AN115 APPLICATION NOTE CS4923/4/5/6/7/8/9 HARDWARE USER'S GUIDE Contents Description l Host The CS4923/4/5/6/7/8/9 is a system on a chip solution for multi-channel audio decompression and digital signal processing. Because the device is RAM-based, a download of application software is required each time the CS4923/4/5/6/7/8/9 is powered up. This document focuses on hardware control of the chip from a functional perspective. Communication (Serial and Parallel Host Communication) l Boot Procedures for Host Boot and Autoboot l Resetting the CS492x l Connecting External Memory (Using Paged and Non-Paged Memory) l Understanding Configuration Messages l Input & Output Hardware Configuration l Pseudocode examples for SPI and I2C Communication with the CS492x l Pseudocode Outlining a Typical Download Session with the CS492X l Pseudocode Outlining a Typical Reset Sequence with the CS492X This document takes more of a functional approach to the hardware of the chip. An indepth description of communication, boot procedure, external memory and the hardware configuration are given in this document. This document will be valuable to both the hardware designer and the system programmer. DIGITAL SOUND Input TM Input Configuration C R Y S T A L MPEG Transport DMX (R) Host COMM & Boot PROCESSING Host Controller DIR S/PDIF ADC Analog Input DAC Multi-Channel Analog Output CS492x EXT Memory Output Configuration 64K x 8 ROM P.O. Box 17847, Austin, Texas 78760 (512) 445 7222 FAX: (512) 445 7581 http://www.cirrus.com Copyright Cirrus Logic, Inc. 1999 (All Rights Reserved) AUG `99 AN115REV2 1 TABLE OF CONTENTS 1. OVERVIEW ............................................................................................................................... 5 1.1 Multi-Channel Decoder Family of Parts ............................................................................. 5 1.2 Document Strategy ............................................................................................................ 6 1.2.1 Hardware Documentation ..................................................................................... 6 1.2.2 CS4923/4/5/6/7/8/9 Application Code User's Guides ........................................... 6 1.3 Using the CS4923/4/5/6/7/8/9 ............................................................................................ 7 2. HOST COMMUNICATION ........................................................................................................ 8 2.1 Serial Communication ........................................................................................................ 8 2.1.1 SPI Communication .............................................................................................. 8 2.1.1.1 Writing in SPI ........................................................................................ 9 2.1.1.2 Reading in SPI ...................................................................................... 9 2.1.2 I2C Communication ............................................................................................. 12 2.1.2.1 Writing in I2C ....................................................................................... 12 2.1.2.2 Reading in I2C ..................................................................................... 13 2.1.3 INTREQ Behavior: A Special Case ..................................................................... 14 2.2 Parallel Host Communication ........................................................................................... 17 2.2.1 Intel Parallel Host Communication Mode ............................................................ 18 2.2.1.1 Writing a Byte in Intel Mode ................................................................ 19 2.2.1.2 Reading a Byte in Intel Mode .............................................................. 19 2.2.2 Motorola Parallel Host Communication Mode ..................................................... 20 2.2.2.1 Writing a Byte in Motorola Mode ......................................................... 20 2.2.2.2 Reading a Byte in Motorola Mode ....................................................... 21 2.2.3 Procedures for Parallel Host Mode Communication ........................................... 22 2.2.3.1 Control Write in a Parallel Host Mode ................................................. 22 2.2.3.2 Control Read in a Parallel Host Mode ................................................. 23 3. BOOT PROCEDURE & RESET ............................................................................................. 25 3.1 Host Boot ......................................................................................................................... 25 3.2 Autoboot .......................................................................................................................... 28 3.2.1 Autoboot INTREQ Behavior ................................................................................ 31 3.3 Application Failure Boot Message ................................................................................... 32 3.4 Resetting the CS4923/4/5/6/7/8/9 .................................................................................... 32 4. EXTERNAL MEMORY ............................................................................................................ 34 4.1 Basic Memory Architecture .............................................................................................. 35 4.2 Non-Paged Memory ......................................................................................................... 36 4.3 Paged Memory ................................................................................................................. 36 4.4 Examples ......................................................................................................................... 38 4.4.1 Non-Paged Memory ............................................................................................ 38 4.4.2 32 Kilobyte Paged Autoboot Memory .................................................................. 39 Contacting Cirrus Logic Support For a complete listing of Direct Sales, Distributor, and Sales Representative contacts, visit the Cirrus Logic web site at: http://www.cirrus.com/corporate/contacts/ Preliminary product information describes products which are in production, but for which full characterization data is not yet available. Advance product information describes products which are in development and subject to development changes. Cirrus Logic, Inc. has made best efforts to ensure that the information contained in this document is accurate and reliable. However, the information is subject to change without notice and is provided "AS IS" without warranty of any kind (express or implied). No responsibility is assumed by Cirrus Logic, Inc. for the use of this information, nor for infringements of patents or other rights of third parties. This document is the property of Cirrus Logic, Inc. and implies no license under patents, copyrights, trademarks, or trade secrets. No part of this publication may be copied, reproduced, stored in a retrieval system, or transmitted, in any form or by any means (electronic, mechanical, photographic, or otherwise) without the prior written consent of Cirrus Logic, Inc. Items from any Cirrus Logic website or disk may be printed for use by the user. However, no part of the printout or electronic files may be copied, reproduced, stored in a retrieval system, or transmitted, in any form or by any means (electronic, mechanical, photographic, or otherwise) without the prior written consent of Cirrus Logic, Inc.Furthermore, no part of this publication may be used as a basis for manufacture or sale of any items without the prior written consent of Cirrus Logic, Inc. The names of products of Cirrus Logic, Inc. or other vendors and suppliers appearing in this document may be trademarks or service marks of their respective owners which may be registered in some jurisdictions. A list of Cirrus Logic, Inc. trademarks and service marks can be found at http://www.cirrus.com. 2 AN115REV2 4.4.3 64 Kilobyte Paged Autoboot Memory ................................................................. 40 4.4.4 64 Kilobyte Paged DTS & Autoboot Memory ...................................................... 41 4.5 CRD4923-MEM ............................................................................................................... 42 5. .HARDWARE CONFIGURATION .......................................................................................... 46 5.1 Address Checking ........................................................................................................... 47 5.2 Input and Output .............................................................................................................. 47 5.2.1 Digital Audio Formats .......................................................................................... 47 5.2.2 Digital Input and Output Ports ............................................................................. 49 5.2.3 Parallel Delivery of Data ..................................................................................... 50 5.2.3.1 PCM Data Write in Parallel Host Mode ............................................... 50 5.2.3.2 Compressed Data Write in Parallel Host Mode .................................. 51 5.2.3.2.1 MFB Bit Example ....................................................................... 51 5.2.3.2.2 CMPREQ Example .................................................................... 52 5.3 Configuration Messages .................................................................................................. 52 5.3.1 Address Checking ............................................................................................... 52 5.3.2 Input ................................................................................................................... 53 5.3.2.1 Special Considerations ....................................................................... 53 5.3.3 Output ................................................................................................................. 55 5.3.3.1 Special Considerations ....................................................................... 55 5.3.4 Creating Hardware Configuration Messages ...................................................... 56 6. APPENDIX A - PSEUDOCODE FOR THE CS4923/4/5/6/7/8/9 FAMILY .............................. 58 6.1 SPI Pseudocode .............................................................................................................. 58 6.1.1 SPI Write Operation ............................................................................................ 58 6.1.2 SPI Read Operation ............................................................................................ 59 6.2 I2C Pseudocode .............................................................................................................. 60 6.2.1 I2C Write Operation ............................................................................................. 60 6.2.2 I2C Read Operation ............................................................................................ 62 6.3 Typical Download Session with the CS4923/4/5/6/7/8/9 ................................................. 64 6.4 Typical Reset Sequence for the CS4923/4/5/6/7/8/9 ...................................................... 65 LIST OF FIGURES Figure 1. SPI Write Flow Diagram................................................................................................... 9 Figure 2. SPI Read Flow Diagram ................................................................................................ 10 Figure 3. SPI Timing ..................................................................................................................... 11 Figure 4. I2C Write Flow Diagram ................................................................................................. 12 Figure 5. I2C Read Flow Diagram ................................................................................................. 13 Figure 6. I2C Timing ...................................................................................................................... 15 Figure 7. Intel Mode, One-Byte Write Flow Diagram .................................................................... 19 Figure 8. Intel Mode, One-Byte Read Flow Diagram .................................................................... 20 Figure 9. Motorola Mode, One-Byte Write Flow Diagram ............................................................. 