IEEE Paper Implementation of Local Area Digital Audio Broadcasting System upon I2C Network
The 47th I E E InternationalMidwest
Symposium on C r u t and Systems
Implementation of Local Area Digital Audio Broadcasting System upon 12C Network
Chao-Huang Wei, Yun-Chung Lin
Department of Electrical Engineering
Southern Taiwan University of Technology
No. 1 Nan-Tai St.,Yung-KangTainan, Taiwan 7 10, R.O.C.
Abstract Conventionally, local area audio broadcasting systems
use analog channels for voice transmission. They are constructed with one main voice transmitter and many loudspeakers.
The inherent disadvantages lie in the difficulties to make individual or restricted area communication without additional
complex switching box; besides, a pre-planed wiring are necessary, which makes any on-line re-distribution almost impossible.
To overcome above problems, the system developed here uses
digital voice transmission upon 12C network. This system has
many advantages: enhance the voice quality of broadcasting,
encrypts the voice data for secret communication and reduce the
overall system cost for construction and maintenance. Other
devices such as sensors and actuators can be added or removed
from this I2C bus easily; therefore a cost effective network of
communication with securitylmonitoring functions for buildings,
plants or communities can be built.
Keywords: 12C BUS, DAB, PCM, ADPCM, SoPC
As a basic facility for communications, voice-broadcasting
system is widely applied in industry, medical care, entertainment, sports and commercial activities. The scales of
broadcasting systems range from large integrated system to
small broadcasting station installed on train. The conventional used analogue transceiver suffers from additional noise
and interference from a variety of sources.
Digital Audio Broadcasting (DAB) - the revolutionary broadcasting technology - dramatically improves sound quality and
signal reliability while enabling extra digital data transmission
at the same time. DAB offers high quality, crystal clear sound
wherever the receiver located. Also, unlike analog transmission, DAB reception won't fade at long distance.
Based on today's technology development of various kinds of
audio-visual digitization, the digitization of traditional wired
audio broadcasting system is foreseeable. Such a digitized
broadcasting can address the designated receiver and check
the response of receiver side to guarantee the transmission of
This developed voice-broadcasting system is based on a simple 12C (Inter-Integrated Circuit) bus, and can be mixed with
a control network, features are:
1. Broadcasting network can be built with simple 4 wires
telephone line in series.
2. The receiver's ID is changeable for individual or for group
3. The key of encryption can be changed at any time to insure
4. Additional devices for control or monitoring can be
clipped on the network easily.
ter-Integrated Circuit" for short distance data exchange,
however there is no maximum length specified in the 12C
specification. The limit of the length is a function of several factors including capacitance (usual maximum limit is
400 pF), the minimum value for pull-up resistor, propagation delays along the cables, the type of cable used and the
integrity of the logic signals in the presence of noise [ 1,2].
With a simple bus extender like P82B96 , which provides complete buffering of the long bus and really removes all 12C restrictions about capacitance limits. It can
drive 100-meter cable at max clock of 71 KHz or
1000-meter cable at max clock of 3 1 KHz .
(b)Digitized transmission of analog signal requires a serial of
signal processing. At the sending side, the weak analog
signal should be amplified first and then put through band
pass filtering, modulation, and final conversion to digital
signal. At the receiver side, processing at a reversed order is needed. Since this system is designed for voice
broadcasting, and the frequency range of human vocal
signal lies in between 300 Hz and 3300 Hz; thus, for reduce noise and other interferences, a band pass filter was
used to filter out the signal below 100 Hz and above 4 KHz.
The sample frequency of PCM data is then fixed to 8 KHz
with resolution of 16-bit. This results a basic signal
transmission speed of 128 Kbit/s, a further compression to
32 Kbit/s ADPCM (Adaptive Differential Pulse Code
Modulation) is required to reduce the bandwidth,
(c)To achieve the goal of addressable broadcasting, the voice
data must be transmitted in package; the simplest method
is adding the receiver address in front of a package before
the continuous voice data. The receiver monitors the
header addresses whether the messages that follow are effective or not. The address field is defined as 7-bits wide.
