of Microprocessors

LOW-COST SYNTHESIS TECHNIQUES 769
Obviously, the key to the unit's operation lies in the 64-pin custom IC,
which is also used in the other models. This all-digital IC both scans the
keyboard and performs all of the synthesis calculations for up to eight
simultaneous voices. Its final output is a 16-bit sampled digital signal,
although only 10 bits are actually D-to-A converted in the MT30. The chip is
driven by a 1. 14-MHz tunable L-C oscillator and essentially simulates eight
independent variable sample rate oscillators whose digital outputs are added
together every two clock cycles to produce a composite digital waveform. The
channel sample rates are quantitized in units of two clock cycles (1. 75 jLsec)
and vary between 5 and 20 ks/s according to the note played. The DAC itself
is external and in the MT-30 consists of CMOS buffers (4049) driving an
R-2R resistor ladder network constructed with 1% and 5% resistors. The
dual op-amp is used in the two-section (four-pole) Butterworth low-pass
filter.
Most of the preprogrammed timbres are standard items such as piano,
pipe organ, cello, accordion, glockenspiel, clarinet, plus a couple of
electronic sounds such as "synthe-fuzz." The quality of emulation is certainly
not up to professional standards but at least the sounds are distinctive. A
more bothersome defect on the MT30 is severe distortion near the end of
decay envelopes, which is probably due to having only lO-bit D-to-A
conversion with 7-8-bit linearity likely. This is probably much less of a
problem in rhe more expensive models. The synthesis technique used is
straightforward wavetable scanning. Each note begins by scanning Table A
during the attack and then gradually shifts to Table B for the sustain and
decay. This can be thought of as a keypress actually triggering two fixed
waveform notes at the same pitch but with different amplitude envelopes.
The chip apparently has a small RAM that is capable of holding four pairs of
wavetables at once. A front panel slide switch can instantly select one ofthem
for playing. In "set mode," each white key represents a timbre and playing it
copies a waveform from ROM into the particular wavetable RAM currently
selected by the switch. Thus, any combination of four timbres can be placed
on-line for playing. Waveforms are apparently stored in compressed form in
the ROM, probably as a line-segment approximation, since the actual
waveshapes seem to be quite simple.
Being so cheap, the Casio instruments have encouraged a bit of
experimentation among musician/hobbyists. One. simple modification is to
add a wide range variable clock oscillator to the synthesizer chip. With a
good sound system and the organ voice lowered two or three octaves with the
variable clock, a pretty convincing "mighty Wurlitzer" sound can be
produced. The latent automatic arpeggio feature can also be easily activated.
All kinds of wild distortions are possible by gating off or scrambling the
digital Output words before they reach the DAC. Thus, a suitably modified
Casio keyboard can become quite a special-effects device in a professional
performance as well as an amazingly versatile toy.