Around 1983-84 Sequential Circuits, in order to break into the Japanese market, shipped the Curtis SENTE (CEM 3394) chip and the firmware for the Sequential Six-Trak to an other Japanese manufacturer. It was their plan to have that that manufactuer design a synthesizer more targeted at the Japanese buyer. This Japanese company created an 8-voice bi-timbral version of the Six-Trak. They added a 5-octave, 61-note (C-C) keyboard to it and sold it in Japan, calling it the Pro-8. Sequential, liking what the Japanese manufacturer had done with their design and worrying that they might market their synth with one of their competitors (ie. Korg), asked for some design changes and a name change, and sold the Japanese-made synth as the Split-8.
Features And Cabinet Brief Summary of Features: Relatively uncommon 8-voice polyphonic, bi-timbral analogue synthesizer with the ability to split or layer two patches ("double" mode). 64 patch memories. Patches may be "linked" in memory so that calling up one automatically calls up the second for a split or layer. Unison (monophonic, all 8 oscillators stacked) is possible, and programmable. Polyphonic portamento. MIDI, including system-exclusive. A user-friendly polyphonic analogue synth with a few nice features, acceptable sound, and more than bare-bones MIDI implementation.
Two volume pots, master tune, data entry knob, modulation and pitch wheels, various buttons for mode (double, split) and program selection, 4 x 8 grid for programming (all parameters labelled on the grid, 4 buttons down the left and 8 along the bottom). Some other functions are also listed on the panel as a reminder of which buttons to press. Metal case with light blue graphics and lettering, lots of red LEDs, black wooden end cheeks. Seems well-constructed, although on mine one of the rubber buttons has a tendency to pop out and the volume pots needed cleaning.
Tape jacks (one doubles as footswitch input), MIDI in and out, separate A and B audio outs, mix audio out (A and B combined), power switch, fuse.
61 notes, C to C, unweighted. No velocity or pressure sensitivity.
Voice Architecture and Programming Voice Architecture: One VCO per voice, with on/off toggles for any combination of sawtooth, triangle and square waves. Variable pulse width for the square. Coarse (up to 4 octaves) and fine (to just under a semitone) tune. Programmable "poly mod" (0-15) which allows the audio output of one voice to feed another's VCF when they are layered in "double" mode. This gives 4-voice polyphony with no audio from the second voice's oscillators. One LFO, free-running (i.e. doesn't sync to key on), with programmable rate (0.25 to 20 Hz according to the manual, but see the specs below), depth, and waveform choice (triangle or square). Can be routed to any or all of filter cutoff, pitch, or pulse-width. Programmable chorus with fixed rate and frequency. Filter cutoff, resonance (to self-oscillation), keyboard track (0, half or full) and envelope amount (cannot be inverted). ADSR envelopes for VCA and the 24dB/octave VCF. Programmable polyphonic portamento rate. "Unison" mode (8 oscillators stacked; envelope triggers become "legato"; low-note priority) can be programmed. A program can be "linked" to another so that both are called up and assigned correctly in a split or layer. Note that this does not appear to work on my unit, however, which may be because the battery was dead (I have not tried this since replacing the battery). In unison mode, delay of four oscillators relative to the other four can be programmed (0, 20, 40 or 80 ms). Programmable output level for each patch.
Separate volume pots for channels A and B (i.e. the two patches in a split or layer). Programmable footswitch assignment to momentarily change patches, or change one of several patch parameters (filter cutoff, envelope amount or attack; VCA decay or release; voice volume [the momentary value of any of the preceding is user-defined]; chorus on/off, portamento on/off). Parameter value knob can be used to alter one parameter in real-time, although the value will "jump" to the knob's current setting as soon as it is moved and some parameters stair-step badly when treated this way. Pitch-bend wheel is not spring-loaded and is centre-detented. Oscillator tuning can be initiated from the front panel (takes 20 seconds or so).
OMNI mode, wheel controller reception/transmission and program change can be toggled on/off. All MIDI operation (in and out) can be disabled. Sysex dump can be initiated from the front panel and the unit can receive sysex patch dumps. Receive channel (1-16) can be set for each patch. Four voices can be assigned to one patch that is controlled only via MIDI (a sort of "local off"), with the other four playing a patch controlled only from the keyboard. In this case, wheel effects can be disabled for the MIDI-controlled patch. Recognizes pitch-bend and modulation wheel as well as note number, note on/off, program change, tune request and sysex dump and load (2240 bytes per dump). The sysex dumps are actually a long string of single program dumps, so loading single programs at a time works perfectly. Powers up in OMNI mode, with wheels disabled (send and receive).
Some MIDI functions are accessed through a special "MIDI Expand" mode. For the benefit of users who do not have a manual, this mode is detailed here. To enter MIDI Expand mode, hold down RECORD and press PROGRAM SELECT 5. All the LEDs for the PROGRAM SELECT buttons will now blink. Pressing one of the PROGRAM SELECT buttons now performs one of the following MIDI functions:
OMNI mode ON. Selects MIDI mode 1 (OMNI on/poly). Note that if the synthesizer is in SPLIT mode, the synth will receive MIDI data from the MIDI in port and pass it to both programs (as though the synth were in DOUBLE mode). This is the reverse of using button 3, below.
