Both are six voices, one DCO per voice, a 24dB low-pass filter, one ADSR, 61 unweighted keys and a bucket-brigade chorus, designed by the same people eighteen months apart. One costs roughly twice the other. The cheaper one contains a part Roland potted in black resin in 1984, and which is now failing in essentially every unit ever built.

That is the Juno 60 vs Juno 106 argument. Most of the detail repeated online either is wrong or credits a real difference to the wrong circuit. The chorus is where people say it lives. It mostly does not.

Juno 60 vs Juno 106, in one table

The filter module is the only architectural difference in the audio path worth caring about, and the chorus delay chips are the same part in both. Florian Anwander's survey of Roland chorus circuits lists 2 x MN3009 for the Juno-6, 60 and 106 alike, and SoundForce, who reverse-engineered it to build a clone, describe theirs as using "the same MN3009 audio path as the original Juno 6/60/106 circuit".

Juno 106 chorus vs Juno 60 chorus: is there really a difference?

Yes, but not in the part people blame. The delay line is identical silicon; what differs is the modulation driving it. Anwander has the Juno-6's combined I+II mode at 8Hz and 25 percent depth, while the Juno-60's I+II runs 1Hz at 8 percent: a slow shimmer, not a seasick warble. If you heard the "both buttons on" trick described as a fast wobble and then heard almost nothing on a 60, that is why.

The 106 is where sources fall out. Anwander states that on the 106 "there is no longer the possibility to set the mode I+II", and that Roland dropped the option to disable phase inversion on the second BBD path. Roland's own current documentation disagrees, modelling a mode described as "the effect produced when pressing both chorus button [I] and [II] on the JUNO-106 at the same time". What nobody who has measured these circuits reports is any difference in the BBD path itself.

Noise is age rather than model. MN3009s hiss as they age, on both, and Bell Tone Synth Works sells a noisy chorus chip replacement as a standalone USD 50 job on the 106. Hiss with nothing playing is two BBDs at end of life, and cheap. If it is the effect you want rather than the synth, a Dimension D works on everything else you own, and the Korg Polysix has its own chorus and six voices for less.

The filter question, and why the 60 sounds fatter

Both run a 24dB per octave low-pass built around the Roland IR3109, a four-stage OTA array. On the 60 it sits on the voice board as an ordinary IC beside BA662 OTAs doing resonance and amplitude. On the 106, Roland mounted one IR3109 and two BA662F as surface-mount parts on a ceramic carrier and dipped the lot in black resin. That is the 80017A, one per voice.

Electrically it is the same filter, so the difference comes from elsewhere. The best account on record is a 2008 circuit analysis on the Sequence 15 blog. The 106 adds an output buffer between the high-pass filter and the VCA, where the 60 couples them almost directly. So on the 60 the VCA loads down the passive filter, and as VCA gain falls that loading changes, giving a wide bandpass character and high-frequency rolloff. Single source, flagged as one.

The oscillators differ too. Electric Druid has the 60 clocking a digital counter from an analogue master oscillator around 1.9MHz, which lets global pitch bend move smoothly, while the 106 uses an 8MHz crystal into Intel 8253 timers. The 106 holds tune better. The 60 bends better.

Arpeggiator, DCB and MIDI

The 60 has an arpeggiator: up, up-and-down and down, over one to three octaves, clockable externally. The 106 has none, and for some buyers that omission ends the argument before the chip discussion starts.

DCB was Roland's pre-MIDI protocol from 1981, fitted to exactly two synthesizers: the Jupiter-8 and the Juno-60. It carries note on and off, program change and VCF/VCA control, nothing else. Period DCB sequencers exist and we have one, the JSQ-60, which plugs straight in with no converter. Otherwise DCB is worth something rather than much, because sequencing a 60 from a DAW means an MD-8 or a modern interface on top. The 106 needs nothing, and its MIDI was unusually complete for 1984, every slider transmitting and receiving over sysex. The Prophet-600 got MIDI first, in 1982, with its own problems.

