digital audio seminar - Rice University Electrical and ...duarte/elec542/audio.pdf · Analog and...

Audio

System DesignECE 542 Spring 2005

Dr. T. Tran

Rice University

Introduction: The Wide World of Audio

Analog and Digital Audio

Algorithms for Audio Processing

OMAP Audio Algorithms

DSP Architectures for the Future of Audio

OMAP Audio Design

Class D Audio Amplifier

Agenda

2

Agenda: Introduction

3






OMAP Audio Design


Introduction: Wide World of AudioThe Wide World of Audio

The audio life cycleAudio takes a journey from lips to ears

Audio in the 21st CenturyMore and more technologies are converging onto consumer entertainment electronics

Different phases, different facesNot just a ‘one size fits all’ approach

What we have to doBecome a reliable partner with our customers

How TI can meet the customers’ needsDelivering outstanding audio technologies and audio system solutions

4

The Audio Life Cycle

CaptureCapture

CreateCreate

ManipulateManipulate

ProduceProduceChannelDistribution

ChannelDistribution

StoreStore

Personal DistributionPersonal

Distribution

ConsumeConsume

Audio in the 21st CenturyIn the “Good Old Days”...

A new music delivery format once every ~20 yearsPortability was only for those with very strong arms

Some signs of change in the last five years:Significant new physical formats - MD, SACD, DVDNew non physical formats – Streaming audio over networksNew broadcast sources - DAB, Digital TV, HDTVMore and more people are making their own music

Now every month brings something newThe musical WWW reaches critical massWith some audio standards even set by a software company...Soon, every tune in the world can reach every ear in the world (provided they have IP addresses...)GOOD AUDIO EXPERIENCE MAKES BETTER VIDEO EXPERIENCE, NOT THE OTHER WAY AROUND!

Phonograph

Boom Box

MP3 Player

Agenda: Analog & Digital Audio






OMAP Audio Design



ADC DSP DAC

Typical Audio System

AMP

CD,DVD

Analog Digital Digital Analog High-PowerAnalog

AnalogAudio

DigitalAudio

AnalogAudio

AnalogAudio

AnalogAudio

AnalogAudio

The whole world is analog and digital only plays a bit part

Analog Audio: Inputs

Microphone inputElectret (moving capacitive plates) or dynamic (moving coil) condenser microphone: Electret requires DC bias to functionVery low level input (~100mV): pre-amp of about 20dB requiredMost critical input as far as noise sensitivity: requires special care!Mic performance is very important for good speech recognition, video phone, karaoke, and full-duplex speakerphone

Microphone Front-End IssuesYou need a high performance solution with low component count, small board area and component bulk, simple integration into your digital control environment

PGA2500 programmable gain 10dB to 65dB in 1dB step.Extremely low noise and low distortion, 1nV/√Hz and 0.00002% THD

Simple interface to A-to-D converters: careful matching of supply voltages and signal levels can create an elegant input stage design

Line level inputsLine level amplitude varies greatly depending on the audio source, for example portable CD to DVD player.Typically, Line In = 200mVrms to 2Vrms, where 1Vrms = 2.828Vp-p

Issues related to line level:• Portable devices can’t drive audio system to the

maximum power. • High-end DVD players can overdrive the audio

system inputs and cause severe distortion.Audio Engineering Society (AES) has published many standards to define the audio levels for different types of equipment but only a few manufactures follow this recommendation.


Example of audio line level input problem


ADC

3.3V

ground

L-AudioR-Audio2Vrms/

5.656Vp-p

Input level is clipped at 3.3V, overdriving the CODEC

One solution to the clipping problem


ADC

3.3V

ground

L-AudioR-Audio

Volt.Divider

Add a voltage divider network to limit the AC input signals to theCODEC supply voltage. For example,

R1R2

)21(2656.53.3RR

RVV+

= R2 = 1.4 R1 Let R1 = 20K -> R2 = 28K,Keep input impedance greater than 20k

Analog to Digital Conversion Issues

You need enough dynamic range and performance to cope with almost any situation at almost any sample rate, but without burning lots of power, taking lots of board space or spending lots of money

A clear case of where the best product is in fact the best compromise between competing requirements; in real-world products, you don’t want the extra cost, board area and heat that a few extra dB’s brings - and you’ll lose a lot of versatility and performance if you try to save any more money

This function is the ‘gateway to the soul’ of your equipment, and you want the reassurance of the long experience of the Burr-Brown division of TI

ADCSR

Deci.By N

AnalogFilter

AudioInput

Nfs

DigitalLPF

AIC23 CODEC

fsData rate

Nfs Data Rate

Oversample by N pushes the quantization noise way above the audio spectrum. Sampling at Nyquist rate (2 times the highest frequency component) requires higher order anti-aliasing filter.Digital LPF removes the noise generated by the ADC. Sample Rate (SR) Decimation reduces the data rate for audio processing

Oversampling ADC

Noi

se D

ensi

ty

Frequency

Noise shaping characteristic (sigma-delta A/D)

Nfs/2fs/2

Nfs Sample Rate

Digital LPF Passband

Oversampling ADC

Watch for these potential problemsAnalog input filter is required to limit the audio bandwidth. Without this, out-of-band noise folds back into the audio spectrum and degrades the CODEC performance.Isolate the analog power supply by using a high performance linear voltage regulator.The reference voltage Vref needs to be decoupled properly. This voltage is used for sampling threshold comparison so it needs to be very clean.Routing all the input signals carefully and away from high speed and high current switching signals.

