By definition, a synthesiser produces music by artificial methods, usually by means of electronics. A device of this kind often allows a composer to create sounds that could be made by an unknown form of instrument, either simulating a new ‘acoustic’ device or something entirely different. In addition, it can provide a performance that would be physically impossible for a human musician to play.
An electric or electroacoustic instrument, on the other hand, such as an electric guitar or electric violin, is really a specialised acoustic instrument that uses electrical amplification to make it audible. Oddly enough, electronic effects devices have almost made the electric guitar into a synthesiser. And older electrophonic devices, such as the Compton organ, created sounds using electromagnetic or other kinds of tone generator rather than electronics. They often imitated the traditional pipe organ and were invariably controlled by some kind of keyboard. Examples include the reed organ and the Hammond organ, the latter employing toothed metal wheels and electromagnetic pickups to create the tones.
Modern electronic instruments need electrical energy to create sounds: they don’t use mechanical parts. In fact, the sound produced by such an instrument doesn’t exist until it’s fed into a loudspeaker. However, the traditional pipe organ and the modern synthesiser still have much in common. They’re usually controlled by a piano-style keyboard, are polyphonic (meaning they can play several notes at once) and multitimbral (they can play several different kinds of sound or ‘voice’ at a time).
Nineteenth-century electrical instruments include the Componium, the Electroharmonic Piano and the Electromechanical Piano. The Dynamophone or Telharmonium of 1906, devised by Thaddeus Cahill, used alternating current (AC) dynamos to create its sounds and weighed an amazing 200 tons.
The earliest electronic synthesisers used analogue circuitry, as opposed to modern digital technology. One example is the Theremin (or Thereminovox), invented sometime between 1919 and 1924, and finally developed in the US in 1928 by the Russian, Leon Theremin (b. 1896). The pitch, covering a range of five octaves, was controlled by moving the right hand around a pole, whilst the volume was adjusted by waving the left hand near a loop aerial. The device itself contained two radio frequency (RF) oscillators, closely tuned to each other, which then created a string-like whine. The movements of the operator’s hands influenced the tuned circuits of the oscillators, changing the sound’s pitch and volume. The instrument is shown below being played by its inventor at Carnegie Hall.
Having appeared in some music halls, this instrument was soon forgotten, although it was used by the Beach Boys on their record Good Vibrations and by the rock band known as Led Zeppelin.
Other early electronic devices included the Ondes Martenot, devised in 1928 by Maurice Martenot (1898-1980), and the Trautonium. Unfortunately for a keyboard instrument, the Martenot could only play one note at a time, a characteristic common to many early devices of this kind.
Electronics found its first challenge in the organ, where the traditional pipes were replaced by electronic counterparts. In the most advanced electronic organs, separate oscillators were used for each note, with complex keyboard logic to ‘sound’ the appropriate oscillators for each key and stop.
The domestic electronic organ side-stepped these complications by using a set of simple square wave oscillators for the top octave, with frequency dividers for the lower octaves. The results, particularly at the extremes of the keyboard (where the limitations of output filtering became obvious), were often unsatisfactory. Unfortunately, a decent instrument could never be produced at a realistic price until the appearance of the microprocessor. But, unlike many analogue synthesisers, the electronic organ was fully polyphonic: you could press all the keys you wanted, and get all the sounds.
The schematic for this kind of instrument is shown below:-
The Hammond organ, really an electrophonic device, succeeded where many others failed. It incorporated a set of motorised metal tone wheels, each with teeth machined to give the correct pitch, and positioned near a magnetic pickup that extracted the signal. The result was a simple and reliable instrument, with a very characteristic sound, enhanced by its rotating ‘Leslie’ loudspeaker.
The BBC Radiophonic Workshop created its own unique ‘organ’ from a rack of ordinary test-room oscillators, a small keyboard and a keying unit. As each key was pressed a pentode valve in the keying unit connected an oscillator to the output: each valve had adjustable timing for attack and decay. Limited perhaps, but few instruments today would allow you to tune any key to any pitch.
By 1952, electronic music could be created by purely electronic methods. This required the use of sine wave tone generators, which produced a pure sound at a given fundamental frequency, noise generators, often producing white noise with a wide spectrum of frequencies, and filters, which removed anything above or below a given frequency. Such material was usually manipulated by means of variable-speed tape recorders. The Olsen-Belar Sound Synthesiser, developed in 1955 by the Radio Corporation of America (RCA), contained oscillators that produced waveforms that included other harmonics. The use of such material with real sounds is sometimes known as electroacoustic music, although strictly speaking this refers to an acoustic instrument that’s assisted electronically.
The voltage-controlled synthesiser, as described by Robert Moog (b. 1934) in his 1958 paper to the Audio Engineering Society (AES), combined elements of these earlier devices in a new instrument. This generated audio waveforms scientifically, shaping and modifying them into the required sounds. Pioneering users included Walter (later Wendy) Carlos, who horrified classical musicians with Switched on Bach, although subsequent work, such as Timesteps, showed that the instrument could create truly original material. And recordings such as Lucky Man (Emerson, Lake and Palmer) and The Six Wives of Henry VIII (Rick Wakeman) introduced the synthesiser into rock music.
