You are here

Exploring Analogue, Part 1

Exploration By Steve Howell
Published May 1994

PART 1: The modular, analogue synthesizer, owned by few and coveted by many, is an extraordinarily flexible beast. Many of the effects made possible by these lumbering monsters, however, can also be emulated using far more modest instruments, as Steve Howell explains. This is the first article in a three‑part series.

While compact analogue synths undoubtedly provide the benefit of convenience, modular systems offer more flexibility because there's virtually no limit to the different ways in which the various modules may be interconnected. However, the modules themselves are still based on the filters, envelope generators, oscillators and so forth covered by Paul White in his short series 'Sound Foundation' (see the February and March issues of SOS).

So called because they comprised a series of open‑ended, unconnected modules, a typical modular synth might have eight oscillators, a bank of assorted filters, four or more envelope generators and various other bits and pieces. In some systems, additional modules could be added, rather like like a sonic Lego set, while other systems provided a set number of modules mounted on a common front panel.

With so few modular systems still in circulation, you might wonder at the relevance of this article other than as a nostalgic trip through synthesizer history, but many of the capabilities of a modular system have been carried over to programmable analogue synthesizers such as the Oberheim Matrix series. Even the modestly priced Matrix 1000 (used with suitable editing software), provides enormous routing flexibility while offering the added benefit of polyphony. Needless to say, many of the techniques described in this article translate directly to their more modern counterparts, but with modular synths still being available on the secondhand market, perhaps you'll be tempted to go down that route after reading what I have to say.

Modular History

The different manufacturers (Moog, ARP, and the British company, EMS, were the three main contenders in those days) all had their own ideas of how a modular synth should work. Bob Moog's idea was to leave all modules unconnected — to make any sound, you needed to insert patch cords as appropriate. He also chose to use rotary controls. ARP's founder, Alan R Pearlman, preferred to use sliders, as these give a good visual reference of what a sound may be doing (compare, for example, a graphic equaliser with a parametric — a graphic shows you exactly what EQ curve has been set, even when viewed from distance). However, on the legendary ARP 2600, he hard‑wired the basic connections internally so that a simple patch could be created without cords but you could override these 'preset' connections simply by inserting a patch cord as appropriate for more complex patches. One significant improvement of ARP's designs, however, was that each module had its own voltage mixer. On Moog's designs, if you wanted to mix a variety of controllers (LFO and envelope, for example) to control, say, the filter, you first needed to patch the LFO and envelope into a separate mixer and then take the output of that mixer to the filter's control input. On the 2600, you could patch straight into the module's own control voltage mixer. ARP's designs, as a result, led to a less cluttered control panel even when a complex patch was being programmed; Roland later used this approach on their System 100M and 700.

EMS, on the other hand, preferred a matrix system, where all inputs and outputs were made available along a horizontal and vertical axis and pins were inserted to make the appropriate connections. This was neat and simple, if a little bit fiddly, on the small VCS3 (also known as 'The Putney' in America, so named after its place of birth), but on the large Synth100 (also known as 'The Delaware' because the BBC's Radiophonic Workshop in Delaware Road, Maida Vale, London, were the first to take delivery of one), the two 256 x 256 matrix pin boards (one for audio signals, the other for control signals) were something of a challenge to circumnavigate. Rotary controls were used on all EMS gear.

All three methods have their advantages and disadvantages, and I suppose an ideal method of patching has yet to be found. However, aside from all of this, these large beasties were equipped with some modules that were to be dropped from the scaled‑down MiniMoogs, Odysseys and suchlike that were to follow.