D-to-A Conversion IssuesThe required state-of-the-art in replay keeps rising; better dynamic range, less distortion, lower solution cost, greater filtering flexibility for both DSD and PCM operation over a wide Fs range

You try to make your clocks as good as possible but they are never perfect; you need good tolerance to clock jitter

Not made of money: you need wide dynamic range and THD in real applications using affordable external components, not made out of Unobtanium

Good control of out-of-band noise is critical in broadcast applications, and in general any application where the D-to-A output feeds any kind of converter or modulator

Format flexibility (PCM, DSD) and filter transient response choice to deliver low delay and good sound quality

Oversampling DAC

DAC AnalogLPF

DigitalAudioInput

Nfs

Inter.Filter

AIC23 CODEC

AnalogAudio

Nfs Data Rate

Interpolation filter increases the sample rate by N for the oversample DAC to process the data.Analog lowpass filter is required to remove the sampling noise. This filter is included in the TLV320AIC23 device.

Oversampling DAC

Watch for these potential problems Isolate the analog power supply by using a high performance linear voltage regulator.Isolate the analog filter supply from the DAC supply as recommended by the manufacture.The reference voltage Vref needs to be decoupled properly. This voltage is used for sampling threshold comparison so it needs to be very clean.

TLV320AIC23 Stereo CODEC

1.0X

1.0X

1.0X

1.5X

Ω50k

Ω50k

VADC

VDAC

VMID

2:1MUX

+12 to -34dB,1.5dB Steps

Σ−∆ADC

2:1MUX

+12 to -34dB,1.5dB Steps

Σ−∆ADC

Ω50k

Ω10k

VMID

Mute,0db, 20dB

Σ−∆DAC

+6 to -73dB,1dB Steps

HeadphoneDriver

Σ−∆DAC

+6 to -73dB,1dB Steps

HeadphoneDriver

DigitalFilters

DigitalAudio

Interface

ControlInterface

OSC

Σ

Σ

Divider(1X, 1/2X)

AVDD

VMID

AGND

MICBIAS

RLINEIN

MICIN

LLINEIN

HPGND

HPVDD

RHPOUT

ROUT

LOUT

LHPOUT

XTI/MCLK

XTO

CLKOUT

/CS

SDIN

SCLK

MODE

DVDD

DGND

LRCIN

DIN

LRCOUT

DOUT

BCLK

VADC

VDAC

BVDD

TLV320AIC23 Stereo CODEC

DSP

0.1uF 10uF

3.3V

0.1uF 10uF

10uF

3.3V

5.6K

5.6K 220pF

0.1uF

5.6K

5.6K 220pF

0.1uF

5.6K

47K

1.0uF

3.3V

5.6K

47K

1.0uF

5.6K

47K

220uF

5.6K

47K

220uF

100

0.1uF

3.3V

1.5 - 3.3V

0.1uF10uF

10uF 0.1uF

220pF47K

680

Rmic

1uF

AVDD (14)

AGND (15)

HPVDD (8)

HPGND (11)

LHPOUT (9)

LOUT (12)

ROUT (13)

RHPOUT (10)

CLKOUT (2)

VMID (16)

XTI/MCLK (25)

XTO(26)

BVDD (1)

DGND (28)

LLINEIN (20)

RLINEIN (19)

MICBIAS (17)

MICIN (18)

DVDD (27)

15pF 15pF

LRCIN (5)

DIN (4)

DOUT (6)

LRCOUT (7)

BCLK (3)

CSB (21)

SDIN (23)

SCLK (24)

MODE (22)

SCLK (24)

3.3V

10K

BCLKX1 (33)

BDX1 (36)

BDX0 (24)

BFSX1(35)

BDR0 (30)

BFSR0 (29)BCLKR0 (27)

BFSX0 (23)

DSP PIN NUMBERREFERS TO COMMON80-PIN CONNECTOR

AIC23

FSA 3/21/01

Analog inputfilters

S/PDIF Digital Interfaces

S/PDIF, Sony/Philips Digital Interface, has been adapted as an I/O interface for digital audio and other consumer electronics equipment.

S/PDIF connector comes in two types, optical and/or RCA. Optical provides higher performance due to noise isolation but it costs more because of the additional optical transceiver.

Low jitter is a must for this interface, embedded clock and data on a single conductor.

