[sdiy] Variable rate waveform playback in NED synclavier

[Scott Nordlund]

I've been playing around with this a bit more. I don't think an exact, perfectly faithful emulation is super exciting since my frequency dither idea (eliminating aliasing by restricting waveform periods to integer numbers of samples and alternating between the nearest two frequencies) seems to open up a lot more possibilities. I'm surprised I haven't heard of it before, as it seems extremely useful. I'll have to look around to see if it's been done before. I'll be upset if it wasn't implemented, published or patented before 1980. Maybe I'll have to write a paper on it...

More on topic, I also checked out the Synclavier's FM boards. A set of 5 boards produces 8 voices. Everything is time multiplexed and overall it's actually a lot more sensible and clever than I was expecting. SS1 has 64kbits of RAM, which seems to store the carrier waveforms (but this is 4x the size I was expecting). SS2 has the sine ROM and 7 64 bit RAM chips for phase registers and stuff (these are AM27S03, like the 74S189 used in the PPG Wave and Korg DW-8000 for the same purpose). SS3 has 14 more 64 bit RAMs. SS4 seems to be the master clock, as it has a 32 MHz crystal oscillator. SS7 has DACs, demultiplexers, sample and holds, and other analog stuff. The later stereo FM voices added a second SS7 board to permit dynamic panning of each voice. I haven't delved any deeper than that. I don't know where the multiplier is to scale the modulator, and I don't know if the amplitude envelope generators are on these boards as well. It would be a huge headache to figure out much more than this. The PPG PROZ board looks trivial by comparison. There are 177 ICs total, including some that I can't identify and some that seem to be PALs or ROMs.