How Analogue Synthesizers Work
They've been a mainstay of electronic music for over 20 years, they're less expensive than ever before, and they still make sounds the digital machines can't. Analogue synthesizers have been receiving some digital assistance of late, but at heart they're still analogue synthesizers - and if you want to make them perform magic, you need to speak their language. Craig Anderton is your interpreter...
They've been a mainstay of electronic music for over 20 years, they're less expensive than ever before, and they still make sounds the digital machines can't. Analogue synthesizers have been receiving some digital assistance of late, but at heart they're still analogue synthesizers - and if you want to make them perform magic, you need to speak their language. Craig Anderton acts as your interpreter...
Far from being 'obsolete', with creative programming, analogue synths have much to offer - and they still have the power to create sounds that can amaze and delight us. Like different automobiles, different analogue synthesizers are functionally pretty similar. This is because analogue synth technology has been around for over 20 years now (the modern synthesizer was developed in the mid-'60s independently by Donald Buchla and Robert Moog), and design concepts that have stood the test of time are incorporated in most present-day synths.
What exactly is an analogue synth? The term is becoming somewhat imprecise, since purely analogue synths died out in the late '70s. These days, most synths are hybrids that combine analogue and digital technology to varying degrees. However, a good working definition is that analogue synths contain circuitry that operates on audio signals, rather than treating all audio signals as numbers and manipulating those numbers using digital techniques to change the sound. When you hear a Moog filter open up and give that wonderful, fat, resonant sound, you're hearing a circuit that's working directly on the audio signal - not a computer model that changes data to fit a software concept of what a filter is. Analogue synths tend to sound somewhat less 'sparkly' than digital synths, but make up for that with a richness of sound that even some of the best digital synths cannot duplicate. There's a reason why many musicians swear by their Moogs, Prophets, ARPs, and Oberheims, and why major companies - Ensoniq, Kawai, Korg, Oberheim, Roland, and others - still design synthesizers based on analogue circuitry and concepts. But before we can get the most out of these wonderful instruments, we first need to examine how they work.
The synthesizer creates musical sound through the use of signal generating and signal modifying (or signal processing) circuitry. The concept of having elements that make sounds, and elements that modify sounds, is common to much audio technology. (For example, in the studio, a multitrack tape recorder records and generates sound; these sounds are modified in the mixer via the use of equalisation, reverb, faders to set level, and so on. Similarly, a digital sampler has a digital recorder section that stores audio sounds, along with a bunch of signal processors to alter the stored sound.) Coupled with these two elements are controllers such as keyboards, modulation wheels, and so forth to allow you to interact with the signal generating and processing modules.
With early synthesizers, no one knew what combination of generators and modifiers would sound good and what wouldn't. So, a synthesizer consisted of several signal generator and processing modules, whose inputs and outputs were interconnected by patch cords (which is why particular synth sounds are still called 'patches'). Changing a patch involved manually repositioning patch cords and remembering a patch meant notating the configuration on paper. Over the years, certain combinations of modules seemed to work better than others, and since patch cords were a pain to deal with, eventually these modules were wired together in a 'normalised' configuration. Modern analogue synths use software equivalents of patch cords to interconnect modules and computer power to let you store settings in memory for later recall, but aside from these advances, the basic operational concept has remained virtually unchanged for over two decades.
In a synthesizer, the tone generator is usually an electronic circuit called an oscillator. This naturally raises the question "how do a bunch of wires make sound?"
Sound is variations in air pressure that are picked up by our ears and interpreted by our brain. If these variations happen in a regular (periodic) manner, they are said to have a frequency, and we perceive them as a tone. The frequency is specified in Hertz (abbr. Hz), which means cycles-per-second. Look at a vibrating guitar string under a fluorescent light some time, and you'll see it vibrating in a regular fashion. The sound of rain, on the other hand, is a random phenomenon. Therefore, we don't hear it as a pitched sound, but as noise.
Loudspeakers generate sound by moving a paper cone in response to current flowing through a magnetic voice coil attached to the cone. Thus, if you send a positive voltage into the loudspeaker, the cone moves forward. If you send a negative voltage, the cone moves backwards (assuming the speakers are wired correctly). If this voltage varies at a rapid periodic rate (at a frequency within the range of human hearing), then the speaker will move air that corresponds to a musical tone.