New TI SPACT technology completely eliminates performance degradation through plastic optical connections (long cables, low supply voltage...) even at high sample rate

Effect of Interface Jitter

0.002

0.020

0.004

0.006

0.008

0.010

0.012

0.014

0.016

0.018

20 20k50 100 200 500 1k 2k 5k 10k

Audio Signal Frequency [Hz]

5 ns 2 ns

1 ns

0.5 ns250 ps

THD+N [%]

0 ps

‘Analog’ DIR 1.5 ns

Effect of Interface Jitter

0.002

0.020

0.004

0.006

0.008

0.010

0.012

0.014

0.016

0.018

20 20k50 100 200 500 1k 2k 5k 10k

Audio Signal Frequency [Hz]

5 ns 2 ns 1.5 ns

1 ns

0.5 ns250 ps

THD+N [%]

0 ps

‘Analog’ DIRSPACT DIR







OMAP Audio Design


Lossless vs. Lossy Compression

Coding that is bit exact, i.e., the original signal is recovered bit for bit. Examples of lossless coding include numerical coders:

Arithmetic coding, Ziv-Lempel (LZW) coding / WinZipHuffman coding / FaxRun Length CodingMeridian Lossless Packing (MLP) / DVD Audio

Coders that create an approximate reproduction of the input signal, i.e., “sounds like”the original speech or audio or “looks like” the original image. Examples of lossy coding includes source coders:

Sub-band coding / MP3Transform coding / JPEG2000Vector quantization / MPEG-4

Lossless Coding Lossy Coding

Lossless Coding

Assign the most frequent events to the shortest code words.Non-duplicating prefix, for example 110 and 11011 can’t be code words

Huffman Coding

Huffman coding example:Bus Status Probability Code wordOn time 0.5 0Late 0.35 10Early 0.125 110Wrecked 0.025 111

0

111

00

Run Length CodingSpecial start and stop codes for repeated data.Only send the value and how many times the value is repeated; for example 5555 5555 is coded as 85.

Frequency domainPrevalent when signal is approximately periodic

Expressed using non-uniform frequency scale derived from physiology of cochlea

Subband coding (AC-3, DTS, MP3, etc.)

Time domainPrevalent when signal contains large transient componentsDue to mechanical processes and early neural processingMasking components occurring just before and after a transient

Human Audio PerceptionProperties of auditory perception can be used to exploit irrelevancy in audio that hides noise where you can not perceive itStrategy: Identify signal components for which the ear is less sensitive to distortionEffects are typically signal dependentEffects are broadly categorized into:

critical bands

96 (dB)

22 kHz0

Hearing threshold

Number of quantization bits

Exploiting Perceptual Effects

Typically encoders/decoders perform these steps: 1. Analyzes input in frequency domain (critical subbands).2. The ear masks (does not hear) neighborhood low amplitude signals of a

certain frequency in the presence of a nearby higher-amplitude signal. Thus the psychoacoustic model helps ignore these low amplitude signals well (frequency irrelevancy).

3. The psychoacoustic model also helps gauge which subbands the ear does not hear as well. Scales the number of bits in relation to hearing threshold and masking signals.

SideInformationand optionalAncillary Data

Digital Audio Input

EncodedAudioBitstream32

Sub-bandFilterbank

Bits

trea

mFo

rmat

ting

Run Length

Encoding

Bit Allocation,

Quantization,and Coding

Block Diagram of an MP3 Encoder(by layer)

ExternalControl

Layer 3

Layer 2enhancements

Layer 1

Modified Discrete Cosine

Transform

1024 Points

FFT

PsychoacousticModel

+

HuffmanCoding

Widely Used Audio StandardsWidely Used ISO: MPEG Audio Coders:

MPEG-1 Audio Layers 1, 2 and 3 (MP3 is the popular name for MPEG-1 Audio Layer 3)MPEG-2 Audio Layers 1, 2 and 3: 32-256 kb/s/ch European DVD and HDTVMPEG-2 Advanced Audio Coder (AAC): 8-160 kb/s/ch Japanese HDTVMPEG-4 Speech and Audio Coders - MPEG-2 AAC (with extensions)

Other Audio Coders:Sony ATRAC: Primarily 2-ch coders used in Mini-Disk players at 144 kb/s/chMicrosoft WMA: Designed for 1- and 2-ch streaming, 5 kb/s/ch to 128 kb/s/ch, 8 to 48 kHz sampling Real Audio: Designed for 1- and 2-ch streaming, 5 to 96 kb/s/ch, 8 to 48 kHz samplingDolby Digital AC-3: Theater/Home multichannel standard. 5.1 channels, US DVD (5.1 ch/320 kb/s), US HDTVDTS (Digital Theater System): Home multichannel competitor. 5.1-ch system w/48 kHz sampling.