Thus, 128 separate receiving groups can be addressed with
each group contains from one up to n receivers. Among
them the ID number "0" is reserved for broadcasting to all.
(d)To simplify the connection between the transmitter and receivers, the signal is transmitted over a single 12C bus serially with normal telephone twisted cable.
(e)Even if the data transmission rate on the 12C bus of this
system is about 40Kbitsh merely, the receiver should
check the "START" and "STOP" signal on the bus continuously. To save the efforts of the processor, the 12C
interface is realized with hardware.
(f) Since the data and clock are sent synchronously, transmission rate can be changed to any value, and the time delay
due to the length of cable is not critical either.
Based on above factors, a low-cost local voice network with
12C bus is possible.
Considerations for the construction of this communication
(a)Although the original 12C bus was designed as "In-
This system consists of following components (Fig. I):
1. 12C BUS Communication Network,
0-7803-8346-X/04/$20.00 02004 IEEE
They are pulled up to the high electric potential with resistors,
thus allow Multi-Masters structure on the same bus. Similar
to a hot-plug concept, join or remove components has no interference to the normal bus operation.
To manage the 4-bits ADPCM data transmission in the system,
in addition to the 8 bits address, an acknowledge bit is generated or received after every 4-bits data transmission.
Therefore a 40Kbitsh bandwidth is required on the 12C bus.
3.3.I 12C Master Control Circuit
Fig. 1. 12C BUS voice broadcasting network
2. Digital Voice Transmitter,
3. Digital Voice Receiver.
The transmitter and the receiver implemented in this system
were constructed with only two chips: one is the PCM
CODEC, and the other is a programmable single chip - SoPC,
which contains a 32-bit processor inside. As a result, these
devices are very compact and economical.
3. I PCM CODEC
From the system analysis, a lot of different parts in voice signal processes are required. In order to save parts and the
circuit board space, a PCM CODEC from Texas Instruments
-- TLV32OAICI 110  for interfacing to the analog signal
was selected. This chip has a PCM voice coder and decoder
inside; its main circuit block diagram is shown as in figure 2.
3.2 ADPCM Compressor/De-compressor
The PCM voice is further compressed to ADPCM format to
reduce the data transmission rate; this procedure can be carried out by hardware or software . A programmable logic
device with embedded processor is suitable for this purpose.
The Altera's FPGA -- Cyclone EPIC200F400C7 was used in
this system, which contains a 32bits Nios CPU, a kind of
"soft" processor. The processor occupies about 12% chip
resource, remains enough spare space for other user logics.
While the 12C bus master controller is waiting for voice
transmission, it remains in the idle state. Upon receiving the
"START" command from CPU, it enters an active state and
sends the address of the target receiver first. After a acknowledge signal is confirmed, following 4 bits ADPCM
signals are then transmitted in stream, until a "STOP" command from CPU is received and terminates this communication session. Figure 4 shows the control block diagram; figure
5 is the simulation waveform.
3.3.2 12C Slave Control Circuit
While the 12C bus slave controller waits for "START" signal,
it remains in the idle state. Otherwise it begins to receive the
voice data package, the received target address will be compared with local ID at first, if they match, a confirm signal is
sent back to the transmitter, then 4-bits ADPCM data will be
received repeatedly, until a
"STOP" condition on the 12C
bus is found, this terminate
this communication section.
Figure 7 shows the controls
block diagram, and figure 8
is the simulation waveform.
3.3 12C BUS Communication Network
A digital information communication demands wide-band
transmission medias generally; but in this system, they are too
complicate and too expensive. This system therefore uses
12C bus for voice communication instead. 12C bus is a synchronous, bi-directional serial bus, it offers a simple and uncomplicated wiring to every connected component for efficient data exchange. The bus is constructed with two signal
lines only, one is the SDA, and the other is the SCL; output
signals of every component is wired-AND together on the bus.