So, to generate a sound electronically, all we need to do is send a periodic voltage into a loudspeaker. By controlling the period of this voltage, we can create different frequencies - and yes, it really is that simple.
Getting back to synthesizers, the oscillator is the element responsible for generating the periodic waveform. Periodic waveforms are very easy to generate electronically; there are five basic waveforms commonly used with traditional analogue synthesizers. These waveforms consist of a fundamental frequency (which gives a sound its sense of pitch and is where most of the audio energy is concentrated), and additional audio energy in the form of harmonics (frequencies that are some multiple of the fundamental). The more harmonics, the more high frequency energy, and the brighter the sound. Fewer harmonics means a duller sound. For precise information on the harmonic content of various waveforms, refer to any good book on acoustics; for information on the harmonic content of different instruments, see Howard Massey's book 'A Synthesist's Guide to Acoustic Instruments'. For now, we will describe each waveform in terms of its sound quality.
SINE: This waveform consists, in theory, of purely the fundamental (ie. it has no harmonics). In practice, though, it's very difficult to generate a perfect sine wave, so there will usually be some very weak harmonics that add a slight 'buzz' to the sound. Whistling produces a waveform that is pretty close to a sine, as do guitar strings towards the very end of their decay (especially if you're listening to the bass pickup with the tone control turned down). Most audio test generators emit sine waves.
TRIANGLE: This is similar to the sine wave, but contains more harmonics and sounds very much like a clarinet.
SAWTOOTH: High in harmonic content, the sawtooth is the richest-sounding waveform. It is often used to synthesize orchestral washes and monster bass parts.
PULSE: Most pulse waves let you vary the duty cycle, or the percentage of time the pulse is high compared to the time the pulse is low (see Figure 1). Thinner pulses produce a reedy, nasal sound, while pulses with a 50% duty cycle - called square waves - produce a hollow kind of sound.
NOISE: This is an unpitched sound source that generates a sound like falling rain or escaping steam. It is used a lot for percussive patches.
Some synthesizers take short pieces of periodic waveforms sampled from existing instruments (bass, piano, etc), store them digitally, and use these waveforms in place of the standard oscillator waveforms described above.
These digitally-stored or generated waveforms are not samples in the normal sense; to show why, consider a piano sound. The sound of a piano is very complex and varies considerably over time, so just grabbing a few cycles of the sound will not necessarily sound anything like a piano. However, you will end up with a 'natural-sounding' waveform that is unlike any of the five 'standard' waveforms described above. Instruments like the Korg DW8000, Ensoniq ESQ-1, Kawai K3, Roland D50, and so on, offer multiple digital waveforms in place of, or in addition to, the standard synth waveforms; this translates into a greater variety of sound generators, which means more potential variety in your patches.
Most oscillators let you vary more than just the waveform. If the synth includes two oscillators, there will usually be a detuning control that alters the pitch of one oscillator with respect to the other oscillator. This can create flanging/chorusing type effects that help thicken up the sound. Sometimes you will also have the option of hard syncing one oscillator to the other. This is a little complex to understand. With hard sync, a slave oscillator is always locked to the period of the master oscillator. Thus, even if you alter the frequency of the slave, it has the same apparent pitch as the master since the period remains constant; what does change is the harmonic content of the slave. Figure 2 shows how hard sync can provide unusual waveforms in the slave oscillator. The sound that's produced is very 'nasty' and harmonically complex.
Usually a standard organ-type keyboard controls the oscillator pitch (although there are many alternate controllers other than the keyboard). However, just being able to control the pitch of an oscillator doesn't give us a very musical sound - we at least need some way to gate the sound on and off. The simplest option is to simply hook up an amplifier after the oscillator (Figure 3). When the key goes down, the gain is high and input signals are passed to the output; when the key is up, the gain is zero, so nothing makes it past the amplifier.
This is a start, but we still need more control. The only instrument that cuts on and off like this is the organ; some instruments decay over time (percussion, drums, plucked strings, etc), while some require a bit of attack time to reach full volume (like woodwind, brass, and some bowed instruments). What we need is some way to alter gain over time.
