VS1053 B VS1053b preliminary VS1053b Ogg Vorbis/MP3/AAC/WMA/MIDI AUDIO CODEC Features Description • Decodes Ogg Vorbis; VS1053b is a single-chip Ogg Vorbis/MP3/AAC/MPEG 1 & 2 audio layer III (CBR +VBR WMA/MIDI audio decoder and an IMA ADPCM +ABR); layers I & II optional; and user-loadable Ogg Vorbis encoder. It contains MPEG4 / 2 AAC-LC(+PNS), a high-performance, proprietary low-power DSP HE-AAC v2 (Level 3) (SBR + PS); processor core VS DSP4 , working data memory, WMA 4.0/4.1/7/8/9 all profiles (5-384 kbps); 16 KiB instruction RAM and 0.5+ KiB data RAM WAV (PCM + IMA ADPCM); for user applications running simultaneously with General MIDI 1 / SP-MIDI format 0 files any built-in decoder, serial control and input data • Encodes Ogg Vorbis with software pluinterfaces, upto 8 general purpose I/O pins, an gin (available Q4/2007) UART, as well as a high-quality variable-samplerate stereo ADC (mic, line, line + mic or 2×line) • Encodes IMA ADPCM from mic/line (stereo) and stereo DAC, followed by an earphone ampli• Streaming support for MP3 and WAV fier and a common voltage buffer. • EarSpeaker Spatial Processing • Bass and treble controls VS1053b receives its input bitstream through a • Operates with a single 12..13 MHz clock serial input bus, which it listens to as a system • Can also be used with a 24..26 MHz clock slave. The input stream is decoded and passed through a digital volume control to an 18-bit over• Internal PLL clock multiplier sampling, multi-bit, sigma-delta DAC. The decod• Low-power operation ing is controlled via a serial control bus. In addi• High-quality on-chip stereo DAC with no tion to the basic decoding, it is possible to add phase error between channels application specific features, like DSP effects, to • Zero-cross detection for smooth volume the user RAM memory. change • Stereo earphone driver capable of driving a Optional factory-programmable unique chip ID pro30 Ω load vides basis for digital rights management or unit • Quiet power-on and power-off identification features. • I2S interface for external DAC I2S • Separate voltages for analog, digital, I/O audio VS1053 L Stereo Ear− Stereo Stereo • On-chip RAM for user code and data differential MUX MIC AMP mic / line 1 phone Driver ADC DAC R line 2 output • Serial control and data interfaces 8 GPIO GPIO • Can be used as a slave co-processor X ROM DREQ • SPI flash boot for special applications SO Serial SI X RAM Data/ • UART for debugging purposes 4 SCLK Control VSDSP Interface • New functions may be added with software XCS XDCS Y ROM and upto 8 GPIO pins RX • Lead-free RoHS-compliant package (Green) TX UART Y RAM Clock multiplier Version 0.5, 2007-12-03 Instruction RAM Instruction ROM 1 VLSI VS1053b preliminary y Solution VS1053 B CONTENTS Contents 1 Licenses 9 2 Disclaimer 9 3 Definitions 9 4 Characteristics & Specifications 10 4.1 Absolute Maximum Ratings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 4.2 Recommended Operating Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 4.3 Analog Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 4.4 Power Consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 4.5 Digital Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 4.6 Switching Characteristics - Boot Initialization . . . . . . . . . . . . . . . . . . . . . . . 12 5 Packages and Pin Descriptions 13 5.1 Packages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 5.1.1 13 LQFP-48 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Connection Diagram, LQFP-48 16 7 SPI Buses 18 7.1 General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 7.2 SPI Bus Pin Descriptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 7.2.1 VS1002 Native Modes (New Mode) . . . . . . . . . . . . . . . . . . . . . . . . 18 7.2.2 VS1001 Compatibility Mode (deprecated) . . . . . . . . . . . . . . . . . . . . . 18 Data Request Pin DREQ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 7.3 Version 0.5, 2007-12-03 2 VLSI VS1053b preliminary y Solution 7.4 CONTENTS Serial Protocol for Serial Data Interface (SDI) . . . . . . . . . . . . . . . . . . . . . . . 19 7.4.1 General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 7.4.2 SDI in VS1002 Native Modes (New Mode) . . . . . . . . . . . . . . . . . . . . 19 7.4.3 SDI in VS1001 Compatibility Mode (deprecated) . . . . . . . . . . . . . . . . . 20 7.4.4 Passive SDI Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 Serial Protocol for Serial Command Interface (SCI) . . . . . . . . . . . . . . . . . . . . 20 7.5.1 General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 7.5.2 SCI Read . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 7.5.3 SCI Write . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 7.5.4 SCI Multiple Write . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 7.6 SPI Timing Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 7.7 SPI Examples with SM SDINEW and SM SDISHARED set . . . . . . . . . . . . . . . 24 7.7.1 Two SCI Writes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 7.7.2 Two SDI Bytes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 7.7.3 SCI Operation in Middle of Two SDI Bytes . . . . . . . . . . . . . . . . . . . . 25 7.5 8 VS1053 B Functional Description 26 8.1 Main Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 8.2 Supported Audio Codecs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 8.2.1 Supported MP3 (MPEG layer III) Formats . . . . . . . . . . . . . . . . . . . . 26 8.2.2 Supported MP1 (MPEG layer I) Formats . . . . . . . . . . . . . . . . . . . . . 27 8.2.3 Supported MP2 (MPEG layer II) Formats . . . . . . . . . . . . . . . . . . . . . 27 8.2.4 Supported Ogg Vorbis Formats . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 8.2.5 Supported AAC (ISO/IEC 13818-7 and ISO/IEC 14496-3) Formats . . . . . . . 28 8.2.6 Supported WMA Formats . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 Version 0.5, 2007-12-03 3 VLSI VS1053b preliminary y Solution 9 VS1053 B CONTENTS 8.2.7 Supported RIFF WAV Formats . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 8.2.8 Supported MIDI Formats . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 8.3 Data Flow of VS1053b . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 8.4 EarSpeaker Spatial Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 8.5 Serial Data Interface (SDI) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 8.6 Serial Control Interface (SCI) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 8.7 SCI Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 8.7.1 SCI MODE (RW) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 8.7.2 SCI STATUS (RW) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 8.7.3 SCI BASS (RW) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 8.7.4 SCI CLOCKF (RW) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 8.7.5 SCI DECODE TIME (RW) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 8.7.6 SCI AUDATA (RW) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 8.7.7 SCI WRAM (RW) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 8.7.8 SCI WRAMADDR (W) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 8.7.9 SCI HDAT0 and SCI HDAT1 (R) . . . . . . . . . . . . . . . . . . . . . . . . . 45 8.7.10 SCI AIADDR (RW) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 8.7.11 SCI VOL (RW) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 8.7.12 SCI AICTRL[x] (RW) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 Operation 48 9.1 Clocking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 9.2 Hardware Reset . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 9.3 Software Reset . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 9.4 Low Power Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 Version 0.5, 2007-12-03 4 VLSI VS1053b preliminary y Solution 9.5 VS1053 B CONTENTS Play and Decode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 9.5.1 Playing a Whole File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 9.5.2 Cancelling Playback . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 9.5.3 Fast Play . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 9.5.4 Fast Forward and Rewind without Audio . . . . . . . . . . . . . . . . . . . . . 50 9.5.5 Maintaining Correct Decode Time . . . . . . . . . . . . . . . . . . . . . . . . . 51 9.6 Feeding PCM data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 9.7 Ogg Vorbis Recording . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 9.8 ADPCM Recording . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 9.8.1 Activating ADPCM mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 9.8.2 Reading IMA ADPCM Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 9.8.3 Adding a RIFF Header . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 9.8.4 Playing ADPCM Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 9.8.5 Sample Rate Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 SPI Boot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 9.10 Real-Time MIDI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 9.11 Extra Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 9.11.1 Common Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 9.11.2 WMA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 9.11.3 AAC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 9.11.4 Midi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 9.11.5 Ogg Vorbis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 9.12 SDI Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 9.12.1 Sine Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 9.9 Version 0.5, 2007-12-03 5 VLSI VS1053b preliminary y Solution VS1053 B CONTENTS 9.12.2 Pin Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 9.12.3 SCI Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 9.12.4 Memory Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 9.12.5 New Sine and Sweep Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 10 VS1053b Registers 65 10.1 Who Needs to Read This Chapter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 10.2 The Processor Core . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 10.3 VS1053b Memory Map . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 10.4 SCI Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 10.5 Serial Data Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 10.6 DAC Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 10.7 GPIO Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 10.8 Interrupt Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 10.9 Watchdog v1.0 2002-08-26 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 10.9.1 Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 10.10UART v1.1 2004-10-09 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69 10.10.1 Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69 10.10.2 Status UARTx STATUS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69 10.10.3 Data UARTx DATA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 10.10.4 Data High UARTx DATAH . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 10.10.5 Divider UARTx DIV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 10.10.6 Interrupts and Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 10.11Timers v1.0 2002-04-23 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 10.11.1 Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 Version 0.5, 2007-12-03 6 VLSI VS1053b preliminary y Solution VS1053 B CONTENTS 10.11.2 Configuration TIMER CONFIG . . . . . . . . . . . . . . . . . . . . . . . . . . 72 10.11.3 Configuration TIMER ENABLE . . . . . . . . . . . . . . . . . . . . . . . . . . 73 10.11.4 Timer X Startvalue TIMER Tx[L/H] . . . . . . . . . . . . . . . . . . . . . . . 73 10.11.5 Timer X Counter TIMER TxCNT[L/H] . . . . . . . . . . . . . . . . . . . . . . 73 10.11.6 Interrupts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 10.12VS1053b Audio Path . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 10.13I2S DAC Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 10.13.1 Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 10.13.2 Configuration I2S CONFIG . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 11 VS1053 Version Changes 11.1 Changes Between VS1033c and VS1053a/b Firmware, 2007-03-08 . . . . . . . . . . . . 76 76 12 Document Version Changes 78 13 Contact Information 79 Version 0.5, 2007-12-03 7 VLSI VS1053b preliminary y Solution VS1053 B LIST OF FIGURES List of Figures 1 Pin Configuration, LQFP-48. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 2 VS1053b in LQFP-48 Packaging. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 3 Typical Connection Diagram Using LQFP-48. . . . . . . . . . . . . . . . . . . . . . . . 16 4 BSYNC Signal - one byte transfer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 5 BSYNC Signal - two byte transfer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 6 SCI Word Read . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 7 SCI Word Write . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 8 SCI Multiple Word Write . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 9 SPI Timing Diagram. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 10 Two SCI Operations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 11 Two SDI Bytes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 12 Two SDI Bytes Separated By an SCI Operation. . . . . . . . . . . . . . . . . . . . . . . 25 13 Data Flow of VS1053b. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 14 EarSpeaker externalized sound sources vs. normal inside-the-head sound . . . . . . . . . 35 15 RS232 Serial Interface Protocol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69 16 VS1053b ADC and DAC data paths . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 17 I2S Interface, 192 kHz. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 Version 0.5, 2007-12-03 8 VLSI VS1053b preliminary y Solution 1 VS1053 B 1. LICENSES Licenses MPEG Layer-3 audio decoding technology licensed from Fraunhofer IIS and Thomson. Note: If you enable Layer I and Layer II decoding, you are liable for any patent issues that may arise from using these formats. Joint licensing of MPEG 1.0 / 2.0 Layer III does not cover all patents pertaining to layers I and II. VS1053b contains WMA decoding technology from Microsoft. This product is protected by certain intellectual property rights of Microsoft and cannot be used or further distributed without a license from Microsoft. VS1053b contains AAC technology (ISO/IEC 13818-7 and ISO/IEC 14496-3) which cannot be used without a proper license from Via Licensing Corporation or individual patent holders. VS1053b contains spectral band replication (SBR) and parametric stereo (PS) technologies developed by Coding Technologies. Licensing of SBR is handled within MPEG4 through Via Licensing Corporation. Licensing of PS is handled with Coding Technologies. See http://www.codingtechnologies.com/licensing/aacplus.htm for more information. To the best of our knowledge, if the end product does not play a specific format that otherwise would require a customer license: MPEG 1.0/2.0 layers I and II, WMA, or AAC, the respective license should not be required. Decoding of MPEG layers I and II are disabled by default, and WMA and AAC format exclusion can be easily performed based on the contents of the SCI HDAT1 register. Also PS and SBR decoding can be separately disabled. 2 Disclaimer This is a preliminary datasheet. All properties and figures are subject to change. 3 Definitions B Byte, 8 bits. b Bit. Ki “Kibi” = 210 = 1024 (IEC 60027-2). Mi “Mebi” = 220 = 1048576 (IEC 60027-2). VS DSP VLSI Solution’s DSP core. W Word. In VS DSP, instruction words are 32-bit and data words are 16-bit wide. Version 0.5, 2007-12-03 9 VLSI VS1053b preliminary y Solution 4 4. CHARACTERISTICS & SPECIFICATIONS Characteristics & Specifications 4.1 Absolute Maximum Ratings Parameter Analog Positive Supply Digital Positive Supply I/O Positive Supply Current at Any Digital Output Voltage at Any Digital Input Operating Temperature Storage Temperature 1 VS1053 B Symbol AVDD CVDD IOVDD Min -0.3 -0.3 -0.3 -0.3 -40 -65 Max 3.6 2.0 3.6 ±50 IOVDD+0.31 +85 +150 Unit V V V mA V ◦C ◦C Must not exceed 3.6 V 4.2 Recommended Operating Conditions Parameter Ambient Operating Temperature Analog and Digital Ground 1 Positive Analog Positive Digital I/O Voltage Input Clock Frequency2 Internal Clock Frequency Internal Clock Multiplier3 Master Clock Duty Cycle Symbol AGND DGND AVDD CVDD IOVDD XTALI CLKI Min -40 2.5 1.65 CVDD-0.6V 12 12 1.0× 40 Typ 0.0 2.8 1.8 2.8 12.288 36.864 3.0× 50 Max +85 3.6 2.0 3.6 13 55.3 4.5× 60 Unit ◦C V V V V MHz MHz % 1 Must be connected together as close the device as possible for latch-up immunity. The maximum samplerate is XTALI/256. XTALI must be at least 12.288 MHz to play 48 kHz at correct speed. 