SAA2520 Datasheet, PDF (7/36 Pages) NXP Semiconductors – Stereo filter and codec for MPEG layer 1 audio applications

Philips Semiconductors
Stereo filter and codec for MPEG layer 1
audio applications
Preliminary specification
SAA2520
FUNCTIONAL DESCRIPTION
Coding System
MPEG coding achieves highly efficient digital encoding of
audio signals by using an algorithm based on the
characteristics of the human auditory system.
The broad-band audio signal is split into 32 sub-band
signals during encoding. For each of the sub-band signals
the masking threshold is calculated. The samples of the
sub-bands are incorporated in the signal with an accuracy
that is determined by the signal to masking threshold ratio
for that sub-band.
During decoding, the sub-band signals are reconstructed
and combined into a broadband audio signal. The
integrated filter processor performs the splitting (encoding)
and joining (decoding) including the corresponding
formatting functions.
For encoding, a SAA2521 is necessary to calculate the
masking threshold and required accuracy of the sub-band
samples.
Encoding (See Fig.4)
An encoding algorithm table is used during the coding
process but, due to the Adaptive Allocation functions of the
SAA2521, this may change with every frame. The table is
therefore calculated for each frame by the SAA2521 and
then transferred to the SAA2520.
A frame contains 2 × 384 samples of Left and Right audio
data. This results in 12 samples per sub-band
(32 sub-bands). The samples of the greatest amplitude are
used to determine the scale factor for a given sub-band.
All samples are then scaled to represent a fraction of the
greatest amplitude.
Once scaled, the samples are quantized to reduce the
number of bits to correspond with the allocation table as
calculated by the SAA2521. Synchronization and coding
information data is then added to result in a fully encoded
MPEG signal.
After sync and coding information, allocation data and the
scale factors are used to correctly fill the scale factor array.
This is followed by a process of multiplication to provide
de-quantization and de-scaling of the samples.
The decoded sub-band samples, which are represented in
24-bit two's complement notation, are processed by the
sub-band filters and reconstituted into a single digital audio
signal.
RESET
Reset must be active under the following conditions:
1. From system power-up until CLK24 has executed
more than 24 clock cycles.
2. From the falling edge of PWRDWN for a period
equivalent to 24 cycles of CLK24 + oscillator start-up
time. This is typically >1 ms, however, this value is
crystal dependent.
PWRDWN
A HIGH input applied to this pin will halt all internally
generated clock signals. As a result, chip activity will halt
completely with outputs frozen in the state which was
current at the time of PWRDWN activation.
The bi-directional outputs: LTDATA, FDAC, FDAF, SDA,
SBWS, SBCL and SBDA will be 3-stated.
Crystal Oscillators
A 24.576 MHz crystal together with some external
components form the 24.576 MHz oscillator (pins 42 and
43). Similarly a 22.5792 MHz oscillator (pins 40 and 41) is
formed by similar peripheral components together with an
appropriate crystal (see Fig.6).
The component values shown apply only to crystals from
the Philips 4322 156 series which exhibit an equivalent
series resistance of ≤ 40 Ω.
Decoding (See Fig.5)
All essential information (synchronization, system
information, scale factors and encoded sub-band
samples) are conveyed by incoming data. Decoding is
repeated for every frame.
August 1993
7