21 Figure 10. Motorola Mode, One-Byte Read Flow Diagram ........................................................... 21 Figure 11. Typical Parallel Host Mode Control Write Sequence Flow Diagram............................ 22 Figure 12. Typical Parallel Host Mode Control Read Sequence Flow Diagram............................ 23 Figure 13. Typical Serial Boot and Download Procedure ............................................................. 26 Figure 14. Typical Parallel Boot and Download Procedure........................................................... 27 Figure 15. Autoboot Memory Architecture .................................................................................... 28 Figure 16. Autoboot Timing Diagram ............................................................................................ 29 Figure 17. Autoboot Sequence ..................................................................................................... 30 Figure 18. Autoboot INTREQ Behavior......................................................................................... 31 Figure 19. Performing a Reset ...................................................................................................... 32 Figure 21. Autoboot Timing Diagram ............................................................................................ 35 Figure 22. Run-Time Memory Access........................................................................................... 35 Figure 20. Basic Memory Architecture .......................................................................................... 35 AN115REV2 3 Figure 23. External Memory with 64 Kilobyte Pages..................................................................... 37 Figure 24. External Memory With 32 Kbyte Pages ....................................................................... 37 Figure 25. Non-Paged Memory ..................................................................................................... 39 Figure 26. 32 Kbyte Paged memory.............................................................................................. 39 Figure 27. Autoboot Sequence for 32 Kbyte Paged Memory........................................................ 40 Figure 28. 64 Kbyte Paged Autoboot Memory .............................................................................. 40 Figure 29. Autoboot for 64 Kbyte paged Memory ......................................................................... 41 Figure 30. 64 Kbyte Paged DTS/Autoboot Memory ...................................................................... 42 Figure 31. Autoboot Sequence for DTS System using Symmetrical 64 Kilobyte Pages............... 43 Figure 32. CRD4923-MEM............................................................................................................ 45 Figure 33. Memory Map for CRD4923 Daughter Board................................................................ 46 Figure 34. DTS Autoboot Flow Diagrams...................................................................................... 46 Figure 35. I2S Format ................................................................................................................... 48 Figure 36. Left Justified Format..................................................................................................... 48 Figure 37. Multi-Channel Format (M == 20) .................................................................................. 48 Figure 38. PCM Data Write Sequence in Parallel Host Mode Flow Diagram................................ 51 Figure 39. MFB Bit Status Polling Flow Diagram .......................................................................... 52 Figure 40. CMPREQ Pin Status Polling Flow Diagram ................................................................. 52 LIST OF TABLES Table 1. Serial Host Mode Configurations....................................................................................... 8 Table 2. SPI Communication Signals.............................................................................................. 8 Table 3. I2C Communication Signals ........................................................................................... 12 Table 4. Parallel Host Mode Configurations.................................................................................. 17 Table 5. Intel Mode Communication Signals................................................................................. 18 Table 6. Motorola Mode Communication Signals.......................................................................... 20 Table 7. Boot Write Messages ...................................................................................................... 25 Table 8. Boot Read Messages ...................................................................................................... 25 Table 9. Memory Interface Pins .................................................................................................... 34 Table 10. Memory and Control Requirements for the CS4923/4/5/6/7/8/9 Family........................ 34 Table 11. ROM Speeds................................................................................................................. 36 Table 12. External Memory Configurations ................................................................................... 38 Table 13. DAI - Digital Audio Input Port ........................................................................................ 49 Table 14. CDI - Compressed Digital Input Port ............................................................................. 49 Table 15. DAO - Digital Audio Output Port.................................................................................... 49 Table 16. Output Channel Mapping .............................................................................................. 49 Table 17. Input Data Type Configuration ...................................................................................... 53 Table 18. Input Data Format Configuration ................................................................................... 54 Table 19. Input SCLK Polarity Configuration................................................................................. 54 Table 20. FIFO Setup Configuration ............................................................................................. 54 Table 21. Output Clock Configuration ........................................................................................... 55 Table 22. Output Data Format Configuration ................................................................................ 55 Table 23. Output MCLK Configuration .......................................................................................... 56 Table 24. Output SCLK Configuration........................................................................................... 56 Table 25. Output SCLK Polarity Configuration.............................................................................. 56 Table 26. Example Values to be Sent to CS492X After Download or Soft Reset ......................... 57 4 AN115REV2 1. OVERVIEW The CS4923/4/5/6/7/8/9 is a family of system on a chip solutions for multi-channel audio decompression and digital signal processing. Since the part is RAM-based, a download of application software is required each time the CS4923/4/5/6/7/8/9 is powered up. These parts are generally targeted at two different market segments. The broadcast market where audio/video (A/V) synchronization is required, and the outboard decoder markets where audio/video synchronization is not required. The important differentiation is the format in which the data will be received by the CS4923/4/5/6/7/8/9. In systems where A/V synchronization is required from the CS4923/4/5/6/7/8/9, the incoming data is typically PES encoded. In an outboard decoder application the data typically comes in the IEC61937 format (as specified by the DVD consortium). An important point to remember is that the CS4923/4/5/6/7/8/9 will support both environments, but different downloads are required depending on the input data type. Broadcast applications include (but are not limited to) set top box applications, DVDs and digital TVs. Outboard decoder applications include standalone decoders and audio/video receivers. Often times a system may be a hybrid between an outboard decoder and a broadcast system depending on its functionality. As discussed above, compressed audio can be packed in IEC61937, PES, or elementary formats depending on the decoder environment. Each format is supported by a separate download of application code. Consult the relevant Application Code User's Guide to determine which formats are supported by a particular application. A brief description of each format is presented below. Elementary - an elementary bitstream consists only of compressed audio data (e.g., strictly the Dolby AN115REV2 Digital bitstream); used primarily in broadcast environments. PES - a Packetized Elementary Stream (PES) bitstream contains the elementary compressed audio stream and additional header information which can be used for A/V synchronization; used primarily in broadcast environments. IEC61937 - a method of packing compressed audio such that it can be delivered using a bi-phase encoded signal (e.g., S/PDIF output signal from DVD player); used primarily for outboard decoders where A/V synchronization is not required. 1.1 Multi-Channel Decoder Family of Parts CS4923 - Dolby DigitalTM Audio Decoder. The CS4923 is the original member of the family and is intended to be used if only Dolby Digital decoding is required. For Dolby Digital, post processing includes bass management, delays and Dolby Pro Logic decoding. Separate downloads can also be used to support stereo to 5.1 channel effects processing and stereo MPEG decoding. CS4924 - Dolby DigitalTM Source Product Decoder. The CS4924 is the stereo version of the CS4923 designed for source products such as DVD, HDTV, and set-top boxes. Separate downloads are available for stereo decode of Dolby Digital and MPEG audio. CS4925 - International Multi-Channel DVD Audio Decoder. The CS4925 supports both Dolby Digital and MPEG-2 multi-channel formats. For both Dolby Digital and MPEG-2 multi-channel, post processing includes bass management and Dolby Pro Logic decoding. Separate downloads are available for decode of Dolby Digital and MPEG audio. Another code load can be used to support stereo to 5.1 channel effects processing. CS4926 - DTS/Dolby(R) Multi-Channel Audio Decoder. The CS4926 supports both Dolby Digital and DTS, or Digital Theater Surround. For Dolby Digital, post processing includes bass management 5 and Dolby Pro Logic. The Dolby Digital code and DTS code take separate code downloads. Separate downloads can also be used to support stereo to 5.1 channel effects processing and stereo MPEG decoding. CS4927 - MPEG-2 Multi-Channel Decoder. The CS4927 supports MPEG-2 multi-channel decoding and should be used in applications where Dolby Digital decoding is not necessary. For MPEG-2 multi-channel decoding, post processing includes bass management and Dolby Pro Logic decoding. Another code load can be used to support stereo to 5.1 channel effects processing. CS4928 - DTS Multi-Channel Decoder. The CS4928 supports DTS multi-channel decoding and should be used in applications where Dolby Digital decoding is not necessary. For DTS multi-channel decoding, post processing includes bass management. Separate downloads can also be used to support stereo to 5.1 channel effects processing and stereo MPEG decoding. CS4929 - AAC 2-Channel, (Low Complexity) and MPEG-2 Stereo Decoder. The CS4929 is capable of decoding both 2-channel AAC and MPEG-2 audio. The CS4929 supports elementary and PES formats. 1.2 Document Strategy Multiple documents are needed to fully define, understand and implement the functionality of the CS4923/4/5/6/7/8/9. They can be split up into two basic groups: hardware and application code documentation. It should be noted that hardware and application code are co-dependent and one can not successfully use the part without an understanding of both. The `ANXXX' notation denotes the application note number under which the respective user's guide was released. 