Roland shipped the Juno-6 in 1982 with one oscillator per voice — a cost-cutting decision that should have made it sound thin next to the two-oscillator Prophets and Oberheims of the day. It didn't, and the reason is a circuit most owners never thought about: the chorus. Press one plastic button and a mono synth voice blooms into a wide, swimming stereo field. That button is arguably why the Juno family became one of the most recorded synthesizers ever made — and the circuit behind it is charmingly, audibly imperfect.
The Juno chorus is a bucket-brigade device (BBD) design: two analog delay lines built from Matsushita MN3009 chips — 256 tiny capacitor "buckets" each, passing the signal along like a fire brigade — clocked around 70 kHz by an MN3101 driver. One triangle-wave LFO modulates both delay lines, sweeping them a few milliseconds back and forth. The trick that makes it wide: the right channel's modulation is inverted. When the left delay stretches, the right shrinks. Your ears read the anti-phase motion as space.
What the record actually says
The numbers usually quoted for the Juno chorus come from the service notes — and the most quantitative reverse-engineering of the actual Juno-60 schematic disagrees with them in one famous spot:
| Commonly claimed | What the schematic analysis shows |
|---|---|
| Chorus I+II ("both buttons") is about 1 Hz, per the service notes label | Roughly 9.75 Hz — a fast, shimmering, almost Leslie-like vibrato. The "1 Hz" in the service notes appears to be a mislabel |
| The modes just get "deeper" as you add buttons | I and II sweep wide (~1.7–5.4 ms of delay); I+II sweeps shallow (~0.4 ms peak-to-peak) but fast — a different effect entirely, not a bigger one |
| BBD chorus units all use companders to hide noise | The Juno's chorus has no compander stage — the noise you hear is the design |
Modes I and II are measured at about 0.513 Hz and 0.863 Hz — so the service notes are close there. But that third mode is the one that trips up recreations: build "I+II" as a slightly-faster deep chorus and you get something the hardware never did.
Why it sounds like that
The Juno chorus's character is three "flaws" doing load-bearing work. The 256-stage BBD line and the 12 dB/octave low-pass filter ahead of it (there to manage clock noise) shave the top end off the wet signal — the chorused voice is always slightly darker than the dry one, which keeps the effect lush instead of glassy. The missing compander means real, audible hiss and clock bleed; players learned to hear that noise floor as "analog warmth," and a silent Juno chorus simply reads as wrong. And the anti-phase stereo modulation means the image never sits still — the sound isn't in a place, it's in motion between places, which is why a single sawtooth through it feels orchestral.
It's a compact lesson in why "cleaner" doesn't win: Roland's engineers were managing cost and noise constraints, and every compromise landed somewhere musical.
What to ask next
- The Juno-6, -60, and -106 share the chorus topology but not identical boards — do the three generations chorus identically, or is the 106's slightly different noise behavior audible?
- Boss's CE-1 pedal and the Juno chorus are cousins from the same corporate family — what did Roland change moving the concept from pedal to synth board?
- If the noise is part of the sound, how much hiss is right? (Restorers who replace failing MN3009s report the repaired units sounding subtly different — quieter, and less loved.)
Sources
- pendragon-andyh, Juno-60 chorus reverse-engineering from the Juno-60 schematic (measured rates, delay ranges, phase behavior)
- Roland Juno-60 Service Notes (hi-res scan)
- Matsushita MN3009/MN3101 datasheet references (256-stage BBD, clock driver)
- Plasma Music, "Chorus chip transplant" (chip identification across Juno-6/60/106)
- Gearspace, "Juno-60 noisy chorus (repaired)" (noise as characteristic behavior)