MLP (Meridian Lossless packing): Lossless coder for DVD-Audio. Multichannel system up to 24bit/192kHz

Other Standardized Audio Processes:Dolby PrologicDTS NEO:6THX

Audio Effects Classification

Amplitude based effectsVolume, Panning, Tremolo, Noise gating, Compression/Expansion, …

Time delayModulated delay: Chorus, Flanger, Phaser, Vibrato, …Fixed delay: Echo/Delay, Reverb, …

Waveform shaping effectsVacuum tube simulator, Non-linear effects

Frequency filtersNoise Reduction, Pitch-shifter, Vocoding, Equalizer, …

Compression/Expansion

Reverb

Acoustic Effects Library ExampleEqualizer Adjust the level of signal frequencies components*

Boost

Cut

CutoffFrequency

Gain

Frequency

Lowpass Shelving Filter Frequency Response (Magnitude)

1

CutoffFrequency

Gain

2

Highpass Shelving Filter Frequency Response (Magnitude)

Boost

Cut

Frequency

Σb0

b1

b2

-a1

-a2

z-1

z-1

z-1

X(z) H(z)X(z)

z-1

Gain

Peaking Filter Frequency Response (Magnitude)

CenterFrequency

Cut

Frequency

Boost

*From Parametric Equalization on the TMS320C6000 DSP (SPRA867) app note and code for C6711 DSK







OMAP Audio Design


System Software (Play, Stop, FF, Rew, Repeat)EQ (5-band)Volume (Stereo)SRC(Stereo)AC3 Decode MP3 decodeMP3 encode AAC decodeAAC encodeWAV PlaybackWAV RecordADPCM PlaybackADPCM RecordATRAC-3 decodeACELP.NET decode

TwinVQDolby Digital 2 channel Class - CDolby Digital 5.1 channel Class -CReal G2 decodeWMA decode WMA w/ DRM decodeWMA encodeLucent ePACLiquid Audio SP3 (ver. 1.0) MPEG-1 Layer I EncoderMPEG-1 Layer I DecoderMPEG-1 Layer II EncoderMPEG-1 Layer II DecoderAC-3 2 ch Encoder (DDCE) Class AIntertrust723.1

Audio Algorithms for C54xx/C55xx DSPs

Audio Algorithms

AAC DecoderFeatures

MPEG-4 AAC Decoder with simple and Low complexity Profile.Decoder outputs pcm file format from 16-bit AAC Encoded samplesall sample rates (8 kHz to 48 kHz)all bit rates (8 kbit/s to 288 kbit/s)mono, stereo, dual-channel, joint stereo The given decoder supports both; ADIF (Audio Data Interchange Format) and ADTS (Audio Data Transport Stream) input formats.automatic bitstream synchronization and error recoveryaudible fast-forward/rewind modeTI eXpressDSP compliantC5500 large memory model object code library

PerformanceLow memory library

25.5 MHz (peak) and 23.5 MHz (average) for stereo 44.1 kHz at 128 kbit/s13.5 Kwords of SARAM (const and program) and 11 Kwords of DARAM including real-time I/O buffers and stack

Normal memory library (with more memory)20.5 MHz (peak) and 18.5 MHz (average) for stereo 44.1 kHz at 128 kbit/s21 kwords of SARAM ( const and program ) and 11 Kwords of DARAM including real-time I/O buffers and stack

AAC EncoderFeatures

The given Encoder implements AAC Low complexity (LC) profile of the standard up to 2 channels (stereo).The given encoder supports both; ADIF (Audio Data Interchange Format) and ADTS (Audio Data Transport Stream) input formats.It supports sampling frequency from 8 kHz to 96 kHz as specified by the standard.Encoder outputs AAC file format from 16- bit raw PCM samples.It supports the bit-rate up to maximum depending upon sampling frequency as specified by the standard (see 8.2.2 of ISO/IEC 13818-7). It supports all necessary tools and features, so that the given algorithm is standard compliant.

Performance73 MHz (peak) and 64 MHz (average) for stereo 44.1 kHz at 128 kbit/s27 kwords data RAM including real-time I/O buffers14 kwords data ROM23 kwords program ROM

MP3 DecoderFeatures

layer-3 only16-bit PCM outputall sample rates (8 kHz to 48 kHz)all bit rates (8 kbit/s to 320 kbit/s)variable bit rate (VBR) modemono, stereo, dual-channel, joint stereoCRC error protectionautomatic bitstream synchronization and error recoveryaudible fast-forward/rewind modeTI eXpressDSP compliantC5500 large memory model object code library

Performance20.97 MHz (peak) and 15.38 MHz (average) for stereo 44.1 kHz at 128 kbit/s11 kwords data RAM including real-time I/O buffers and stack8 kwords data ROM10 kwords program ROMTested against MPEG-1 and -2 compliance suites

MP3pro Decoder 2/2Features

Spectral Band Replication Technology using ancillary data field ( designed by -Coding Technologies.)Supports .mp3 filesSupports a maximum of up to two channelsAll mode (stereo/joint stereo/dual channel/single channel)Ancillary data extractionVariable bit-rate decoding