1.2.1 Hardware Documentation CS4923/4/5/6/7/8/9 Family Data Sheet - This document describes the electrical characteristics of 6 the device from timing to base functionality. This is the hardware designers tool to learn the part's electrical and systems requirements. AN115 - CS4923/4/5/6/7/8/9 Hardware User's Guide - describes the functional aspects of the device. An in depth description of communication, boot procedure, external memory and hardware configuration are given in this document. This document will be valuable to both the hardware designer and the system programmer. 1.2.2 CS4923/4/5/6/7/8/9 Application Code User's Guides The following application notes describe the application codes used with the CS4923/4/5/6/7/8/9. Whenever an application code user's guide is referred to, it should be assumed that one or more of the below documents are being referenced. The following list covers currently released application notes. This list will grow with each new application released. For a current list of released user's guides please see www.crystal.com and search for the part number. AN120 - Dolby Digital User's Guide for the CS4923/4/5/6. This document covers the features available in the Dolby Digital code including delays, pink noise, bass management, Pro Logic, PCM pass through and Dolby Digital processing features. Optional appendices are available that document code for Dolby Virtual, Q-Surround and VMAx. AN121 - MPEG User's Guide for the CS4925. This document covers the features available in the MPEG Multi-Channel code including delays, bass management, Pro Logic, and MPEG processing features. AN122 - DTS User's Guide for the CS4926, CS4928. This document covers the features available in the DTS code including bass management and DTS processing features. AN115REV2 AN123 - Surround User's Guide for the CS4923/4/5/6/7/8. This code covers the different Stereo PCM to surround effects processing code. Optional appendices are available that document Crystal Original Surround, Circle Surround and Logic 7. AN140 - Broadcast Systems Guide for the CS4923/4/5/6/7/8/9. This guide describes all application code (e.g. Dolby Digital, MPEG, AAC) designed for broadcast systems such as HDTV and set-top box receivers. This document also provides a discussion of broadcast system considerations and dependencies such as A/V synchronization and channel change procedures. 1.3 Using the CS4923/4/5/6/7/8/9 No matter what application is being used on the chip, the following four steps are always followed to use the CS4923/4/5/6/7/8/9 in system. AN115REV2 1) Reset and/or Download Code - Detailed information in AN115 2) Hardware Configuration - Detailed information in AN115 3) Application configuration Detailed information in the appropriate Application Code User's guide 4) Kickstart - This is the "Go" command to the CS492X once the system is properly configured. Information can be found in the appropriate Application Code User's guide. For this document, CS4923/4/5/6/7/8/9 has been replaced in certain places with CS492X for readability. Unless otherwise specified CS492X should be interpreted as applying to the CS4923, CS4924, CS4925 and CS4926. 7 2. HOST COMMUNICATION The host communication port of the CS4923/4/5/6/7/8/9 is used for downloading application code to the DSP and it is used for communicating with the DSP during run-time. The CS492X supports two parallel host communication modes (Intel mode and Motorola mode) and two serial host communication modes (I2C and SPI). Please note that when a parallel host communication mode has been selected, the external memory interface cannot be used. This constraint has two significant implications: * Autoboot cannot be used in a system using parallel host communication * Parallel host communication modes cannot be used when processing DTS (CS4926 or CS4928) Each of the host communication modes supported by the CS492X family will be discussed in subsequent sections. The following information will be provided for each mode: * How to configure the CS492X for each host communication mode * Which pins of the CS492X must be used * The protocol used for writing to the CS492X * The protocol used for reading from the CS492X 2.1 Serial Communication The CS4923/4/5/6/7/8/9 has a serial control port that supports both SPI and I2C forms of communication. The mode of communication is determined by the states of the RD (pin 5) and WR (pin 4) pins at the rising edge of RESET (pin 36). Table 1 below shows the two possible mode configurations: Other modes are not supported at this time and should not be used. If the mode pins are driven dynamically by the host, then set up (trstsu) and hold 8 RD 0 WR 1 1 0 MODE I2C SPI Table 1. Serial Host Mode Configurations (trsthld) times must be satisfied around the rising edge of reset as specified in the RESET Switching characteristics portion of the CS492X Family Datasheet. The following sections will explain each communication mode in more detail. Flow diagrams will illustrate read and write cycles. Pseudocode is presented in "Appendix A Pseudocode For The CS4923/4/5/6/7/8/9 Family" 58 to demonstrate communication with the chip from a programming perspective. Timing diagrams will be shown to demonstrate relative edge positions of signal transitions for read and write operations. Only the subsection describing the communication mode being used needs to be read by the system designer. 2.1.1 SPI Communication SPI communication with the CS4923/4/5/6/7/8/9 is accomplished with 5 communication lines: chip select, serial control clock, serial data in, serial data out and an interrupt request line to signal that the DSP has data to transmit to the host. Table 2 shows the mnemonic, pin name and pin number of each of these signals on the CS4923/4/5/6/7/8/9. Mnemonic Chip Select Serial Clock Serial Data In Serial Data Out Interrupt Request Pin Name CS SCCLK SCDIN SCDOUT INTREQ Pin Number 18 7 6 19 20 Table 2. SPI Communication Signals AN115REV2 2.1.1.1 Writing in SPI When writing to the device in SPI the same protocol will be used whether writing a byte, a message or even an entire executable download image. The examples shown in this document can be expanded to fit any write situation. Figure 1 shows a typical write sequence: 4) When all of the bytes have been transferred, chip select should be raised to signify an end of write. Once again it is crucial that the serial clock transitions from high to low on the last bit of the last byte before chip select is raised, or a loss of data will occur. SPI START: CS (LOW) WRITE ADDRESS BYTE WITH MODE BIT SET TO 0 FOR WRITE The pseudocode in Section 6.1.1 "SPI Write Operation" -- page 58 demonstrates a write operation for the SPI mode of communication. SEND DATABYTE Y MORE DATA? N CS (HIGH) Figure 1. SPI Write Flow Diagram The following is a detailed description of an SPI write sequence with the CS492X. 1) An SPI transfer is initiated when chip select (CS) is driven low. 2) This is followed by a 7-bit address and the read/write bit set low for a write. The address for the CS492X defaults to 0000000b. It is necessary to clock this address in prior to any transfer in order for the CS492X to accept the write. In other words a byte of 0x00 should be clocked into the device preceding any write. The 0x00 byte represents the 7 bit address 0000000b, and the least significant bit set to 0 to designate a write. AN115REV2 3) The host should then clock data into the device most significant bit first, one byte at a time. The data byte is transferred to the DSP on the falling edge of the eighth serial clock. For this reason, the serial clock should be default low so that eight transitions from low to high to low will occur for each byte. The same write routine could be used to send a single byte, message or an entire application code image. From a hardware perspective, it makes no difference whether communication is by byte or multiple bytes of any length as long as the correct hardware protocol is followed. 2.1.1.2 Reading in SPI A read operation is necessary when the CS4923/4/5/6/7/8/9 signals that it has data to be read. The CS492X does this by dropping its interrupt request line (INTREQ) low. When reading from the device in SPI, the same protocol will be used whether reading a single byte or multiple bytes. The examples shown in this document can be expanded to fit any read situation. Figure 2 shows a typical read sequence: The following is a detailed description of an SPI read sequence with the CS492X. 1) An SPI read transaction is initiated by the CS492X dropping INTREQ, signaling that it has data to be read. 2) The host responds by driving chip select (CS) low. 9 NO INTREQ LOW? 5) If INTREQ is still low, another byte should be clocked out of the CS492X. Please see the discussion below for a complete description of INTREQ behavior. YES CS (LOW) 6) When INTREQ has risen, the chip select line of the CS492X should be raised to end the read transaction. WRITE ADDRESS BYTE WITH MODE BIT SET TO 1 FOR READ READ DATA BYTE YES INTREQ STILL LOW? NO CS (HIGH) Figure 2. SPI Read Flow Diagram 3) This is followed by a 7-bit address and the read/write bit set high for a read. The address for the CS492X defaults to 0000000b. It is necessary to clock this address in prior to any transfer in order for the CS492X to acknowledge the read. In other words a byte of 0x01 should be clocked into the device preceding any read. The 0x01 byte represents the 7 bit address 0000000b, and the least significant bit set to 1 to designate a read. 4) After the falling edge of the serial control clock (SCCLK) for the read/write bit, the data is ready to be clocked out on the control data out pin (CDOUT). Data clocked out by the host is valid on the rising edge of SCCLK and data 10 transitions occur on the falling edge of SCCLK. The serial clock should be default low so that eight transitions from low to high to low will occur for each byte. Understanding the role of INTREQ is important for successful communication. INTREQ is guaranteed to remain low (once it has gone low) until the second to last rising edge of SCCLK of the last byte to be transferred out of the CS492X. If there is no more data to be transferred, INTREQ will go high at this point. For SPI this is the rising edge for the second to last bit of the last byte to be transferred. After going high, INTREQ is guaranteed to stay high until the next rising edge of SCCLK. This end of transfer condition signals the host to end the read transaction by clocking the last data bit out and raising CS. If INTREQ is still low after the second to last rising edge of SCCLK, the host should continue reading data from the serial control port. It should be noted that all data should be read out of the serial control port during one cycle or a loss of data will occur. In other words, all data should be read out of the chip until INTREQ signals the last byte by going high as described above. Please see Section 2.1.3 "INTREQ Behavior: A Special Case" -- page 10 for a more detailed description of INTREQ behavior. The pseudocode in Section 6.1.2 "SPI Read Operation" -- page 59 demonstrates a read operation for the SPI mode of communication. The Figure 3 timing diagram shows the relative edges of the control lines for an SPI read and write. AN115REV2 AN115REV2 SCCLK SCDIN AD6 AD5 AD4 AD3 AD2 AD1 AD0 R/W D7 D6 D5 D4 D3 D2 D1 D0 D7 D6 D5 D4 D3 D2 D1 D0 D7 D6 D5 D4 D3 D2 D1 D0 D3 D2 D1 D0 D7 D6 D5 D4 D3 D2 D1 D0 CS SPI W rite Functional Tim ing SCCLK SCDIN SCDOUT AD6 AD5 AD4 AD3 AD2 AD1 AD0 R/W D7 D6 D5 D4 D3 D2 D1 D0 D7 D6 D5 D4 CS INTREQ SPI Read Functional Tim ing Note 1 Notes: 1. INTREQ is guaranteed to stay LOW until the rising edge of SCCLK for bit D1 of the last byte to be transferred out of the CS4923/4/5/6/7/8/9. 