Program dump. Initiates a 2240 byte system exclusive data dump. The LED for PROGRAM SELECT 2 remains lit while this takes place (it takes a second or two).
Mode 3 select. Selects MIDI mode 3 (OMNI off/poly). Note that this is the reverse of using button 1, above.
Enable/disable program changes. Toggles between the two settings. Transmission and reception of program changes via MIDI is disabled when the synthesizer is switched on.
Enable/disable wheel changes. Affects pitch-bend and modulation wheels. On power-up, MIDI for these is disabled (transmit and receive).
A Subjective Opinion Opinion:
I'm very fond of mine, but remember, this isn't a Prophet-5. Analogue fanatics mercilessly bash this and related SCI synths (the Six-Trak, Multi- Trak, and MAX) because of the synth chip that is used (the Curtis CEM 3394). It sounds a little thin, with somewhat harsh resonance on the filter. However, the sound is dramatically improved in "double" mode, especially if the two patches are detuned slightly relative to each other, and really grew on me as I played with it. It is capable of very nice pads, brasses and mono basses and leads (eight oscillators at once!). The chorus, located on a separate circuit board from the main synthesis machinery, is not nearly as noisy as you might expect. Programmable footswitch assignment and separate volume pots for the two patches are nice features, as is the support for sysex dump and load. The programming interface is better than almost any other single-parameter-at-a-time synth I've used, although increment/decrement buttons would be handy. MIDI implementation is difficult to figure out without a manual, although some other non-obvious functions are listed on the front panel. The keyboard has a very light feel to it - not unpleasant, but nothing special. Overall, I would say that it provides convincing analogue sound with the potential for some lovely warmth, and the ability to produce tremendous bass sounds in "Unison" mode. For what it's worth, I'm very happy with my Split-8.
Annoying features and bugs:
Sysex dump-request is possible, but the manual doesn't tell you what the request bytes are, nor does it give you a byte-map of the sysex data. Powers up in OMNI mode, with MIDI response to external wheel controllers disabled. Mod. wheel gives a large increase in a very short "throw" distance, either from the unit itself or via MIDI. Doesn't respond to "all notes off" message. Pitch-bend depth is not adjustable (+/- major third, although mine seems to be +4/-3 semitones from its own wheel). Only one VCO per voice unless in "double" mode (with 4-voice polyphony). No noise source or sample/hold. No LFO pre-delay or key-sync. No pitch envelope. Filter envelope cannot be inverted. Only one LFO. Stair-stepping precludes real-time filter cutoff knob twiddling.
The only obvious operating system bugs are in the envelope generators. A decay value of 10 (the range is 0 to 15) gives a very short decay, comparable to a setting of 1. This is true of both the VCA and VCF envelope generators. I do not know if this is common to all Split-8s, or is simply a flaw in mine. Also, the envelopes do not re-trigger correctly under some circumstances. I haven't rigorously tested to determine exactly why and when this happens.
When sending MIDI program (system-exclusive) information to the Split-8, the programs are sent as a string of single-program dumps. The synthesizer will sporadically "lose" some programs if a short delay is not inserted between each program, presumably because it cannot handle MIDI data transmitted at the speed that modern computers can send it. Most sequencers and librarians will let you specify such a delay between packets of sysex information. I have found that a 50 millisecond delay is adequate. Shorter delays may also work; I didn't test rigorously.
Rumours and speculations:
Built in Japan, this was one of the last synths marketed by SCI. It was released in 1985 at a list price of $1199 (or $1195) U.S., with an optional footswitch. I have read that a "handful" of Split-8s were sold, which may have been called "Pro-8"s in Japan and possibly also in the U.K. This and some other SCI synths (Six-Trak, MAX, Multi-Trak) were built in Japan around a synth chip that was reputedly originally designed by Curtis for video games (the "Bally/Sente synth on a chip"; in other words, the Curtis CEM 3394). The rumour is that a Japanese company built the Split-8, then bullied SCI into marketing it by threatening to sell it to Korg instead. (Another version of this story suggests that SCI was very happy with the product and requested only minor, cosmetic modifications, and decided to market the synth simply because they were worried about competition from Korg and other Japanese companies.) Regardless of the truth, SCI lost on this, because it was introduced two years after the Yamaha DX7 made everyone want to buy digital synths instead. Ironically, Yamaha later acquired SCI, then turned around and sold them to Korg anyway. I believe the Six-Trak was a more successful instrument; more people seem to own them and they command higher prices in the used market, as well as having wonderful sysex control of many parameters and 6-way multitimbral operation. It could be argued that Sequential made another marketing blunder by listing the Split-8 at $300 more than the Six-Trak. According to Keyboard magazine, June 1985, John Bowen wrote the factory presets for the Split-8. The Split-8's model number is 608.
About the memory battery:
The existing 3 volt nickel-sized lithium memory battery can be replaced with any equivalent lithium battery. I replaced mine (which was reading 0.1 volts) with a large 3.6 volt battery that had long, flexible axial leads, but probably any similar battery that you could shoehorn into the case would work fine. This repair requires opening the case, removing the keyboard assembly, and pulling out a few circuit boards in order to get at the battery location. It's not too difficult but is more involved than simply popping the case open and inserting a new battery.