Alternative Filter Types

A filter is an audio sieve that allows you to remove certain elements of the sound but let others pass through. The most useful filter found on any analogue synth (or digital synth, for that matter) is the resonant low‑pass filter. This, as explained in earlier issues, allows you to filter out upper harmonics, letting lower ones pass through unaffected. The low‑pass filter is so useful because it mimics real life quite well — a sound naturally loses upper harmonics as it dies away in level. This is due to the fact that upper frequencies have less physical energy than lower ones and simply cannot 'live' as long as lower frequencies, so they tend to fade out progressively during the course of a note. It follows, therefore, that if we control such a filter, sweeping down its cut‑off frequency with an envelope generator that has the same or a similar envelope to the amplitude envelope, we can get a synthesizer to replicate this acoustic phenomenon... ish!

The low‑pass filters on analogue synths had resonance (a feat not so easily achieved with digital simulations). This boosts the area around the cut‑off frequency, allowing you to highlight a particular band of harmonics in the signal. With high resonance settings and envelope sweeps, the characteristic synth sounds currently so trendy can easily be created, but by setting the cut‑off frequency manually and not applying any modulation other than keyboard tracking, you can actually create something approximating vowel sounds. This allows some interesting vocal effects to be created, and using sequencer control you can almost get your filter to 'speak'!

The low‑pass filter, therefore, is an all‑round good egg, and for that reason is found on all analogue synths, without exception. However, other filter types exist, which can be called upon to perform a variety of other tasks, and some of these occasionally turn up on very low cost analogue machines, including some early Yamaha models, and even the infamous EDP Wasp.

    As the name suggests, this type of filter allows high frequencies to pass through unaffected, whilst filtering out lower frequencies. The effect, depending on the filter's settings, will be to remove (or at least attenuate) the fundamental, resulting in a thin, nasal sound which can be put to good use in the creation of reed instrument sounds (like oboe and bassoon) or for synthesizing instruments in the harpsichord family, the tonal qualities of which are thin and brittle. The high‑pass filter may be used in moderation to attenuate the fundamental very slightly, and I have found that this is a good way to balance a mix. For example, when synthesizing a brass section, you may have a situation where the fundamental frequencies of some harmony parts are competing too strongly with the main melody part. Reducing the volume of the harmony parts simply upsets the overall balance. However, using just a hint of high‑pass filtering can remove their fundamental slightly, so that they do not conflict with the main melody. The same technique may be applied to strings and the like where, during ensemble playing, strong fundamental frequencies in the less important instruments cloud the overall mix.

Only a few 'portable' synths that were made from scaling down the big modulars have high‑pass filters (the ARP Odyssey immediately springs to mind), but these are simple, fixed filters, little more than simple bass‑cut tone controls, and not very useful. A full‑blown high‑pass filter will be voltage controlled, so that it may track the keyboard and be driven by other controllers for special musical (and non‑musical) effects. Ideally, it may also come equipped with variable resonance to further expand its usefulness.

    On the old synths, you could patch low‑pass and high‑pass filters together, in parallel, to create band‑pass and notch‑reject filters. Some synths, however, actually featured these as separate filter modules.

A band‑pass filter will remove frequencies below and above its cut‑off frequency, with an aural effect similar to that of a wah‑wah pedal. A notch‑reject filter will keep frequencies below and above the cut‑off point but will remove those that fall between the filter's roll‑off area. The notch‑reject filter seems like a useful module to have around, but the effects it creates can only be described as subtle. Of course, the usefulness of these two filter types is greatly enhanced if they can be voltage controlled.

    Fixed filter banks, which allow the selective modification of certain frequencies, were found on some of the larger modular synths — the Moog synths featured these (sometimes as optional modules), as did the EMS Synthi 100. You could regard fixed filter banks as resembling graphic equalisers, except that their frequencies were very tightly spaced, in intervals of half an octave, over a small range of between 125Hz and 5.6kHz. Whilst you might be tempted to think that slapping a graphic on the output of your synth would achieve much the same thing (and, up to a point, you'd be right), the beauty of modular synthesis is that the oscillators could be processed through these fixed filter banks before passing through to the other filters, amplifiers and so on, thereby increasing the tonal versatility of the fixed palette of preset waveforms supplied by the oscillators. Uses for these filters include creating fixed‑frequency resonances (sometimes referred to as 'formants') to simulate acoustic instruments. For example, the oboe has two such resonances at 500Hz and 1.5kHz, which remain constant regardless of pitch. Using a fixed filter bank to boost these frequencies in a pulse wave would help give a more lifelike oboe sound, which could be further improved by judicious use of a high‑pass filter to help create the oboe's characteristic 'thinness'. The fixed filter bank could be applied to other reed instruments, like bassoons and saxes, and also to stringed instruments.