2. INTREQ is guaranteed to remain HIGH until the next rising edge of SCCLK at which point it may go LOW again if there is new data to be read. The condition of INTREQ going LOW at this point should be treated as a new read condition. After a stop condition, a new start condition and an address byte should be sent Figure 3. SPI Timing Note 2 11 2.1.2 I2C Communication I2C communication with the CS4923/4/5/6/7/8/9 is accomplished with 3 communication lines: serial control clock, a bi-directional serial data input/output line and an interrupt request line to signal that the DSP has data to transmit to the host. Table 3 shows the mnemonic, pin name and pin number of each of these signals on the CS492X. Mnemonic Serial Clock Bi-Directional Data Interrupt Request Pin Name SCCLK SCDIO INTREQ Pin Number 7 19 20 Table 3. I2C Communication Signals Typically in I2C communication SCDIO is an open drain line with a pull-up. A logic one is placed on the line by tri-stating the output and allowing the pull-up to raise the line. At this point another device can drive the line low if necessary. Tristating SCDIO can have two effects: 1. To send out a one when writing data or sending a "no acknowledge"; 2. release the line when another chip is writing data. For our pseudocode examples, driving SCDIO high effectively tri-states this signal since it is open drain and SCDIO (HIGH) should be interpreted as such. 2.1.2.1 Writing in I2C When writing to the device in I2C the same protocol will be used whether writing a byte, a message or even an application code image. The examples shown in this document can be expanded to fit any write situation. Figure 4 shows a typical write sequence: The following is a detailed description of an I2C write sequence with the CS492X. 1) An I2C transfer is initiated with an I2C start condition which is defined as the data (SCDIO) line falling while the clock (SCCLK) is held high. 12 SEND I2C START: DROP SCDIO LOW WHILE SCCLK IS HIGH WRITE ADDRESS BYTE WITH MODE BIT SET TO 0 FOR WRITE GET ACK SEND DATABYTE GET ACK MORE DATA? Y N I2C STOP: RAISE SCDIO HIGH WHILE SCCLK IS HIGH Figure 4. I2C Write Flow Diagram 2) Next a 7-bit address with the read/write bit set low for a write should be sent to the CS492X. The address for the CS492X defaults to 0000000b. It is necessary to clock this address in prior to any transfer in order for the CS492X to accept the write. In other words a byte of 0x00 should be clocked into the device preceding any write. The 0x00 byte represents the 7 bit of address (0000000b) and the read/write bit set to 0 to designate a write. AN115REV2 3) After each byte (including the address and each data byte) the host must release the data line and provide a ninth clock for the CS492X to acknowledge. The CS492X will drive the data line low during the ninth clock to acknowledge. If for some reason the CS492X does not acknowledge, it means that the last byte sent was not received and should be resent. If the resent byte fails to produce an acknowledge, a stop condition should be sent and the device should be reset. 4) The host should then clock data into the device most significant bit first, one byte at a time. The CS492X will (and must) acknowledge each byte that it receives which means that after each byte the host must provide an acknowledge clock pulse on SCCLK and release the data line, SCDIO. 5) At the end of a data transfer a stop condition must be sent. The stop condition is defined as the rising edge of SCDIO while SCCLK is high. The pseudocode in Section 6.2.1 "I2C Write Operation" -- page 60 demonstrates a write operation for the I2C mode of communication. 2.1.2.2 Reading in I2C A read operation is necessary when the CS4923/4/5/6/7/8/9 signals that it has data to be read. It does this by dropping its interrupt request line (INTREQ) low. When reading from the device in I2C, the same protocol will be used whether reading a single byte or multiple bytes. The examples shown in this document can be expanded to fit any read situation. Figure 5 shows a typical I2C read sequence 1) An I2C read transaction is initiated by the CS492X dropping INTREQ, signaling that it has data to be read. AN115REV2 NO INTREQ LOW? YES SEND I2C START: DROP SCDIO LOW WHILE SCCLK IS HIGH WRITE ADDRESS BYTE WITH MODE BIT SET TO 1 FOR READ GET ACK READ DATABYTE YES INTREQ STILL LOW? SEND ACK NO SEND NACK SEND I2C STOP: RISING EDGE OF SCDIO WHILE SCLK IS HIGH Figure 5. I2C Read Flow Diagram 2) The host responds by sending an I2C start condition which is SCDIO dropping while SCCLK is held high. 13 3) The start condition is followed by a 7-bit address and the read/write bit set high for a read. The address for the CS492X defaults to 0000000b. It is necessary to clock this address in prior to any transfer in order for the CS492X to acknowledge the read. In other words a byte of 0x01 should be clocked into the device preceding any read. The 0x01 byte represents the 7 bit address 0000000b and a read/write bit set to 1 to designate a read. 4) After the falling edge of the serial control clock (SCCLK) for the read/write bit of the address byte, an acknowledge must be read in by the host. The CS492X will drive SCDIO low to acknowledge the address byte and to indicate that it is ready for a read operation. If an acknowledge is not sent by the CS492X, a stop condition should be issued and the read sequence should be restarted. 5) The data is ready to be clocked out on the SCDIO line at this point. Data clocked out by the host is valid on the rising edge of SCCLK and data transitions occur on the falling edge of SCCLK. 6) If INTREQ is still low after a byte transfer, an acknowledge (SCDIO clocked low by SCCLK) must be sent by the host to the CS492X and another byte should be clocked out of the CS492X. Please see the discussion below for a complete description of INTREQ's behavior. 7) When INTREQ has risen, a no acknowledge should be sent by the host (SCDIO clocked high by the host) to the CS492X. This, followed by an I2C stop condition (SCDIO raised, while SCCLK is high) signals an end of read to the CS492X. Understanding the role of INTREQ is important for successful communication. INTREQ is guaranteed to remain low (once it has gone low), until the rising edge of SCCLK for the last bit of the last byte to be transferred out of the CS492X (i.e. the rising 14 edge of SCCLK before the ACK SCCLK). If there is no more data to be transferred, INTREQ will go high at this point. After going high, INTREQ is guaranteed to stay high until the next rising edge of SCCLK (i.e. it will stay high until the rising edge of SCCLK for the ACK/NACK bit). This end of transfer condition signals the host to end the read transaction by clocking the last data bit out of the CS492X and then sending a no acknowledge to the CS492X to signal that the read sequence is over. At this point the host should send an I2C stop condition to complete the read sequence. If INTREQ is still low after the rising edge of SCCLK on the last data bit of the current byte, the host should send an acknowledge and continue reading data from the serial control port. It should be noted that all data should be read out of the serial control port during one cycle or a loss of data will occur. In other words, all data should be read out of the chip until INTREQ signals the last byte by going high as described above. Please see Section 2.1.3 "INTREQ Behavior: A Special Case" -- page 10 for a more detailed description of INTREQ behavior. The pseudocode in Section 6.2.2 "I2C Read Operation" -- page 62 demonstrates a read operation for the I2C mode of communication. The timing diagram in Figure 6 shows the relative edges of the control lines for an I2C read and write. 2.1.3 INTREQ Behavior: A Special Case When communicating with the CS4923/4/5/6/7/8/9 there are two types of messages which force INTREQ to go low. These messages are known as solicited messages and unsolicited messages. For more information on the specific types of messages that require a read from the host, one of the application code user's guides should be referenced. In general, when communicating with the CS492X, INTREQ will not go low unless the host first sends AN115REV2 AN115REV2 I 2 C Start I 2 C Stop SCCLK SCDIO AD6 AD5 AD4 AD3 AD2 AD1 AD0 R/W ACK D7 D6 D5 D4 D3 D2 D1 D0 ACK D7 D6 D5 D4 D3 D2 D1 D0 ACK D7 D6 D5 D4 D3 D2 D1 D0 ACK I 2 C Write Functional Timing I 2 C Stop I 2 C Start SCCLK SCDIO AD6 AD5 AD4 AD3 AD2 AD1 AD0 R/W ACK D7 D6 D5 D4 D3 D2 D1 D0 ACK D7 D6 D5 D4 D3 D2 D1 D0 ACK D7 D6 D5 D4 D3 D2 D1 D0 NACK INTREQ Note 1 Note 2 Note 5 Note 3 Note 4 I 2 C Read Functional Timing Notes: 1. The ACK for the address byte is driven by the CS4923/4/5/6/7/8/9. 2. The ACKs for the data bytes being read from the CS4923/4/5/6/7/8/9 should be driven by the host. 3. INTREQ is guaranteed to stay LOW until the rising edge of SCCLK for bit D0 of the last byte to be transferred out of the CS4923/4/5/6/7/8/9. 4. A NACK should be sent by the host after the last byte to indicate the end of the read cycle. 5. INTREQ is guaranteed to stay HIGH until the next rising edge of SCCLK (for the ACK/NACK bit) at which point it may go LOW again if there is new data to be read. The condition of INTREQ going LOW at this point should be treated as a new read condition. After a stop condition, a new start condition followed by an address byte should be sent. Figure 6. I2C Timing 15 a read request command message. In other words the host must solicit a response from the DSP. In this environment, the host must read from the CS492X until INTREQ goes high again. Once the INTREQ pin has gone high it will not be driven low until the host sends another read request. When unsolicited messages, such as those used for Autodetect, have been enabled, the behavior of INTREQ is noticeably different. The CS492X will drop the INTREQ pin whenever the DSP has an outgoing message, even though the host may not have requested data. There are three ways in which INTREQ can be affected by an unsolicited message: 1) During normal operation, while INTREQ is high, the DSP could drop INTREQ to indicate an outgoing message, without a prior read request. 2) The host is in the process of reading from the CS492X, meaning that INTREQ is already low. An unsolicited message arrives which forces INTREQ to remain low after the solicited message is read. 3) The host is reading from the CS492X when the unsolicited message is queued, but INTREQ goes high for one period of SCCLK and then goes low again before the end of the read cycle. In case (1) the host should perform a read operation as discussed in the previous sections. In case (2) an unsolicited message arrives before the second to last SCCLK of the final byte transfer of a read, forcing the INTREQ pin to remain low. In this scenario the host should continue to read from the CS492X without a stop/start condition or data will be lost. In case (3) an unsolicited message arrives between the second to last SCCLK and the last SCCLK of the final byte transfer of a read. In this scenario, INTREQ will transition high for one clock (as if the read transaction has ended), and then back low (indicating that more data has queued). This final 16 case is the most complicated and shall be explained in detail. There are two constraints which completely characterize the behavior of the INTREQ pin during a read. The first constraint is that the INTREQ pin is guaranteed to remain low until the second to last SCCLK (SCCLK number N-1) of the final byte being transferred from the CS492X (not necessarily the second to last bit of the data byte). The second constraint is that once the INTREQ pin has gone high it is guaranteed to remain high until the rising edge of the last SCCLK (SCCLK number N) of the final byte being transferred from the CS492X (not necessarily the last bit of the data byte). If an unsolicited message arrives in the window of time between the rising edge of the second to last SCCLK and the final SCCLK, INTREQ will drop low on the rising edge of the final SCCLK as illustrated in the functional timing diagrams shown for I2C and SPI read cycles. INTREQ behavior for I2C communication is illustrated in figure 6. When using I2C communication the INTREQ pin will remain low until the rising edge of SCCLK for the data bit D0 (SCCLK N-1), but it can go low at the rising edge of SCCLK for the NACK bit (SCCLK N) if an unsolicited message has arrived. If no unsolicited messages arrive, the INTREQ pin will remain high after rising. INTREQ behavior for SPI communication is illustrated in figure 3. When using SPI communication, the INTREQ pin will remain low until the rising edge of SCCLK for the data bit D1 (SCCLK N-1), but it can go low at the rising edge of SCCLK for data bit D0 (SCCLK N) if an unsolicited message has arrived. If no unsolicited messages arrive, the INTREQ pin will remain high after rising. Ideally, the host will sample INTREQ on the falling edge of SCCLK number N-1 of the final byte of each read response message. If INTREQ is AN115REV2 sampled high, the host should conclude the current read cycle using the stop condition defined for the communication mode chosen. The host should then begin a new read cycle complete with the appropriate start condition and the chip address. If INTREQ is sampled low, the host should continue reading the next message from the CS492X without ending the current read cycle. When using automated communication ports, however, the host is often limited to sampling the status of INTREQ after an entire byte has been transferred. In this situation a low-high-low transition (case 3) would be missed and the host will see a constantly low INTREQ pin. Since the host should read from the CS492X until it detects that INTREQ has gone high, this condition will be treated as a multiple-message read (more than one read response is provided by the CS492X). Under these conditions a single byte of 0x00 will be read out before the unsolicited message. The length of every read response is defined in the user's manual for each piece of application code. Thus, the host should know how many bytes to expect based on the first byte (the OPCODE) of a read response message. It is guaranteed that no read responses will begin with 0x00, which means that a NULL byte (0x00) detected in the OPCODE position of a read response message should be discarded. Please see an Application Code User's Guide for an explanation of the OPCODE. must be shifted out of the CS492X before the first byte of the unsolicited message can be read. In other words, if a system can only sample INTREQ after an entire byte transfer the following routine should be used if INTREQ is low after the last byte of the message being read: 1) Read one byte 2) If the byte == 0x00 discard it and skip to step 3. If the byte != 0x00 then it is the OPCODE for the next message. For this case skip to step 4. 