Moog’s invention consisted of simple elements that influenced each other by control voltages. How they were used and connected was left to the composer. Musically, the results were often awful: considerable skill and artistry was needed to rise above the mundane. Worse still, since analogue voltages determined the pitch of the oscillators, there were problems with tuning, often due to temperature changes as the equipment warmed up. Despite this, the device remained a dominant force into the age of the microprocessor. In fact, such machines still exist and are even fitted with MIDI.
The main elements, as typically connected, are shown below:-
This was usually monophonic, meaning only one note could be played at a time. However, some synths had dual keyboards or used special electronics to give duophonic operation. The control voltage (CV) produced by the keyboard represented the pitch, usually based on a scale of one volt per octave, 0.5 volt per octave or one millivolt per hertz (mV/Hz). CV signals weren’t standardised. For example, zero volts could correspond to the bottom note (higher keys giving a positive voltage) or the centre note (keys to each side going positive or negative). The latter was convenient for tuning, since zero volts then coincided with middle C or A, and gave linear results across the keyboard. The pitch of a decaying note was kept in a ‘sample and hold’ circuit until the next key was pressed. The keyboard also produced a gate signal (positive 5, 12 or 15 volts) to show that a note had been pressed.
Some keyboards also generated a trigger pulse at the instant a key was pressed. In some instruments, such as the ARP Odyssey, a small change in CV also created a trigger pulse. A second note could then be played whilst another was held (for playing in legato), providing a form of duophonic operation.
The following elements, in various numbers, could be found in most synthesisers:-
This device usually provided square, sine and triangular audio waveforms, often with a variable mark-to-space ratio. The initial pitch of each VCO had to be adjusted, also the sensitivity to voltage control, ensuring satisfactory tuning across the entire keyboard. An extra control voltage input was often used in conjunction with a low-frequency oscillator (LFO) to provide vibrato.
This was used for ‘tuning’ the frequency spectrum of any sound (often the output of the synthesiser’s white noise generator) under control of the keyboard. Each VCF usually had controls for the initial frequency and filter quality (Q), both of which were usually voltage-controlled. Some VCFs could also oscillate, allowing them to be used as extra VCOs.
This device accepted a gate signal, usually from the keyboard, and used it to create a varying control voltage. ADSR shapers had controls for attack (the initial rise), decay (the fall as a key is held), sustain (the level as a key is held) and release (the fall after a key is lifted). An ADR shaper was a simpler version with three parameters. Most shapers were fully voltage-controlled.
This was often used to regulate the volume of the output of a VCO, under the control of an envelope shaper. VCAs worked at 10 to 20 decibels per volt (dB/V), although the polarity varied. With a zero voltage at the control input, most VCAs were at their ‘initial’ gain, usually between 0 dB and +10 dB.
This simple circuit could be used to retain a control voltage after a note on the keyboard had been released, thereby ‘memorising’ the pitch of any note that required a long decay. Its inputs were usually connected to the keyboard’s CV and gate (or trigger) outputs.
This provided random noise over wide spectrum of frequencies. Its output could be fed into a VCF to create impulsive or atmospheric sounds.
The control voltages created by this component could introduce an element of chaos into any work. The voltage swing and frequency was usually set by ‘Range’ and ‘Variance’ controls.
A trapezoid could be used to slow down rapidly changing signals, whilst inverters would reverse control voltages. In addition, other voltage modifiers could introduce control offsets or level changes.
Various connection methods were used, including 3.5 mm jack cords, popular on American and Japanese products. However, Electronic Music Systems (EMS), who took the analogue synthesiser to its zenith, employed patch-pins on matrix boards. The latter had virtual earth inputs, whilst the patch-pins were fitted with resistors. This allowed any number of sources to feed a single destination, with the value of a patch-pin’s resistor (denoted by the colour of the pin) setting the sensitivity of an input.
The voltage-controlled synthesiser offered more than anything that came before, but was difficult to use. Larger machines were expensive and clumsy, the best results often being obtained by accident. The arrival of the microprocessor was to see both its final development and demise.
At the end of the seventies, the speed of most processors was inadequate for creating or modifying real sounds. Instead, new synthesisers employed such a device to act as a ‘front end’ for conventional voltage-controlled technology. One early machine of this type, the Yamaha CS80, looked like a large electronic organ, but was really processor-controlled. Whenever you pressed a note, that key would be assigned to a specific voltage-controlled oscillator, so providing true polyphonic operation.
Other synthesisers in this category must include the famous Prophet 5 from Sequential Circuits and the Oberheim OBX8. These hybrid machines came as a shock. In appearance they were unlike any ‘classic’ synthesiser, but behind that new front an analogue machine was lurking.