One option would be to simply turn a knob that varies the gain of the output amplifier. Unfortunately, this method is neither fast nor repeatable, and requires that you use one of your hands on the knob instead of the keyboard. A much better option is to use a Voltage-Controlled Amplifier (VCA) whose gain responds to a voltage applied to a special control terminal (more voltage gives higher gain, less voltage gives lower gain). Then all we need is a device that, in response to pressing a keyboard key, generates a series of voltage changes that alter the VCA gain exactly as we want. If the voltage ramps up slowly from full off to full on, the gain will increase slowly over time and produce an attack effect. If the voltage goes from full on to full off, we'll hear a decay.
The circuit that generates this voltage change is called the envelope generator (EG). Normally, it produces no voltage; but as soon as you strike a key, this circuit leaps into action and generates a varying voltage to control the VCA. The most common type of envelope generator is the AD5R, so named because there are four voltage stages called Attack, Decay, Sustain, and Release. Figure 4 shows the effect and timing of the various envelope parameters; refer to these as you read the following descriptions of the four basic parameters.
Pressing a key initiates the attack phase, where the envelope output goes from full off to full on. This can take anywhere from a few milliseconds to a second or more, depending on the attack time control setting.
After reaching maximum level, the decay phase kicks in. This determines how long it will take for the envelope to decay from its highest level down to the sustain level. The sustain setting is not a time, but rather, a voltage level (corresponding to an amplitude/volume level) that is maintained for as long as the key is held down. Setting the sustain to zero gives a signal that attacks and then decays down to nothing; this is called an AD (for the two stages: Attack and Decay) type of response.
Releasing the key initiates, not surprisingly, the release phase. This is the time it takes for the envelope to go from the sustain level back to zero.
The four-stage ADSR is not the only type of envelope generator; newer technology has brought us the multi-stage Rate/Level type of envelope. This envelope type offers more than just four stages, and while the sustain parameter is the same as that in the ADSR, attacks and decays are created by specifying a level you want the envelope to attain, and the rate of time it takes to reach that level. Figure 5 shows how using a six-stage version of this type of envelope can give a 'double attack' by setting a moderately fast rate to the maximum level of the first stage; a slower rate to a lower level in the second; a fast rate to almost the maximum in the third; a slower rate to the sustain level; and finally, a slow rate from the sustain level back to zero. The end may be specified as an End level, or simply, a level setting of 00, depending on the make of synth. While more complex, these types of envelopes offer greater flexibility than the 'old standby' ADSR.
By the way, many analogue synth functions - envelopes, VCAs, and so forth - are now implemented digitally (often through software programmed into a computer) instead of with dedicated integrated circuits. However, from a user standpoint, an envelope is an envelope, no matter how it was generated. These days, an 'analogue synth' may well consist of mostly digital circuits, but since it produces the same sounds and uses the same principles of operation as traditional analogue synthesizers, it requires similar programming techniques and can be treated similarly from a conceptual standpoint.
There's more to sound than dynamics: harmonic structure for example. A lowpass filter can vary the harmonic content of a signal by blocking all audio energy above a specified cut-off frequency. Setting a high cut-off frequency means that most of the high frequencies will remain, but setting a low cut-off frequency will dull the sound considerably by removing harmonics. Although static filter settings are useful, it's often more desirable to vary the filter setting over time. We use the same approach as with the VCA - make the filter cut-off respond to a voltage, and use an envelope to vary (control) the cut-off frequency.
The typical filter will have controls for the initial cut-off frequency and envelope depth. Envelope depth determines the range over which the filter will be swept (the greater the depth, the greater the sweep range), while the initial cut-off control determines at what frequency the sweep will start. So these two controls interact; if you want to sweep a specific frequency range, first set the lowest frequency with the initial cut-off control, then increase the depth until the envelope sweeps the desired maximum range.
Some filters allow you to invert the polarity of the envelope so that the filter sweeps downward instead of upward (Figure 6). In this case, the initial cut-off frequency determines the upper limit of the sweep, while the depth control determines how far down the cut-off frequency will be swept. Many people find this mode confusing, but just remember - if you're using an inverted envelope, then use the initial cut-off control to set the upper limit of the sweep, not the lower.
Note, by the way, that filter and VCA envelope settings interact. For example, if the VCA is set for an extremely short decay, then you won't hear the effects of setting a long filter decay because the volume of the sound will reach zero before the filter decay is complete.