3) Read one more byte. This is the OPCODE for the next message. 4) Read the rest of the message as indicated in the previous sections. 2.2 The parallel host communication modes of the CS4923/4/5/6/7/8/9 provide an 8-bit interface to the DSP. An Intel-style parallel mode and a Motorola-style parallel mode are supported. The mode of communication is determined by the states of the RD (pin 5), WR (pin 4), and PSEL (pin 19) pins at the rising edge of RESET (pin 36). Each time the CS492X is reset, the RD, WR, and PSEL pins are sampled to determine how the host interface port will be configured. Table 4 shows the necessary pin configurations for selecting a parallel configuration mode. RD 1 1 It is important that the host be aware of the presence of NULL bytes, or the communication channel could become corrupted. When case (3) occurs and the host issues a stop condition before starting a new read cycle, the first byte of the unsolicited message is loaded directly into the shift register and 0x00 is never seen. Alternatively, if case (3) occurs and the host continues to read from the CS492X without a stop condition (a multiple message read), the 0x00 byte AN115REV2 Parallel Host Communication WR 1 1 PSEL 0 1 MODE Intel Mode Motorola Mode Table 4. Parallel Host Mode Configurations The host interface is implemented using four communication registers within the CS492X: * The Host Message register (A[1:0]==00b): receives incoming control data bytes and provides outgoing response data bytes. 17 * * * The Host Control register (A[1:0]=01b): provides information about the state of the communication interface. The PCM Data Input register (A[1:0]=10b): accepts bytes of linear PCM audio data (WRITE ONLY). The Compressed Data Input register (A[1:0]=11b): accepts bytes of compressed audio data (WRITE ONLY). When the host is downloading code to the CS492X or configuring the application code, control messages will be written to (and read from) the Host Message register. The Host Control register is used during messaging sessions to determine when the CS492X can accept another byte of control data, and when the CS492X has an outgoing byte that may be read. The PCM Data and Compressed Data registers are used strictly for the transfer of audio data. The host cannot read from these two registers. Audio data written to registers 11b and 10b are transferred directly to the internal FIFOs of the CS492X. When the level of the PCM FIFO reaches the FIFO threshold level, the MFC bit of the Host Control register will be set. When the level of the Compressed Data FIFO reaches the FIFO threshold level, the MFB bit of the Host Control register will be set. It is important to remember that the parallel host interface requires the DATA[7:0] pins of the CS492X. The external memory interface also requires the DATA[7:0] pins. This conflict results in the following constraint: * Parallel host communication modes cannot be used when processing DTS (CS4926 and CS4928) Systems that require DTS capability and systems utilizing the autoboot capabilities of the CS492X must use a serial host communication protocol. A detailed description for each parallel host mode will now be given. The following information will be provided for the Intel mode and Motorola mode: 18 * The pins of the CS492X which must be used for proper communication * Flow diagram and description for a parallel byte write * Flow diagram and description for a parallel byte read The four registers of the CS492X's parallel host mode are not used identically. The algorithm used for communicating with each register will be given as a functional description, building upon the basic read and write protocols defined in the Motorola and Intel sections. The following will be covered: * Flow diagram and description for a control write * Flow diagram and description for a control read 2.2.1 Intel Parallel Host Communication Mode The Intel parallel host communication mode is implemented using the pins given in Table 5. Mnemonic Chip Select Write Enable Output Enable Register Address Bit 1 Register Address Bit 0 Interrupt Request DATA7 DATA6 DATA5 DATA4 DATA3 DATA2 DATA1 DATA0 Pin Name CS WR RD A1 A0 INTREQ DATA7 DATA6 DATA5 DATA4 DATA3 DATA2 DATA1 DATA0 Pin Number 18 4 5 6 7 19 8 9 10 11 14 15 16 17 Table 5. Intel Mode Communication Signals The INTREQ pin is controlled by the application code when a parallel host communication mode has been selected. When the code supports INTREQ notification, the INTREQ pin is asserted whenever the DSP has an outgoing message for the host. This AN115REV2 same information is reflected by the HOUTRDY bit of the Host Control Register (A[1:0] = 01b). INTREQ is useful for informing the host of unsolicited messages. An unsolicited message is defined as a message generated by the DSP without an associated host read request. Unsolicited messages can be used to notify the host of conditions such as a change in the incoming audio data type (e.g. PCM --> AC-3). 2.2.1.1 Writing a Byte in Intel Mode Information provided in this section is intended as a functional description of how to write control information to the CS492X. The system designer must insure that all of the timing constraints of the Intel Parallel Host Mode Write Cycle are met. The timing specifications for the Intel Parallel Host Mode can be found in the CS4923/4/5/6/7/8/9 Family Datasheet. The flow diagram shown in Figure 7 illustrates the sequence of events that define a one-byte write in Intel mode. ADDRESS A PARALLEL I/O REGISTER (A[1:0] SET APPROPRIATELY CS (LOW) WR (LOW) Host Message: A[1:0]==00b. Host Control: A[1:0]==01b. PCMDATA: A[1:0]==10b. CMPDATA: A[1:0]==11b. 2) The host then indicates that the selected register will be written. The host initiates a write cycle by driving the CS and WR pins low. 3) The host drives the data byte to the DATA[7:0] pins of the CS492X. 4) Once the setup time for the write has been met, the host ends the write cycle by driving the CS and WR pins high. 2.2.1.2 Reading a Byte in Intel Mode Information provided in this section is intended as a functional description of how to write control information to the CS492X. The system designer must insure that all of the timing constraints of the Intel Parallel Host Mode Read Cycle are met. The timing specifications for the Intel Parallel Host Mode can be found in the CS4923/4/5/6/7/8/9 Family Datasheet The flow diagram shown in Figure 8 illustrates the sequence of events that define a one-byte read in Intel mode. WRITE BYTE TO DATA [7:0] The protocol presented in Figure 8 will now be described in detail. CS (HIGH) WR (HIGH) 1) The host must first drive the A1 and A0 register address pins of the CS492X with the address of the desired Parallel I/O Register. Note that only the Host Message register and the Host Control register can be read. Figure 7. Intel Mode, One-Byte Write Flow Diagram The protocol presented in Figure 7 will now be described in detail. AN115REV2 1) The host must first drive the A1 and A0 register address pins of the CS492X with the address of the desired Parallel I/O Register. Host Message: A[1:0]==00b. Host Control: A[1:0]==01b. 19 ADDRESS A PARALLEL I/O REGISTER (A[1:0] SET APPROPRIATELY CS (LOW) INTREQ is useful for informing the host of unsolicited messages. An unsolicited message is defined as a message generated by the DSP without an associated host read request. Unsolicited messages can be used to notify the host of conditions such as a change in the incoming audio data type (e.g. PCM --> AC-3). RD (LOW) READ BYTE FROM DATA [7:0] CS (HIGH) RD (HIGH) Figure 8. Intel Mode, One-Byte Read Flow Diagram 2) The host now indicates that the selected register will be read. The host initiates a read cycle by driving the CS and RD pins low. 3) Once the data is valid, the host can read the value of the selected register from the DATA[7:0] pins of the CS492X. 4) The host should now terminate the read cycle by driving the CS and RD pins high. 2.2.2 Motorola Parallel Host Communication Mode The Motorola parallel host communication mode is implemented using the pins given in Table 6.The INTREQ pin is controlled by the application code when a parallel host communication mode has been selected. When the code supports INTREQ notification, the INTREQ pin is asserted whenever the DSP has an outgoing message for the host. This same information is reflected by the HOUTRDY bit of the Host Control Register (A[1:0] = 01b). 20 Mnemonic Chip Select Data Strobe Read or Write Select Register Address Bit 1 Register Address Bit 0 Interrupt Request DATA7 DATA6 DATA5 DATA4 DATA3 DATA2 DATA1 DATA0 Pin Name CS DS R/W A1 A0 INTREQ DATA7 DATA6 DATA5 DATA4 DATA3 DATA2 DATA1 DATA0 Pin Number 18 4 5 6 7 19 8 9 10 11 14 15 16 17 Table 6. Motorola Mode Communication Signals 2.2.2.1 Writing a Byte in Motorola Mode Information provided in this section is intended as a functional description of how to write control information to the CS492X. The system designer must insure that all of the timing constraints of the Motorola Parallel Host Mode Write Cycle are met. The timing specifications for the Motorola Parallel Host Mode can be found in the CS4923/4/5/6/7/8/9 Family Datasheet. The flow diagram shown in Figure 9 illustrates the sequence of events that define a one-byte write in Motorola mode. The protocol presented in figure 9 will now be described in detail. AN115REV2 R/W (LOW) ADDRESS A PARALLEL I/O REGISTER (A[1:0] SET APPROPRIATELY R/W (HIGH) ADDRESS A PARALLEL I/O REGISTER (A[1:0] SET APPROPRIATELY CS (LOW) DS (LOW) CS (LOW) DS (LOW) WRITE BYTE TO DATA [7:0] READ BYTE FROM DATA [7:0] CS (HIGH) DS (HIGH) CS (HIGH) DS (HIGH) Figure 9. Motorola Mode, One-Byte Write Flow Diagram 1) The host must drive the A1 and A0 register address pins of the CS492X with the address of the address of the desired Parallel I/O Register. Host Message: A[1:0]==00b. Host Control: A[1:0]==01b. PCMDATA: A[1:0]==10b. CMPDATA: A[1:0]==11b. The host indicates that this is a write cycle by driving the R/W pin low. 2) The host initiates a write cycle by driving the CS and DS pins low. 3) The host drives the data byte to the DATA[7:0] pins of the CS492X. Figure 10. Motorola Mode, One-Byte Read Flow Diagram The protocol presented Figure 10 will now be described in detail. 1) The host must drive the A1 and A0 register address pins of the CS492X with the address of the desired Parallel I/O Register. Note that only the Host Message register and the Host Control register can be read. Host Message: A[1:0]==00b. Host Control: A[1:0]==01b. The host indicates that this is a read cycle by driving the R/W pin high. 2) The host initiates the read cycle by driving the CS and DS pins low. 4) Once the setup time for the write has been met, the host ends the write cycle by driving the CS and DS pins high. 3) Once the data is valid, the host can read the value of the selected register from the DATA[7:0] pins of the CS492X. 2.2.2.2 4) The host should now terminate the read cycle by driving the CS and DS pins high. Reading a Byte in Motorola Mode The flow diagram shown in Figure 10 illustrates the sequence of events that define a one-byte read in Motorola mode. AN115REV2 21 2.2.3 2.2.3.1 Procedures for Parallel Host Mode Communication Control Write in a Parallel Host Mode When writing control data to the CS492X, the same protocol is used whether the host is writing a control message or an entire executable download image. Messages sent to the CS492X should be written most significant byte first. Likewise, downloads of the application code should also be performed most significant byte first. The example shown in this section can be generalized to fit any control write situation. The generic function `Read_Byte_*()' is used in the following example as a generalized reference to either Read_Byte_MOT() or Read_Byte_INT(), and `Write_Byte_*()' is a generic reference to Write_Byte_MOT() or Write_Byte_INT(). Figure 11 shows a typical write sequence. The protocol presented in figure 11 will now be described in detail. 