Performance66 MHz (peak) and 59 MHz (average) for stereo 44.1 kHz at 96 kbit/s25.5 Kwords of Program memorymp3 part of mp3PRO decoder = 10Kwords , PRO part of mp3PRO decoder = 15.5Kwords35 Kwords of Data memorymp3 part of mp3PRO decoder = 17Kwords,PRO part of mp3PRO decoder = 18Kwords

MP3 Encoder

Performance93 MHz (peak) and 61 MHz (average) for stereo 44.1 kHz at 128 kbit/s24.2 kwords data RAM including real-time I/O buffers8.3 kwords data ROM30 kwords program ROM

ADPCM Decoder

DescriptionThe ADPCM decoder supports all the bit rates and sampling frequencies, which are specified in ITU G726 Standard. I.e. Sample Rates 8, 11.025, 16000, 22.050,32 and 44.1 KHz and 2,3,4 and 5 bit encoding/decoding of Mono streams.

Performance3.78 MIPS (peak) and 3.76 MIPS (average) for mono 8 kHz20.72 MIPS (peak) and 20.5 MIPS (average) for mono 44.1 kHz977 words data RAM1069 words program ROM

ADPCM Encoder

DescriptionThe ADPCM encoder supports all the bit rates and sampling frequencies, which are specified in ITU G726 Standard i.e. Sample Rates 8,11.025,16000,22.050,32 and 44.1 KHz and at 2,3,4 and 5 bit encoding/decoding. Mono encoding and decoding is supported.

Performance4.71 MIPS (peak) and 4.68 MIPS (average) for mono 8 kHz25.137 MIPS (peak) and 24.86 MIPS (average) for mono 44.1 kHz990 words data RAM1750 words program ROM

WMA Decoder

DescriptionThe WMA C55x all bit rate decoder supports all the bit rates, which are specified in WMA decoder Standard. It supports version 2, version 7 as well as version 8 encoded bit streams.

Performance44.4 MHz (peak) and 17.5 MHz (average) for stereo 44.1 kHz at 128 kbit/s27 kwords data RAM11.5 kwords data ROM16.5 kwords program ROM

5-band EqualizerFeatures

N bandsGains adjustable from –15db to +15 db in 1db steps Distortion free flat spectrum at 0 dB gainNo unpleasant audible artifacts during gain changes16-bit mono PCM input and output audio dataCenter frequencies and bandwidths are configurable off-lineImplemented by 2nd order IIR sections in cascadeTI eXpressDSP xDAIS compliant

Performance7.64 MIPS (peak) and 7.58 MIPS (average) for 5 bands4.24 MIPS (peak) and 4.24 MIPS (average) for 3 bands208 words data RAM including real-time I/O buffers266 words data ROM1427 words program ROM

Speech RecognitionSolutions from Scansoft: VoCon ASR 1/2

Key Features

VoCon is optimized for small vocabularies and offers the following advanced features:Continuous speech inputVoCon can recognize natural speech, so there is no need to leave gaps between words.Speaker Dependent RecognitionUser creates own vocabulary via user word training.Speaker Independent RecognitionVocabulary consists of optimized acoustic models corresponding to pre-defined grammar and word list.Whole-word and Phoneme-based RecognitionEnsures high recognition accuracy and the design of specific words and acronyms.Keyword ActivationFilters and recognizes keywords out of natural spoken sentences.Out-of-Vocabulary RejectionRejects words that are not part of the vocabulary.Speaker AdaptationAdapts speaker-specific language characteristics to speaker-independent systems.Dialogue CapabilitiesSupport the design of the user interface via grammars and integration of word probabilities.Context-switching capabilitiesEnsures shortest turnaround time when switching vocabulary.

Speech Recognition

Solutions from Scansoft: VoCon ASR 2/2

Performance (implementation on C5402)about 80 MIPS100 kwords program ROM112 kwords data RAM

Scansoft ASR software VoCon, ASR3200

Speech RecognitionSolutions from Advanced Recognition Technologies: smART Speak XG

Key FeaturesName dialing - speaker independentIncorporates core modules (smARTspeak NG):Name dialing - speaker dependentContinuous digit dialing - speaker independent/ dependentMenu navigation and device control – speaker independent / dependentUnique adaptation to user’s voice for enhanced performance with accents, while maintaining an excellent out of the box experience

Performanceruns on ARM or DSPabout 8 MIPS46.7 kwords program ROM167 kwords constant data (continuous digit dialing, 20 fixed commands, 100names)30.5 kwords data RAM