As you may know, the human voice also has pronounced fixed resonances (hence the reason why transposing vocal samples up and down sounds so unnatural — you're actually pitching fixed resonances which should stay at a constant frequency). Using fixed filter banks allows you to go some way towards capturing these resonances, thereby creating more realistic vocal effects.

Such a wealth of sound sources (the many oscillators a large modular is blessed with) and this plethora of filters goes some way to explaining why these synths are so damned versatile and so sought after by certain individuals, whose sound creating aspirations usually stretch much further than their bank balance!

Ring Modulator

This curious device takes two inputs and produces an output that is the sum and difference of the two frequencies fed into it. For example, feeding a sine wave frequency of 500Hz into one input and 750Hz into the other would produce 250Hz and 1,250Hz at the ring modulator's output. The fact that these frequencies are (usually) not harmonically related to the input gives rise to discordant results — thus the ring modulator's prime use is in the creation of complex (and often very realistic) bell sounds and other metallic clangs.

The effect of the ring modulator depends very much on the inputs fed to it. A 'harmonic‑less' sine wave gives a more predictable result, but if you feed in a harmonically rich sawtooth or square wave, the effect is even more clangorous as the harmonics are processed by this 'sum and difference' technique. For example, feeding in two sawtooths at 500 and 750Hz would not only give you the 250 and 1,250Hz result from the fundamental but also the frequencies 500Hz and 2.5kHz (the sum and difference of the first harmonics, 1kHz and 1.5kHz), 1kHz and 3kHz (the sum and difference of the second harmonics, 2kHz and 3kHz) and so on. In practice, you would normally use simple harmonic structures for a bit more control over the ring mod's output, typically using sine or triangle waves.

Another application for the ring modulator is to produce the famous 'dalek' voices. This is achieved by feeding in a frequency of around 100Hz to one input and a mic signal to the other (usually via some form of preamp) to create the characteristic effect. The voice is then chopped by the 100Hz signal to produce that nasty, rasping effect. Of course, ring modulators can be used to create other robotic and machine‑like vocal sounds by varying the modulation frequency; these sounds are as relevant to today's experimental dance music as they were to Doctor Who.

Yet another application for the ring modulator is as an octave splitter. Bearing in mind that a ring modulator will produce the sum and difference frequencies of the input signals, splitting a signal into two and feeding identical signals into both inputs would create an upwards octave split. For example, splitting a 500Hz signal and feeding both inputs would produce the sum (1kHz — an octave higher) and the difference (0Hz). This effect can be used to augment your oscillator complement; instead of using two oscillators set at octave intervals, use one via the ring modulator to achieve a similar effect. Via a pre‑amp, the ring modulator could be used to process external instruments such as guitars. Unlike dedicated pitch shifters, the output of a ring modulator will not have the usual wobble and delay normally associated with all but the most expensive pitch shifters, and so you could use your modular synth as an outboard effects unit during mixdown or track laying to process sounds other than those originating from the synth itself. Most modern synths forego the luxury of a ring modulator, but several early 'compact' analogue synths had one, including the Korg 700S, so ring modulation is by no means limited to those using modular systems.

If your synth is sans ring modulator, similar effects can be achieved by modulating the filter and/or amplifiers with signals from audio oscillators. We will look at these techniques next month.