1) When the host is communicating with the CS492X, the host must verify that the DSP is ready to accept a new control byte. If the DSP is in the midst of an interrupt service routine, it will be unable to retrieve control data from the Host Message Register. Please note that `Read_Byte_*()' and `Write_Byte_*()' are generic references to either the Intel or Motorola communication protocol. If the most recent control byte has not yet been read by the DSP, the host must not write a new byte. 2) In order to determine whether the CS492X is ready to accept a new control byte the host must check the HINBSY bit of the Host Control 22 READ_*(HOST CONTROL REGISTER) YES HINSBY==1 NO WRITE_*(HOST MESSAGE REGISTER) YES MORE BYTES TO WRITE? NO FINISHED Figure 11. Typical Parallel Host Mode Control Write Sequence Flow Diagram Register (bit 2). If HINBSY is high, then the DSP is not prepared to accept a new control byte, and the host should poll the Host Control Register again. If HINBSY is low, then the host may write a control byte into the Host Message Register. 3) The host knows that the DSP is ready for a new control byte at this point and should write the control byte to the Host Message Register (A[1:0] = 00b). 4) If the host would like to write any more control bytes to the CS492X, the host should once again poll the Host Control Register (return to step 1). AN115REV2 2.2.3.2 Control Read in a Parallel Host Mode When reading control data from the CS4923/4/5/6/7/8/9, the same protocol is used whether the host is reading a single byte or a 6 byte message. INTREQ = 0 YES READ_*(HOST CONTROL REGISTER) During the boot procedure, a handshaking protocol is used by the CS492X. This handshake consists of a 3 byte write to the CS492X followed by a 1 byte response from the DSP. The host must read the response byte and act accordingly. The boot procedure is discussed in Section 3.1 "Host Boot" - page 25. During regular operation (at run-time), the responses from the CS492X will always be 6 bytes in length. The example shown in this section can be used for any control read situation. The generic function `Read_Byte_*()' is used in the following example as a generalized reference to either Read_Byte_MOT() or Read_Byte_INT(). Figure 12 shows a typical read sequence. NO HOUTRDY==1 YES READ_*(HOST MESSAGE REGISTER) YES MORE BYTES TO READ? NO WAIT 100 uS The protocol presented in Figure 12 will now be described in detail. 1) Optionally, INTREQ going low may be used as an interrupt to the host to indicate that the CS492X has an outgoing message. Even with the use of INTREQ, HOUTRDY must be checked to insure that bytes are ready for the host during the read process. Please note that INTREQ does not go low to indicate an outgoing message during boot. 2) The host reads the Host Control Register (A[1:0] = 01b) in order to determine the state of the communication interface. Please note that `Read_Byte_*()' is a generalized reference to either Read_Byte_MOT() or Read_Byte_INT(). AN115REV2 READ_*(HOST CONTROL REGISTER) YES HOUTRDY==1 NO FINISHED Figure 12. Typical Parallel Host Mode Control Read Sequence Flow Diagram 23 3) In order to determine whether the CS492X has an outgoing control byte that is valid, the host must check the HOUTRDY bit of the Host Control Register (bit 1). If HOUTRDY is high, then the Host Message Register contains a valid message byte for the host. If HOUTRDY is low, then the DSP has not placed a new control byte in the Host Message Register, and the host should poll the Host Control Register again. 4) The host knows that the DSP is ready to provide a new response byte at this point. The host can safely read a byte from the Host Message Register (A[1:0] = 00b). 5) If the host expects to read any more response 24 bytes, the host should once again check the HOUTRDY bit (return to step 1). Please refer to one of the application code user's guides to determine the length of messages to read from the CS492X. Typically this length is 1, 3 or 6 bytes, and can be deduced from the message OPCODE. 6) After the response has been read the host should wait at least 100 uS and check HOUTRDY one final time. If HOUTRDY is high once again this means that an unsolicited message has come during the read process and the host has another message to read (i.e. skip back to step 4 and read out the new message). AN115REV2 3. BOOT PROCEDURE & RESET In this section the process of booting and downloading to the CS492X will be covered as well as how to perform a soft reset. Both host boot and autoboot are covered in this section. 3.1 Host Boot A flow diagram of a typical serial download sequence and a typical parallel download sequence will be presented, as well as pseudocode representing a download sequence from the programmers perspective. The pseudocode is written in a general sense where function calls are made to Write_* and Read_*. The * can be replaced by I2C, SPI, INTEL, or MOTO depending on the mode of host communication. For each case the general download algorithm is the same. The download and boot procedure is accomplished with RESET (pin 36), and the communication pins discussed in Section 2.1 "Serial Communication" - page 8. The flow diagrams in Figures 13 and 14 illustrate a typical boot and download procedure. Table 7 defines the boot write messages and Table 8 defines the boot read messages in mnemonic and actual hex value. These messages will be used in the boot sequence. The following is a detailed description of a download sequence for the CS492X. All writes and reads with the CS492X should follow the protocol given in Section 2 "Host Communication" -- page 8, and timing given in the CS492X Datasheet. NOTE: When reading from the chip in a serial communication mode, the host must wait for the interrupt request (INTREQ) to fall before starting the read cycle. 1) A download sequence is started when the host issues a hard reset and holds the mode pins appropriately (WR, RD, and PSEL) as described in Section 2 "Host Communication" -- page 8 and in the CS492X Datasheet. It is assumed that timing is satisfied as per the CS492X Datasheet. AN115REV2 2) The host should then send the boot message DOWNLOAD_BOOT (0x000004). This causes the CS492X to initialize itself for download. 3) If the initialization was successful the CS492X sends out the boot message BOOT_START (0x01) and the host should proceed to step 5. MNEMONIC SOFT_RESET RESERVED RESERVED DOWNLOAD_BOOT BOOT_SUCCESS_RECEIVED VALUE 0x000001 0x000002 0x000003 0x000004 0x000005 Table 7. Boot Write Messages MNEMONIC BOOT_START BOOT_SUCCESS APPLICATION_FAILURE BOOT_ERROR INVALID_MSG BOOT_ERROR INIT_FAILURE INIT_FAILURE BAD_CHECKSUM VALUE 0x01 0x02 0xF0 0xFA 0xFB 0xFC 0xFD 0xFE 0xFF Table 8. Boot Read Messages 4) If initialization fails, the CS492X sends out an INIT_FAILURE boot message byte (0xFD or 0xFE), INVALID_MSG byte (0xFB), or BOOT_ERROR byte (0xFA or 0xFC) and spins waiting for a hard reset. The host should re-try steps 1 through 3 and if failure is met again, the serial communication timing and protocol should be inspected. 5) After receiving the BOOT_START byte, the host should write the downloadable image (from the .LD file). 6) The end of the .LD file contains a three byte checksum. If the checksum is good after download, the CS492X will send a BOOT_SUCCESS message (0x02) to the host. 25 RESET(LOW) (NOTE 1) RESET(HIGH) (NOTE 2) WAIT 500 NS WRITE_*(DOWNLOAD_ BOOT, MSG_SIZE) INTREQ LOW? N TIMEOUT AFTER 20MS (NOTE 3) Y READ_*(MESSAGE) N MESSAGE == BOOTSTART? EXIT(ERROR) Y WRITE_*(.LD FILE, DOWNLOAD FILE SIZE) INTREQ LOW? N TIMEOUT AFTER 20MS (NOTE 3) Y READ_*(MESSAGE) MESSAGE == BOOT_SUCCESS? Notes: 1. RESET must be held LOW for at least 100 ns to satisfy trstl 2. It should be noted that mode pins are used to configure the CS492X serial communication mode. These mode pins are latched internally on the rising edge of reset. The pins can be set dynamically by a microprocessor or can be statically pulled HIGH or LOW. If these pins are driven dynamically, setup and hold times must be satisfied as stated in the CS492X datasheet. More information about the function of the mode pins can be found in the CS492X datasheet and in Section 2 "Host Communication" -- page 8. 3. Time-out values reflect worst case response time for the CS492X. The values shown may be used for the host's time-out control loop. 4. Hardware configuration messages are covered in Section 5 ".Hardware Configuration" -- page 46. Application configuration messages are covered in each application code user's manual. N EXIT(ERROR) Y WRITE_*(BOOT_ SUCCESS_RECEIVED, MSG-SIZE) WAIT 5 MS WRITE_*(CONFIGURATION_ MESSAGES, CONFIG_ MSG_SIZE) (NOTE 4) DOWNLOAD COMPLETE Figure 13. Typical Serial Boot and Download Procedure 26 AN115REV2 RESET(LOW) (NOTE 1) RESET(HIGH) (NOTE 2) WAIT 500 NS CONTROL_WRITE_*(DOWNLOAD_ BOOT, MSG_SIZE) HOUTRDY HIGH? N TIMEOUT AFTER 20MS (NOTE 3) Y READ_*(MESSAGE) MESSAGE == BOOTSTART? N EXIT(ERROR) Y CONTROL_WRITE_*(.LD FILE, DOWNLOAD FILE SIZE) HOUTRDY HIGH? N TIMEOUT AFTER 20MS (NOTE 3) Y READ_*(MESSAGE) MESSAGE == BOOT_SUCCESS? Notes: 1. RESET must be held LOW for at least 100 ns to satisfy trstl 2. It should be noted that mode pins are used to configure a CS492X parallel communication mode. These mode pins are latched internally on the rising edge of reset. The pins can be set dynamically by a microprocessor or can be statically pulled HIGH or LOW. If these pins are driven dynamically, setup and hold times must be satisfied as stated in the CS492X Datasheet. More information about the function of the mode pins can be found in the CS492X Datasheet and in Section 2 "Host Communication" -- page 8. 3. Time-out values reflect worst case response time for the CS492X. The values shown may be used for the host's time-out control loop. 4. Hardware configuration messages are covered in Section 5 ".Hardware Configuration" -- page 46. Application configuration messages are covered in each application code user's manual. N EXIT(ERROR) Y CONTROL_WRITE_*(BOOT_ SUCCESS_RECEIVED, MSG-SIZE) WAIT 5 MS CONTROL_WRITE_*(CONFIGURATION_ MESSAGES, CONFIG_MSG_SIZE) (NOTE 4) DOWNLOAD COMPLETE Figure 14. Typical Parallel Boot and Download Procedure AN115REV2 27 3.3V 8 EMAD[7:0] D 3.3V 3.3V CS4923/4/5/6/7/8/9 HOST uCONTROLLER Q 8 BIT '574 DFF ADDR[7:0] D Q 8 BIT '574 DFF ADDR[15:8] ADDR[7:0] EMOE OE ABOOT EXTMEM CS RESET EMWR 3.3V 3.3V ABOOT is an open-drain pin which requires the pull-up shown. ABOOT and INTREQ are multiplexed onto the same pin. Therefore the host should drive ABOOT with an open-drain driver. R1 R3 R2 R4 D A T A [7:0] ADDR[15:8] 64K X 8 ROM Only one of R1 and R2 should be stuffed. Only one of R3 and R4 should be stuffed. The state of EMOE and EMWR at the rising edge of RESET will determine the serial mode that the part comes up in while using external memory. Please see the Serial Communication section for more details. Figure 15. Autoboot Memory Architecture If the checksum was bad, the CS492X responds with the BAD_CHECKSUM message byte (0xFF) and spins, waiting for hard reset. 