3 Reset value is 1.0×. Recommended SC MULT=3.5×, SC ADD=1.0× (SCI CLOCKF=0x8800). 2 Version 0.5, 2007-12-03 10 VLSI VS1053b preliminary y Solution 4.3 VS1053 B 4. CHARACTERISTICS & SPECIFICATIONS Analog Characteristics Unless otherwise noted: AVDD=2.5..2.85V, CVDD=1.8V, IOVDD=CVDD-0.6V..3.6V, TA=-40..+85◦ C, XTALI=12..13MHz, Internal Clock Multiplier 3.5×. DAC tested with 1307.894 Hz full-scale output sinewave, measurement bandwidth 20..20000 Hz, analog output load: LEFT to GBUF 30Ω, RIGHT to GBUF 30Ω. Microphone test amplitude 50 mVpp, fs =1 kHz, Line input test amplitude 1.1 V, fs =1 kHz. Parameter DAC Resolution Total Harmonic Distortion Dynamic Range (DAC unmuted, A-weighted) S/N Ratio (full scale signal) Interchannel Isolation (Cross Talk) Interchannel Isolation (Cross Talk), with GBUF Interchannel Gain Mismatch Frequency Response Full Scale Output Voltage (Peak-to-peak) Deviation from Linear Phase Analog Output Load Resistance Analog Output Load Capacitance Microphone input amplifier gain Microphone input amplitude Microphone Total Harmonic Distortion Microphone S/N Ratio Line input amplitude Line input Total Harmonic Distortion Line input S/N Ratio Line and Microphone input impedances Symbol Min THD IDR SNR 50 -0.5 -0.1 1.3 AOLR 16 Typ 18 0.03 100 94 75 40 1.51 Max 0.1 0.5 0.1 1.7 5 302 100 MICG MTHD MSNR 50 LTHD LSNR 26 50 0.02 62 2200 0.06 92 100 1403 0.10 28003 0.10 Unit bits % dB dB dB dB dB dB Vpp ◦ Ω pF dB mVpp AC % dB mVpp AC % dB kΩ 1 3.0 volts can be achieved with +-to-+ wiring for mono difference sound. AOLR may be much lower, but below Typical distortion performance may be compromised. 3 Above typical amplitude the Harmonic Distortion increases. 2 4.4 Power Consumption Tested with an MPEG 1.0 Layer-3 128 kbps sample and generated sine. Output at full volume. Internal clock multiplier 3.0×. Parameter Power Supply Consumption AVDD, Reset Power Supply Consumption CVDD = 1.8V, Reset Power Supply Consumption AVDD, sine test, 30Ω + GBUF Power Supply Consumption CVDD = 1.8V, sine test Power Supply Consumption AVDD, no load Power Supply Consumption AVDD, output load 30Ω Power Supply Consumption AVDD, 30Ω + GBUF Power Supply Consumption CVDD = 1.8V Version 0.5, 2007-12-03 Min Typ 0.6 4 36.9 6 7 11 16 10 Max 5.0 50.0 Unit µA µA mA mA mA mA mA mA 11 VLSI VS1053b preliminary y Solution 4.5 2 4. CHARACTERISTICS & SPECIFICATIONS Digital Characteristics Parameter High-Level Input Voltage Low-Level Input Voltage High-Level Output Voltage at IO = -2.0 mA Low-Level Output Voltage at IO = 2.0 mA Input Leakage Current SPI Input Clock Frequency 2 Rise time of all output pins, load = 50 pF 1 VS1053 B Symbol Must not exceed 3.6V Value for SCI reads. SCI and SDI writes allow Min 0.7×IOVDD -0.2 0.7×IOVDD Typ Max IOVDD+0.31 0.3×IOVDD 0.3×IOVDD 1.0 -1.0 CLKI 6 50 Unit V V V V µA MHz ns CLKI 4 . 4.6 Switching Characteristics - Boot Initialization Parameter XRESET active time XRESET inactive to software ready Power on reset, rise time to CVDD 1 Symbol Min 2 22000 10 Max 500001 Unit XTALI XTALI V/s DREQ rises when initialization is complete. You should not send any data or commands before that. Version 0.5, 2007-12-03 12 VLSI VS1053b preliminary y Solution 5 VS1053 B 5. PACKAGES AND PIN DESCRIPTIONS Packages and Pin Descriptions 5.1 Packages LPQFP-48 is a lead (Pb) free and also RoHS compliant package. RoHS is a short name of Directive 2002/95/EC on the restriction of the use of certain hazardous substances in electrical and electronic equipment. 5.1.1 LQFP-48 48 1 Figure 1: Pin Configuration, LQFP-48. LQFP-48 package dimensions are at http://www.vlsi.fi/ . Figure 2: VS1053b in LQFP-48 Packaging. Version 0.5, 2007-12-03 13 VLSI VS1053b preliminary y Solution Pad Name MICP / LINE1 MICN XRESET DGND0 CVDD0 IOVDD0 CVDD1 DREQ GPIO2 / DCLK1 GPIO3 / SDATA1 GPIO6 / I2S SCLK3 GPIO7 / I2S SDATA3 XDCS / BSYNC1 IOVDD1 VCO DGND1 XTALO XTALI IOVDD2 DGND2 DGND3 DGND4 XCS CVDD2 GPIO5 / I2S MCLK3 RX TX SCLK SI SO CVDD3 XTEST GPIO0 GPIO1 GND GPIO4 / I2S LROUT3 AGND0 AVDD0 RIGHT AGND1 AGND2 GBUF AVDD1 RCAP AVDD2 LEFT AGND3 LINE2 LQFP Pin 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 Pin Type AI AI DI DGND CPWR IOPWR CPWR DO DIO DIO DIO DIO DI IOPWR DO DGND AO AI IOPWR DGND DGND DGND DI CPWR DIO DI DO DI DI DO3 CPWR DI DIO DIO DGND DIO APWR APWR AO APWR APWR AO APWR AIO APWR AO APWR AI VS1053 B 5. PACKAGES AND PIN DESCRIPTIONS Function Positive differential mic input, self-biasing / Line-in 1 Negative differential mic input, self-biasing Active low asynchronous reset, schmitt-trigger input Core & I/O ground Core power supply I/O power supply Core power supply Data request, input bus General purpose IO 2 / serial input data bus clock General purpose IO 3 / serial data input General purpose IO 6 / I2S SCLK General purpose IO 7 / I2S SDATA Data chip select / byte sync I/O power supply For testing only (Clock VCO output) Core & I/O ground Crystal output Crystal input I/O power supply Core & I/O ground Core & I/O ground Core & I/O ground Chip select input (active low) Core power supply General purpose IO 5 / I2S MCLK UART receive, connect to IOVDD if not used UART transmit Clock for serial bus Serial input Serial output Core power supply Reserved for test, connect to IOVDD Gen. purp. IO 0 (SPIBOOT), use 100 kΩ pull-down resistor2 General purpose IO 1 I/O Ground General purpose IO 4 / I2S LROUT Analog ground, low-noise reference Analog power supply Right channel output Analog ground Analog ground Common buffer for headphones, do NOT connect to ground! Analog power supply Filtering capacitance for reference Analog power supply Left channel output Analog ground Line-in 2 (right channel) 1 First pin function is active in New Mode, latter in Compatibility Mode. Unless pull-down resistor is used, SPI Boot is tried. See Chapter 9.9 for details. 3 If I2S CF ENA is ’0’ the pins are used for GPIO. See Chapter 10.13 for details. 2 Version 0.5, 2007-12-03 14 VLSI VS1053b preliminary y Solution VS1053 B 5. PACKAGES AND PIN DESCRIPTIONS Pin types: Type DI DO DIO DO3 AI Version 0.5, Description Digital input, CMOS Input Pad Digital output, CMOS Input Pad Digital input/output Digital output, CMOS Tri-stated Output Pad Analog input 2007-12-03 Type AO AIO APWR DGND CPWR IOPWR Description Analog output Analog input/output Analog power supply pin Core or I/O ground pin Core power supply pin I/O power supply pin 15 VLSI VS1053b preliminary y Solution 6 VS1053 B 6. CONNECTION DIAGRAM, LQFP-48 Connection Diagram, LQFP-48 Figure 3: Typical Connection Diagram Using LQFP-48. Figure 3 shows a typical connection diagram for VS1053. Figure Note 1: Connect either Microphone In or Line In, but not both at the same time. Note: This connection assumes SM SDINEW is active (see Chapter 8.7.1). If also SM SDISHARE is used, xDCS should be tied low or high (see Chapter 7.2.1). Version 0.5, 2007-12-03 16 VLSI VS1053b preliminary y Solution VS1053 B 6. CONNECTION DIAGRAM, LQFP-48 The common buffer GBUF can be used for common voltage (1.24 V) for earphones. This will eliminate the need for large isolation capacitors on line outputs, and thus the audio output pins from VS1053b may be connected directly to the earphone connector. GBUF must NOT be connected to ground under any circumstances. If GBUF is not used, LEFT and RIGHT must be provided with coupling capacitors. To keep GBUF stable, you should always have the resistor and capacitor even when GBUF is not used. See application notes for details. Unused GPIO pins should have a pull-down resistor. Unused line and microphone inputs should not be connected. If UART is not used, RX should be connected to IOVDD and TX be unconnected. Do not connect any external load to XTALO. Version 0.5, 2007-12-03 17 VLSI VS1053b preliminary y Solution 7 VS1053 B 7. SPI BUSES SPI Buses 7.1 General The SPI Bus - that was originally used in some Motorola devices - has been used for both VS1053b’s Serial Data Interface SDI (Chapters 7.4 and 8.5) and Serial Control Interface SCI (Chapters 7.5 and 8.6). 7.2 SPI Bus Pin Descriptions 7.2.1 VS1002 Native Modes (New Mode) These modes are active on VS1053b when SM SDINEW is set to 1 (default at startup). DCLK and SDATA are not used for data transfer and they can be used as general-purpose I/O pins (GPIO2 and GPIO3). BSYNC function changes to data interface chip select (XDCS). SDI Pin XDCS SCI Pin XCS SCK SI - 7.2.2 SO Description Active low chip select input. A high level forces the serial interface into standby mode, ending the current operation. A high level also forces serial output (SO) to high impedance state. If SM SDISHARE is 1, pin XDCS is not used, but the signal is generated internally by inverting XCS. Serial clock input. The serial clock is also used internally as the master clock for the register interface. SCK can be gated or continuous. In either case, the first rising clock edge after XCS has gone low marks the first bit to be written. Serial input. If a chip select is active, SI is sampled on the rising CLK edge. Serial output. In reads, data is shifted out on the falling SCK edge. In writes SO is at a high impedance state. VS1001 Compatibility Mode (deprecated) This mode is active when SM SDINEW is set to 0. In this mode, DCLK, SDATA and BSYNC are active. SDI Pin - SCI Pin XCS BSYNC DCLK SCK SDATA - SI SO Version 0.5, 2007-12-03 Description Active low chip select input. A high level forces the serial interface into standby mode, ending the current operation. A high level also forces serial output (SO) to high impedance state. SDI data is synchronized with a rising edge of BSYNC. Serial clock input. The serial clock is also used internally as the master clock for the register interface. SCK can be gated or continuous. In either case, the first rising clock edge after XCS has gone low marks the first bit to be written. Serial input. SI is sampled on the rising SCK edge, if XCS is low. Serial output. In reads, data is shifted out on the falling SCK edge. In writes SO is at a high impedance state. 18 VLSI VS1053b preliminary y Solution 7.3 VS1053 B 7. SPI BUSES Data Request Pin DREQ The DREQ pin/signal is used to signal if VS1053b’s 2048-byte FIFO is capable of receiving data. If DREQ is high, VS1053b can take at least 32 bytes of SDI data or one SCI command. DREQ is turned low when the stream buffer is too full and for the duration of a SCI command. Because of the 32-byte safety area, the sender may send upto 32 bytes of SDI data at a time without checking the status of DREQ, making controlling VS1053b easier for low-speed microcontrollers. Note: DREQ may turn low or high at any time, even during a byte transmission. Thus, DREQ should only be used to decide whether to send more bytes. It does not need to abort a transmission that has already started. Note: In VS10XX products upto VS1002, DREQ was only used for SDI. In VS1053b DREQ is also used to tell the status of SCI. There are cases when you still want to send SCI commands when DREQ is low. Because DREQ is shared between SDI and SCI, you can not determine if a SCI command has been executed if SDI is not ready to receive. In this case you need a long enough delay after every SCI command to make certain none of them is missed. The SCI Registers table in section 8.7 gives the worst-case handling time for each SCI register write. 7.4 7.4.1 Serial Protocol for Serial Data Interface (SDI) General The serial data interface operates in slave mode so DCLK signal must be generated by an external circuit. Data (SDATA signal) can be clocked in at either the rising or falling edge of DCLK (Chapter 8.7). VS1053b assumes its data input to be byte-sychronized. SDI bytes may be transmitted either MSb or LSb first, depending of contents of SCI MODE (Chapter 8.7.1). The firmware is able to accept the maximum bitrate the SDI supports. 7.4.2 SDI in VS1002 Native Modes (New Mode) In VS1002 native modes (SM NEWMODE is 1), byte synchronization is achieved by XDCS. The state of XDCS may not change while a data byte transfer is in progress. To always maintain data synchronization even if there may be glitches in the boards using VS1053b, it is recommended to turn XDCS every now and then, for instance once after every disk data block, just to make sure the host and VS1053b are in sync. If SM SDISHARE is 1, the XDCS signal is internally generated by inverting the XCS input. For new designs, using VS1002 native modes are recommended. Version 0.5, 2007-12-03 19 VLSI y Solution 7.4.3 VS1053 B VS1053b preliminary 7. SPI BUSES SDI in VS1001 Compatibility Mode (deprecated) BSYNC SDATA D7 D6 D5 D4 D3 D2 D1 D0 DCLK Figure 4: BSYNC Signal - one byte transfer. When VS1053b is running in VS1001 compatibility mode, a BSYNC signal must be generated to ensure correct bit-alignment of the input bitstream. The first DCLK sampling edge (rising or falling, depending on selected polarity), during which the BSYNC is high, marks the first bit of a byte (LSB, if LSB-first order is used, MSB, if MSB-first order is used). If BSYNC is ’1’ when the last bit is received, the receiver stays active and next 8 bits are also received. BSYNC SDATA D7 D6 D5 D4 D3 D2 D1 D0 D7 D6 D5 D4 D3 D2 D1 D0 DCLK Figure 5: BSYNC Signal - two byte transfer. 7.4.4 Passive SDI Mode If SM NEWMODE is 0 and SM SDISHARE is 1, the operation is otherwise like the VS1001 compatibility mode, but bits are only received while the BSYNC signal is ’1’. Rising edge of BSYNC is still used for synchronization. 7.5 Serial Protocol for Serial Command Interface (SCI) 7.5.1 General The serial bus protocol for the Serial Command Interface SCI (Chapter 8.6) consists of an instruction byte, address byte and one 16-bit data word. Each read or write operation can read or write a single register. Data bits are read at the rising edge, so the user should update data at the falling edge. Bytes are always send MSb first. XCS should be low for the full duration of the operation, but you can have pauses between bits if needed. The operation is specified by an 8-bit instruction opcode. The supported instructions are read and write. See table below. Instruction Name Opcode Operation READ 0b0000 0011 Read data WRITE 0b0000 0010 Write data Note: VS1053b sets DREQ low after each SCI operation. The duration depends on the operation. It is not allowed to finish a new SCI/SDI operation before DREQ is high again. Version 0.5, 2007-12-03 20 VLSI y Solution 7.5.2 VS1053 B VS1053b preliminary 7. SPI BUSES SCI Read XCS 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 0 0 0 0 0 0 1 1 0 0 0 30 31 SCK 3 SI instruction (read) 2 1 0 don’t care 0 data out address 15 14 SO 0 0 0 0 0 0 0 0 0 0 0 0 0 don’t care 0 0 1 0 0 X execution DREQ Figure 6: SCI Word Read VS1053b registers are read from using the following sequence, as shown in Figure 6. First, XCS line is pulled low to select the device. Then the READ opcode (0x3) is transmitted via the SI line followed by an 8-bit word address. After the address has been read in, any further data on SI is ignored by the chip. The 16-bit data corresponding to the received address will be shifted out onto the SO line. XCS should be driven high after data has been shifted out. DREQ is driven low for a short while when in a read operation by the chip. This is a very short time and doesn’t require special user attention. 7.5.3 SCI Write XCS 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 SI 0 0 0 0 0 0 1 0 0 0 0 SO 0 0 0 0 0 0 0 30 31 SCK 3 instruction (write) 0 0 0 0 2 1 0 15 14 1 data out address 0 0 X 0 0 0 0 0 0 0 0 0 X execution DREQ Figure 7: SCI Word Write VS1053b registers are written from using the following sequence, as shown in Figure 7. First, XCS line is pulled low to select the device. Then the WRITE opcode (0x2) is transmitted via the SI line followed by an 8-bit word address. Version 0.5, 2007-12-03 21 VLSI y Solution VS1053 B VS1053b preliminary 7. SPI BUSES After the word has been shifted in and the last clock has been sent, XCS should be pulled high to end the WRITE sequence. After the last bit has been sent, DREQ is driven low for the duration of the register update, marked “execution” in the figure. The time varies depending on the register and its contents (see table in Chapter 8.7 for details). If the maximum time is longer than what it takes from the microcontroller to feed the next SCI command or SDI byte, status of DREQ must be checked before finishing the next SCI/SDI operation. 