Pr

1
NCCR-MICS Project
MP3 on Btnode
2
Main Idea

• Btnode designed as clever  sensor 
• Goal Use it as audio sensor (AudioNode) but
• Audio requires computing power
• Audio is real-time
• Raw audio material takes memory space

3
Main Idea (2)

• Corollary
• We need more computing strength
• (to process the audio stream)
• We need a clever algorithm
• (to manage the data in real time)
• We need compression
• (to be able to deal with the Bluetooth bitrate)

4
Solution

• Use a dedicated Audio MP3 chip
• Provides specific computing strength, integrated
  functions (D/A conversion, Filtering)
• Provides audio data compression__________________
  _________
• But does not solve the real-time problem (data
  always needs to be processed on the fly)

5
Solution (2)

• Develop a software appliation that will manage in
  real-time
• The sending or the reception of MP3 stream via
  the Bluetooth connection
• The control and the management of the MP3 chip

Bluetooth
Software
MP3 chip
6
First part MP3 Chip

• Use MAS3587f audio chip
• Encode and decode MP3
• Multiple bitrates
• Microphone or Line Inputs
• Line Outputs
• Other features

7
PCB

• MAS Chip requires many external components
• A Printed Composite Board is needed
• Three succesive PCB
• Revision 1 Buggy design
• Revision 2 Large surface, expensive
• Revision 3 Received, not yet tested

8
Interfacing

• MAS-Chip must be controlled
• I2C interface is needed
• Atmel controller as a dedicated feature to do
  that
• MAS-Chip must receive data
• Parallel or Serial Interface possible
• Atmel controller as an automated serial
  connection

9
Configuration

• Many things in Atmel are configurable
• pins, I2C-interface, serial interface
• MAS-Chip is a complex device, it must be
  configured
• By sending many commands via I2C
• This part is OK, functional

10
Sending stream

• Once MAS-Chip is configured, one can send the
  stream
• But stream needs a (cyclic) buffer
• External SRAM of the Btnode used
• 60 KB buffer achieved

11
Bluetooth link

• Stream shall come over Bluetooth
• From a PC
• From another Btnode
• Easier to develop first with the PC
• Large hard disk
• Better Development Environment

12
First shot

• Advantages
• No buffer overrun
• Simple code
• Works (at 96 kbps)
• Problems
• Waste of time on Btnode
• May fail easily

• Simple algorithm

BTnode
PC
13
Second solution

• Stream algorithm

• Advantages
• Smarter !
• Lets time to BTnode
• Problems
• Hard to implement

BTnode
PC
Start
Slower
Quicker
14
Results

• MAS side
• Decoding is ok
• Configuration is ok
• PCB is ok
• Bluetooth side
• Works at 96 kbps but blocks
• Setup problems for the streaming protocol

15
Further development

• Encoder part
• Reliable protocol(may not be reliable in the
  data direction but must be the case in the
  control direction)
• Use of RFCOMM
• Btnode to Btnode application
• Nicer PC client Program
• Winamp integration

16
Thank you for your attention

• Sébastien Rumley (sebastien.rumley_at_epfl.ch)
• Harald Vogt (vogt_at_inf.ethz.ch)