OMAP Audio Design


DSP Architectures

DSP & Analog AV Solutions

ROOM #1 – 7.1 ANALOG AMPLIFICATION

PCM1738/1791/1792

OrPCM1608/DSD1608

Or PCM1606

A/V Receiver System Diagram

Tuner

2-6 channelsof A to D

Conversion

5.1 Analog In

OpticalDigital

Receiver

1-3 Optical Digital InDVD/DVP

Set op BoxCD

S/PDIF In (high Quality)DVD/DVPSet Top Box

Mul

tiple

xer

16-20 Bit 48 Khz

R L

OPA134

FILTER8 Ch DAC VOLUME

PGA2310 AM

PLIFIER

OPA134 PGA2310

1

8

ROOM #2 – 7.1 DIGITAL AMPLIFICATIONPOWER STAGE DIGITAL PWM

TAS5110

TAS5110

TAS5110

TAS5110

OR

REMOTE SPEAKER

PCM1804

-Or-

PCM1802

-Or-

PCM1800/2

S/P DIF Transceiver

DIX1700

PCM1738/1791/1792

PCM1608/DSD1608

PCM1606

HIGH

MID

LOW

TAS5015

TAS5012

TAS5010

Or

Or

OMAP5910

Encoder/DecoderDolby Digital (AC-3)

WMAMultichannel AAC

MP3Effects

Bass Management Treble Control

Dolby HeadphoneVirtual 3-D

DownmixingHall Effects

S/PDIF Out eg. Mini Disc

mjc11

mjc11 I added 'one big arrow from DA610/C6714 to Room #1 and Room#2marilyn connell, 4/30/2003

A/V Receiver System Diagram

150W x 5.1 Channels(0.09% THD @ 8 ohms BTL)

1 x $2.32 3 x $1.93 24 x $0.20

$12.91 ($2.15 per channel)

TAS50466-Channel

PWM Processor100 dB

TAS51922-Channel

Gate Driver

TAS51922-Channel

Gate Driver

TAS51922-Channel

Gate Driver

TAS31066-Channel

Digital AudioPost-Processor

AudioDecode

OMAP5910Processor

L

R

LS

RS

C

LFE

mjc11

mjc11 I added 'one big arrow from DA610/C6714 to Room #1 and Room#2marilyn connell, 4/30/2003

Typical requirement is 32-48 bit dynamic range with a 48-76 bit precision for accumulator

AudioProcessing

Speaker EqualizerRoom Equalizer

Typical requirement is 24-30 bit dynamic range with 48-64 bit precision for accumulator

Typical requirement is 16-24 bit dynamic range with 32-48 bit precision for accumulator

AudioSource

Audio HiFiPlayback

Typical A/D16~24-bit @ 11.25k~48kHz

Typical DAC24-bit @96kHz

Interface format to typical DAC’s are fixed point.

No Floating-point DAC!!

A/V Receiver Applications: The Whole Gamut

AudioEncode/Decode

Dolby Digital (AC-3)DTSMultichannel AACEtc…

AudioProcessing

Dolby Pro LogicDTS NEO:6GraphicEqualizer

MP3 Jukebox System Diagram

Mass StorageFlash

Hard DriveIDE CD

Stereo A/D, D/A+ Amplifier

TLV320AIC23

3V VoltageRegulator

Power Supply

TPS61010

1.5V VoltageRegulatorTPS62000

OMAP5910

STEREO 16Ω, S.E.

40mW

Connectivity

WiFi 802.11

USBConnector

LCD

EEPROM24WC256

Crystal12 MHz

Crystal32.768 KHz

Batteries

MonitorButtons

IIC

PLL

RTCGPIOParallel

Bus

USB

LCD

MMC/SD

McBSP

GPIO

MMCSD

Audio Application MP3 Jukebox: Reduced Gamut

HiFi AudioHome TheaterSuper Audio CDDVD AudioHDD serverAV receiverDVD Receiver

Outdoor, MobileHome

Automotive AudioCar Audio and Entertainment

Portable AudioPortable CD, MP3PDA AudioWireless Phone AudioPortable Digital Radio

General AudioMini Stereo systemBoombox

Hi Fidelity

Casual

16-bit words, 44.1 kHz is the recreation goal of the MP3 decoderTradeoff bit rate vs. storage space Typically in a noisier environment than the A/V Receiver (i.e. what SNR do you need when you are jogging?)


CDQuality

And… Often other concerns like cost, power consumption, portability etc. outweigh performance for a « leaner » system

2002

Dolby™• Dolby Digital™ (AC3)• ProLogic™• ProLogic™ II• Dolby Headphone™

Fraunhoffer™• MPEG AAC (LC)

DTS™• Consumer 5.1™• ES 6.1™• Neo 6.1™• DTS 96/24™

Func

tiona

lity

Code Composer Studio

DSP/BIOS

xDAIS Algorithm Standard

Third Parties

Fully Programmable

Fixed Function devices

Low pin count

GUI tools

DA610/C6713

C6712C5509

C5401C5402

TAS3103

TAS3002/3004

Floating Point C67xx Core

Fixed Point C54xx & C55xx Core

Fixed Function Digital Audio Processors

Digital Audio Devices

TAS3001

Speaker EqualizationEffects

MP3 (encode/decode)AAC (encode/decode)Windows Media Audio™ (decode)ATRAC3™ (decode)Security features