Many filters also have a tracking control. This oft-misunderstood control is very useful; it varies the filter cut-off frequency with respect to which notes are being played on the keyboard. Usually it is referenced to a setting of 1.00, where the filter cut-off precisely tracks the keyboard pitch. Thus, if you have a certain harmonic structure when you play one key, playing a different key will shift the filter frequency in order to maintain the same harmonic structure. This generally gives the most realistic synthesizer and instrument sounds. With 0.00 tracking, the filter will not be affected at all by keyboard pitch. This is useful for bass patches where you want the lower notes to have more highs, and the high notes to become progressively more muted. With tracking settings higher than 1.00, the filter cut-off will increase at a faster rate than pitch as you play up the keyboard. For example, with a tracking setting of 2.00, increasing the pitch by one octave will raise the filter's cut-off frequency by two octaves. This technique comes in handy when you want upper keyboard register leads that really 'cut' while the lower registers sound more muted and sedate.
The lowpass filter is only one member of the filter family, but it is the mainstay of virtually all synthesizers. One exception is the Oberheim Xpander, whose multi-mode filter offers 15 different filter types, including several variations on the traditional lowpass.
Synthesizer keyboards used to be a series of simple on-off switches. The action was therefore like playing an organ, which limited the synth's dynamic capabilities. Fortunately, many keyboards now boast velocity- and pressure-sensitivity. Velocity measures the time it takes for a key to go from the up position to the down position, and (usually) translates shorter times into louder dynamics. Pressure (also called Aftertouch) measures the pressure exerted on a key after it has already been pressed down. This is great when you want to introduce vibrato, pitch bend, or other types of expressive effects by just pressing on the keys. A few keyboards offer polyphonic pressure, where pressure is measured for each key; but most keyboards simply take an average pressure reading of all keys that are down and use that to provide the pressure data.
We've almost completed our tour of the synthesizer, but we need to consider the Low Frequency Oscillator (LFO). This module provides cyclic (periodic) modulation of synthesizer parameters such as pitch, volume, or filter cut-off. Applying a periodic modulating signal to the oscillator (pitch) creates vibrato; this is deemed such an important application that virtually all synthesizers include a modulation wheel designed for real-time manipulation that adds vibrato, as desired, under the player's control. Applying modulation to the VCA (amplitude) produces tremolo; modulating the filter cut-off with an LFO signal produces a 'waa-waa' effect or, if used subtly in the higher registers, a shimmering type of sound. Using an LFO can create a more animated sound, particularly if it is used to vary the pulse width (duty cycle) of a pulse wave. This produces a sound that is very similar to flanging, even with single-oscillator synthesizers.
A popular LFO feature is LFO delay, where modulation is not introduced immediately upon playing a key but rather, only after a programmable amount of time has elapsed (Figure 7). This provides a more natural modulation effect, as most musicians do not introduce vibrato immediately upon playing a note.
Adding all these elements together - an oscillator or two to create sounds, a filter to vary harmonic structure, a VCA to vary dynamics, an envelope generator or two, and an LFO or other modulation circuit-produces a single note we refer to as a voice. Early synths, like the famous Minimoog, could only play one note at a time and were called monophonic instruments. Newer synthesizers can play several notes at once and are therefore considered polyphonic, these synths duplicate the modules listed above for each voice (well, some might only have one LFO for all voices, but we're quibbling here). As you can probably figure out, more voices means more circuitry and more expense but also enables you to play more notes simultaneously. Most analogue synths are eight-voice machines, although some of the lower-priced models may only have six voices or so.
Of course, there are many other synthesizer 'bells and whistles' unique to individual synthesizers - too many to go into here, in fact. That's why owner's manuals are well worth reading; it pays to discover what particular little goodies are built into your synth, since taking advantage of these will enable you to create a more individualised sound. Still, even with just the basic elements mentioned above, an analogue synthesizer can generate a stunning array of warm, expressive sounds. Experiment, and your efforts will be well-rewarded.
Craig Anderton is the author of many books on the subject of MIDI, synthesizers, and home recording. He is also Editor of America's leading hi-tech music magazine Electronic Musician.
© 1987 Electronic Musician magazine (Contact Details).