[sdiy] Synth Chips (OPL)

Nils Pipenbrinck np at inverse-entertainment.de
Sun Jun 3 18:19:09 CEST 2001


> How about a knob and switch for every function that the FM chip offers?!
> I'm too scared to ask how many controls there are :)

That's not worth the trouble. Most of the "controls" of the chips are
encoded in 4 bits which gives you 16 different values to choose from.

Internally the chip (speaking of the OPL2 in fm-mode) had 11 voices each
using two "operators". Each of this operator had:

 * ADSR (4 bits for each control)
 * Feedback to the "other" operator (3 bits)
 * 4 different but very boring waveforms (built out of sine-waves)
 * harmonic processor (adding octave, fifth and so on, 16 different
   possibilities.. I remember only few of them sounded good)
 * 10 bits for the frequency per octave (8 octaves where possible)
 * volume (6 bit)
 * cheap vibrato and tremolo with fixed frequency and depth
  (all operators using the very same lfo of cause).

The biggest problem of the chip is the one LFO for everything. That made the
chip sound boring. Also the limited choice of operator waveforms make most
of the sound alike. They changed that later on the OPL3 chip which
introduced the "square wave"
 (without pulse modulation) However, I was able to get some really nice
sounds out of this chip. I faked around the shortcommings using slightly
detuned voices. Modifing values per software (software-lfo) was impossible
because the programming of it's internal registers was incredible slow. It's
also not a good idea to modulate the value in a 4-bit register (clicks).

ok.. back to work,

    Nils Pipenbrinck





More information about the Synth-diy mailing list