Yamaha DX7: FM synthesis and the sound of the 1980s

How FM synthesis works

An analog synthesizer starts with a harmonically rich wave and filters some of it away. Frequency modulation (FM) works the other way round: it starts with plain sine waves and lets one change the frequency of another. The DX7 calls each sine-wave source an operator — an oscillator with its own envelope. An operator that is heard is a carrier; one that only modulates another is a modulator.

When the modulator's frequency is a simple multiple of the carrier's, the result is harmonic, like a string or brass tone; other ratios give inharmonic, bell-like and metallic sounds. Turning up the modulator's level adds more and more overtones, and because each operator has its own envelope, the brightness can change through a note — which is how FM imitates the sharp attack of a struck or plucked instrument. The DX7 has six operators per voice and 32 algorithms, each a different arrangement of carriers and modulators; algorithm 32 makes all six carriers, which is additive synthesis rather than FM.

From a university lab to a best-seller

FM synthesis was developed at Stanford University in California in the 1960s; in 1971 it was used to imitate organs and brass to show its commercial potential. Stanford patented the method, but American companies including Hammond and Wurlitzer turned down a licence. Yamaha, then the world's largest maker of musical instruments, licensed it for a year in 1973 to build a prototype and secured exclusive rights in 1975.

Yamaha's first commercial FM synthesizer, the GS1 of 1980, needed some 50 chips and was expensive to build. Yamaha's new very-large-scale integration (VLSI) chips cut that to two in the DX7 — the YM2128 operator chip and the YM2129 envelope generator — and its implementation ran at a higher sampling rate than Stanford's systems. A prototype was shown in 1982 as the CSDX. Two specialists then had less than four days to create the DX7's 128 preset sounds.

DX7 specifications

Compared with the warm sound of analog synthesizers, the DX7 sounds bright, glassy and clean, with complex attacks. It was praised for convincing tubular bells, metallophones and harpsichord, and it took over the electric piano market.

Presets, sales and the E Piano sound

Yamaha had hoped to sell more than 20,000 DX7s. Within a year orders exceeded 150,000, and 200,000 were sold in three years — the first synthesizer to pass 100,000 units, made on a scale its American competitors could not match. A whole industry grew up selling cartridges of extra sounds.

With menus on a small display and no knobs or sliders for the sound, few owners learned to program it in depth, so the factory presets were everywhere. The E Piano 1 preset, an imitation of the Rhodes piano, was used in 40% of the US number-one singles of 1986 — and 40% of country and 60% of R&B number ones — and some players gave up their Rhodes for it. The "Bass 1" and "Bass 2" presets were almost as familiar. In later years the sounds came to be seen as dated, and second-hand digital synthesizers sold for less than analog ones, until software FM revived interest.

Successors and recreations

Because the DX7 is digital, its sound can be recreated accurately in software — free emulations such as Dexed load original DX7 patches — without the complex modelling analog instruments need.

Using a vintage DX7

1. Back up the memory. The 32 internal patches live in battery-backed memory. Save them to a cartridge or send them over MIDI as system-exclusive (SysEx) data — the same format the cartridges use — before the session.
2. Program from a computer. Editing on the panel is slow; a software editor or an emulation that reads DX7 patches makes it far easier to shape sounds.
3. Use the expression inputs. Velocity, aftertouch, foot controllers and a breath controller can all be routed to operator levels and pitch, which brings the presets to life.
4. No filter to sweep. Brightness comes from modulator levels and envelopes, so a "filter sweep" means changing those.