix them. Doing it involves higher, and more precisely controlled, output impedances than are customary in the Eurorack world. The trade-off is that then the impedance issues in the opposite direction, splitting a single output to multiple inputs, and the need for buffered multiples to split signals, become even more significant.