7) After reading out the BOOT_SUCCESS byte, the host should send the BOOT_SUCCESS_RECEIVED message (0x000005) which will cause an internal application code reset and allow the downloaded application to run. 8) After waiting 5ms to allow the downloaded application to initialize, the host can send configuration messages for both hardware and software configuration. The hardware configuration messages are described in Section 5 ".Hardware Configuration" -- page 46. For more information about software application messages please refer to the Application Code User's Guide for the code(s) that are being used. The pseudocode in Section 6.3 "Typical Download Session with the CS4923/4/5/6/7/8/9" -- page 64 demonstrates a typical serial host download session with the CS492X. 28 3.2 Autoboot Autoboot is a feature available on all DSPs in the CS492X family which gives the decoder the ability to load application code into itself. Because external memory is accessed with the 8-bit GPIO interface, autoboot restricts host control to serial communication. Figure 15 shows that an autoboot system can be built with a CS492X decoder, two octal latches and an external ROM. In this system RESET and ABOOT are the control pins which are used to initiate autoboot. It is important to be aware that the ABOOT pin also serves as the INTREQ pin for the decoder, which means that it will be driven by the decoder when out of the reset condition. Due to this constraint, ABOOT should be connected to an open-drain output of the microcontroller so as to allow the specified pull-up resistor to generate the high value. At the completion of a successful download INTREQ (ABOOT) becomes an output and the host should no longer drive it. The EMAD[7:0] pins serve as a multiplexed data and address bus. Note that the pins are connected to AN115REV2 RESET ABOOT EXTMEM EMOE EMWR EMAD7:0 MA23:16 MA15:8 MA7:0 Data7:0 Figure 16. Autoboot Timing Diagram both the input of the first latch, and the output of the ROM. The two latches are cascaded such that on each clock pulse a new address byte is latched into the first latch, and the previous ADDR[7:0] byte is latched into the second register becoming ADDR[15:8]. The timing for an autoboot sequence is illustrated in Figure 16. The sequence is initiated by driving RESET low and placing the decoder into a reset state. At the rising edge of RESET the ABOOT, WR, and RD pins are sampled. If ABOOT is low when sampled, and the WR and RD pins are set to configure the device for serial communications, the device will begin to autoboot. Section 2.1 discusses the procedure required for placing the CS492X into a serial communication mode. For a more thorough description of ABOOT's behavior after the rising edge of RESET please see section 3.2.1 The EMOE pin of the CS492X is used for two purposes. It generates clock pulses for the latches, and it is used in conjunction with EXTMEM to enable the outputs of the ROM. The first three rising edges of EMOE are used to latch address bytes, as shown in the diagram. The fourth low pulse of EMOE is used to enable the ROM outputs. When both EXTMEM and EMOE go low, the EMAD[7:0] pins of the DSP become inputs and await the data coming from the ROM. When comparing the memory system in Figure 15 to the timing diagram of Figure 16 there may AN115REV2 appear to be a discrepancy. The timing diagram shows three address cycles, but there are only two latches in the illustration of the memory architecture. This difference is a result of code size limitations. The application code is guaranteed to fit into a 32 Kilobyte space, which means that only 15 address bits will actually be used for retrieving code from the ROM. Thus, the two latches catch the least significant bytes, and the most significant byte is dropped. In autoboot mode, latching the most significant byte would be perfectly valid since the most significant bits are guaranteed to be zeros (the three bytes represent a true 24-bit address). If the external memory is to be used for access during run-time (such as in DTS decode), however, a third latch would break the memory interface. Different external memory interface schemes are discussed in more detail in Section 4. The flow chart given in Figure 17 demonstrates the interaction required by the microcontroller when placing the DSP into autoboot mode. The host must first drive the RESET line low. The host also drives ABOOT low and hold it in a low state until the rising edge of RESET. The low state of ABOOT at the rising edge of RESET initiates autoboot. As noted on the diagram, the host control mode must be configured for serial communications, and the appropriate setup (Trstsu) and hold (Trsthld) times must be observed. 29 RESET(LOW) (NOTE 1) ABOOT(LOW) RESET(HIGH) (NOTE 2) ABOOT(HIGH) WAIT 175 MS (NOTE 3) READ_*(VARIABLE) (NOTE 4) CORRECT VALUE? N WAIT 5 MS Y AUTOBOOT COMPLETE WRITE_*(HW_CONFIG_MSG, HW_MSG_SIZE) (NOTE 4) WRITE_*(SW_CONFIG_MSG, SW_MSG_SIZE) (NOTE 4) WRITE_*(KICKSTART, MSG_SIZE) (NOTE 4) Notes: 1. RESET must be held LOW for at least 100 ns to satisfy the Trstl as specified in the CS4923/4/5/6/7/8/9 Datasheet. 2. The RD and WR pins must be configured to select a serial communication mode as defined in the CS4923/4/5/6/7/8/9 Datasheet. The setup (Trstsu = 50 ns) and hold (Trsthld = 15 ns) times must be observed for the RD, WR, and AUTOBOOT pins. 3. INTREQ should be ignored during this period. 4. The READ_* and WRITE_* functions are placeholders for the READ_I2C/READ_SPI and WRITE_I2C/WRITE_SPI functions defined in the Serial Communication section. Figure 17. Autoboot Sequence 30 AN115REV2 Because ABOOT must be driven by the host with an open-drain pin, the statement ABOOT(HIGH) should be understood to mean ABOOT(RELEASE). The pull-up resistor required on the ABOOT (or INTREQ) line will always be responsible for producing a high value on the pin. ABOOT should never be driven high by the host. After waiting for 175ms, the download should have completed. During the wait period, the host should ignore all INTREQ behavior (mask the INTREQ interrupt). The host can then verify that the code has successfully initialized itself by reading a variable from the application and checking the returned value against the known default value. Any variable can be used for the verification step, but a robust design will select a variable whose value is neither all 0's nor all 1's. If the first read attempt returns an incorrect value, a 5ms wait should be inserted and the read should be repeated. If a second invalid number is read, the entire boot process should be repeated. When the number returned matches the default value for the variable read, the host can be confident that the application is resident in the DSP and awaiting further instruction. Hardware configuration messages are used to define the behavior of the DSP's audio ports. A more detailed description of the different hardware configurations can be found in the Section 5 ".Hardware Configuration" -- page 46. The software configuration messages are specific to each application (AC-3, MPEG, Crystal Original Surround, and DTS). The user's guide for each application provides a list of all pertinent configuration messages. Writing the KICKSTART message to the CS492X begins the audio decode process. The KICKSTART message will also be described in the user's guide for each application. Until the KICKSTART has been sent, the decoder is in a wait state. 3.2.1 Autoboot INTREQ Behavior It is important to note that ABOOT and INTREQ are multiplexed on pin 20 of the CS492X. Because this pin serves as an input before reset, and an output after reset, the host should release the ABOOT line after RESET has gone high. As shown in Figure 18, the host must drive ABOOT low around the rising edge of RESET, while observing the Trstsu and Trsthld timing parameters given in the CS492X Family Data Sheet in order to initiate an autoboot sequence. After the host has released the ABOOT line, it will remain high while the DSP prepares to load code into itself. When the DSP has begun the boot process INTREQ (ABOOT) will be driven low and it will remain low during the entire download procedure. INTREQ should be ignored during download, i.e. interrupts should be masked on the host. The download time will vary according to the size of the download image and the frequency of Driven Low by Host Trstsu Driven Low by CS492X RESET Download in Progress ABOOT Trsthld Figure 18. Autoboot INTREQ Behavior AN115REV2 31 the main DSP clock. At the conclusion of autoboot, the DSP issues an internal reset which will cause INTREQ to rise, indicating that boot has completed. The autoboot sequence is guaranteed to complete in 175ms (from rising edge of RESET to the internal reset of the CS492X). 3.3 Application Failure Boot Message Each piece of application code is specifically tailored for either the CS4923, CS4924, CS4925, CS4926, CS4927, CS4928 or CS4929. Although it is possible to load a piece of code into the wrong chip and receive a BOOT_SUCCESS byte, the code will not initialize itself. In order to facilitate the debug of designs which can accept many members of the CS492X family, an APPLICATION_FAILURE message is provided. As mentioned earlier, the host should wait for at least 5ms after download before sending configuration messages to the CS492X. This provides time for the code to initialize itself. If the INTREQ pin is low 1ms after the download process has completed, the host should read from the CS492X. The byte 0xF0 indicates APPLICATION_FAILURE. This byte informs the host that the application code was loaded into an incompatible DSP. As an example, loading DTS application code into the CS4923 will generate an APPLICATION_FAILURE byte. Although most boot messages are essentially ignored for autoboot, it should be noted that the APPLICATION_FAILURE message is applicable whether serial boot or autoboot is used. 3.4 Resetting the CS4923/4/5/6/7/8/9 Resetting the CS492X uses a combination of software and hardware. To reset the device, a previous application must have been downloaded. The flow diagram in Figure 19 shows the procedure for performing a reset. RESET(LOW) (NOTE 1) RESET(HIGH) (NOTE 2) Notes: 1. RESET must be held LOW for at least 100 ns to satisfy trstl WAIT 500 ns WRITE_* (SOFTRESET, MSG_SIZE) WAIT 5 ms WRITE_* (CONFIGURATION_MESSAGES, CONFIG_MSG_SIZE) (NOTE 3) 2. It should be noted that mode pins are used to configure the CS492X communication mode. These mode pins are latched internally on the rising edge of reset and can be set dynamically by a microprocessor or can be statically pulled HIGH or LOW. If these pins are driven dynamically, setup and hold times must be satisfied as stated in the CS492X Datasheet. More information about the function of the mode pins can be found in the CS492X Datasheet and in Section 2 "Host Communication" -- page 8, 3. Configuration messages determine both hardware and software configuration. Hardware configurations are described in section 5 of this manual. Software application configuration messages are described in the Application Code User's Guide for the code being used. Figure 19. Performing a Reset 32 AN115REV2 The following is a detailed description of a reset sequence to the CS492X. All writes and reads with the CS492X should follow the protocol given in Section 2 "Host Communication" -- page 8, and timing given in the CS492X Datasheet. 1) Reset begins when the host issues a hard reset and holds the mode pins appropriately (WR, RD, and PSEL) as described in Section 2 "Host Communication" -- page 8 and in the CS492X Datasheet. It is assumed that the communication protocol is followed for whichever communication mode is chosen by the host and that timing is satisfied per the CS492X Datasheet. 2) The host should then send the message SOFT_RESET (0x000001). This will restart the previously downloaded application with all AN115REV2 of the hardware configurations in their default states. The Application Code User's Guide for each application lists those parameters which are affected by a SOFT_RESET. 