Z89340 Datasheet, PDF (14/30 Pages) Zilog, Inc. – Digital Wavetable Engine

Z89340
Digital Wavetable Engine
FUNCTIONAL DESCRIPTION
Digital Audio Input and Output
The Z89340 has eight output registers with signals that
can be sent to a DAC or CODEC. Four of these can be
used for quadraphonic output and have a panning mecha-
nism called Polar Pan that supports motion in all four quad-
rants. The other four output registers are used internally as
effects channels, but can still send their serial streams to a
DAC, a second Z89340, or other digital signal processor.
The Z89340 also has eight serial input registers whose sig-
nals can be accessed and processed.
Submix Registers
The Z89340 has 16 stereo submix registers. The submix
registers are needed to chain together the various compo-
nents of the reverb, chorus, echo, and flange algorithms.
When the data in a submix register is no longer needed, it
is cleared with bit 7 of the Data Source register in the pa-
rameter block of the pertinent oscillator types. The host
CPU can read and write all oscillator parameter RAM, in-
cluding the submix registers.
Oscillators
The Z89340 is a special-purpose audio processor with an
instruction set designed for music synthesis. Fundamental
to the Z89340 are the 64 full-speed oscillators, the first 48
of which have a half-speed counterpart. Control of an os-
cillator is handled by setting the parameters in the oscilla-
tor’s 24-byte parameter block. An oscillator can be used to
generate sound, or it can be used to perform other opera-
tions-input device, tape-loop reverberator, dual tap-reader,
input mixer, and so on. Typical implementations would in-
clude an Z89340, waveform ROM and RAM, and a CPU to
control the Z89340. The CPU is normally connected via
the ISA bus interface. (Refer to Figure 10 for General Pur-
pose Oscillator Address Map.)
The following subsections briefly introduce each of the os-
cillator types. (A detailed description of the Oscillator Pa-
rameter Blocks for each oscillator type follows.)
Half-Speed Oscillators
Half-Speed Oscillators are best suited for playing lower
notes where the upper bandwidth is not critical, since they
operate at half the sampling rate.
Oscillators 0 through 47 (0x2f) can each be split into two
oscillators operating at half the clock rate. This yields 112
oscillators total (2 ^. 48 + 16). Oscillators 48 through 63
(0x30 through 0x3f) can only operate at the full clock rate
since the parameter RAM that would be needed for their
half-speed counterparts is unavailable
They can also be used for higher notes with simple wave-
forms that have few significant upper harmonics. For ex-
ample, at a sampling rate of 48 kHz, the Half-Speed Oscil-
lators will have a sampling rate of 24 kHz, so the maximum
frequency that can be produced by these oscillators with-
out aliasing is under 12 kHz. This bandwidth is adequate
for many sounds, but not suitable for high strings, cymbals,
or other crisp or bright sounds. An empirical listening test
will help determine if a Half-Speed Oscillator can be used
for a particular situation. The obvious advantage to using
Half-Speed Oscillators wherever possible is that more
notes can be played at the same time.
Sample Loop Oscillators/Wavetable Mode
Oscillators
For music synthesis, the Z89340 Sample Loop and Wavet-
able Mode Oscillators are the main workhorses. They are
interpolating wavetable look-up oscillators and perform the
tasks of fetching two adjacent samples from waveform
memory. They interpolate between them based on the cur-
rent phase angle (frequency and time), filtering, scaling,
and routing the resulting signal to multiple effects sends
and output channels. These oscillators can use 8- or 16-bit
linear PCM samples. The ADPCM oscillators described
later are similar in function and use ADPCM (IMA and DVI
4- or 8-bit samples).
Sample Loop Oscillators
Sampling
One of the more popular methods of re-creating or re-syn-
thesizing the sound of a traditional acoustic musical instru-
ment is through a process referred to as sampling or PCM
(pulse-code modulation) synthesis. A sample, in the strict-
est sense, is a value taken at a specified point in time that
represents the instantaneous amplitude of the subject
waveform. A digital recording consists of a sequence of
amplitude values sampled at evenly spaced intervals of
time. The term “sample” sometimes refers to the sequence
of samples found in a digital recording. (This usage is es-
pecially popular throughout the music industry.) This digi-
tal recording is much like the recording that would be cap-
tured with a tape recorder, except that it can be stored in
digital memory, and as such, can be randomly accessed.
Looping
To reduce the length of the recording to make it fit in a lim-
ited memory space, the most common form of processing
used with sampling is looping. The process of looping can
be briefly described as follows: The synthesizer plays the
original recording of a note up to a designated time point,
whereupon it plays a short sequence of samples that de-
scribe one or more periods of the temporally varying wave-
form-the “loop.” The loop is then repeated until the note
stops. A decaying amplitude envelope is often imposed
upon the loop so that the sound will decay naturally. The
Sample Loop Oscillator is designed to facilitate looping al-
gorithms with single or multiple loops.
1-14
PRELIMINARY
DS96DSP0201