7.5.4 SCI Multiple Write XCS 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 SI 0 0 0 0 0 0 1 0 0 0 0 SO 0 0 0 0 0 0 0 29 30 31 32 33 m−2m−1 SCK 3 instruction (write) 0 0 0 0 2 1 0 15 14 1 0 data out 1 address 0 15 14 1 0 X 0 0 0 0 0 0 0 X data out 2 d.out n 0 0 0 execution 0 0 0 X execution DREQ Figure 8: SCI Multiple Word Write VS1053b allows for the user to send multiple words to the same SCI register, which allows fast SCI uploads, shown in Figure 8. The main difference to a single write is that instead of bringing XCS up after sending the last bit of a data word, the next data word is sent immediately. After the last data word, XCS is driven high as with a single word write. After the last bit of a word has been sent, DREQ is driven low for the duration of the register update, marked “execution” in the figure. The time varies depending on the register and its contents (see table in Chapter 8.7 for details). If the maximum time is longer than what it takes from the microcontroller to feed the next SCI command or SDI byte, status of DREQ must be checked before finishing the next SCI/SDI operation. Version 0.5, 2007-12-03 22 VLSI y Solution 7.6 VS1053 B VS1053b preliminary 7. SPI BUSES SPI Timing Diagram tWL tXCSS tWH tXCSH XCS 0 1 14 15 30 16 31 tXCS SCK SI tH tSU SO tZ tV tDIS Figure 9: SPI Timing Diagram. Symbol tXCSS tSU tH tZ tWL tWH tV tXCSH tXCS tDIS 1 Min 5 -26 2 0 2 2 Max 2 (+ 25ns1 ) -26 2 10 Unit ns ns CLKI cycles ns CLKI cycles CLKI cycles CLKI cycles ns CLKI cycles ns 25ns is when pin loaded with 100pF capacitance. The time is shorter with lower capacitance. Note: tWL and tWH, as well as tH require at least 2 clock cycles, so the maximum speed for the SPI bus that can easily be used is 1/6 of VS1053b’s internal clock speed CLKI. Slightly higher speed can be achieved with very careful timing tuning. For details, see Application Notes for VS10XX. Note: The system always starts up in 1.0× mode, thus CLKI=XTALI at startup. Note: Negative numbers mean that the signal can change in different order from what is shown in the diagram. Version 0.5, 2007-12-03 23 VLSI y Solution 7.7 7.7.1 VS1053 B VS1053b preliminary 7. SPI BUSES SPI Examples with SM SDINEW and SM SDISHARED set Two SCI Writes SCI Write 1 SCI Write 2 XCS 0 1 2 3 30 31 1 0 32 33 61 62 63 2 1 0 SCK SI 0 0 0 0 X 0 0 X DREQ up before finishing next SCI write DREQ Figure 10: Two SCI Operations. Figure 10 shows two consecutive SCI operations. Note that xCS must be raised to inactive state between the writes. Also DREQ must be respected as shown in the figure. 7.7.2 Two SDI Bytes SDI Byte 1 SDI Byte 2 XCS 0 1 2 3 7 6 5 4 6 7 8 9 1 0 7 6 13 14 15 2 1 0 SCK 3 SI 5 X DREQ Figure 11: Two SDI Bytes. SDI data is synchronized with a raising edge of xCS as shown in Figure 11. However, every byte doesn’t need separate synchronization. Version 0.5, 2007-12-03 24 VLSI y Solution 7.7.3 VS1053 B VS1053b preliminary 7. SPI BUSES SCI Operation in Middle of Two SDI Bytes SDI Byte SDI Byte SCI Operation XCS 0 1 7 8 9 39 40 41 7 6 46 47 1 0 SCK 7 SI 6 5 1 0 0 5 X 0 DREQ high before end of next transfer DREQ Figure 12: Two SDI Bytes Separated By an SCI Operation. Figure 12 shows how an SCI operation is embedded in between SDI operations. xCS edges are used to synchronize both SDI and SCI. Remember to respect DREQ as shown in the figure. Version 0.5, 2007-12-03 25 VLSI y Solution 8 VS1053 B VS1053b preliminary 8. FUNCTIONAL DESCRIPTION Functional Description 8.1 Main Features VS1053b is based on a proprietary digital signal processor, VS DSP. It contains all the code and data memory needed for Ogg Vorbis, MP3, AAC, WMA and WAV PCM + ADPCM audio decoding, MIDI synthesizer, together with serial interfaces, a multirate stereo audio DAC and analog output amplifiers and filters. Also ADPCM audio encoding is supported using a microphone amplifier and/or line-level inputs and a stereo A/D converter. A UART is provided for debugging purposes. 8.2 Supported Audio Codecs Mark + ? - 8.2.1 Conventions Description Format is supported Format is supported but not thoroughly tested Format exists but is not supported Format doesn’t exist Supported MP3 (MPEG layer III) Formats MPEG 1.01 : Samplerate / Hz 48000 44100 32000 32 + + + 40 + + + 48 + + + 56 + + + 64 + + + 80 + + + Bitrate / kbit/s 96 112 128 + + + + + + + + + 160 + + + 192 + + + 224 + + + 256 + + + 320 + + + 8 + + + 16 + + + 24 + + + 32 + + + 40 + + + 48 + + + Bitrate / kbit/s 56 64 80 + + + + + + + + + 96 + + + 112 + + + 128 + + + 144 + + + 160 + + + 8 + + + 16 + + + 24 + + + 32 + + + 40 + + + 48 + + + Bitrate / kbit/s 56 64 80 + + + + + + + + + 96 + + + 112 + + + 128 + + + 144 + + + 160 + + + MPEG 2.01 : Samplerate / Hz 24000 22050 16000 MPEG 2.51 : Samplerate / Hz 12000 11025 8000 1 Also all variable bitrate (VBR) formats are supported. Version 0.5, 2007-12-03 26 VLSI y Solution 8.2.2 VS1053 B VS1053b preliminary 8. FUNCTIONAL DESCRIPTION Supported MP1 (MPEG layer I) Formats Note: Layer I / II decoding must be specifically enabled from register SCI MODE. MPEG 1.0: Samplerate / Hz 48000 44100 32000 32 + + + 64 + + + 96 + + + 128 + + + 160 + + + Bitrate / kbit/s 192 224 256 288 + + + + + + + + + + + + 320 + + + 352 + + + 384 + + + 416 + + + 448 + + + 32 ? ? ? 48 ? ? ? 56 ? ? ? 64 ? ? ? 80 ? ? ? 96 ? ? ? Bitrate / kbit/s 112 128 144 ? ? ? ? ? ? ? ? ? 160 ? ? ? 176 ? ? ? 192 ? ? ? 224 ? ? ? 256 ? ? ? MPEG 2.0: Samplerate / Hz 24000 22050 16000 8.2.3 Supported MP2 (MPEG layer II) Formats Note: Layer I / II decoding must be specifically enabled from register SCI MODE. MPEG 1.0: Samplerate / Hz 48000 44100 32000 32 + + + 48 + + + 56 + + + 64 + + + 80 + + + 96 + + + Bitrate / kbit/s 112 128 160 + + + + + + + + + 192 + + + 224 + + + 256 + + + 320 + + + 384 + + + 8 + + + 16 + + + 24 + + + 32 + + + 40 + + + 48 + + + Bitrate / kbit/s 56 64 80 + + + + + + + + + 96 + + + 112 + + + 128 + + + 144 + + + 160 + + + MPEG 2.0: Samplerate / Hz 24000 22050 16000 8.2.4 Supported Ogg Vorbis Formats Parameter Channels Window size Samplerate Bitrate Min 64 Max 2 4096 48000 500 Unit samples Hz kbit/sec Only floor 1 is supported. No known current encoder uses floor 0. All one- and two-channel Ogg Vorbis files should be playable with this decoder. Version 0.5, 2007-12-03 27 VLSI VS1053b preliminary y Solution 8.2.5 VS1053 B 8. FUNCTIONAL DESCRIPTION Supported AAC (ISO/IEC 13818-7 and ISO/IEC 14496-3) Formats VS1053b decodes MPEG2-AAC-LC-2.0.0.0 and MPEG4-AAC-LC-2.0.0.0 streams, i.e. the low complexity profile with maximum of two channels can be decoded. If a stream contains more than one element and/or element type, you can select which one to decode from the 16 single-channel, 16 channelpair, and 16 low-frequency elements. The default is to select the first one that appears in the stream. Dynamic range control (DRC) is supported and can be controlled by the user to limit or enhance the dynamic range of the material that contains DRC information. Both Sine window and Kaiser-Bessel-derived window are supported. For MPEG4 pseudo-random noise substitution (PNS) is supported. Short frames (120 and 960 samples) are not supported. Spectral Band Replication (SBR) level 3, and Parametric Stereo (PS) level 3 are supported (HE-AAC v2). Level 3 means that maximum of 2 channels, samplerates upto and including 48 kHz without and with SBR (with or without PS) are supported. Also, both mixing modes (Ra and Rb ), IPD/OPD synthesis and 34 frequency bands resolution are implemented. The downsampled synthesis mode (core coder samplerates > 24 kHz and <= 48 kHz with SBR) is implemented. SBR and PS decoding can also be disabled. Also different operating modes can be selected. See config1 and sbrAndPsStatus in section 9.11 : ”Extra parameters”. If enabled, the internal clock (CLKI) is automatically increased if AAC decoding needs a higher clock. PS and SBR operation can also be automatically switched off if the internal clock is too slow for correct decoding. For AAC the streaming ADTS format is recommended. This format allows easy rewind and fast forward because resynchronization is easily possible. In addition to ADTS (.aac), MPEG2 ADIF (.aac) and MPEG4 AUDIO (.mp4 / .m4a) files are played, but these formats are less suitable for rewind and fast forward operations. You can still implement these features by using the safe jump points table, or using slightly less robust but much easier automatic resync mechanism (see Section 9.5.4). Because 3GPP (.3gp) and 3GPPv2 (.3g2) files are just MPEG4 files, those that contain only HE-AAC or HE-AACv2 content are played. Note: To be able to play the .3gp, .3g2, .mp4 and .m4a files, the mdat atom must be the last atom in the MP4 file. Because VS1053b receives all data as a stream, all metadata must be available before the music data is received. Several MP4 file formatters do not satisfy this requirement and some kind of conversion is required. This is also why the streamable ADTS format is recommended. Programs exist that optimize the .mp4 and .m4a into so-called streamable format that has the mdat atom last in the file, and thus suitable for web servers’ audio streaming. You can use this kind of tool to process files for VS1053b too. For example mp4creator -optimize file.mp4. Version 0.5, 2007-12-03 28 VLSI y Solution VS1053 B VS1053b preliminary 8. FUNCTIONAL DESCRIPTION AAC12 : Samplerate / Hz 48000 44100 32000 24000 22050 16000 12000 11025 8000 ≤96 + + + + + + + + + Maximum Bitrate kbit/s - for 2 channels 132 144 192 264 288 384 529 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + 576 + 1 64000 Hz, 88200 Hz, and 96000 Hz AAC files are played at the highest possible samplerate (48000 Hz with 12.288 MHz XTALI). 2 Also all variable bitrate (VBR) formats are supported. Note that the table gives the maximum bitrate allowed for two channels for a specific samplerate as defined by the AAC specification. The decoder does not actually have a lower or upper limit. Version 0.5, 2007-12-03 29 VLSI y Solution 8.2.6 VS1053 B VS1053b preliminary 8. FUNCTIONAL DESCRIPTION Supported WMA Formats Windows Media Audio codec versions 2, 7, 8, and 9 are supported. All WMA profiles (L1, L2, and L3) are supported. Previously streams were separated into Classes 1, 2a, 2b, and 3. The decoder has passed Microsoft’s conformance testing program. Windows Media Audio Professional is a different codec and is not supported. WMA 4.0 / 4.1: Samplerate / Hz 8000 11025 16000 22050 32000 44100 48000 5 + 6 + 8 + + 10 + + 12 + 16 20 Bitrate / kbit/s 22 32 40 + + + + + + + + + + + + 48 64 80 96 128 160 192 + + + + + + + + + + 48 64 80 96 128 160 192 + + + + + + + + + + 48 64 80 96 128 160 192 + + + + + + + + + + + WMA 7: Samplerate / Hz 8000 11025 16000 22050 32000 44100 48000 5 + 6 + 8 + + 10 + + 12 + 16 20 + + + + + + Bitrate / kbit/s 22 32 40 + + + + + WMA 8: Samplerate / Hz 8000 11025 16000 22050 32000 44100 48000 5 + 6 + 8 + + 10 + + 12 + 16 20 + + + + + + Bitrate / kbit/s 22 32 40 + + + + + WMA 9: Samplerate / Hz 8000 11025 16000 22050 32000 44100 48000 5 + 6 + 8 + + 10 + + 12 + 16 20 + + + + + + + 22 + Bitrate / kbit/s 32 40 48 64 80 96 128 160 192 256 320 + + + + + + + + + + + + + + + + + + + In addition to these expected WMA decoding profiles, all other bitrate and samplerate combinations are supported, including variable bitrate WMA streams. Note that WMA does not consume the bitstream as evenly as MP3, so you need a higher peak transfer capability for clean playback at the same bitrate. Version 0.5, 2007-12-03 30 VLSI VS1053b preliminary y Solution 8.2.7 VS1053 B 8. FUNCTIONAL DESCRIPTION Supported RIFF WAV Formats The most common RIFF WAV subformats are supported, with 1 or 2 audio channels. Format 0x01 0x02 0x03 0x06 0x07 0x10 0x11 0x15 0x16 0x30 0x31 0x3b 0x3c 0x40 0x41 0x50 0x55 0x64 0x65 Version 0.5, Name PCM ADPCM IEEE FLOAT ALAW MULAW OKI ADPCM IMA ADPCM DIGISTD DIGIFIX DOLBY AC2 GSM610 ROCKWELL ADPCM ROCKWELL DIGITALK G721 ADPCM G728 CELP MPEG MPEGLAYER3 G726 ADPCM G722 ADPCM 2007-12-03 Supported + + + - Comments 16 and 8 bits, any samplerate ≤ 48kHz Any samplerate ≤ 48kHz For supported MP3 modes, see Chapter 8.2.1 31 VLSI VS1053b preliminary y Solution 8.2.8 VS1053 B 8. FUNCTIONAL DESCRIPTION Supported MIDI Formats General MIDI and SP-MIDI format 0 files are played. Format 1 and 2 files must be converted to format 0 by the user. The maximum polyphony is 64, the maximum sustained polyphony is 40. Actual polyphony depends on the internal clock rate (which is user-selectable), the instruments used, whether the reverb effect is enabled, and the possible global postprocessing effects enabled, such as bass enhancer, treble control or EarSpeaker spatial processing. The polyphony restriction algorithm makes use of the SP-MIDI MIP table, if present, and uses smooth note removal. 43 MHz (3.5× input clock) achieves 19-31 simultaneous sustained notes. The instantaneous amount of notes can be larger. This is a fair compromise between power consumption and quality, but higher clocks can be used to increase polyphony. Reverb effect can be controlled by the user. In addition to reverb automatic and reverb off modes, 14 different decay times can be selected. These roughly correspond to different room sizes. Also, each midi song decides how much effect each instrument gets. Because the reverb effect uses about 4 MHz of processing power the automatic control enables reverb only when the internal clock is at least 3.0×. In VS1053b both EarSpeaker and MIDI reverb can be on simultaneously. This is ideal for listening MIDI songs with headphones. New instruments have been implemented in addition to the 36 that are available in VS1003. VS1053b now has unique instruments in the whole GM1 instrument set and one bank of GM2 percussions. Version 0.5, 2007-12-03 32 VLSI VS1053b preliminary y Solution 1 Acoustic Grand Piano 2 Bright Acoustic Piano 3 Electric Grand Piano 4 Honky-tonk Piano 5 Electric Piano 1 6 Electric Piano 2 7 Harpsichord 8 Clavi 9 Celesta 10 Glockenspiel 11 Music Box 12 Vibraphone 13 Marimba 14 Xylophone 15 Tubular Bells 16 Dulcimer 17 Drawbar Organ 18 Percussive Organ 19 Rock Organ 20 Church Organ 21 Reed Organ 22 Accordion 23 Harmonica 24 Tango Accordion 25 Acoustic Guitar (nylon) 26 Acoustic Guitar (steel) 27 Electric Guitar (jazz) 28 Electric Guitar (clean) 29 Electric Guitar (muted) 30 Overdriven Guitar 31 Distortion Guitar 32 Guitar Harmonics 27 High Q 28 Slap 29 Scratch Push [EXC 7] 30 Scratch Pull [EXC 7] 31 Sticks 32 Square Click 33 Metronome Click 34 Metronome Bell 35 Acoustic Bass Drum 36 Bass Drum 1 37 Side Stick 38 Acoustic Snare 39 Hand Clap 40 Electric Snare 41 Low Floor Tom 42 Closed Hi-hat [EXC 1] Version 0.5, 2007-12-03 VS1053 B 8. FUNCTIONAL DESCRIPTION VS1053b Melodic Instruments (GM1) 33 Acoustic Bass 65 Soprano Sax 34 Electric Bass (finger) 66 Alto Sax 35 Electric Bass (pick) 67 Tenor Sax 36 Fretless Bass 68 Baritone Sax 37 Slap Bass 1 69 Oboe 38 Slap Bass 2 70 English Horn 39 Synth Bass 1 71 Bassoon 40 Synth Bass 2 72 Clarinet 41 Violin 73 Piccolo 42 Viola 74 Flute 43 Cello 75 Recorder 44 Contrabass 76 Pan Flute 45 Tremolo Strings 77 Blown Bottle 46 Pizzicato Strings 78 Shakuhachi 47 Orchestral Harp 79 Whistle 48 Timpani 80 Ocarina 49 String Ensembles 1 81 Square Lead (Lead 1) 50 String Ensembles 2 82 Saw Lead (Lead) 51 Synth Strings 1 83 Calliope Lead (Lead 3) 52 Synth Strings 2 84 Chiff Lead (Lead 4) 53 Choir Aahs 85 Charang Lead (Lead 5) 54 Voice Oohs 86 Voice Lead (Lead 6) 55 Synth Voice 87 Fifths Lead (Lead 7) 56 Orchestra Hit 88 Bass + Lead (Lead 8) 57 Trumpet 89 New Age (Pad 1) 58 Trombone 90 Warm Pad (Pad 2) 59 Tuba 91 Polysynth (Pad 3) 60 Muted Trumpet 92 Choir (Pad 4) 61 French Horn 93 Bowed (Pad 5) 62 Brass Section 94 Metallic (Pad 6) 63 Synth Brass 1 95 Halo (Pad 7) 64 Synth Brass 2 96 Sweep (Pad 8) VS1053b Percussion Instruments (GM1+GM2) 43 High Floor Tom 59 Ride Cymbal 2 44 Pedal Hi-hat [EXC 1] 60 High Bongo 45 Low Tom 61 Low Bongo 46 Open Hi-hat [EXC 1] 62 Mute Hi Conga 47 Low-Mid Tom 63 Open Hi Conga 48 High Mid Tom 64 Low Conga 49 Crash Cymbal 1 65 High Timbale 50 High Tom 66 Low Timbale 51 Ride Cymbal 1 67 High Agogo 52 Chinese Cymbal 68 Low Agogo 53 Ride Bell 69 Cabasa 54 Tambourine 70 Maracas 55 Splash Cymbal 71 Short Whistle [EXC 2] 56 Cowbell 72 Long Whistle [EXC 2] 57 Crash Cymbal 2 73 Short Guiro [EXC 3] 58 Vibra-slap 74 Long Guiro [EXC 3] 97 Rain (FX 1) 98 Sound Track (FX 2) 99 Crystal (FX 3) 100 Atmosphere (FX 4) 101 Brightness (FX 5) 102 Goblins (FX 6) 103 Echoes (FX 7) 104 Sci-fi (FX 8) 105 Sitar 106 Banjo 107 Shamisen 108 Koto 109 Kalimba 110 Bag Pipe 111 Fiddle 112 Shanai 113 Tinkle Bell 114 Agogo 115 Pitched Percussion 116 Woodblock 117 Taiko Drum 118 Melodic Tom 119 Synth Drum 120 Reverse Cymbal 121 Guitar Fret Noise 122 Breath Noise 123 Seashore 124 Bird Tweet 125 Telephone Ring 126 Helicopter 127 Applause 128 Gunshot 75 Claves 76 Hi Wood Block 77 Low Wood Block 78 Mute Cuica [EXC 4] 79 Open Cuica [EXC 4] 80 Mute Triangle [EXC 5] 81 Open Triangle [EXC 5] 82 Shaker 83 Jingle bell 84 Bell tree 85 Castanets 86 Mute Surdo [EXC 6] 87 Open Surdo [EXC 6] 33 VLSI VS1053b preliminary y Solution 8.3 VS1053 B 8. FUNCTIONAL DESCRIPTION Data Flow of VS1053b SDI Bitstream FIFO MP3 MP2 MP1 WAV ADPCM WMA AAC MIDI Vorbis SM_ADPCM=0 AIADDR = 0 User Application AIADDR != 0 SB_AMPLITUDE=0 Bass enhancer Treble control SB_AMPLITUDE!