C6711

C5410

5910C5510C5416

mjc12

mjc12 Rita I added C5502 but it needs positioningmarilyn connell, 4/30/2003






OMAP Audio Design


OMAP Audio Design

R610K

U1TLV320AIC23PW

LOUT 12

ROUT13

LLINEIN19

RLINEIN20

MICIN18

MICBIAS17

DIN4

LRCIN5

BCLK3

XTL_MCLK25XTO

26

CS#21

MODE22

BVDD 1

BVDD 27

DG

ND

28

LHPOUT9

RHPOUT 10

HP

GN

D11

AG

ND

15

VMID16

HPVDD8

AVDD14

CLKOUT 2

LRCOUT7

DOUT6

SDIN23

SCLK24

R710K

C4000.1UF S0T-23

Q1DK/MMBT3904ICT-ND

VCC_I/O

R4220K

MCBSP1_BCLK

R32.2K

R2100

+ C2100UF_(D)

+ C510.0UF

+ C410.0UF

RLINE_IN

LLINE_IN

MCBSP1_SYNC

R130

R90

VCC_I/O MCBSP1_DOUT

I2C_DATA

MCBSP1_DIN

I2C_CLK

2

3

1

3

1

2

MK1DK/P9925-ND

12

OMAP Audio Design

Microphone Preamp-Need to bypass the biasvoltage-Use op amp instead of transistor-> much betternoise rejection-Need to terminate McBSPbus

L210uH

GND AUDIO

GND AUDIO

C?

100nFGND AUDIO

C135B

470nF

+C13910UF16V

C142100nF

GND AUDIO

C143100nF

C144100nF

+C14110UF16V

+C14010UF16V

GND AUDIO

R187F

10K

GND AUDIO

VCC_AUDIO

C135B 100pF

R187C 100K

L110uH

VCC_AUDIO

+

C127B

1UF16V

R159B 10K

+

-

U119A

OPA348

1

34

52

RT120K

RT220K

+CT110UF16V

ampbias

R187E

5.6K

R187D 5.6K

+

C141B

10UF16V

GND AUDIO

J7

MIC I/F

123

GND AUDIO

ampbias

U43

TLV320AIC23

SCLK 24

XTO26 XTI/MCLK25

CLKOUT2

DGND 28

LRCOUT7

DOUT6

DIN4

LRCIN5

BCLK3

BVDD1

LHPOUT9

HPGND 11

HPVDD8

RHPOUT10LOUT12 ROUT13

AVDD 14

AGND15

MICBIAS17

VMID 16

MICIN18

RLINEIN19

LLINEIN 20

DVDD 27

MODE22 CS/21

SDIN23

OMAP Audio Design

Low noise micPre-amp circuit

OMAP Audio DesignU1TLV320AIC23PW

LOUT12

ROUT13

LLINEIN19

RLINEIN20

MICIN18

MICBIAS17

DIN4

LRCIN5

BCLK3

XTL_MCLK25

XTO26

CS#21

MODE22

BVDD1

BVDD 27

DG

ND

28

LHPOUT9

RHPOUT 10

HP

GN

D1

1A

GN

D1

5

VMID16

HPVDD 8

AVDD14

CLKOUT2

LRCOUT7

DOUT6

SDIN23

SCLK24

L6

100NH D/K#PCD1201CT-ND

SG-710(CA)

Q200DK/MMBT3904ICT-ND


U35DK/SG8002CAPCB-ND(16.9344MHZ)

OUT3

GND2

EN1

VCC4

R1210K

VCC_I/O

C9050.1UF


OUT3

GND2

EN1

VCC4

TP20TP

1

U22SN74AHC1G04DCKR

24

53

1

C9060.1UF

CLK_SEL

MCLK, Oversampling Clock = OS x Frame Sync-> OS = 256, 250, 272 or 384 oversampleFrame Sync = 44.1KHz, 48KHz or 32KHzBCLK has to be sync’ed with MCLKMaster CODEC drives BCLK & Frame Sync

Audio Clock Design

OMAP Audio DesignU1TLV320AIC23PW

LOUT12

ROUT13

LLINEIN19

RLINEIN20

MICIN18

MICBIAS17

DIN4

LRCIN5

BCLK3

XTL_MCLK25

XTO26

CS#21

MODE22

BVDD1

BVDD27

DGND

28

LHPOUT9

RHPOUT10

HPGN

D11

AGND

15

VMID16

HPVDD8

AVDD14

CLKOUT2

LRCOUT7

DOUT6

SDIN23

SCLK24

L6

100NH D/K#PCD1201CT-ND

SG-710(CA)