Namely, the signal pins are either open-drain or tri-state type.
Fig. 3. Control flow of master 12C BUS signal
Fig. 4. Control diagram of master 12C BUS
Fig. 2. TI TLV320AICll I O main functions
PCM-CLK and PCM-SYN signals are generated from the
user's logic in the FPGA.
ual amplifier, filtering methods, PCM format, DTMF signal
and power consumption. Secondary 12C bus controller of
the FPGA controls these settings. While theses settings are
required only when power is turned on, in order to save the
chip resources, the internal CPU can process this 12C bus
function with software.
Fig. 6 . Control flow of slave 12C BUS signal
The voice compression uses ADPCM method. Basically, it
utilizes the dependence of succession digitized analog signal
while sampling. It produces the next value according to the
present output value. The output code represents the difference between two successes signals, instead of signals itself.
The sample rate is 16 or 32Kbps usually.
Figure 10 and 11 show the compression from PCM to
ADPCM. This procedure was implemented by hardware
circuit and software program in SoPC separately. The results indicate a process time of 4.4pS using pure software
method, which is equivalent to 227 KHz sample rate, far beyond the necessary 8 KHz sample rate for this system. On
the other hand, 193 logical elements are required while realizing with a pure hardware. Hence, it is relatively better to
carry out this part of procedures with the software program.
Fig. 7 . Control diagram of slave 12C BUS
3.4 Digital Voice Transmitter
The circuit's block diagram of the digital voice transmitter is
shown in figure 9. After the analog signal enters a microphone, it will be amplified and filtered by TI'S PCM CODEC,
then converted to digital signal. Serially shift this signal into
the FPGA, and compress it to 4-bits ADPCM. The SoPC
Fig. 10. ADPCM compression flow
D(2)= I ;
PCM -= Stepsize
Fig. 13. ADPCM de-compression flow
audiohide0 information, a lot of extra control functions can
be supported. The experimental results of this system have
attained the following functions successfully:
Long distance of digital voice communication up to about
650 meter with an 12C bus extender.
Voice broadcasting can be addressed to individual, to particular groups or to all receivers.
The main transmitter can get the status of receiver side in
advance, and know beforehand whether the receiver is
ready to take this information or not.
Content of the voice signal can be encrypted to avoid
Clear tone quality without interference.
Additional controlling and monitoring devices with 12C
interface can be added to this network easily.
The combination of software and hardware designs in the
SoPC has an advantage to solve problems quickly. Implementation of ADPCM signal compression and decompression
require extra 193 and 21 1 logical elements when realizing
with pure hardware circuit. On the other hand, 4.4pS and
4.0pS time-consuming are necessary while carrying out with
the pure software methods. This fulfills the real time requirement of voice communication, so the design except of
12C bus interface is optimized with the software realization.
Currently, only one transmitter was implemented in the system; but according the specification or 12C bus, it allows
multi-master on the same bus. Development with an architecture of multiple to multiple points voice communication on a
single 12C network is clearly worthy of further efforts.
This work was partially supported by the National Science
Council, Taiwan, Republic of China, under grant number
NSC92-22 1 8-E-2 18-018-.
“THE I2C-BUS SPECIFICATION VERSION 2. l”, Philips
Semiconductors, Jan. 2000.
“The I2C-bus and how to use it”, Philips Semiconductors,Apr.
“AN460: Using the P82B96 for bus interface”, Application
note, Philips Semiconductors,Jul. 1998.
P. Tracy, A. Anderson, J.-M. Irazabal, S. Blozis, “AN255-02:
12C/SMBusREPEATERS, HUBS AND EXPANDERS’, Dec.
“PCM CODEC TLV320AIClllO Datasheet”, Texas Instruments Inc., Dec. 200 1.
J. Reimer, M. McMahan, M. Arjmand, “32-Kbit/s ADPCM
with the TMS32010”, APPLICATION REPORT, SPRA131,
Texas Instruments. 1989.