=0 Audio FIFO 2048 stereo samples ST_AMPLITUDE=0 Ear Speaker ST_AMPLITUDE!=0 L S.rate.conv. R and DAC Volume SCI_VOL control Figure 13: Data Flow of VS1053b. First, depending on the audio data, and provided ADPCM encoding mode is not set, Ogg Vorbis, MP3, WMA, AAC, PCM WAV, IMA ADPCM WAV, or MIDI data is received and decoded from the SDI bus. After decoding, if SCI AIADDR is non-zero, application code is executed from the address pointed to by that register. For more details, see Application Notes for VS10XX. Then data may be sent to the Bass Enhancer and Treble Control depending on the SCI BASS register. Next, headphone processing is performed, if the EarSpeaker spatial processing is active. After that the data to the Audio FIFO, which holds the data until it is read by the Audio interrupt and fed to the samplerate converter and DACs. The size of the audio FIFO is 2048 stereo (2×16-bit) samples, or 8 KiB. The samplerate converter upsamples all different samplerates to XTALI/2, or 128 times the highest usable samplerate with 18-bit precision. Volume control is performed in the upsampled domain. New volume settings are loaded only when the upsampled signal crosses the zero point (or after a timeout). This zerocrossing detection almost completely removes all audible noise that occurs when volume is suddenly changed. The samplerate conversion to a common samplerate removes the need for complex PLL-based clocking schemes and allows almost unlimited sample rate accuracy with one fixed input clock frequency. With a 12.288 MHz clock, the DA converter operates at 128 × 48 kHz, i.e. 6.144 MHz, and creates a stereo in-phase analog signal. The oversampled output is low-pass filtered by an on-chip analog filter. This signal is then forwarded to the earphone amplifier. Version 0.5, 2007-12-03 34 VLSI VS1053b preliminary y Solution 8.4 VS1053 B 8. FUNCTIONAL DESCRIPTION EarSpeaker Spatial Processing While listening to headphones the sound has a tendency to be localized inside the head. The sound field becomes flat and lacking the sensation of dimensions. This is an unnatural, awkward and sometimes even disturbing situation. This phenomenon is often referred in literature as ‘lateralization’, meaning ’in-the-head’ localization. Long-term listening to lateralized sound may lead to listening fatigue. All real-life sound sources are external, leaving traces to the acoustic wavefront that arrives to the ear drums. From these traces, the auditory system of the brain is able to judge the distance and angle of each sound source. In loudspeaker listening the sound is external and these traces are available. In headphone listening these traces are missing or ambiguous. EarSpeaker processes sound to make listening via headphones more like listening to the same music from real loudspeakers or live music. Once EarSpeaker processing is activated, the instruments are moved from inside to the outside of the head, making it easier to separate the different instruments (see figure 14). The listening experience becomes more natural and pleasant, and the stereo image is sharper as the instruments are widely on front of the listener instead of being inside the head. Figure 14: EarSpeaker externalized sound sources vs. normal inside-the-head sound Note that EarSpeaker differs from any common spatial processing effects, such as echo, reverb, or bass boost. EarSpeaker accurately simulates the human auditory model and real listening environment acoustics. Thus is does not change the tonal character of the music by introducing artificial effects. EarSpeaker processing can be parameterized to a few different modes, each simulating a little different type of acoustical situation, suiting different personal preferences and types of recording. See section 8.7.1 for how to activate different modes. • Off: Best option when listening through loudspeakers or if the audio to be played contains binaural preprocessing. • minimal: Suited for listening to normal musical scores with headphones, very subtle. • normal: Suited for listening to normal musical scores with headphones, moves sound source further away than minimal. • extreme: Suited for old or ’dry’ recordings, or if the audio to be played is artificial, for example generated MIDI. Version 0.5, 2007-12-03 35 VLSI VS1053b preliminary y Solution 8.5 VS1053 B 8. FUNCTIONAL DESCRIPTION Serial Data Interface (SDI) The serial data interface is meant for transferring compressed data for the different decoders of VS1053b. If the input of the decoder is invalid or it is not received fast enough, analog outputs are automatically muted. Also several different tests may be activated through SDI as described in Chapter 9. 8.6 Serial Control Interface (SCI) The serial control interface is compatible with the SPI bus specification. Data transfers are always 16 bits. VS1053b is controlled by writing and reading the registers of the interface. The main controls of the serial control interface are: • • • • Version 0.5, control of the operation mode, clock, and builtin effects access to status information and header data receiving encoded data in recording mode uploading and controlling user programs 2007-12-03 36 VLSI VS1053b preliminary y Solution 8.7 VS1053 B 8. FUNCTIONAL DESCRIPTION SCI Registers VS1053b sets DREQ low when it detects an SCI operation (this delay is 16 to 40 CLKI cycles depending on whether an interrupt service routine is active) and restores it when it has processed the operation. The duration depends on the operation. If DREQ is low when an SCI operation is performed, it also stays low after SCI operation processing. If DREQ is high before a SCI operation, do not start a new SCI/SDI operation before DREQ is high again. If DREQ is low before a SCI operation because the SDI can not accept more data, make certain there is enough time to complete the operation before sending another. Reg 0x0 0x1 0x2 0x3 0x4 0x5 0x6 0x7 0x8 0x9 0xA 0xB 0xC 0xD 0xE 0xF Type rw rw rw rw rw rw rw rw r r rw rw rw rw rw rw Reset 0x4800 0x000C3 0 0 0 0 0 0 0 0 0 0 0 0 0 0 SCI registers, prefix SCI Time1 Abbrev[bits] 80 CLKI4 MODE 80 CLKI STATUS 80 CLKI BASS 1200 XTALI5 CLOCKF 100 CLKI DECODE TIME 450 CLKI2 AUDATA 100 CLKI WRAM 100 CLKI WRAMADDR 80 CLKI HDAT0 80 CLKI HDAT1 210 CLKI2 AIADDR 80 CLKI VOL 80 CLKI2 AICTRL0 80 CLKI2 AICTRL1 80 CLKI2 AICTRL2 80 CLKI2 AICTRL3 Description Mode control Status of VS1053b Built-in bass/treble control Clock freq + multiplier Decode time in seconds Misc. audio data RAM write/read Base address for RAM write/read Stream header data 0 Stream header data 1 Start address of application Volume control Application control register 0 Application control register 1 Application control register 2 Application control register 3 1 This is the worst-case time that DREQ stays low after writing to this register. The user may choose to skip the DREQ check for those register writes that take less than 100 clock cycles to execute and use a fixed delay instead. 2 In addition, the cycles spent in the user application routine must be counted. 3 Firmware changes the value of this register immediately to 0x48 (analog enabled), and after a short while to 0x40 (analog drivers enabled). 4 When mode register write specifies a software reset the worst-case time is 22000 XTALI cycles. 5 Writing to CLOCKF register may force internal clock to run at 1.0 × XTALI for a while. Thus it is not a good idea to send SCI or SDI bits while this register update is in progress. Reads from all SCI registers complete in under 100 CLKI cycles, except a read from AIADDR in 200 cycles. In addition the cycles spent in the user application routine must be counted to the read time of AIADDR, AUDATA, and AICTRL0..3. Version 0.5, 2007-12-03 37 VLSI VS1053b preliminary y Solution 8.7.1 VS1053 B 8. FUNCTIONAL DESCRIPTION SCI MODE (RW) SCI MODE is used to control the operation of VS1053b and defaults to 0x0800 (SM SDINEW set). Bit 0 Name SM DIFF Function Differential 1 SM LAYER12 Allow MPEG layers I & II 2 SM RESET Soft reset 3 SM CANCEL Cancel decoding current file 4 SM EARSPEAKER LO EarSpeaker low setting 5 SM TESTS Allow SDI tests 6 SM STREAM Stream mode 7 SM EARSPEAKER HI EarSpeaker high setting 8 SM DACT DCLK active edge 9 SM SDIORD SDI bit order 10 SM SDISHARE Share SPI chip select 11 SM SDINEW VS1002 native SPI modes 12 SM ADPCM ADPCM recording active 13 - - 14 SM LINE1 MIC / LINE1 selector 15 SM CLK RANGE Input clock range Value 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 Description normal in-phase audio left channel inverted no yes no reset reset no yes off active not allowed allowed no yes off active rising falling MSb first MSb last no yes no yes no yes right wrong MICP LINE1 12..13 MHz 24..26 MHz When SM DIFF is set, the player inverts the left channel output. For a stereo input this creates virtual surround, and for a mono input this creates a differential left/right signal. SM LAYER12 enables MPEG 1.0 and 2.0 layer I and II decoding in addition to layer III. If you enable Layer I and Layer II decoding, you are liable for any patent issues that may arise. Joint licensing of MPEG 1.0 / 2.0 Layer III does not cover all patents pertaining to layers I and II. Software reset is initiated by setting SM RESET to 1. This bit is cleared automatically. If you want to stop decoding a in the middle, set SM CANCEL, and continue sending data honouring DREQ. When SM CANCEL is detected by a codec, it will stop decoding and return to the main loop. The stream buffer content is discarded and the SM CANCEL bit cleared. SCI HDAT1 will also be cleared. See Chapter 9.5.2 for details. Bits SM EARSPEAKER LO and SM EARSPEAKER HI control the EarSpeaker spatial processing. If both are 0, the processing is not active. Other combinations activate the processing and select 3 different effect levels: LO = 1, HI = 0 selects minimal, LO = 0, HI = 1 selects normal, and LO = 1, HI = 1 selects extreme. EarSpeaker takes approximately 12 MIPS at 44.1 kHz samplerate. Version 0.5, 2007-12-03 38 VLSI VS1053b preliminary y Solution VS1053 B 8. FUNCTIONAL DESCRIPTION If SM TESTS is set, SDI tests are allowed. For more details on SDI tests, look at Chapter 9.12. SM STREAM activates VS1053b’s stream mode. In this mode, data should be sent with as even intervals as possible and preferable in blocks of less than 512 bytes, and VS1053b makes every attempt to keep its input buffer half full by changing its playback speed upto 5%. For best quality sound, the average speed error should be within 0.5%, the bitrate should not exceed 160 kbit/s and VBR should not be used. For details, see Application Notes for VS10XX. This mode only works with MP3 and WAV files. SM DACT defines the active edge of data clock for SDI. When ’0’, data is read at the rising edge, when ’1’, data is read at the falling edge. When SM SDIORD is clear, bytes on SDI are sent MSb first. By setting SM SDIORD, the user may reverse the bit order for SDI, i.e. bit 0 is received first and bit 7 last. Bytes are, however, still sent in the default order. This register bit has no effect on the SCI bus. Setting SM SDISHARE makes SCI and SDI share the same chip select, as explained in Chapter 7.2, if also SM SDINEW is set. Setting SM SDINEW will activate VS1002 native serial modes as described in Chapters 7.2.1 and 7.4.2. Note, that this bit is set as a default when VS1053b is started up. By activating SM ADPCM and SM RESET at the same time, the user will activate IMA ADPCM recording mode (see section 9.8). SM LINE IN is used to select the left-channel input for ADPCM recording. If ’0’, differential microphone input pins MICP and MICN are used; if ’1’, line-level MICP/LINEIN1 pin is used. SM CLK RANGE activates a clock divider in the XTAL input. When SM CLK RANGE is set, the clock is divided by 2 at the input. From the chip’s point of view e.g. 24 MHz becomes 12 MHz. SM CLK RANGE should be set as soon as possible after a chip reset. Version 0.5, 2007-12-03 39 VLSI VS1053b preliminary y Solution 8.7.2 VS1053 B 8. FUNCTIONAL DESCRIPTION SCI STATUS (RW) SCI STATUS contains information on the current status of VS1053b. It also controls some low-level things that the user does not usually have to care about. Name SS DO NOT JUMP SS SWING SS VCM OVERLOAD SS VCM DISABLE SS SS SS SS SS VER APDOWN2 APDOWN1 AD CLOCK REFERENCE SEL Bits 15 14:12 11 10 9:8 7:4 3 2 1 0 Description Header in decode, do not fast forward/rewind Set swing to +0 dB, +0.5 dB, .., or +3.5 dB GBUF overload indicator ’1’ = overload GBUF overload detection ’1’ = disable reserved Version Analog driver powerdown Analog internal powerdown AD clock select, ’0’ = 6 MHz, ’1’ = 3 MHz Reference voltage selection, ’0’ = 1.2 V, ’1’ = 1.65 V SS DO NOT JUMP is set when WAV, Ogg Vorbis, WMA, MP4, or AAC-ADIF header is being decoded and jumping to another location in the file is not allowed. If AVDD is higher at least 3.3 V, SS REFERENCE SEL can be set to select 1.65 V reference voltage to increase the analog output swing. SS AD CLOCK can be set to divide the AD modulator frequency by 2 if XTALI/2 is too much. SS VER is 0 for VS1001, 1 for VS1011, 2 for VS1002, 3 for VS1003, 4 for VS1053, 5 for VS1033, and 7 for VS1103. SS APDOWN2 controls analog driver powerdown. SS APDOWN1 controls internal analog powerdown. These bit are meant to be used by the system firmware only. If the user wants to powerdown VS1053b with a minimum power-off transient, set SCI VOL to 0xffff, then wait for at least a few milliseconds before activating reset. VS1053b contains GBUF protection circuit which disconnects the GBUF driver when too much current is drawn, indicating a short-circuit to ground. SS VCM OVERLOAD is high while the overload is detected. SS VCM DISABLE can be set to disable the protection feature. SS SWING allows you to go above the 0 dB volume setting. Value 0 is normal mode, 1 gives +0.5 dB, and 2 gives +1.0 dB. Although the range of the register is upto 7, higher settings than 2 do not work and should not be used. Version 0.5, 2007-12-03 40 VLSI VS1053b preliminary y Solution 8.7.3 VS1053 B 8. FUNCTIONAL DESCRIPTION SCI BASS (RW) Name ST AMPLITUDE ST FREQLIMIT SB AMPLITUDE SB FREQLIMIT Bits 15:12 11:8 7:4 3:0 Description Treble Control in 1.5 dB steps (-8..7, 0 = off) Lower limit frequency in 1000 Hz steps (1..15) Bass Enhancement in 1 dB steps (0..15, 0 = off) Lower limit frequency in 10 Hz steps (2..15) The Bass Enhancer VSBE is a powerful bass boosting DSP algorithm, which tries to take the most out of the users earphones without causing clipping. VSBE is activated when SB AMPLITUDE is non-zero. SB AMPLITUDE should be set to the user’s preferences, and SB FREQLIMIT to roughly 1.5 times the lowest frequency the user’s audio system can reproduce. For example setting SCI BASS to 0x00f6 will have 15 dB enhancement below 60 Hz. Note: Because VSBE tries to avoid clipping, it gives the best bass boost with dynamical music material, or when the playback volume is not set to maximum. It also does not create bass: the source material must have some bass to begin with. Treble Control VSTC is activated when ST AMPLITUDE is non-zero. For example setting SCI BASS to 0x7a00 will have 10.5 dB treble enhancement at and above 10 kHz. Bass Enhancer uses about 2.1 MIPS and Treble Control 1.2 MIPS at 44100 Hz samplerate. Both can be on simultaneously. In VS1053b bass and treble initialization and volume change is delayed until the next batch of samples are sent to the audio FIFO. Thus, unlike with earlier VS10XX chips, audio