PW




OUT3

GND2

EN1

VCC4

C9000.1UF

+C90110.0UF

MCBSP1_CLKS

R19833

R1210K

MCBSP1_BCLK VCC_I/O

LLINE_IN

RLINE_IN

VCC_3V3MCBSP1_SYNC

VCC_I/O

C9050.1UF


OUT3

GND2

EN1

VCC4

U2TLV320AIC23PW

LOUT 12

ROUT13

LLINEIN19

RLINEIN20

MICIN18

MICBIAS17

DIN4

LRCIN5

BCLK3

XTL_MCLK25XTO

26

CS#21

MODE22

BVDD1

BVDD27

DGND

28

LHPOUT 9

RHPOUT10

HPGN

D11

AGND

15

VMID16

HPVDD8

AVDD14

CLKOUT2

LRCOUT7

DOUT6

SDIN23

SCLK24

PW

C9000.1UF

+C90110.0UF

MCBSP1_CLKS

R19833

VCC_I/O

LLINE_IN

RLINE_IN

VCC_3V3

TP20TP

1

MCBSP2_DOUT

I2C_DATA

MCBSP2_DIN

I2C_CLK

TP20TP

1R130

R90

DO NOT POPULATE

U22SN74AHC1G04DCKR

24

53

1

C9060.1UF

VCC_I/O MCBSP1_DOUT

I2C_DATA

MCBSP1_DIN

CLK_SEL

I2C_CLK

MULTIPLE CODECs DESIGN-Master device drives all FS and BCLKinputs of slave devices.-Only McBSP1 has SRG, generating different BCLK from one CLKS input.-Only one master is allowed for Sync CODECs






OMAP Audio Design



Power Amplification Issues

All equipment designers are striving for more compact, lighter, cheaper products with less heat generation and uncritical power require-ments, while preserving great sound quality

Switching amplifiers, both digital-input and analog-input, are the way to go for efficiency improvement and bulk reduction (smaller, or even no, heatsinks). Customers nervous about switching designs; TI has world-class experts to advise.

Small form factor is sexier - and cheaper to ship! Ultra low profile PCB design enables sub-50mm total product height

TI is pioneering Power Supply Compensation for digital-input amplifiers

Ease-of-use features of the latest analog-input switching amplifiers make them the solution of choice for sleek products like LCD television

The “Linear Danger Zone”

Pdiss (W)555045

353025201510

50

0.1 1 10Po each channel (W)

Power Dissipation (Sine Wave)

40Vs = 14.4 VRL = 4 x 4 Ohm

CLASS AB

Danger Zone!

TDAATDAA

Legacy Audio Amplifiers (Linear Class AB)

Efficiency: 40-60%• Bulky and heavy• Hot and noisy• Shorter battery life• Power hogs

Emerging Audio Amplifiers (TDAA Class D)

Efficiency: ~ 90%• Light and streamlined• Cool and quiet• Longer battery life • Energy efficient• Consumer friendly• Cosmetically pleasing

Class D Amplifier

Class D Amplifier

Higher Efficiency>90% Theoretical, / >85% Achievable

• Lower Current RequirementLonger Battery LifeLess Power Input for same Power Output

• Less Heat DissipationSmaller PackagesLess Heat Sinking

Smaller Power Supply Requirement• Less Weight

In General – “Greener”

Class D Amplifier

Superior Distortion CharacteristicsNo Slew-rate induced TIM (Transient Inter-Modulation Distortion)Very Low Feedback Required

Excellent Transient ResponseExcellent Open-loop Characteristics

Low Open-loop Output ImpedanceLess Back-EMF Interference

Psycho-acoustically Speaking“Tighter Bass Response”“Crisper/Cleaner High End”

TI Class D Amplifier

PCM to PWM

Processor

CD/DVD/DAB

/SACD/MP3

SwitchingOutput stage

24 bit32-96kHz

PCM DigitalPWMControl

V+

TAS50XXfamily

TAS51XXfamily

Digital Upsampler,Linearity Corrector

and Modulator

DigitalSource

TI Class D Amplifier

“Thunderbird”Power Amplifier

AC Power Strip TAS5100Digital Audio PWM Power Output Stage

TAS5100Digital Audio PWM Power Output Stage

TAS5010Digital Audio

PWM Processor

TAS3002 Digital Audio

Processor

Volume Control

Parallel Port Interface

DIR1703 S/PDIF

Receiver

I2S Interface

I2S Bus

I2S BusI2S Input

I2C BusAnalog Input

Analog Output

PC Software

32kHz - 192kHz

32kHz - 96kHz

Thunderbird Power Amplifier

Toslink CableCD

Player

24 bit44.1 kHz

PCM8 ohm Speaker

8 ohm Speaker

AC Power Strip 75 Watt Power Supply

TI Class D Amplifier EVM

digital audio seminar - Rice University Electrical and ...duarte/elec542/audio.pdf · Analog and...

Documents

Transcript of digital audio seminar - Rice University Electrical and ...duarte/elec542/audio.pdf · Analog and...