Roland Jupiter-4, Jupiter-6 and Jupiter-8 compared

The three Jupiters side by side

None of them responds to velocity or aftertouch. All three use digital control of analog circuits, which Roland called compuphonic on the Jupiter-4: a microprocessor scans the panel and generates the control voltages, which is what makes memory, arpeggiators and voice assignment possible.

Jupiter-4 (1978): Roland's first polysynth

The Jupiter-4 was dramatically cheaper than rival polyphonic synthesizers of its day, and it was designed partly as an accompaniment to organs — an optional music stand could be fitted. Two NEC 8048 microcontrollers handle its digital features: an arpeggiator (up, down, up/down or random) that can follow an external clock such as the output of Roland's CR-78 drum machine, a hold button, polyphonic portamento, and four voice-assignment modes, including a unison mode that stacks all four oscillators on one key.

Each of its four voice cards holds a VCO, a sub-oscillator, a self-oscillating resonant low-pass filter and a VCA, with an ADSR envelope each for filter and amplifier. The LFO reaches audio frequencies for crude FM and AM effects, and a stereo chorus built on now-rare MN3004 delay chips widens the output. Roland built four revisions of the voice cards, using either BA662 or IR3109 filter chips, which changes the calibration procedure.

Its monophonic sibling, the Promars (1979), has two voice cards and 37 keys, without the arpeggiator or chorus.

Jupiter-8 (1981): the flagship

The Jupiter-8 was Roland's flagship for the first half of the 1980s; about 3,300 were made. Each voice has two discrete VCOs with cross-modulation and sync, pulse-width modulation, a non-resonant high-pass filter, a resonant low-pass filter switchable between two-pole and four-pole, an LFO and two envelope generators, one of which can be inverted. Every parameter has its own control on the panel.

• Split and layers: the keyboard divides into two zones with a different patch on each, and separate Upper and Lower outputs (balanced XLR and unbalanced quarter-inch).
• Polyphonic unison: all 16 oscillators can stack on one note and divide as more keys are played — no other polysynth of the time could do this.
• Fast envelopes: generated in hardware by Roland's IR3R01 chip, with a 1 ms attack, much quicker than the software envelopes of the later Jupiter-6.
• Brains: a Zilog Z80 processor stores patches, scans the panel and runs the auto-tune.

DCB (Digital Control Bus) is Roland's pre-MIDI interface; it lets the Jupiter-8 talk to DCB equipment such as Roland's MC-4 and MC-8 MicroComposer sequencers. The memory battery is soldered in and typically lasts ten years or more.

Jupiter-6 (1983): cheaper, different, and with MIDI

Introduced in January 1983 as a less expensive alternative, the Jupiter-6 is not simply a cut-down Jupiter-8. To lower costs Roland used Curtis integrated circuits (CEM3340 oscillators, CEM3360 amplifiers) instead of discrete circuits, which gives it its own character. In return it has a true multimode filter, lets each oscillator play several waveforms at once, offers unison detune, and splits its six voices 4/2 or 2/4 between two sounds.

It was among the first instruments with MIDI, alongside Roland's JX-3P and the Sequential Prophet-600. At the January 1983 Winter NAMM Show, MIDI was demonstrated by connecting a Prophet-600 to a Jupiter-6. A popular third-party firmware upgrade called Europa modernises its MIDI, memory and arpeggiator.

Successors and recreations

Which Jupiter for which job

• Need MIDI without extra boxes? Only the Jupiter-6 has it. A Jupiter-8 needs DCB-equipped gear or a converter; a Jupiter-4 has neither, but its arpeggiator can follow a clock.
• Big pads and splits: the Jupiter-8's eight voices, layered zones and separate outputs.
• Arpeggios on a budget: the Jupiter-4, but allow warm-up time and expect older tuning circuits.
• Early Jupiter-8? Check the serial number: units below 171700 have the 12-bit converter that some players blame for tuning problems.