interrupts can no longer be missed when SCI BASS or SCI VOL is written to. Version 0.5, 2007-12-03 41 VLSI VS1053b preliminary y Solution 8.7.4 VS1053 B 8. FUNCTIONAL DESCRIPTION SCI CLOCKF (RW) The operation of SCI CLOCKF has changed slightly in VS1053b compared to VS1003 and VS1033. Multiplier 1.5× and addition 0.5× have been removed to allow higher clocks to be configured. Name SC MULT SC ADD SC FREQ SCI CLOCKF bits Bits Description 15:13 Clock multiplier 12:11 Allowed multiplier addition 10: 0 Clock frequency SC MULT activates the built-in clock multiplier. This will multiply XTALI to create a higher CLKI. The values are as follows: SC MULT 0 1 2 3 4 5 6 7 MASK 0x0000 0x2000 0x4000 0x6000 0x8000 0xa000 0xc000 0xe000 CLKI XTALI XTALI×2.0 XTALI×2.5 XTALI×3.0 XTALI×3.5 XTALI×4.0 XTALI×4.5 XTALI×5.0 SC ADD tells, how much the decoder firmware is allowed to add to the multiplier specified by SC MULT if more cycles are temporarily needed to decode a WMA or AAC stream. The values are: SC ADD 0 1 2 3 MASK 0x0000 0x0800 0x1000 0x1800 Multiplier addition No modification is allowed 1.0× 1.5× 2.0× SC FREQ is used to tell if the input clock XTALI is running at something else than 12.288 MHz. XTALI is set in 4 kHz steps. The formula for calculating the correct value for this register is XT ALI−8000000 4000 (XTALI is in Hz). Note: The default value 0 is assumed to mean XTALI=12.288 MHz. Note: because maximum samplerate is MHz. XT ALI 256 , all samplerates are not available if XTALI < 12.288 Note: Automatic clock change can only happen when decoding WMA and AAC files. Automatic clock change is done one 0.5× at a time. This does not cause a drop to 1.0× clock and you can use the same SCI and SDI clock throughout the file. Example: If SCI CLOCKF is 0x9BE8, SC MULT = 4, SC ADD = 3 and SC FREQ = 0x3E8 = 1000. This means that XTALI = 1000×4000+8000000 = 12 MHz. The clock multiplier is set to 3.5×XTALI = 42 MHz, and the maximum allowed multiplier that the firmware may automatically choose to use is (3.5 + 2.0)×XTALI = 66 MHz. Version 0.5, 2007-12-03 42 VLSI VS1053b preliminary y Solution 8.7.5 VS1053 B 8. FUNCTIONAL DESCRIPTION SCI DECODE TIME (RW) When decoding correct data, current decoded time is shown in this register in full seconds. The user may change the value of this register. In that case the new value should be written twice to make absolutely certain that the change is not overwritten by the firmware. A write to SCI DECODE TIME also resets the byteRate calculation. SCI DECODE TIME is reset at every hardware and software reset. It is no longer cleared when decoding of a file ends to allow the decode time to proceed automatically with looped files and with seamless playback of multiple files. With fast playback (see the playSpeed extra parameter) the decode time also counts faster. Some codecs (WMA and Ogg Vorbis) can also indicate the absolute play position, see the positionMsec extra parameter in section 9.11. 8.7.6 SCI AUDATA (RW) When decoding correct data, the current samplerate and number of channels can be found in bits 15:1 and 0 of SCI AUDATA, respectively. Bits 15:1 contain the samplerate divided by two, and bit 0 is 0 for mono data and 1 for stereo. Writing to SCI AUDATA will change the samplerate directly. Example: 44100 Hz stereo data reads as 0xAC45 (44101). Example: 11025 Hz mono data reads as 0x2B10 (11024). Example: Writing 0xAC80 sets samplerate to 44160 Hz, stereo mode does not change. To reduce the digital power consumption when in idle, you can write a low samplerate to SCI AUDATA. 8.7.7 SCI WRAM (RW) SCI WRAM is used to upload application programs and data to instruction and data RAMs. The start address must be initialized by writing to SCI WRAMADDR prior to the first write/read of SCI WRAM. As 16 bits of data can be transferred with one SCI WRAM write/read, and the instruction word is 32 bits long, two consecutive writes/reads are needed for each instruction word. The byte order is big-endian (i.e. most significant words first). After each full-word write/read, the internal pointer is autoincremented. 8.7.8 SCI WRAMADDR (W) SCI WRAMADDR is used to set the program address for following SCI WRAM writes/reads. Address offset of 0 is used for X, 0x4000 for Y, and 0x8000 for instruction memory. Peripheral registers can also be accessed. Version 0.5, 2007-12-03 43 VLSI VS1053b preliminary y Solution SM WRAMADDR Start. . . End 0x1800. . . 0x18XX 0x5800. . . 0x58XX 0x8040. . . 0x84FF 0xC000. . . 0xFFFF Dest. addr. Start. . . End 0x1800. . . 0x18XX 0x1800. . . 0x18XX 0x0040. . . 0x04FF 0xC000. . . 0xFFFF Bits/ Word 16 16 32 16 VS1053 B 8. FUNCTIONAL DESCRIPTION Description X data RAM Y data RAM Instruction RAM I/O Only user areas in X, Y, and instruction memory are listed above. Other areas can be accessed, but should not be written to unless otherwise specified. Version 0.5, 2007-12-03 44 VLSI VS1053b preliminary y Solution 8.7.9 VS1053 B 8. FUNCTIONAL DESCRIPTION SCI HDAT0 and SCI HDAT1 (R) For WAV files, SCI HDAT1 contains 0x7665 (“ve”). SCI HDAT0 contains the data rate measured in bytes per second for all supported RIFF WAVE formats: mono and stereo 8-bit or 16-bit PCM, mono and stereo IMA ADPCM. To get the bitrate of the file, multiply the value by 8. For AAC ADTS streams, SCI HDAT1 contains 0x4154 (“AT”). For AAC ADIF files, SCI HDAT1 contains 0x4144 (“AD”). For AAC .mp4 / .m4a files, SCI HDAT1 contains 0x4D34 (“M4”). SCI HDAT0 contains the average data rate in bytes per second. To get the bitrate of the file, multiply the value by 8. For WMA files, SCI HDAT1 contains 0x574D (“WM”) and SCI HDAT0 contains the data rate measured in bytes per second. To get the bitrate of the file, multiply the value by 8. For MIDI files, SCI HDAT1 contains 0x4D54 (“MT”) and SCI HDAT0 contains the average data rate in bytes per second. To get the bitrate of the file, multiply the value by 8. For Ogg Vorbis files, SCI HDAT1 contains “Og”. SCI HDAT0 contains the average data rate in bytes per second. To get the bitrate of the file, multiply the value by 8. For MP3 files, SCI HDAT1 is between 0xFFE0 and 0xFFFF. SCI HDAT1 / 0 contain the following: Version 0.5, 2007-12-03 Bit HDAT1[15:5] HDAT1[4:3] Function syncword ID HDAT1[2:1] layer HDAT1[0] protect bit HDAT0[15:12] HDAT0[11:10] bitrate samplerate HDAT0[9] pad bit HDAT0[8] HDAT0[7:6] private bit mode HDAT0[5:4] HDAT0[3] extension copyright HDAT0[2] original HDAT0[1:0] emphasis Value 2047 3 2 1 0 3 2 1 0 1 0 3 2 1 0 1 0 3 2 1 0 1 0 1 0 3 2 1 0 Explanation stream valid ISO 11172-3 MPG 1.0 ISO 13818-3 MPG 2.0 (1/2-rate) MPG 2.5 (1/4-rate) MPG 2.5 (1/4-rate) I II III reserved No CRC CRC protected see bitrate table reserved 32/16/ 8 kHz 48/24/12 kHz 44/22/11 kHz additional slot normal frame not defined mono dual channel joint stereo stereo see ISO 11172-3 copyrighted free original copy CCITT J.17 reserved 50/15 microsec none 45 VLSI y Solution VS1053 B VS1053b preliminary 8. FUNCTIONAL DESCRIPTION When read, SCI HDAT0 and SCI HDAT1 contain header information that is extracted from MP3 stream currently being decoded. After reset both registers are cleared, indicating no data has been found yet. The “samplerate” field in SCI HDAT0 is interpreted according to the following table: “samplerate” 3 2 1 0 ID=3 32000 48000 44100 ID=2 16000 24000 22050 ID=0,1 8000 12000 11025 The “bitrate” field in HDAT0 is read according to the following table. Notice that for variable bitrate stream the value changes constantly. “bitrate” 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 Layer I ID=3 ID=0,1,2 kbit/s forbidden forbidden 448 256 416 224 384 192 352 176 320 160 288 144 256 128 224 112 192 96 160 80 128 64 96 56 64 48 32 32 - Layer II ID=3 ID=0,1,2 kbit/s forbidden forbidden 384 160 320 144 256 128 224 112 192 96 160 80 128 64 112 56 96 48 80 40 64 32 56 24 48 16 32 8 - Layer III ID=3 ID=0,1,2 kbit/s forbidden forbidden 320 160 256 144 224 128 192 112 160 96 128 80 112 64 96 56 80 48 64 40 56 32 48 24 40 16 32 8 - The average data rate in bytes per second can be read from memory, see the byteRate extra parameter. This variable contains the byte rate for all codecs. To get the bitrate of the file, multiply the value by 8. The bitrate calculation is not automatically reset between songs, but it can also be reset without a software or hardware reset by writing to SCI DECODE TIME. 8.7.10 SCI AIADDR (RW) SCI AIADDR indicates the start address of the application code written earlier with SCI WRAMADDR and SCI WRAM registers. If no application code is used, this register should not be initialized, or it should be initialized to zero. For more details, see Application Notes for VS10XX. Version 0.5, 2007-12-03 46 VLSI VS1053b preliminary y Solution 8.7.11 VS1053 B 8. FUNCTIONAL DESCRIPTION SCI VOL (RW) SCI VOL is a volume control for the player hardware. The most significant byte of the volume register controls the left channel volume, the low part controls the right channel volume. The channel volume sets the attenuation from the maximum volume level in 0.5 dB steps. Thus, maximum volume is 0x0000 and total silence is 0xFEFE. Note, that after hardware reset the volume is set to full volume. Resetting the software does not reset the volume setting. Setting SCI VOL to 0xFFFF will activate analog powerdown mode. Example: for a volume of -2.0 dB for the left channel and -3.5 dB for the right channel: (2.0/0.5) = 4, 3.5/0.5 = 7 → SCI VOL = 0x0407. Example: SCI VOL = 0x2424 → both left and right volumes are 0x24 * -0.5 = -18.0 dB In VS1053b bass and treble initialization and volume change is delayed until the next batch of samples are sent to the audio FIFO. Thus, audio interrupts can no longer be missed during a write to SCI BASS or SCI VOL. This delays the volume setting slightly, but because the volume control is now done in the DAC hardware instead of performing it to the samples going into the audio FIFO, the overall volume change response is better than before. Also, the actual volume control has zero-cross detection, which almost completely removes all audible noise that occurs when volume is suddenly changed. 8.7.12 SCI AICTRL[x] (RW) SCI AICTRL[x] registers ( x=[0 .. 3] ) can be used to access the user’s application program. The AICTRL registers are also used with IMA ADPCM encoding mode. Version 0.5, 2007-12-03 47 VLSI VS1053b preliminary y Solution 9 9.1 VS1053 B 9. OPERATION Operation Clocking VS1053b operates on a single, nominally 12.288 MHz fundamental frequency master clock. This clock can be generated by external circuitry (connected to pin XTALI) or by the internal clock crystal interface (pins XTALI and XTALO). This clock is used by the analog parts and determines the highest available samplerate. With 12.288 MHz clock all samplerates upto 48000 Hz are available. VS1053b can also use 24..26 MHz clocks when SM CLK RANGE in the SCI MODE register is set to 1. The system clock is then divided by 2 at the clock input and the chip gets a 12..13 MHz input clock. 9.2 Hardware Reset When the XRESET -signal is driven low, VS1053b is reset and all the control registers and internal states are set to the initial values. XRESET-signal is asynchronous to any external clock. The reset mode doubles as a full-powerdown mode, where both digital and analog parts of VS1053b are in minimum power consumption stage, and where clocks are stopped. Also XTALO is grounded. When XRESET is asseted, all output pins go to their default states. All input pins will go to highimpedance state (to input state), except SO, which is still controlled by the XCS. After a hardware reset (or at power-up) DREQ will stay down for around 22000 clock cycles, which means an approximate 1.8 ms delay if VS1053b is run at 12.288 MHz. After this the user should set such basic software registers as SCI MODE, SCI BASS, SCI CLOCKF, and SCI VOL before starting decoding. See section 8.7 for details. If the input clock is 24..26 MHz, SM CLK RANGE should be set as soon as possible after a chip reset without waiting for DREQ. Internal clock can be multiplied with a PLL. Supported multipliers through the SCI CLOCKF register are 1.0 × . . . 5.0× the input clock. Reset value for Internal Clock Multiplier is 1.0×. If typical values are wanted, the Internal Clock Multiplier needs to be set to 3.5× after reset. Wait until DREQ rises, then write value 0x9800 to SCI CLOCKF (register 3). See section 8.7.4 for details. 9.3 Software Reset In some cases the decoder software has to be reset. This is done by activating bit SM RESET in register SCI MODE (Chapter 8.7.1). Then wait for at least 2 µs, then look at DREQ. DREQ will stay down for about 22000 clock cycles, which means an approximate 1.8 ms delay if VS1053b is run at 12.288 MHz. After DREQ is up, you may continue playback as usual. As opposed to all earlier VS10XX chips, it is not recommended to do a software reset between songs. This way the user may be sure that even files with low samplerates or bitrates are played right to their end. Version 0.5, 2007-12-03 48 VLSI VS1053b preliminary y Solution 9.4 VS1053 B 9. OPERATION Low Power Mode If you need to keep the system running while not decoding data, but need to lower the power consumption, you can use the following tricks. • Select the 1.0× clock by writing 0x0000 to SCI CLOCKF. This disables the PLL and saves some power. • Write a low non-zero value, such as 0x0010 to SCI AUDATA. This will reduce the samplerate and the number of audio interrupts required. Between audio interrupts the VSDSP core will just wait for an interrupt, thus saving power. • Turn off all audio post-processing (tone controls and EarSpeaker). • If possible for the application, write 0xffff to SCI VOL to disable the analog drivers. To return from low-power mode, revert register values in reverse order. Note: The low power mode consumes significantly more electricity than hardware reset. 9.5 Play and Decode This is the normal operation mode of VS1053b. SDI data is decoded. Decoded samples are converted to analog domain by the internal DAC. If no decodable data is found, SCI HDAT0 and SCI HDAT1 are set to 0. When there is no input for decoding, VS1053b goes into idle mode (lower power consumption than during decoding) and actively monitors the serial data input for valid data. 9.5.1 Playing a Whole File This is the default playback mode. 1. 2. 3. 4. 5. 6. Send an audio file to VS1053b. Read extra parameter value endFillByte (Chapter 9.11). Send at least 2052 bytes of endFillByte[7:0]. Set SCI MODE bit SM CANCEL. Send at least 32 bytes of endFillByte[7:0]. Read SCI MODE. If SM CANCEL is still set, go to 5. If SM CANCEL hasn’t cleared after sending 2048 bytes, do a software reset (this should be extremely rare). 7. The song has now been successfully sent. HDAT0 and HDAT1 should now both contain 0 to indicate that no format is being decoded. Return to 1. Version 0.5, 2007-12-03 49 VLSI VS1053b preliminary y Solution 9.5.2 VS1053 B 9. OPERATION Cancelling Playback Cancelling playback of a song is a normal operation when the user wants to jump to another song while doing playback. 1. Send a portion of an audio file to VS1053b. 2. Set SCI MODE bit SM CANCEL. 3. Continue sending audio file, but check SM CANCEL after every 32 bytes of data. If it is still set, goto 3. If SM CANCEL doesn’t clear after 2048 bytes or one second, do a software reset (this should be extremely rare). 4. When SM CANCEL has cleared, read extra parameter value endFillByte (Chapter 9.11). 5. Send 2052 bytes of endFillByte[7:0]. 6. HDAT0 and HDAT1 should now both contain 0 to indicate that no format is being decoded. You can now send the next audio file. 9.5.3 Fast Play VS1053b allows fast audio playback. If your microcontroller can feed data fast enough to the VS1053b, this is the preferred way to fast forward audio. 1. 2. 3. 4. Start sending an audio file to VS1053b. To set fast play, set extra parameter value playSpeed (Chapter 9.11). Continue sending audio file. To exit fast play mode, write 1 to playSpeed. To estimate whether or not your microcontroller can feed enough data to VS1053b in fast play mode, see contents of extra parameter value byteRate (Chapter 9.11). Note that byteRate contains the data speed of the file played back at nominal speed even when fast play is active. Note: Play speed is not reset when song is changed. 