Since it’s a matter of occasional confusion, even amongst people doing synth repairs for a living, I thought I might try a simple description of how divide-down polyphony in analogue instruments works.

The basis of the technology is a circuit called a frequency divider. ^[1] This can be done differently with digital manipulation, but a typical analogue frequency divider responds to a (significant enough) change in input voltage by waiting until a similar change recurs one or more times before changing its output. The simplest and commonest division is the one-half or first suboctave, where the count is every two cycles. (A binary divider. n.b. divider circuits are often also known as counters.)

In electronic instruments, input to dividers typically comes from an oscillator circuit. Originally, each of the notes in the top octave of divide-down instruments were generated by a separate individually tuned oscillator. The collection of oscillators this required is usually referred to as an oscillator bank.

This is quite different from the wide-spectrum Voltage-Controlled Oscillators (VCOs) used in monophonic synthesisers. (For today’s purposes VCOs will represent all controlled oscillator types. ^[2]) Once tuned, a bank of oscillators may not change pitch at all, though in some designs they can be modulated to produce some degree of vibrato or pitch bend. But VCOs are designed specifically to change their pitch over multiple octaves under control by an external voltage source e.g. a keyboard. ^[3]

Binary dividers can be chained to produce successively lower octaves:

And a complete working electronic instrument can be built by connecting a divider chain to each of the top octave oscillators:

In early electronic instruments, the divider circuits were constructed from discrete components. With the advent of integrated circuits these were speedily replaced in most new designs with multi-stage divider chips such as the RCA 4024 in the picture above.

The division count does not have to be two. For example, if the count is three, the third subharmonic will be produced. In principle, it can be any number. In the early 1970s, more complex integrated circuits were designed to generate something close to a chromatic scale over one octave from a single high frequency input, using higher counts.

However, as the chromatic scale is not based on harmonic intervals, but divisions produces subharmonics, divider circuits are never precisely chromatic. For multi-tone divider ICs, the input frequency is usually in the hundreds of thousands or even millions of Hertz to allow the outputs to approximate the desired frequency.

Effectively this is an electronic version of the electromechanical frequency division of the tonewheel organ, e.g. the Telharmonium or original Hammond organs, which had similar inaccuracies.

Whatever the generation method, there is no limit in principle to how many lower octaves can be generated, though in practice the lower limit of human hearing limits their usefulness. But some electronic instruments do go lower, as did some acoustic pipe organs before them; the lower pitches have uses including modulation and bass reinforcement.

It is important to note when dealing with ‘divide-down’ systems that what is being divided, arithmetically, is frequency, not wavelength. This seems to be a frequent point of confusion. Division produces subharmonics, not overtones. In other words, you don’t produce higher notes with dividers, only lower ones. This comes from the fact that divider circuits are basically counters, changing their output state when a set number of inputs has occurred. To divide a wavelength you need to know the length of the wave in advance so you could switch the output before the next input was in fact counted. (There are circuits which can approximate this but only after a delay while the input waveform is measured. They have other uses.)

Whereas wide-spectrum VCOs often begin with a sawtooth, ^[7] in divide-down instruments, typically all the signals in the top octave and divider chains will be square waves. The signals being generated here may not themselves the audio signal; they may be used as a timing signal to determine the pitch of one or more tone generation circuits, or they may be passed through a waveshaping circuit to generate additional waveforms (this is more likely to be a feature of synthesisers). But in some instruments these original square waves are the only basic tone signals.

Irrespective of the original generation method, divide-down systems produce all pitches all the time, i.e. you have a fully polyphonic set of pitches continuously available. (Whereas a VCO has a single varying pitch.) Thereafter each pitch is commonly passed separately through an envelope circuit triggered by its specific key contact, and into one or more filters, and usually one or more output amplifiers, to generate polyphonic voices.

These envelope circuits are actual envelopes, which directly set the loudness profile of the tone signal for each note. (Whereas in synthesisers, “envelope generators” (EGs) are usually modulation signal generators which control independent VCAs and may also be used to control other sections of the instrument, including VCOs and VCFs.)

This approach works for these instruments because they do not vary the tone of the individual note, only its amplitude. They can get away with being paraphonic — feeding all notes into one filter for each voice. But synthesisers are instruments which can vary tone over the course of each note, so they have a problem here. How do you allow a tone source producing all possible pitches to only pass the notes played into the next stage, if you don’t have an individual fixed envelope for each note on the keyboard?

In some simpler divide-down synthesiser designs the individual notes from a divide-down network are sent directly to key contacts, and connected to a common bus for output. This works where you have only one filter. ^[9]

In more complex designs, each note can have its own VCF circuit, though they are likely to have at least some controls in common. These were rare and rather expensive synthesisers.

For comparison, monosynths, and polysynths with multiple independent voice circuits which are switched between by a key assigner, can route each VCO into a VCF and then a VCA independently, under control of one or more envelope generators.

In this sense the only necessary differences between basic organs and synthesisers are the pitch generation method, and the synthesiser’s independent envelope, able to control several stages of sound production. However, there can be many variations on the above routes in both divide-down and VCO-based instruments. And this is complicated by the point that the majority of divide-down synthesisers appear to have been multi-section instruments which function both as organs and synthesisers, perhaps with a shared tone source.

Later DD polyphonic instruments allowed the possibility of modulating pitch. This was expensive with oscillator banks, but became more common with the use of single-oscillators with divider chips, as only one oscillator pitch modulation circuit is needed.. However, this still does not permit individual single-note pitch modulation, which had to wait until duophonic and eventually more polyphonic synthesisers came along. This does require a separate oscillator for each note that could be changed. For example, a Roland Juno 6 (1982) has six separate complete voices, each with its own pitch oscillator, which can be modulated independently by its envelope circuit, triggered in turn by a key assigner. This in some ways is far from the full polyphony of organs and some earlier synthesisers, but it allows some specific kinds of detail in the sound which was not possible with the earlier DD instruments.

Comment or Question about this page? write

Article text ©2025 Electropict .
Click images for individual licences.