Table of Contents
- Roland Juno-106: a budget polysynth with a MIDI socket
- What the 80017A chip does
- How a failing voice sounds
- Why the chips fail: three explanations
- The replacement market
- Who kept the Juno-106 in demand
- Final perspective
Roland Juno-106: a budget polysynth with a MIDI socket
According to Wikipedia, Roland released the Juno-106 in February 1984. It had six voices, each with a single digitally controlled oscillator, plus a resonant 24 dB per octave low-pass filter, a non-resonant high-pass filter, an LFO and a chorus. Its patch memory held 128 sounds, compared with 56 on the Juno-60, and it added MIDI. The pitch-bend and modulation lever that Roland introduced on it later became a standard feature on the company’s instruments.
A Roland UK advertisement feature in Electronic Soundmaker & Computer Music in June 1984 gives the British price as 800 pounds, essentially the same as the 799 pounds the original Juno-6 had cost, and below the 1,199 pounds listed for the Juno-60. Because the text is a promotional piece, its praise should be discounted, but its description of the sound is useful: it calls the Juno-106 a budget synth that does not cut corners on build quality, with a sound suited to hard-edged funk and electronic music.
Sound on Sound, looking back in 1998, said the Juno-106 remains one of the best-selling synths of all time. Wikipedia describes it as Roland’s bestselling synthesizer until the D-50 later in the decade.
What the 80017A chip does
The part at the center of the story is a Roland custom chip, usually called the 80017A. A post on Matrixsynth describes it as a single chip combining a voltage-controlled low-pass filter and a voltage-controlled amplifier. Roland used it in the Juno-106, the MKS-30 module, the GR-700 guitar synthesizer and the HS60, according to the same post, to keep an analog sound while lowering production costs.
Because the Juno-106 has six voices, it has six of these chips. That is the economy and the risk in one design. The custom part was chosen, per the Matrixsynth post, to help keep costs down. It also means that when one wears out, the owner cannot simply buy a standard component to replace it. The Matrixsynth post on replacements says the original is no longer manufactured and calls its failure the most common Juno-106 fault.
How a failing voice sounds
The symptoms are well documented by repairers. Syntaur lists crackling whenever the affected voice plays, a thin or distorted tone, notes that hang after the key is released, or a voice that makes no sound at all. The Matrixsynth post on the chip describes sudden loud noise after about ten minutes of power-up, and notes that fail to play, typically every sixth note, because one of six voices is dead.
There is a built-in way to find the culprit. Both Syntaur and SynthTweaks describe a test mode: hold the Key Transpose button while switching on, release it, then press Poly 1 and Poly 2 together. The display then shows which of the six voices is sounding as you play a key, so the faulty chip can be identified by elimination. That diagnostic feature is useful, but it also suggests how familiar the problem has been for a long time.
Why the chips fail: three explanations
Here the sources part company. Syntaur says the chips were epoxy-coated when made in the early 1980s and that, decades later, the epoxy becomes conductive with age, which causes the failure. A video summarized on Synthtopia in 2016 says something similar, that the coating Roland used begins to conduct electricity, and that removing it can extend a chip’s life but does not work in every case. The Stompbox Electronics repair guide soaks the chips in acetone for 24 to 48 hours to remove the resin.
SynthTweaks reaches a different conclusion. Its author states that the main cause of faulty chips is poor soldering, meaning dry or cracked joints on the surface-mount components under the black resin. An acetone soak can cure the fault temporarily, he writes, but the underlying solder problem usually returns. Meanwhile the engineer who designed one of the clone chips told Matrixsynth in 2006 that the original was poorly manufactured.
A seller’s post from 2007 adds that Roland changed its production methods after manufacturing problems appeared, and that chips from a later production lot were more reliable, though a commenter said he had seen later date codes fail too. These explanations may all be partly true, and none of the sources consulted for this article offers a definitive answer.