9.5.4 Fast Forward and Rewind without Audio To do fast forward and rewind you need the capability to do random access to the audio file. Unfortunately fast forward and rewind isn’t available at all times, like when file headers are being read. 1. Send a portion of an audio file to VS1053b. 2. When random access is required, read SCI STATUS bit SS DO NOT JUMP. If that bit is set, random access cannot be performed, so go back to 1. 3. Read extra parameter value endFillByte (Chapter 9.11). 4. Send at least 2048 bytes of endFillByte[7:0]. 5. Jump forwards or backwards in the file. 6. Continue sending the file. Version 0.5, 2007-12-03 50 VLSI VS1053b preliminary y Solution VS1053 B 9. OPERATION Note: It is recommended that playback volume is decreased by e.g. 10 dB when fast forwarding/rewinding. Note: Register DECODE TIME does not take jumps into account. Note: Midi is not suitable for random-access. You can implement fast forward using the playSpeed extra parameter to select 1-128× play speed. SCI DECODE TIME also speeds up. If necessary, rewind can be implemented by restarting decoding of a MIDI file and fast playing to the appropriate place. SCI DECODE TIME can be used to decide when the right place has been reached. 9.5.5 Maintaining Correct Decode Time When fast forward and rewind operations are performed, there is no way to maintain correct decode time for most files. However, WMA and Ogg Vorbis offer exact time information in the file. To use accurate time information whenever possible, use the following algorithm: 1. Start sending an audio file to VS1053b. 2. Read extra parameter value pair positionMsec (Chapter 9.11). 3. If positionMsec is -1, show you estimation of decoding time using DECODE TIME (and your estimate of file position if you have performed fast forward / rewind operations). 4. If positionMsec is not -1, use this time to show the exact position in the file. Version 0.5, 2007-12-03 51 VLSI y Solution 9.6 VS1053 B VS1053b preliminary 9. OPERATION Feeding PCM data VS1053b can be used as a PCM decoder by sending a WAV file header. If the length sent in the WAV header is 0 or 0xFFFFFFFF, VS1053b will stay in PCM mode indefinitely (or until SM CANCEL has been set). 8-bit linear and 16-bit linear audio is supported in mono or stereo. A WAV header looks like this: File Offset 0 4 8 12 16 20 22 24 28 32 34 52 56 Field Name ChunkID ChunkSize Format SubChunk1ID SubChunk1Size AudioFormat NumOfChannels SampleRate ByteRate BlockAlign BitsPerSample SubChunk2ID SubChunk2Size Size 4 4 4 4 4 2 2 4 4 2 2 4 4 Bytes "RIFF" 0xff 0xff 0xff 0xff "WAVE" "fmt " 0x10 0x0 0x0 0x0 0x1 0x0 C0 C1 S0 S1 S2 S3 R0 R1 R2 R3 A0 A1 B0 0xB1 "data" 0xff 0xff 0xff 0xff Description 16 Linear PCM 1 for mono, 2 for stereo 0x1f40 for 8 kHz 0x3e80 for 8 kHz 16-bit mono 2 for mono, 4 for stereo 16-bit 16 for 16-bit data Data size The rules to calculate the four variables are as follows: • • • • • S = sample rate in Hz, e.g. 44100 for 44.1 kHz. For 8-bit data B = 8, and for 16-bit data B = 16. For mono data C = 1, for stereo data C = 2. A = C×B 8 . R = S × A. Example: A 44100 Hz 16-bit stereo PCM header would read as follows: 0000 0100 0200 52 49 46 46 ff ff ff ff 10 00 00 00 01 00 02 00 04 00 10 00 64 61 74 61 57 41 56 45 66 6d 74 20 44 ac 00 00 10 b1 02 00 ff ff ff ff |RIFF....WAVEfmt | |........D.......| |....data....| 9.7 Ogg Vorbis Recording Ogg Vorbis is an open file format that allows for very high sound quality with low to medium bitrates. Ogg Vorbis recording is activated by loading the Ogg Vorbis Encoder Application to the 16 KiB program RAM memory of the VS1053b. After activation, encoder results can be read from registers SCI HDAT0 and SCI HDAT1, much like when using ADPCM recording (Chapter 9.8). Three profiles are provided: one for high-quality stereo recording at a bitrate of approx. 140 kbit/s, and two for speech-quality mono recording at a bitrates between 15 and 30 kbit/s. To use the Ogg Vorbis Encoder application, please load the application from VLSI Solution’s Web page http://www.vlsi.fi/software/plugins/plugins.shtml and read the accompanying documentation. Version 0.5, 2007-12-03 52 VLSI VS1053b preliminary y Solution 9.8 VS1053 B 9. OPERATION ADPCM Recording This chapter explains how to create RIFF/WAV file with IMA ADPCM format. This is a widely supported ADPCM format and many PC audio playback programs can play it. IMA ADPCM recording gives roughly a compression ratio of 4:1 compared to linear, 16-bit audio. This makes it possible to record for example ono 8 kHz audio at 32.44 kbit/s. VS1053 has a stereo ADC, thus also two-channel (separate AGC, if AGC enabled) and stereo (common AGC, if AGC enabled) modes are available. Mono recording mode selects either left or right channel. Left channel is either MIC or LINE1 depending on the SCI MODE register. 9.8.1 Activating ADPCM mode Register SCI MODE SCI AICTRL0 SCI AICTRL1 SCI AICTRL2 SCI AICTRL3 Bits 2, 12, 14 15..0 15..0 15..0 1..0 Description Start ADPCM mode, select MIC/LINE1 Sample rate 8000..48000 Hz (read at recording startup) Recording gain (1024 = 1×) or 0 for automatic gain control Maximum autogain amplification (1024 = 1×, 65535 = 64×) 0 = joint stereo (common AGC), 1 = dual channel (separate AGC), 2 = left channel, 3 = right channel 2 0 = IMA ADPCM mode, 1 = LINEAR PCM mode 15..3 reserved, set to 0 IMA ADPCM recording mode is activated by setting bits SM RESET and SM ADPCM in SCI MODE. Line input 1 is used instead of differential mic input if SM LINE1 is set. Before activating ADPCM recording, user must write the right values to SCI AICTRL0 and SCI AICTRL3. These values are only read at recording startup. SCI AICTRL1 and SCI AICTRL2 can be altered anytime, but it is preferable to write good init values before activation. SCI AICTRL1 controls linear recording gain. 1024 is equal to digital gain 1, 512 is equal to digital gain 0.5 and so on. If the user wants to use automatic gain control (AGC), SCI AICTRL1 should be set to 0. Typical speech applications usually are better off using AGC, as this takes care of relatively uniform speech loudness in recordings. SCI AICTRL2 controls the maximum AGC gain. This can be used to limit the amplification of noise when there is no signal. If SCI AICTRL2 is zero, the maximum gain is initialized to 65535 (64×), i.e. whole range is used. For example: WriteVS10xxRegister(SCI_AICTRL0, 16000U); WriteVS10xxRegister(SCI_AICTRL1, 0); WriteVS10xxRegister(SCI_AICTRL2, 16384U); WriteVS10xxRegister(SCI_AICTRL3, 0); WriteVS10xxRegister(SCI_MODE, ReadVS10xxRegister(SCI_MODE) | SM_RESET | SM_ADPCM | SM_LINE1); selects 16 kHz, stereo mode with automatic gain control and maximum amplification of 16×. Version 0.5, 2007-12-03 53 VLSI y Solution 9.8.2 VS1053 B VS1053b preliminary 9. OPERATION Reading IMA ADPCM Data After IMA ADPCM recording has been activated, registers SCI HDAT0 and SCI HDAT1 have new functions. The IMA ADPCM sample buffer is 1024 16-bit words. The fill status of the buffer can be read from SCI HDAT1. If SCI HDAT1 is greater than 0, you can read as many 16-bit words from SCI HDAT0. If the data is not read fast enough, the buffer overflows and returns to empty state. Note: if SCI HDAT1 ≥ 896, it may be better to wait for the buffer to overflow and clear before reading samples. That way you may avoid buffer aliasing. Each IMA ADPCM block is 128 words, i.e. 256 bytes. If you wish to interrupt reading data and possibly continue later, please stop at a 128-word boundary. This way whole blocks are skipped and the encoded stream stays valid. 9.8.3 Adding a RIFF Header To make your IMA ADPCM file a RIFF / WAV file, you have to add a header before the actual data. The following shows a header for mono file. Note that 2- and 4-byte values are little-endian (lowest byte first) in this format: File Offset 0 4 8 12 16 20 22 24 28 32 34 36 38 40 44 48 52 56 60 316 Field Name ChunkID ChunkSize Format SubChunk1ID SubChunk1Size AudioFormat NumOfChannels SampleRate ByteRate BlockAlign BitsPerSample ByteExtraData ExtraData SubChunk2ID SubChunk2Size NumOfSamples SubChunk3ID SubChunk3Size Block1 ... Size 4 4 4 4 4 2 2 4 4 2 2 2 2 4 4 4 4 4 256 Bytes "RIFF" F0 F1 F2 F3 "WAVE" "fmt " 0x14 0x0 0x0 0x0 0x11 0x0 C0 C1 R0 R1 R2 R3 B0 B1 B2 B3 0x0 0x1 0x4 0x0 0x2 0x0 0xf9 0x1 "fact" 0x4 0x0 0x0 0x0 S0 S1 S2 S3 "data" D0 D1 D2 D3 Description File size - 8 20 0x11 for IMA ADPCM 1 for mono, 2 for stereo 0x1f40 for 8 kHz 0xfd7 for 8 kHz mono 0x100 4-bit ADPCM 2 Samples per block (505) 4 Data size (File Size-60) First ADPCM block More ADPCM data blocks If we have n audio blocks, the values in the table are as follows: F = n × C × 256 + 52 R = Fs (see Chapter 9.8.1 to see how to calculate Fs ) B = Fs ×C×256 505 S = n × 505. D = n × C × 256 Version 0.5, 2007-12-03 54 VLSI VS1053b preliminary y Solution VS1053 B 9. OPERATION If you know beforehand how much you are going to record, you may fill in the complete header before any actual data. However, if you don’t know how much you are going to record, you have to fill in the header size datas F , S and D after finishing recording. The 128 words (256 bytes) of an ADPCM block are read from SCI HDAT0 and written into file as follows. The high 8 bits of SCI HDAT0 should be written as the first byte to a file, then the low 8 bits. Note that this is contrary to the default operation of some 16-bit microcontrollers, and you may have to take extra care to do this right. A way to see if you have written the file in the right way is to check bytes 2 and 3 (the first byte counts as byte 0) of each 256-byte block. Byte 3 should always be zero. Below is an example of a valid header for a 44.1 kHz stereo IMA ADPCM file that had a final length of 10038844 (0x992E3C) bytes: 0000 0010 0020 0030 9.8.4 52 14 00 14 49 00 02 15 46 00 04 97 46 00 00 00 34 11 02 64 2e 00 00 61 99 02 f9 74 00 00 01 61 57 44 66 00 41 ac 61 2e 56 00 63 99 45 66 6d 74 20 00 a7 ae 00 00 74 04 00 00 00 00 |RIFF4...WAVEfmt | |........D.......| |........fact....| |....data....| Playing ADPCM Data In order to play back your IMA ADPCM recordings, you have to have a file with a header as described in Chapter 9.8.3. If this is the case, all you need to do is to provide the ADPCM file through SDI as you would with any audio file. 9.8.5 Sample Rate Considerations VS10xx chips that support IMA ADPCM playback are capable of playing back ADPCM files with any sample rate. However, some other programs may expect IMA ADPCM files to have some exact sample rates, like 8000 or 11025 Hz. Also, some programs or systems do not support sample rates below 8000 Hz. If you want better quality with the expense of increased data rate, you can use higher sample rates, for example 16 kHz. Version 0.5, 2007-12-03 55 VLSI VS1053b preliminary y Solution 9.9 VS1053 B 9. OPERATION SPI Boot If GPIO0 is set with a pull-up resistor to 1 at boot time, VS1053b tries to boot from external SPI memory. SPI boot redefines the following pins: Normal Mode GPIO0 GPIO1 DREQ GPIO2 SPI Boot Mode xCS CLK MOSI MISO The memory has to be an SPI Bus Serial EEPROM with 16-bit or 24-bit addresses. The serial speed used by VS1053b is 245 kHz with the nominal 12.288 MHz clock. The first three bytes in the memory have to be 0x50, 0x26, 0x48. 9.10 Real-Time MIDI If GPIO0 is low and GPIO1 is high during boot, real-time MIDI mode is activated. In this mode the PLL is configured to 4.0×, the UART is configured to the MIDI data rate 31250 bps, and real-time MIDI data is then read from UART and SDI. Both input methods should not be used simultaneously. If you use SDI, first send 0xff and then send the MIDI data byte. EarSpeaker setting can be configured with GPIO2 and GPIO3. The state of GPIO2 and GPIO3 are only read at startup. Real-Time MIDI can also be started with a small patch code using SCI. Note: The real-time MIDI parser in VS1053b does not know how to skip SysEx messages. An improved version can be loaded into IRAM if needed. Version 0.5, 2007-12-03 56 VLSI VS1053b preliminary y Solution 9.11 VS1053 B 9. OPERATION Extra Parameters The following structure is in X memory at address 0x1e00 (note the different location than in VS1033) and can be used to change some extra parameters or get useful information. The chip ID is also easily available. #define PARAMETRIC_VERSION 0x0003 struct parametric { /* configs are not cleared between files */ u_int32 chipID; /*1e00/01 Initialized at reset for your convenience */ u_int16 version; /*1e02 - structure version */ u_int16 config1; /*1e03 ---- ---- ppss RRRR PS mode, SBR mode, Reverb */ u_int16 playSpeed; /*1e04 0,1 = normal speed, 2 = twice, 3 = three times etc. */ u_int16 byteRate; /*1e05 average byterate */ u_int16 endFillByte; /*1e06 byte value to send after file sent */ u_int16 reserved[16]; /*1e07..15 file byte offsets */ u_int32 jumpPoints[8]; /*1e16..25 file byte offsets */ u_int16 latestJump; /*1e26 index to lastly updated jumpPoint */ u_int32 positionMsec /*1e27-28 play position, if known (WMA, Ogg Vorbis) */ s_int16 resync; /*1e29 > 0 for automatic m4a, ADIF, WMA resyncs */ union { struct { u_int32 curPacketSize; u_int32 packetSize; } wma; struct { u_int16 sceFoundMask; /*1e2a SCE’s found since last clear */ u_int16 cpeFoundMask; /*1e2b CPE’s found since last clear */ u_int16 lfeFoundMask; /*1e2c LFE’s found since last clear */ u_int16 playSelect; /*1e2d 0 = first any, initialized at aac init */ s_int16 dynCompress; /*1e2e -8192=1.0, initialized at aac init */ s_int16 dynBoost; /*1e2f 8192=1.0, initialized at aac init */ u_int16 sbrAndPsStatus; /*0x1e30 1=SBR, 2=upsample, 4=PS, 8=PS active */ } aac; struct { u_int32 bytesLeft; } midi; struct { s_int16 gain; /* 0x1e2a proposed gain offset in 0.5dB steps, default = -12 */ } vorbis; } i; }; Notice that reading two-word variables through the SCI WRAMADDR and SCI WRAM interface is not protected in any way. The variable can be updated between the read of the low and high parts. The problem arises when both the low and high parts change values. To determine if the value is correct, you should read the value twice and compare the results. The following example shows what happens when bytesLeft is decreased from 0x10000 to 0xffff and the update happens between low and high part reads or after high part read. Read Invalid Address Value 0x1e2a 0x0000 change after this 0x1e2b 0x0000 0x1e2a 0xffff 0x1e2b 0x0000 Version 0.5, 2007-12-03 Address 0x1e2a 0x1e2b 0x1e2a 0x1e2b Read Valid Value 0x0000 0x0001 change after this 0xffff 0x0000 No Update Address Value 0x1e2a 0x0000 0x1e2b 0x0001 0x1e2a 0x0000 0x1e2b 0x0001 57 VLSI VS1053b preliminary y Solution VS1053 B 9. OPERATION You can see that in the invalid read the low part wraps from 0x0000 to 0xffff while the high part stays the same. In this case the second read gives a valid answer, otherwise always use the value of the first read. The second read is needed when it is possible that the low part wraps around, changing the high part, i.e. when the low part is small. bytesLeft is only decreased by one at a time, so a reread is needed only if the low part is 0. 