The Sub Sub Milestone

Nowhere can I find a reference to the synthesiser that early 90s Haçienda favourite Sub Sub employed for the lead of Space Face, but I’m convinced it’s an OSCar, as I think this is the sound that Chris introduced in the Version 6 ROM:

As all the wave tables in OSCar are procedurally generated, here’s how that waveform is calculated in Chris’s own words (Original OSCar source code, so he really called it ‘mess’!):

The first 256 bytes of ROM are copied into the wave table memory. What we’re listening to is the Z80 initialisation code, the interrupt service routine, and some of the key scanning algorithm looped into a wave. This is why, whenever Doves (same band, different songs) plays Space Face live, it never quite sounds right.

It’s a short post this week because I’m getting on with it. Getting to the Sub Sub milestone required a few things to be wired in:

• The presets database works so I can meaningfully set up a patch.
• The shifted keybed functions work so I can select the correct wave tables, which you can’t reach from the rotary switch because Chris hadn’t thought of them.
• Most of the filter parameters now work properly, and we’re using those refresh bits from last week. All the stuff that’s modulated still needs to be wired in, though. (Doesn’t the filter drive sound beautiful? I can take no credit for that.)
• There aren’t any envelopes, but fortunately Space Face doesn’t use them: I can just turn the wave tables on and off at the beginning and end of every note. Smoke and mirrors …
• As a bit of falsework, I’ve added the Mantis voice management routines so that OSCar can be played in one- or two-voice mode.

I call the Mantis code falsework to imply that it’s holding things together nicely at the moment, but I expect to be able to take it away when the synth is finished.

Ah! Would that it were so simple. The Mantis voice manager is an improvement on OSCar because we had plenty of memory. More data enables more sophisticated decisions, and we were designing a four-voice polysynth in the twenty-first century.

If I were to try to turn my curatorial point of view in to a set of working principles, they’d go something like this:

• If a decision is going to affect the sound, we keep it as faithful to the original as we can, including deliberately keeping deficiencies in the code and hardware: that’s the only reason to make a physical OSCar.
• If a decision affects the user interface and workflow, we think very hard about the trade-off because a musician has to feel like they’re having an authentic dialogue with an old instrument. Some of OSCar’s quirks are a consequence of its rush to production. We added the ‘filter drive’ knob because it was the only shifted control: leaving it off the panel design was clearly just a mistake. In other cases, it is much harder to know what to leave alone.
• If a change reduces the impediment to actually playing music, we do it. First, because a reissue must be as good as you remember the original to be, and that involves aligning to expectations that have increased over four decades. Second, because musicians will want to approach this instrument as they would any electronic keyboard. If they find that, say, DUO mode cannot return to playing one voice after playing two, or the wrong voice gets borrowed back when they’re playing legato, it’s not going to feel merely quirky: it’s going to get in the way of their muscle memory and just be annoying.
• Usually, though, ‘better’ is obvious and we do it. Of course we’ll improve the noise performance and stop it picking up AM radio (the distortion performance is left alone though). Of course we’ll remove the set-up delays in the additive synthesis section because the processor was slow and couldn’t do multiplication. And of course we’ll let it be powered from a USB supply, give it USB MIDI, and make MIDI work properly.

I reserve the right to change my working principles every time I make a decision: ultimately it’s down to my taste and Paul’s. If taste could be turned into a list of hard-and-fast rules, expertise would be far easier to acquire. All this feels exactly like the ‘Should you play Baroque music on Baroque instruments?’ argument. There’s more than one correct answer.

We’ll know a lot more once we get to the arpeggiator and sequencer: that’s when the note routing through the system gets very convoluted. OSCar now has the things a modern musician would expect: a sustain pedal input, keyboard velocity, a ‘local off’ mode, and full MIDI compatibility. The most complex features will really tell us what kind of trouble we’re in.

In the meantime, lots of the progress-o-gram is turning green at last, but we’re still a little way from the Ultravox milestones. They used it all.