Build, and the variants

The 60 is the bigger box, 1060mm and 12kg against 992mm and 10kg on Roland's own figures, which is chassis rather than features. It uses real panel switches where the 106 uses membranes, and membrane failure is a genuine 106 fault; a full front-panel service runs around USD 200. The keybed is the same on both, no velocity, no aftertouch, usual fault dirty contacts.

Two variants. The HS-60 Synth Plus and Juno-106S are the 106 with an amplifier and speakers, so the same voice chips, and usually cheaper. The Alpha Juno 1 and 2 of 1985 are a different instrument, driven by soft buttons and one dial.

Juno 106 voice chip failure: the 80017A

The 80017A is a module, not a chip. A teardown by Obsoletetechnology confirms the contents, one IR3109, two BA662, SMD capacitors and printed resistors on a ceramic substrate, and notes the printed resistors cannot be replaced. Rosen Sound, who sell a replacement, state that "the original 80017A from Roland has a 100 percent failure rate" and only sell them in sets of six. A vendor with an incentive, but no repair shop we found disputes the direction.

On the mechanism the sources conflict. ECH Repairs quotes the common account, that the resin becomes conductive over time, then adds, to its credit, that the explanation "seems a little vague". SynthTweaks puts the cause at cracked surface-mount solder joints under the resin, which is why acetone stripping is temporary. Bell Tone describes capacitance in the resin, dried joints and moisture in the printed resistors. Polynominal blames internal heat. Anyone who says it is settled has read one page.

What does a bad 80017A sound like?

Specific and recognisable. The symptoms reported consistently by Analogue Renaissance's test procedure, Syntaur and the repair blogs:

• Silence from one voice in the rotation.
• Crackle and pop on one voice, worse as the synth warms up.
• Wrong pitch, one voice out with the other five, or thin.
• A hanging note continuing quietly after key release, or shifting pitch as you play. That is a VCA stuck partly open.
• Continuous noise with nothing played, which is the same stuck VCA passing the filter output.

How to test one before you hand over money

The 106 assigns voices round-robin, so the fast version takes a minute. Play one note slowly, seven or eight times, listening for the strike in six that is quieter, duller, sharper or crackly. Repeat with resonance high, then leave the synth on twenty minutes and do it again, because heat is when marginal chips declare themselves.

The proper version uses test mode. Hold KEY TRANSPOSE, power on, release it, and the display shows the test indicator; POLY 1 and POLY 2 together gives rotary mode, where the display names which voice, 1 to 6, sounds each note. Syntaur's addition is the best single test on the machine: take the oscillators out of the mix, push resonance to full, and the filter should self-oscillate into a clean sine on every voice. A dying chip loses that first, so silence there is a diagnosis rather than a suspicion.

80017A replacement, and what it costs

Original 80017As are out of production and surviving stock is old resin with the same clock running, so nobody serious fits them. Almost everyone does all six clones at once, and that is not an upsell: same age, same batch, and the labour is in getting the boards out rather than in the sixth chip. Doing one and waiting buys the whole teardown again next year.

This is not a recapping job

Worth saying loudly, because listings conflate the two. The 80017A is not a capacitor problem. Electrolytics dry out, lose capacitance and leak, a separate issue on any forty-year-old machine, and we covered what that costs in our piece on whether vintage gear needs recapping. A recap does not fix a failing 80017A, and a chip set does not fix a tired power supply. "Fully recapped" tells you nothing about the voice chips, so ask which chips, by name, and who fitted them.

What goes wrong on a Juno-60

Less, but not nothing. The memory battery is first: a CR-1/3N lithium cell with solder tabs, and when it dies the patches go with it. Modern cells have no tabs, so the repair is a holder rather than solder onto a cell. Make the seller save and recall a patch in front of you.