9.11.1 Common Parameters These parameters are common for all codecs. Other fields are only valid when the corresponding codec is active. The currently active codec can be determined from SCI HDAT1. Parameter chipID version config1 playSpeed byteRate endFillByte jumpPoints[8] latestJump positionMsec resync Address 0x1e00-01 0x1e02 0x1e03 0x1e04 0x1e05 0x1e06 0x1e16-25 0x1e26 0x1e27-28 0x1e29 Usage Fuse-programmed unique ID (cosmetic copy of the fuses) Structure version – 0x0003 Miscellaneous configuration 0,1 = normal speed, 2 = twice, 3 = three times etc. average byterate byte to send after file Packet offsets for WMA and AAC Index to latest jumpPoint File position in milliseconds, if available Automatic resync selector The fuse-programmed ID is read at startup and copied into the chipID field. If not available, the value will be all zeros. The version field can be used to determine the layout of the rest of the structure. The version number is changed when the structure is changed. For VS1053b the structure version is 3. config1 controls MIDI Reverb and AAC’s SBR and PS settings. playSpeed makes it possible to fast forward songs. Decoding of the bitstream is performed, but only each playSpeed frames are played. For example by writing 4 to playSpeed will play the song four times as fast as normal, if you are able to feed the data with that speed. Write 0 or 1 to return to normal speed. SCI DECODE TIME will also count faster. All current codecs support the playSpeed configuration. byteRate contains the average bitrate in bytes per second for every code. The value is updated once per second and it can be used to calculate an estimate of the remaining playtime. This value is also available in SCI HDAT0 for all codecs except MP3, MP2, and MP1. endFillByte indicates what byte value to send after file is sent before SM CANCEL. jumpPoints contain 32-bit file offsets. Each valid (non-zero) entry indicates a start of a packet for WMA or start of a raw data block for AAC (ADIF, .mp4 / .m4a). latestJump contains the index of the entry that was updated last. If you only read entry pointed to by latestJump you do not need to read the entry twice to ensure validity. Jump point information can be used to implement perfect fast forward and rewind for WMA and AAC (ADIF, .mp4 / .m4a). Version 0.5, 2007-12-03 58 VLSI VS1053b preliminary y Solution VS1053 B 9. OPERATION positionMsec is a field that gives the current play position in a file in milliseconds, regardless of rewind and fast forward operations. The value is only available in codecs that can determine the play position from the stream itself. Currently WMA and Ogg Vorbis provide this information. If the position is unknown, this field contains -1. resync field is used to force a resynchronization to the stream for WMA and AAC (ADIF, .mp4 / .m4a) instead of ending the decode at first error. This field can be used to implement almost perfect fast forward and rewind for WMA and AAC (ADIF, .mp4 / .m4a). The user should set this field before performing data seeks if they are not in packet or data block boundaries. The field value tells how many tries are allowed before giving up. The value 32767 gives infinite tries. The resync field is set to 32767 after a reset to make resynchronization the default action, but it can be cleared after reset to restore the old action. When resync is set, every file decode should always end as described in Chapter 9.5.1. Seek fields no longer exist. When resync is required, WMA and AAC codecs now enter broadcast/stream mode where file size information is ignored. Also, the file size and sample size information of WAV files are ignored when resync is non-zero. The user must use SM CANCEL or software reset to end decoding. Note: WAV, WMA, ADIF, and .mp4 / .m4a files begin with a metadata or header section, which must be fully processed before any fast forward or rewind operation. SS DO NOT JUMP (in SCI STATUS) is clear when the header information has been processed and jumps are allowed. 9.11.2 WMA Parameter curPacketSize packetSize Address 0x1e2a/2b 0x1e2c/2d Usage The size of the packet being processed The packet size in ASF header The ASF header packet size is available in packetSize. With this information and a packet start offset from jumpPoints you can parse the packet headers and skip packets in ASF files. WMA decoder can also increase the internal clock automatically when it detects that a file can not be decoded correctly with the current clock. The maximum allowed clock is configured with the SCI CLOCKF register. Version 0.5, 2007-12-03 59 VLSI VS1053b preliminary y Solution 9.11.3 VS1053 B 9. OPERATION AAC Parameter config1 sceFoundMask cpeFoundMask lfeFoundMask playSelect dynCompress dynBoost sbrAndPsStatus Address 0x1e03(7:4) 0x1e2a 0x1e2b 0x1e2c 0x1e2d 0x1e2e 0x1e2f 0x1e30 Usage SBR and PS select Single channel elements found Channel pair elements found Low frequency elements found Play element selection Compress coefficient for DRC, -8192=1.0 Boost coefficient for DRC, 8192=1.0 SBR and PS available flags playSelect determines which element to decode if a stream has multiple elements. The value is set to 0 each time AAC decoding starts, which causes the first element that appears in the stream to be selected for decoding. Other values are: 0x01 - select first single channel element (SCE), 0x02 - select first channel pair element (CPE), 0x03 - select first low frequency element (LFE), S ∗ 16 + 5 - select SCE number S, P ∗ 16 + 6 - select CPE number P, L ∗ 16 + 7 - select LFE number L. When automatic selection has been performed, playSelect reflects the selected element. sceFoundMask, cpeFoundMask, and lfeFoundMask indicate which elements have been found in an AAC stream since the variables have last been cleared. The values can be used to present an element selection menu with only the available elements. dynCompress and dynBoost change the behavior of the dynamic range control (DRC) that is present in some AAC streams. These are also initialized when AAC decoding starts. sbrAndPsStatus indicates spectral band replication (SBR) and parametric stereo (PS) status. Bit 0 1 2 3 Usage SBR present upsampling active PS present PS active Bits 7 to 4 in config1 can be used to control the SBR and PS decoding. Bits 5 and 4 select SBR mode and bits 7 and 6 select PS mode. These configuration bits are useful if your AAC license does not cover SBR and/or PS. config1(5:4) ’00’ ’01’ ’10’ ’11’ Version 0.5, Usage normal mode, upsample <24 kHz AAC files do not automatically upsample <24 kHz AAC files, but enable upsampling if SBR is encountered never upsample disable SBR (also disables PS) 2007-12-03 60 VLSI VS1053b preliminary y Solution config1(7:6) ’00’ ’01’ ’10’ ’11’ VS1053 B 9. OPERATION Usage normal mode, process PS if it is available process PS if it is available, but in downsampled mode reserved disable PS processing AAC decoder can also increase the internal clock automatically when it detects that a file can not be decoded correctly with the current clock. The maximum allowed clock is configured with the SCI CLOCKF register. If even the highest allowed clock is too slow to decode an AAC file with SBR and PS components, the advanced decoding features are automatically dropped one by one until the file can be played. First the parametric stereo processing is dropped (the playback becomes mono). If that is not enough, the spectral band replication is turned into downsampled mode (reduced bandwidth). As the last resort the spectral band replication is fully disabled. Dropped features are restored at each song change. 9.11.4 Midi Parameter config1 Address 0x1e03 bits [3:0] bytesLeft 0x1e2a/2b Usage Miscellaneous configuration Reverb: 0 = auto (ON if clock >= 3.0×) 1 = off, 2 - 15 = room size The number of bytes left in this track The lowest 4 bits of config1 controls the reverb effect. 9.11.5 Ogg Vorbis Parameter gain Address 0x1e2a Usage Preferred replay-gain offset Ogg Vorbis decoding supports Replay Gain technology. The Replay Gain technology is used to automatically give all songs a matching volume so that the user does not need to adjust the volume setting between songs. If the Ogg Vorbis decoder finds a Replay Gain tag in the song header, the tag is parsed and the decoded gain setting can be found from the gain parameter. For a song without any Replay Gain tag, a default of -6 dB (gain value -12) is used. For more details about Replay Gain, see http://en.wikipedia.org/wiki/Replay Gain and http://www.replaygain.org/. The player software can use the gain value to adjust the volume level. Negative values mean that the volume should be decreased, positive values mean that the volume should be increased. For example gain = -11 means that volume should be decreased by 5.5 dB (−11/2 = −5.5), and left and right attenuation should be increased by 11. When gain = 2 volume should be increased by 1 dB (2/2 = 1.0), and left and right attenuation should be decreased by 2. Because volume setting can not go above +0 dB, the value should be saturated. Version 0.5, 2007-12-03 61 VLSI y Solution Gain -11 (-5.5 dB) -11 (-5.5 dB) +2 (+1.0 dB) +2 (+1.0 dB) +2 (+1.0 dB) 9.12 VS1053 B VS1053b preliminary Volume 0 (+0.0 dB) 3 (-1.5 dB) 0 (+0.0 dB) 1 (-0.5 dB) 4 (-2.0 dB) 9. OPERATION SCI VOL (Volume-Gain) 0x0b0b (-5.5 dB) 0x0e0e (-7.0 dB) 0x0000 (+0.0 dB) 0x0000 (+0.0 dB) 0x0202 (-1.0 dB) SDI Tests There are several test modes in VS1053b, which allow the user to perform memory tests, SCI bus tests, and several different sine wave tests. All tests are started in a similar way: VS1053b is hardware reset, SM TESTS is set, and then a test command is sent to the SDI bus. Each test is started by sending a 4-byte special command sequence, followed by 4 zeros. The sequences are described below. 9.12.1 Sine Test Sine test is initialized with the 8-byte sequence 0x53 0xEF 0x6E n 0 0 0 0, where n defines the sine test to use. n is defined as follows: Name F s Idx S n bits Bits Description 7:5 Samplerate index 4:0 Sine skip speed F s Idx 0 1 2 3 Fs 44100 Hz 48000 Hz 32000 Hz 22050 Hz F s Idx 4 5 6 7 The frequency of the sine to be output can now be calculated from F = F s × Fs 24000 Hz 16000 Hz 11025 Hz 12000 Hz S 128 . Example: Sine test is activated with value 126, which is 0b01111110. Breaking n to its components, Fs Idx = 0b011 = 3 and thus Fs = 22050Hz. S = 0b11110 = 30, and thus the final sine frequency 30 F = 22050Hz × 128 ≈ 5168Hz. To exit the sine test, send the sequence 0x45 0x78 0x69 0x74 0 0 0 0. Note: Sine test signals go through the digital volume control, so it is possible to test channels separately. 9.12.2 Pin Test Pin test is activated with the 8-byte sequence 0x50 0xED 0x6E 0x54 0 0 0 0. This test is meant for chip production testing only. Version 0.5, 2007-12-03 62 VLSI VS1053b preliminary y Solution 9.12.3 VS1053 B 9. OPERATION SCI Test Sci test is initialized with the 8-byte sequence 0x53 0x70 0xEE n 0 0 0 0, where n − 48 is the register number to test. The content of the given register is read and copied to SCI HDAT0. If the register to be tested is HDAT0, the result is copied to SCI HDAT1. Example: if n is 48, contents of SCI register 0 (SCI MODE) is copied to SCI HDAT0. 9.12.4 Memory Test Memory test mode is initialized with the 8-byte sequence 0x4D 0xEA 0x6D 0x54 0 0 0 0. After this sequence, wait for 1100000 clock cycles. The result can be read from the SCI register SCI HDAT0, and ’one’ bits are interpreted as follows: Bit(s) 15 14:10 9 8 7 6 5 4 3 2 1 0 Mask 0x8000 0x0200 0x0100 0x0080 0x0040 0x0020 0x0010 0x0008 0x0004 0x0002 0x0001 0x83ff Meaning Test finished Unused Mux test succeeded Good MAC RAM Good I RAM Good Y RAM Good X RAM Good I ROM 1 Good I ROM 2 Good Y ROM Good X ROM 1 Good X ROM 2 All ok Memory tests overwrite the current contents of the RAM memories. 9.12.5 New Sine and Sweep Tests A more frequency-accurate sine test can be started and controlled from SCI. SCI AICTRL0 and SCI AICTRL1 set the sine frequencies for left and right channel, respectively. These registers, volume (SCI VOL), and samplerate (SCI AUDATA) can be set before or during the test. Write 0x4020 to SCI AIADDR to start the test. SCI AICTRLn can be calculated from the desired frequency and DAC samplerate by: SCI AICT RLn = Fsin × 65536/Fs The maximum value for SCI AICTRLn is 0x8000U. For the best S/N ratio for the generated sine, three Version 0.5, 2007-12-03 63 VLSI VS1053b preliminary y Solution VS1053 B 9. OPERATION LSb’s of the SCI AICTRLn should be zero. The resulting frequencies Fsin can be calculated from the DAC samplerate Fs and SCI AICTRL0 / SCI AICTRL1 using the following equation. Fsin = SCI AICT RLn × F s /65536 Sine sweep test can be started by writing 0x4022 to SCI AIADDR. Both these tests use the normal audio path, thus also SCI BASS, differential output mode, and EarSpeaker settings have an effect. Version 0.5, 2007-12-03 64 VLSI y Solution 10 10.1 VS1053 B VS1053b preliminary 10. VS1053B REGISTERS VS1053b Registers Who Needs to Read This Chapter User software is required when a user wishes to add some own functionality like DSP effects to VS1053b. However, most users of VS1053b don’t need to worry about writing their own code, or about this chapter, including those who only download software plug-ins from VLSI Solution’s Web site. 10.2 The Processor Core VS DSP is a 16/32-bit DSP processor core that also had extensive all-purpose processor features. VLSI Solution’s free VSKIT Software Package contains all the tools and documentation needed to write, simulate and debug Assembly Language or Extended ANSI C programs for the VS DSP processor core. VLSI Solution also offers a full Integrated Development Environment VSIDE for full debug capabilities. 10.3 VS1053b Memory Map X-memory Address Description 0x0000..0x17ff System RAM 0x1800..0x187f User RAM 0x1880..0x197f Stack 0x1980..0x3fff System RAM 0x4000..0xbfff ROM 32k 0xc000..0xc0ff Peripherals 0xc100..0xffff ROM 15.75k 10.4 Y-memory Address Description 0x0000..0x17ff System RAM 0x1800..0x187f User RAM 0x1880..0x197f Stack 0x1980..0x3fff System RAM 0x4000..0xdfff ROM 40k 0xe000..0xffff System RAM I-memory Address Description 0x0000..0x004f System RAM 0x0050..0x0fff User RAM 0x1000..0x1fff 0x2000..0xffff ROM 56k and banked 0xc000..0xffff ROM4 16k SCI Registers SCI registers described in Chapter 8.7 can be found here between 0xC000..0xC00F. In addition to these registers, there is one in address 0xC010, called SCI CHANGE. Reg 0xC010 Type r Name SCI CH WRITE SCI CH ADDR Version 0.5, 2007-12-03 Reset 0 SCI registers, prefix SCI Abbrev[bits] Description CHANGE[5:0] Last SCI access address Bits 4 3:0 SCI CHANGE bits Description 1 if last access was a write cycle SCI address of last access 65 VLSI VS1053b preliminary y Solution 10.5 10. VS1053B REGISTERS Serial Data Registers Reg 0xC011 0xC012 10.6 VS1053 B Type r w Reset 0 0 SDI registers, prefix SER Abbrev[bits] Description DATA Last received 2 bytes, big-endian DREQ[0] DREQ pin control DAC Registers Reg 0xC013 0xC014 0xC015 0xC016 Type rw rw rw rw Reset 0 0 0 0 DAC registers, prefix DAC Abbrev[bits] Description FCTLL DAC frequency control, 16 LSbs FCTLH DAC frequency control 4MSbs, PLL control LEFT DAC left channel PCM value RIGHT DAC right channel PCM value Every fourth clock cycle, an internal 26-bit counter is added to by (DAC FCTLH & 15) × 65536 + DAC FCTLL. Whenever this counter overflows, values from DAC LEFT and DAC RIGHT are read and a DAC interrupt is generated. 10.7 GPIO Registers Reg 0xC017 0xC018 0xC019 Type rw r rw Reset 0 0 0 GPIO registers, prefix GPIO Abbrev[bits] Description DDR[7:0] Direction IDATA[7:0] Values read from the pins ODATA[7:0] Values set to the pins GPIO DIR is used to set the direction of the GPIO pins. 1 means output. GPIO ODATA remembers its values even if a GPIO DIR bit is set to input. GPIO registers don’t generate interrupts. Note that in VS1053b the VSDSP registers can be read and written through the SCI WRAMADDR and SCI WRAM registers. You can thus use the GPIO pins quite conveniently. Version 0.5, 2007-12-03 66 VLSI VS1053b preliminary y Solution 10.8 VS1053 B 10. VS1053B REGISTERS Interrupt Registers Reg 0xC01A 0xC01B 0xC01C 0xC01D Type rw w w rw Reset 0 0 0 0 Interrupt registers, prefix INT Abbrev[bits] Description ENABLE[7:0] Interrupt enable GLOB DIS[-] Write to add to interrupt counter GLOB ENA[-] Write to subtract from interrupt counter COUNTER[4:0] Interrupt counter INT ENABLE controls the interrupts. The control bits are as follows: Name INT EN INT EN INT EN INT EN INT EN INT EN INT EN TIM1 TIM0 RX TX SDI SCI DAC Bits 7 6 5 4 2 1 0 INT ENABLE bits Description Enable Timer 1 interrupt Enable Timer 0 interrupt Enable UART RX interrupt Enable UART TX interrupt Enable Data interrupt Enable SCI interrupt Enable DAC interrupt Note: It may take upto 6 clock cycles before changing INT ENABLE has any effect. Writing any value to INT GLOB DIS adds one to the interrupt counter INT COUNTER and effectively disables all interrupts. It may take upto 6 clock cycles before writing to this register has any effect. Writing any value to INT GLOB ENA subtracts one from the interrupt counter (unless INT COUNTER already was 0). If the interrupt counter becomes zero, interrupts selected with INT ENABLE are restored. An interrupt routine should always write to this register as the last thing it does, because interrupts automatically add one to the interrupt counter, but subtracting it back to its initial value is the responsibility of the user. It may take upto 6 clock cycles before writing this register has any effect. By reading INT COUNTER the user may check if the interrupt counter is correct or not. If the register is not 0, interrupts are disabled. Version 0.5, 2007-12-03 67 VLSI VS1053b preliminary y Solution 10.9 VS1053 B 10. VS1053B REGISTERS Watchdog v1.0 2002-08-26 The watchdog consist of a watchdog counter and some logic. After reset, the watchdog is inactive. The counter reload value can be set by writing to WDOG CONFIG. The watchdog is activated by writing 0x4ea9 to register WDOG RESET. Every time this is done, the watchdog counter is reset. Every 65536’th clock cycle the counter is decremented by one. If the counter underflows, it will activate vsdsp’s internal reset sequence. Thus, after the first 0x4ea9 write to WDOG RESET, subsequent writes to the same register with the same value must be made no less than every 65536×WDOG CONFIG clock cycles. Once started, the watchdog cannot be turned off. Also, a write to WDOG CONFIG doesn’t change the counter reload value. After watchdog has been activated, any read/write operation from/to WDOG CONFIG or WDOG DUMMY will invalidate the next write operation to WDOG RESET. This will prevent runaway loops from resetting the counter, even if they do happen to write the correct number. Writing a wrong value to WDOG RESET will also invalidate the next write to WDOG RESET. Reads from watchdog registers return undefined values. 