Most hybrid machines used a single microprocessor (micro) for all the tasks, although some instruments employed a separate processor, often a customised device, to provide polyphonic key assignment. A typical schematic is shown below:-
In most instances, the operational switches and keyboard contacts were connected to a parallel port on the microprocessor via a scanning ‘XY’ matrix. The analogue signals from the variable controls, including bend wheels and swell pedals, were multiplexed and sent through a digital to analogue converter (DAC) to the micro. From the micro, switching and control data was sent to an analogue to digital converter (ADC) and on to a demultiplexer that directed signals to an appropriate destination.
Most machines kept the micro’s operating system (OS) and preset voices in read-only memory (ROM) whilst user data was kept in battery-backed random-access memory (RAM). Also, user information could be dumped to or loaded from an audio cassette recorder, although this was often unreliable.
One very important job for the micro was key assignment, the process of choosing which oscillator and envelope shaper was operated by which key. Typically, pressing one key selected, at random, any one of eight oscillators. Pressing another key simultaneously ‘fired’ the next oscillator, and so on, until the maximum number of oscillators, usually eight, were sounding. If another key was pressed (and all the other oscillators were operating, perhaps with their envelope level decaying) the very first oscillator had to be ‘grabbed’ for the new key, possibly truncating the first note.
Tuning such a machine was often very difficult. Most oscillators (once you identified the one that was in operation) had at least three adjustments, and an iterative process was required to give good results across the keyboard. Typically, tuning a Yamaha CS80 would take two or three hours.
The high-speed processor of the eighties could manipulate audio data directly, and so removed the need for analogue circuitry overnight. Robert Moog’s dream had come to an untimely (or timely) end.
Wavetable synthesisers, such as the PPG Wave, kept one complete cycle of each sound in RAM or ROM. When a key was pressed, the data was clocked out into a DAC at the desired rate. The contents of a wavetable could be modified, or parts of different wavetables could be joined to create new tables. The Wave also sampled real sounds that could then be edited and looped. Unlike earlier synthesisers, it included floppy disk drives and a visual display unit (VDU) for graphical editing.
The Fairlight Computer Musical Instrument (CMI) was far more advanced. In fact, this synthesiser was a specialised variety of mini-computer, complete with a graphical display controlled via a light pen. In its final form, this machine incorporated real-time audio recording on eight ‘tracks’.
These two machines were large, powerful and expensive. Moreover, they needed considerable operational skills to fully exploit their potential. The products that followed were completely different. The revolution in consumer musical products had begun.
Digital synthesisers of the later generation often contain customised integrated circuits. Even so, the typical schematic, as shown below, is similar in many ways to a desktop computer:-
Almost invariably, such machines use two processors. The main device works with those elements common to any computer, such as RAM, ROM, disk drives, serial ports, MIDI ports and VDU. The sound processor, which needs to be very fast, is often a custom chip. This provides digital signal processing (DSP) and often has its own RAM and ROM. It can connect to the real world via an ADC or a digital interface (DIF) and may employ a separate hard disk drive to store sound samples.
One early example of DSP is frequency modulation (FM), which first appeared in Yamaha’s DX7 and its successors. The process of FM between two or more tones, known as operators, is simulated by digital computation to create complex sounds, often very musical and penetrating. The results seem to have little in common with the actual operators, making it difficult to create sounds ‘from scratch’. But Yamaha satisfied most of its customers by supplying the machine with an extensive range of preset sounds. It also accepted RAM and ROM cartridges that expanded the choice still further.
Many digital machines, often using highly innovative ideas, have followed on from these pioneers. But perhaps the greatest advance has been the increase in ease of use for the musician.
By incorporating an analogue to digital converter (ADC), the digital synthesiser effectively became an audio sampling machine. The E-mu Systems Emulator II was one of the first machines of this kind. Although it only used 12-bit technology, the sound quality was very impressive. Its place in ancient history is confirmed by the fact that it used two cumbersome 5¼ inch floppy disk drives.
The Roland S50, with its provision for a VDU, was a pioneering keyboard machine that kept samples on a 3½ inch floppy disk. This was followed by the rack-mounted S550, also giving multiple outputs. However, the ‘industry standard’ for many years in the world of sampling was the Akai S1000.
A new generation of machines soon appeared, with samples that matched the quality of digital audio. They included the E-mu Proteus and Procussion, both playback-only samplers containing a vast repertoire of instruments, showing a high technical quality and sampling artistry. Also, they exploited MIDI to the full, allowing sixteen sounds to be played at any time, each with sixteen-note polyphony.
Today, many older synthesisers, including those ancient analogue devices, can be emulated by using suitable software on a fast desktop computer. Often, these new applications don’t require special hardware, apart from standard audio inputs and outputs. For those who want to go further, synthesisers, samplers and digital audio recording can be installed as plug-in PCI cards. As processor speeds continue to increase, even these cards are likely to be replaced by more advanced software. The days of the synthesiser are over: we’re now in the world of ‘desktop composing’.
©Ray White 2001.