That brings us to a timely end for this month. There's still a lot more to cover because, as well as a liberal sprinkling of audio processing knobbage, modular synths also came with a wealth of voltage processors and modulation possibilities, which we shall be covering in due course.

In the meantime, if you need any confirmation of the power of the modules we have looked at this time, get a hold of some Tomita albums and marvel at the breadth of tones (particularly vocal sounds) he was able to create through inspired use and control of these different filter types.

Different Approaches To Layout

This is Bob Moog's idea of a good layout. Here, to mix audio or control signals, you must first patch in either the mixer or the attenuator, leading to a sprawl of cables across the front panel.

This is ARP's approach. Each module has its own audio inputs (where appropriate) and control voltage mixer. Furthermore, each connection shown here would have a preset connection hardwired internally but which could be overidden by the insertion of a patch cord. You will note that whilst Moog used quarter‑inch jacks for patching, ARP used 3.5mm plugs, helping to keep the control panel compact and bijou.

Creating Alternative Filters

When a high‑pass and low‑pass filter are used in parallel as above, when the high‑pass filter's cut‑off is lower than the low‑pass's, you create a bandpass filter, the width of which is determined by the respective settings of the two cut‑off frequencies.

When two filters are used in parallel, as shown in the diagram, and the high‑pass filter's cut‑off is higher than the low‑pass's, you create a notch reject filter, the width of which is determined by the settings of both filter's cut‑off frequencies.

Add to both of these examples the ability to control the cut‑off frequencies of both filters and the possibilities for tonal silliness abound. Consider also that each one may be modulated separately (one sweeping up whilst the other sweeps down, for example) further expanding the possibilities for tonal shaping.

Ring Modulation

Here we see the typical block layout of a ring modulator. Two inputs are used to create sum and difference frequencies (see main text). Some ring modulators mixed the original signal into the output so that all four frequencies were heard. Others only fed the sum and difference frequencies — you mixed in the originals as required. Modular synths that sported ring modulators were the ARP 2600, the EMS VCS3 and Synthi series (A, AKS, 100) and the Roland 700M and 100M, to mention but a few. Although not a modular synth, the Yamaha CS80 also had a fine ring modulator. Ring modulation was not a feature of Moog synths but I believe that one Harold Bode was a chum of Bob's and he made a wonderful (but expensive!) unit which Moog made available. Tomita was a great exponent of the Bode ring modulator, and it can be heard to good effect on several of his classical albums — in particular at the beginning of his rendition of Mussorgski's Pictures at an Exhibition.

Using a ring modulator patched as shown in the diagram, with the same signal feeding both inputs, allows you to create upwards octave splits, where the sum is exactly double the input frequency (see main text for a full description).

Roll‑Off Slopes

Different filters have different roll‑off slopes. The roll‑off slope is the amount of attenuation per octave after the cut‑off frequency. The higher the roll‑off, the more dramatic the effect of the filter, especially when being modulated or swept by the envelope. You can see what is happening in the diagram. With a 24dB/octave filter, you can see that more of the upper harmonics are being cut off, whilst with the 12dB/octave filter, even though the cut‑off frequency is in exactly the same position, more of the upper harmonics pass through. This results in a 12dB/octave synth having a slightly 'fizzier' tone as more harmonics pass through than on a 24dB/octave filter. A 24dB/octave filter is sometimes referred to as a '4‑pole filter' and a 12dB/octave filter a '2‑pole filter'.

In the past, American synths usually used a 4‑pole filter (Oberheim being the only exception), whilst the early Japanese synths from Roland and Yamaha and the like sported a 2‑pole filter; this gave rise to the feeling that American synths had more guts, purely because 2‑pole filters are a bit brighter in character and modulation effects are not quite as dramatic. Having said this, the Oberheim synths, largely regarded as being pretty ballsy, have measly 2‑pole filters! Recently, the newly developed SE1 (reviewed in the January issue of SOS) allows you to select between the two.