10.9.1 Registers Reg 0xC020 0xC021 0xC022 Version 0.5, Watchdog, prefix WDOG Type Reset Abbrev Description w 0 CONFIG Configuration w 0 RESET Clock configuration w 0 DUMMY[-] Dummy register 2007-12-03 68 VLSI y Solution VS1053 B VS1053b preliminary 10.10 10. VS1053B REGISTERS UART v1.1 2004-10-09 RS232 UART implements a serial interface using rs232 standard. Start bit D0 D1 D2 D3 D4 D5 D6 Stop D7 bit Figure 15: RS232 Serial Interface Protocol When the line is idling, it stays in logic high state. When a byte is transmitted, the transmission begins with a start bit (logic zero) and continues with data bits (LSB first) and ends up with a stop bit (logic high). 10 bits are sent for each 8-bit byte frame. 10.10.1 Registers Reg 0xC028 0xC029 0xC02A 0xC02B UART registers, prefix UARTx Type Reset Abbrev Description r 0 STATUS[4:0] Status r/w 0 DATA[7:0] Data r/w 0 DATAH[15:8] Data High r/w 0 DIV Divider 10.10.2 Status UARTx STATUS A read from the status register returns the transmitter and receiver states. Name UART UART UART UART UART ST ST ST ST ST FRAMEERR RXORUN RXFULL TXFULL TXRUNNING UARTx STATUS Bits Bits Description 4 Framing error (stop bit was 0) 3 Receiver overrun 2 Receiver data register full 1 Transmitter data register full 0 Transmitter running UART ST FRAMEERR is set if the stop bit of the received byte was 0. UART ST RXORUN is set if a received byte overwrites unread data when it is transferred from the receiver shift register to the data register, otherwise it is cleared. UART ST RXFULL is set if there is unread data in the data register. UART ST TXFULL is set if a write to the data register is not allowed (data register full). UART ST TXRUNNING is set if the transmitter shift register is in operation. Version 0.5, 2007-12-03 69 VLSI VS1053b preliminary y Solution VS1053 B 10. VS1053B REGISTERS 10.10.3 Data UARTx DATA A read from UARTx DATA returns the received byte in bits 7:0, bits 15:8 are returned as ’0’. If there is no more data to be read, the receiver data register full indicator will be cleared. A receive interrupt will be generated when a byte is moved from the receiver shift register to the receiver data register. A write to UARTx DATA sets a byte for transmission. The data is taken from bits 7:0, other bits in the written value are ignored. If the transmitter is idle, the byte is immediately moved to the transmitter shift register, a transmit interrupt request is generated, and transmission is started. If the transmitter is busy, the UART ST TXFULL will be set and the byte remains in the transmitter data register until the previous byte has been sent and transmission can proceed. 10.10.4 Data High UARTx DATAH The same as UARTx DATA, except that bits 15:8 are used. 10.10.5 Divider UARTx DIV Name UART DIV D1 UART DIV D2 UARTx DIV Bits Bits Description 15:8 Divider 1 (0..255) 7:0 Divider 2 (6..255) The divider is set to 0x0000 in reset. The ROM boot code must initialize it correctly depending on the master clock frequency to get the correct bit speed. The second divider (D2 ) must be from 6 to 255. The communication speed f = TX/RX speed in bps. fm (D1 +1)×(D2 ) , where fm is the master clock frequency, and f is the Divider values for common communication speeds at 26 MHz master clock: Example UART Speeds, fm = 26M Hz Comm. Speed [bps] UART DIV D1 UART DIV D2 4800 85 63 9600 42 63 14400 42 42 19200 51 26 28800 42 21 38400 25 26 57600 1 226 115200 0 226 Version 0.5, 2007-12-03 70 VLSI VS1053b preliminary y Solution VS1053 B 10. VS1053B REGISTERS 10.10.6 Interrupts and Operation Transmitter operates as follows: After an 8-bit word is written to the transmit data register it will be transmitted instantly if the transmitter is not busy transmitting the previous byte. When the transmission begins a TX INTR interrupt will be sent. Status bit [1] informs the transmitter data register empty (or full state) and bit [0] informs the transmitter (shift register) empty state. A new word must not be written to transmitter data register if it is not empty (bit [1] = ’0’). The transmitter data register will be empty as soon as it is shifted to transmitter and the transmission is begun. It is safe to write a new word to transmitter data register every time a transmit interrupt is generated. Receiver operates as follows: It samples the RX signal line and if it detects a high to low transition, a start bit is found. After this it samples each 8 bit at the middle of the bit time (using a constant timer), and fills the receiver (shift register) LSB first. Finally the data in the receiver is moved to the reveive data register, the stop bit state is checked (logic high = ok, logic low = framing error) for status bit[4], the RX INTR interrupt is sent, status bit[2] (receive data register full) is set, and status bit[2] old state is copied to bit[3] (receive data overrun). After that the receiver returns to idle state to wait for a new start bit. Status bit[2] is zeroed when the receiver data register is read. RS232 communication speed is set using two clock dividers. The base clock is the processor master clock. Bits 15-8 in these registers are for first divider and bits 7-0 for second divider. RX sample frequency is the clock frequency that is input for the second divider. Version 0.5, 2007-12-03 71 VLSI VS1053b preliminary y Solution 10.11 VS1053 B 10. VS1053B REGISTERS Timers v1.0 2002-04-23 There are two 32-bit timers that can be initialized and enabled independently of each other. If enabled, a timer initializes to its start value, written by a processor, and starts decrementing every clock cycle. When the value goes past zero, an interrupt is sent, and the timer initializes to the value in its start value register, and continues downcounting. A timer stays in that loop as long as it is enabled. A timer has a 32-bit timer register for down counting and a 32-bit TIMER1 LH register for holding the timer start value written by the processor. Timers have also a 2-bit TIMER ENA register. Each timer is enabled (1) or disabled (0) by a corresponding bit of the enable register. 10.11.1 Reg 0xC030 0xC031 0xC034 0xC035 0xC036 0xC037 0xC038 0xC039 0xC03A 0xC03B Registers Timer registers, prefix TIMER Type Reset Abbrev Description r/w 0 CONFIG[7:0] Timer configuration r/w 0 ENABLE[1:0] Timer enable r/w 0 T0L Timer0 startvalue - LSBs r/w 0 T0H Timer0 startvalue - MSBs r/w 0 T0CNTL Timer0 counter - LSBs r/w 0 T0CNTH Timer0 counter - MSBs r/w 0 T1L Timer1 startvalue - LSBs r/w 0 T1H Timer1 startvalue - MSBs r/w 0 T1CNTL Timer1 counter - LSBs r/w 0 T1CNTH Timer1 counter - MSBs 10.11.2 Configuration TIMER CONFIG Name TIMER CF CLKDIV TIMER CONFIG Bits Bits Description 7:0 Master clock divider TIMER CF CLKDIV is the master clock divider for all timer clocks. The generated internal clock fm frequency fi = c+1 , where fm is the master clock frequency and c is TIMER CF CLKDIV. Example: With a 12 MHz master clock, TIMER CF DIV=3 divides the master clock by 4, and the output/sampling Hz clock would thus be fi = 12M 3+1 = 3M Hz. Version 0.5, 2007-12-03 72 VLSI VS1053b preliminary y Solution VS1053 B 10. VS1053B REGISTERS 10.11.3 Configuration TIMER ENABLE Name TIMER EN T1 TIMER EN T0 TIMER ENABLE Bits Bits Description 1 Enable timer 1 0 Enable timer 0 10.11.4 Timer X Startvalue TIMER Tx[L/H] The 32-bit start value TIMER Tx[L/H] sets the initial counter value when the timer is reset. The timer fi interrupt frequency ft = c+1 where fi is the master clock obtained with the clock divider (see Chapter 10.11.2 and c is TIMER Tx[L/H]. Example: With a 12 MHz master clock and with TIMER CF CLKDIV=3, the master clock fi = 3M Hz. Hz If TIMER TH=0, TIMER TL=99, then the timer interrupt frequency ft = 3M 99+1 = 30kHz. 10.11.5 Timer X Counter TIMER TxCNT[L/H] TIMER TxCNT[L/H] contains the current counter values. By reading this register pair, the user may get knowledge of how long it will take before the next timer interrupt. Also, by writing to this register, a one-shot different length timer interrupt delay may be realized. 10.11.6 Interrupts Each timer has its own interrupt, which is asserted when the timer counter underflows. Version 0.5, 2007-12-03 73 VLSI y Solution 10.12 VS1053 B VS1053b preliminary 10. VS1053B REGISTERS VS1053b Audio Path MICN MICP LINE1 MIC AMP ADC MUX Stereo ADC LINE2 Audio FIFO Sample-Rate Converter Sigma-Delta Modulator + Analog Drivers LEFT RIGHT CBUF Volume Control SRC I2S SDM Figure 16: VS1053b ADC and DAC data paths In IMA ADPCM encoding mode the data from Analog-to-Digital conversion is first processed in 48 kHz or 24 kHz samplerate. The firmware performs DC offset removal and gain control (automatic or fixed), then redirects the data to the audio FIFO. From there the data goes to the samplerate converter with a delay of only a couple of samples. The samplerate converter upsamples the data to XTALI/2 (6.144 MHz with the default clock), from where it is resampled to either 1×, 2×, or 3× the requested samplerate. The additional decimation is performed in software to get the final data at the right frequency for IMA ADPCM encoding or for PCM samples. Version 0.5, 2007-12-03 74 VLSI y Solution 10.13 VS1053 B VS1053b preliminary 10. VS1053B REGISTERS I2S DAC Interface The I2S Interface makes it possible to attach an external DAC to the system. Note: in VS1053b the I2S pins share different GPIO pins than in VS1033 to be able to use SPI boot and I2S in the same application. 10.13.1 Registers Reg 0xC040 I2S registers, prefix I2S Type Reset Abbrev Description r/w 0 CONFIG[3:0] I2S configuration 10.13.2 Configuration I2S CONFIG Name I2S CF MCLK ENA I2S CF ENA I2S CF SRATE I2S CONFIG Bits Description Enables the MCLK output (12.288 MHz) Enables I2S, otherwise pins are GPIO I2S rate, ”10” = 192, ”01” = 96, ”00” = 48 kHz Bits 3 2 1:0 I2S CF ENA enables the I2S interface. After reset the interface is disabled and the pins are used for GPIO. I2S CF MCLK ENA enables the MCLK output. The frequency is either directly the input clock (nominal 12.288 MHz), or half the input clock when mode register bit SM CLK RANGE is set to 1 (2426 MHz input clock). I2S CF SRATE controls the output samplerate. When set to 48 kHz, SCLK is MCLK divided by 8, when 96 kHz SCLK is MCLK divided by 4, and when 192 kHz SCLK is MCLK divided by 2. MCLK SCLK LROUT SDATA MSB LSB Left Channel Word MSB Right Channel Word Figure 17: I2S Interface, 192 kHz. To enable I2S first write 0xc017 to SCI WRAMADDR and 0xf3 to SCI WRAM, then write 0xc040 to SCI WRAMADDR and 0x0c to SCI WRAM. See application notes for more information. Version 0.5, 2007-12-03 75 VLSI VS1053b preliminary y Solution 11 VS1053 B 11. VS1053 VERSION CHANGES VS1053 Version Changes This chapter describes the lastest and most important changes done to VS1053. 11.1 Changes Between VS1033c and VS1053a/b Firmware, 2007-03-08 Completely new or major changes: • No software reset required between files when used correctly. • Ogg Vorbis decoding added. Non-fatal ogg or vorbis decode errors cause automatic resync. This allows easy rewind and fast forward. Decoding ends if the ”last frame” flag is reached or SM CANCEL is set. • HE-AAC v2 Level 3 decoding added. It is possible to disable PS and SBR processing and control the upsampling modes through parametric x.control1. • Like the WMA decoder, the AAC decoder uses the clock adder (see SCI CLOCKF) if it needs more clock to decode the file. HE-AAC features are dropped one by one, if the file can not be decoded correctly even with the highest allowed clock. Parametric stereo is the first feature to be dropped, then downsampled mode is used, and as the final resort Spectral Band Replication is disabled. Features are automatically restored for the next file. • Completely new volume control with zero-cross detection prevents pops when volume is changed. • Audio FIFO underrun detection (with slow fade to zero) instead of looping the audio buffer content. • Average bitrate calculation (byteRate) for all codecs. • All codecs support fast play mode with selectable speeds for the best-quality fast forward operation. Fast play also advances DECODE TIME faster. • WMA and Ogg Vorbis provide an absolute decode position in milliseconds. • When SM CANCEL is detected, the firmware also discards the stream buffer contents. • Bit SCIST DO NOT JUMP in SCI STATUS is ’1’ when jumps in the file should not be done: during header processing and with Midi files. • IMA ADPCM encode now supports stereo encoding and selectable samplerate. • I2S pins are now GPIO4-GPIO7 and do not overlap with SPI boot pins. Other changes or additions: • Delayed volume and bass/treble control calculation reduces the time the corresponding SCI operations take. This delayed handling and the new volume control hardware prevents audio samples from being missed during volume change. • SCI DECODE TIME only cleared at hardware and software reset to allow files to be played backto-back or looped. • Read and write to YRAM at 0xe000..0xffff added to SCI WRAMADDR/SCI WRAM. • The resync parameter (parametric x.resync) is set to 32767 after reset to allow inifinite resynchronization attempts (or until SM CANCEL is set). Old operation can be restored by writing 0 to resync after reset. Version 0.5, 2007-12-03 76 VLSI VS1053b preliminary y Solution VS1053 B 11. VS1053 VERSION CHANGES • WMA,AAC: more robust resync. • WMA,AAC: If resync is performed, broadcast mode is automatically activated. The broadcast mode disables file size checking, and decoding continues until SM CANCEL is set or reset is performed. • Treble control fixed (volume change could cause bad artefacts). • MPEG Layer I mono fixed. • MPEG Layer II half-rate decoding fixed (frame size was calculated wrong). • MPEG Layer II accuracy problem fixed, invalid grouped values set to 0. • WAV parser now skips unknown RIFF chunks. • IMA ADPCM: Maximum blocksize is now 4096 bytes (4088 samples stereo, 8184 mono). Thus, now also plays 44100Hz stereo. • Rt-midi: starts if in reset GPIO0=’0’, GPIO1=’1’, GPIO2&3 give earSpeaker setup. • NewSinTest() and NewSinSweep() added (AIADDR = 0x4020/0x4022) AICTRL0 and AICTRL1 set sin frequency for left/right. • Clears memory before SPI boot and not in InitHardware(). Version 0.5, 2007-12-03 77 VLSI VS1053b preliminary y Solution 12 VS1053 B 12. DOCUMENT VERSION CHANGES Document Version Changes This chapter describes the most important changes to this document. Version 0.5 for VS1053b, 2007-12-03 • Ogg Vorbis recording now documented in Chapter 9.7. • Added stereo ADPCM recording and an example to Chapter 9.8. • Added WAV PCM header example to Chapter 9.6. • Simplified LQFP-48 Typical Connection Diagram (Chapter 6) by removing one of the images. Version 0.4 for VS1053b, 2007-09-06 • First published version. • Completely rewritten Chapter 9.5: Operation / Play and Decode. New designs should be based on this new version. • Updated Image 3: Typical Connection Diagram Using LQFP-48. • Rename SM OUTOFWAV SM CANCEL. • Many minor changes like typo corrections. Version 0.3 for VS1053a, 2007-01-16 • I2S pins are now in GPIO4-GPIO7, they no longer overlap with GPIO0 and GPIO1. • Extra parameter structure updated (version 3), location changed to X:0x1e00. Version 0.2 for VS1053a, 2006-10-03 • Updated codec and feature lists. • Mention of quiet power-off added to feature list. • EarSpeaker documentation added. • ADPCM encoding updated. Version 0.5, 2007-12-03 78 VLSI Solution VS1053b preliminary y 13 VS1053 B 13. CONTACT INFORMATION Contact Information VLSI Solution Oy Entrance G, 2nd floor Hermiankatu 8 FIN-33720 Tampere FINLAND Fax: +358-3-3140-8288 Phone: +358-3-3140-8200 Email: [email protected] URL: http://www.vlsi.fi/ Version 0.5, 2007-12-03 79