Second, the logic and memory section corrodes. Repair threads trace erratic bank behaviour to corrosion on memory chip pins, driven by the atmosphere inside the case and the backup voltage. Looks like a dying synth, usually a cleaning job.

Third, the 60's voice parts do fail, one at a time: BA662s die, IR3109s occasionally. The difference is not that they are immortal, it is that a tech can source and swap them without buying a proprietary module. That is the reliability argument, and it is real.

Is the Juno 60 worth double?

Bands, hedged, because this market moves. In September 2026 Reverb's price guide put the Juno-60 midpoint around USD 2,485 and the 106 around USD 2,000, while a specs archive tracking eBay sales gave the 106 a working range of USD 800 to 1,400. That archive also claims the 106 has velocity and aftertouch, which it does not, so treat its numbers as indicative. Call it USD 2,000 to 3,200 for a 60 and USD 1,200 to 2,000 for a 106: roughly 1.6 to 2 times, not a clean double, and a different sum again landed in South Africa, which is its own arithmetic.

Here is what should change your bidding. A 106 with unknown or original voice chips has a USD 300 to 600 job attached and the job is coming, so deduct it. A 106 with six clones fitted by a named shop, with an invoice, is worth more by roughly the cost of the work, and it is the only version that is a settled purchase. A seller who cannot tell you which chips are in it has not looked, so price it as original.

Which Juno should I buy?

• You want the arpeggiator and you play it rather than sequence it. Juno-60, not close. The arpeggiator into its own chorus is why the machine holds its price.
• It lives in a DAW and you want six voices with no drama. Juno-106 with the chips already done: MIDI in, sliders out over sysex, tuning that does not wander, failure mode spent.
• First vintage polysynth, money tight. Juno-106, budgeting the chip set on day one rather than meeting it in month four. A 106 at the bottom of the band plus a set of clones still lands under a clean 60.
• You want the sound, not a forty-year-old polysynth. Neither. The Matrix-1000 gives six analogue voices in a rack for less, and if the part you want is fat single-line bass rather than pads, an SH-101 or an SH-2 does it better than either Juno did. The System-100 is where you go to move backwards instead of sideways.

What we could not verify

• The 80017A failure mechanism. Four credible repair sources give four different primary causes, and the widely repeated claim that moisture attacks the die or bond wires we could not trace to anyone who had opened a chip and documented it. Unsettled mechanism, settled outcome.
• Patch memory counts. Wikipedia gives the 60 56 patches and the 106 128; Roland's own archived spec pages imply 40 and 64.
• The 106's I+II chorus mode. Anwander says it was removed; Roland's current documentation models it as an existing mode.

We have both on the floor: a Juno-60 and a Juno-106. If you are looking at the 106, ask about the voice chips before anything else, because that is the question we would ask. The same discipline applies to anything with a service history, which is why the tape machine checklist exists too.

Sources

• Wikipedia: Juno-60
• Wikipedia: Juno-106
• Roland: Juno-60 specifications
• Roland: Juno-106 specifications
• Florian Anwander: Roland choruses
• SoundForce: uChorus 6
• Roland JX-08: JUNO-106 CHORUS
• AMSynths: the IR3109 chip
• Obsoletetechnology: 80017A teardown
• Sequence 15: why the Juno-60 sounds different
• Electric Druid: Roland Juno DCOs
• Analogue Renaissance: Juno-106 test procedure
• Syntaur: does your Juno-106 need voice chips
• SynthTweaks: 80017A chip repair
• ECH Repairs: 80017A chip repair
• Bell Tone: Juno-106 service packages
• Rosen Sound: RS80017A clone
• Polynominal: Roland 80017A
• Wikipedia: Digital Control Bus
• Vintage Synth Explorer: HS-60
• Wikipedia: Alpha Juno
• Reverb price guide: Juno-60
• Reverb price guide: Juno-106
• Vintage Technology Archive: Juno-106