3) After waiting 5 ms to allow the downloaded application to initialize, the host can send configuration messages for both hardware and software configuration. Hardware configuration messages are described in Section 5 ".Hardware Configuration" -- page 46. Software application configuration messages are described in the Application Code User's Guide for the code being used. The pseudocode in Section 6.4 "Typical Reset Sequence for the CS4923/4/5/6/7/8/9" -- page 65 demonstrates a typical reset sequence for the CS492X. 33 4. EXTERNAL MEMORY The CS492X family of DSPs provide a byte-wide interface for accessing external memory. The basic memory interface is implemented with the following pins: EMAD[7:0], EXTMEM, EMOE, and EMWR. Each pin has been described in Table 9. The host communication mode is important when considering the external memory interface. External memory can only be connected to those systems which implement a serial control host interface. Any systems using the parallel host cannot use the external memory interface. The entire family of decoders has the capability of autobooting, as discussed in Section 3.2 External memory is used to hold the application code that the DSP will load into itself. Because the CS4923, CS4924, CS4925 and CS4929 do not require Autoboot capability, external memory is optional. Pin Name EMAD0 EMAD1 EMAD2 EMAD3 EMAD4 EMAD5 EMAD6 EMAD7 The CS4926 and CS4928, however, require the use of external memory. An external ROM must be used for holding the DTS look-up tables employed by the CS4926 and CS4928 during the decode of a DTS bit-stream. Table 10 lists the memory configurations for each decoder. In a simple system, a non-paged memory can be used to hold a single piece of autoboot code or the DTS tables alone (e.g., the CS4928). In more complex systems, a paged memory architecture can be used to hold multiple pieces of application code and the DTS tables (e.g., the CS4926). Both memory architectures, paged and non-paged, will be presented in detail later in this section. Pin Number 17 16 15 14 11 10 9 8 21 Pin Function Multiplexed Address and Data Bit 0 Multiplexed Address and Data Bit 1 Multiplexed Address and Data Bit 2 Multiplexed Address and Data Bit 3 Multiplexed Address and Data Bit 4 Multiplexed Address and Data Bit 5 Multiplexed Address and Data Bit 6 Multiplexed Address and Data Bit 7 External Memory Select EMOE 5 * External Memory Output Enable &Address Latch Strobe EMWR 4 * External Memory Write Strobe EXTMEM * - These pins must be configured appropriately to select a serial host communication mode for the CS4923/4/5/6/7/8/9 at the rising edge of RESET Table 9. Memory Interface Pins Part Number CS4923,4,5,7,9 CS4926,8 Host Control Mode if using External ROM I 2C 2 or SPI I C or SPI Memory Usage OPTIONAL - Autoboot mode REQUIRED - DTS Tables OPTIONAL - Autoboot mode Table 10. Memory and Control Requirements for the CS4923/4/5/6/7/8/9 Family 34 AN115REV2 4.1 Basic Memory Architecture 3.3V The simplest external memory system consists of the DSP's memory interface, the external memory, and two octal latches to hold the memory address. This configuration is a non-paged memory architecture, and a block diagram of this system is shown in Figure 20. Non-paged memories are ideal for autobooting a single piece of application code such as AC-3. Because the application code is guaranteed to fit within a 32 Kilobyte space, it is only necessary to provide 15 address bits. The 16th address bit coming from the DSP may, however, be connected to the memory since the most significant bit is guaranteed to always be 0 during autoboot. Figure 21 shows the functional timing of an autoboot sequence in which three address cycles are illustrated. Due to the code size limitation, however, only the lower 15 bits of this address are relevant. The CS4926 and CS4928 both have special memory requirements since they must access lookup tables while processing DTS encoded audio. These tables will reside in a 64 Kilobyte page of external memory. The memory architecture illustrated in Figure 20 is also applicable for the 8 EMAD[7:0] D 3.3V CS4923/4/5/6/7/8/9 Q 8 BIT '574 DFF ADDR[7:0] D Q 8 BIT '574 DFF ADDR[15:8] ADDR[7:0] EMOE OE EXTMEM CS EMWR 3.3V 3.3V R1 R3 R2 R4 D A T A[7:0] ADDR[15:8] 64K X 8 ROM Only one of R1 and R2 shou ld be stuffed. Only one of R3 and R4 shou ld be stuffed. The state of EMOE and EMWR at the rising edge of RESET will determine the serial mode that the part comes up in while using exter nal memory. Please see section 2, Serial Communication for more d etails. Figure 20. Basic Memory Architecture DTS tables since only 16 bits are necessary for addressing all 64 Kilobytes. The timing requirements for the memory used in a DTS system, however, are different from the timing requirements of the memory used for autoboot only. Accesses made during run-time occur at a higher frequency, and there are only two address cycles as can be seen in Figure 22. Consequently, a DTS system requires a faster ROM. RESET ABOOT EXTMEM EMOE EMWR EMAD7:0 MA23:16 MA15:8 MA7:0 Data7:0 Figure 21. Autoboot Timing Diagram Figure 22. Run-Time Memory Access AN115REV2 35 Table 11 lists the memory speed requirements based on the ROM content. ROM Content AUTOBOOT Code DTS Tables Only AUTOBOOT Code & DTS Tables ROM Speed 330ns 110ns 110ns Table 11. ROM Speeds 4.2 Non-Paged Memory A non-paged memory architecture should be used in systems which will need to access 64 Kilobytes of data or less. One example of such a system would be a decoder designed for AC-3 autoboot only. In this case the only code ever needed would be the application code that would run on the CS4923 or the CS4924. The code is constrained to occupy less than 32 Kilobytes of memory, which means that only 15 bits would be required to access the entire image. Therefore a single-code autoboot system could use the 16-bit addressing scheme shown in Figure 20. Another possible scenario is a CS4926 or CS4928 DTS decoder which will be booted by the host. In this system only a 64 Kilobyte ROM would be necessary for holding the look-up tables. Once again, the basic memory architecture presented in Figure 20 would be adequate. The DSP always considers its address space to range from 0x0000 to 0xFFFF. This means that the decoder is unaware of any data which falls outside of these 64 Kilobytes. In autoboot mode there is yet another constraint - the autoboot process always begins with address 0x0000. The implication is that the host microcontroller must somehow be involved in memory accesses which exceed the 64 Kilobyte scope of the CS492X, and the host must also manage access to all pieces of autoboot code which do not physically reside at location 0x0000. The limitations of a non-paged memory are easily seen, and they can be circumvented using paged memory designs as discussed in the next section. 36 4.3 Paged Memory A paged memory architecture is necessary for those systems which provide autoboot for multiple code loads (e.g. when using Autodetect), or in any CS4926 or CS4928 system which utilizes autoboot. Paged memory is defined as a large memory partitioned into smaller blocks. The easiest partitioning scheme for the CS4923/4/5/6/7/8/9 family is on 64 Kilobyte boundaries because of the sixteen bits of address provided by the DSP. If memory space is at a premium, and the system under design does NOT use the CS4926 or the CS4928, it is possible to perform paging on 32 Kilobyte boundaries. The host microcontroller plays a crucial role in paged memory systems, since it is directly responsible for addressing each page. In a memory composed of 64 Kilobyte pages, the high order address bits of the ROM, A16 and A17, would be controlled directly by the microcontroller as shown in Figure 23. These two address lines give the host the capability to select any of four 64 Kilobyte pages in memory. A memory with 32 Kilobyte pages can be used only to hold autoboot code. Such a memory would be paged with bits A15 and A16. Using this design template, it is vital that the most significant address bit latched from the DSP, A15, not be connected to the memory. The host microcontroller is responsible for controlling address bits A15 and greater. A 32 Kilobyte design is presented in Figure 24. The main variable in organizing the ROM is the page size. The easiest way to determine the page size for your system is to look at the largest block of memory required for any operation. If the external ROM is designed to hold only autoboot code, the designer has the option of using either 32 Kilobyte or 64 Kilobyte pages. The final decision would depend upon the size of the memory which will be used in the final design. AN115REV2 3.3V 8 EMAD[7:0] D 3.3V CS4923/4/5/6/7/8/9 ADDR[7:0] Q 8 BIT '574 DFF D Q 8 BIT '574 DFF ADDR[7:0] ADDR[15:8] uCADDR16** uCADDR17** ADDR16 ADDR17 EMOE OE EXTMEM CS EMWR 3.3V 3.3V R1 R3 R2 R4 D A T A[7:0] ADDR[15:8] 256K X 8 ROM Only one of R1 and R2 shou ld be stuffed. Only one of R3 and R4 shou ld be stuffed. The state of EMOE and EMWR at the rising edge of RESET will determine the serial mode that the part comes up in while using exter nal memory. Please see section 2, Serial Communication for more d etails. **The address lines uCADDR16 and uCADDR17 should come from an external microcontroller and be used to page t he memory for the CS4923/4/5/ 6. Figure 23. External Memory with 64 Kilobyte Pages 3.3V 8 EMAD[7:0] D 3.3V CS4923/4/5/7/9 Q 8 BIT '574 DFF ADDR[7:0] D Q 8 BIT '574 DFF ADDR[14:8]** uCADDR15** uCADDR16** ADDR[7:0] ADDR15 ADDR16 EMOE OE EXTMEM CS EMWR 3.3V 3.3V R1 R3 R2 R4 D A T A[7:0] ADDR[14:8] 128K X 8 ROM Only one of R1 and R2 shou ld be stuffed. Only one of R3 and R4 shou ld be stuffed. The state of EMOE and EMWR at the rising edge of RESET will determine the serial mode that the part comes up in while using exter nal memory. Please see section 2, Serial Communication for more d etails. **The address lines uCAD DR15 and uCADDR16 should come from an external microcontroller and be used to page the memory for the CS4923/4/5/7/9. The high order bit from the second D flip flop should be left as a no connect to allow for an external microcontroller to drive ADDR1 5 of the memory. B ecause the CS4923/4/5/7/9 will not drive ADDR15, only ADDR[14:8] should be connected from the second D flip flop to the memor y. Figure 24. External Memory With 32 Kbyte Pages AN115REV2 37 ROM Content Image Size Number of Pages Required Recommended Page Size CS4923 AC-3 32 Kbytes AC-3,Crystal Original Surround 32 + 32 = 64 Kbytes CS4924 AC-3 32 Kbytes AC-3 with Q-Sound 32 Kbytes CS4925 AC-3, MPEG Multi-Channel, 32 + 32 + 32 = Crystal Original Surround 96 Kbytes CS4926 DTS Tables 64 Kbytes DTS Tables, DTS, AC-3, Crystal 64 + 32 + 32+ 32 = Original Surround 160 Kbytes CS4927 MPEG Multi-Channel 32 Kbytes MPEG Multi-Channel, Crystal 32 + 32 = 64 Kbytes Original Surround CS4928 DTS Tables 64 Kbytes DTS Tables, DTS, Crystal 64 + 32 + 32 = Original Surround 128 Kbytes CS4929 AAC 2-Channel 32 Kbytes AAC 2-Channel, MPEG Stereo 32 + 32 = 64 Kbytes 1 2 Non-Paged 32 Kbytes or 64 Kbytes 1 1 Non-Paged Non-Paged 3 32 Kbytes or 64 Kbytes 1 4 Non-Paged 64 Kbytes 1 2 Non-Paged 32 Kbytes or 64 Kbytes 1 3 Non-Paged 64 Kbytes 1 2 Non-Paged 32 Kbytes or 64 Kbytes Table 12. External Memory Configurations A system using a CS4926 or CS4928, however, will always require a 64 Kilobyte page for the DTS look-up tables. Thus, the most straightforward externally paged ROM for an autobooting CS4926 or CS4928 would have 64 Kilobyte pages. The CS4926 or CS4928 would control the lowest sixteen bits of address, and the host microcontroller would control the most significant address bits. Table 12 lists possible external memory configurations for each DSP. The table provides a list of the ROM content, the size of the combined memory images, the recommended page size, and the number of discrete pages required. The page sizes were calculated using the methodology discussed previously. A more complex memory scheme is given in the CRD4923-MEM example (Figure 32), which demonstrates how to pack the 38 memory more efficiently in a DTS system. The examples also include several figures which present the different ROM configurations as composite memory images. 4.4 4.4.1 Examples Non-Paged Memory The most rudimentary memory design discussed above is the non-paged memory. In a non-paged design, the DSP can only access one item in memory which could be either a single autoboot code load, or the DTS tables for the CS4926 or CS4928. The memory image given in Figure 25 is an example of a non-paged memory image. The memory image shown above would be designed into a system using the memory architecture laid out in Figure 20. All 16 output bits AN115REV2 0x00000 0x00000 0x0FFFF AC-3 Code or DTS Tables AC-3 Code 0x07FFF 0x08000 Figure 25. Non-Paged Memory 0x0FFFF of the address latches would be connected to address bits A0-A15 of the external ROM. The host is completely isolated from memory accesses in this situation. Once the hardware has been designed, the DSP itself will be responsible for all communication with the ROM. 4.4.2 32 Kilobyte Paged Autoboot Memory An external memory architecture which is paged on 32 Kilobyte boundaries will necessarily contain only autoboot code. Therefore, this architecture can only be used by the CS4923/4/5/7/9. Because of the 64 Kilobyte look-up table that must be used during the decode of a DTS stream, the CS4926 and CS4928 could never be used with a fixed 32 Kilobyte page size. Figure 26 shows an example of a 32 Kilobyte paged memory image. The flow diagram given in Figure 27 demonstrates the interaction required by the microcontroller during autoboot. After placing the decoder into a reset state, the host selects the page in memory containing AC-3 Code by driving uC15 to a low state. The host also drives ABOOT low and holds it in a low state until the rising edge of RESET to initiate autoboot. As noted in the autoboot section, the ABOOT pin should be connected to an opendrain output of the microcontroller so as to allow the specified pull-up resistor to generate the high AN115REV2 Crystal Original Surround Code Address line A15 used for paging Figure 26. 32 Kbyte Paged memory value. The open-drain driver is required because the DSP will begin using the pin as an output after a successful download (INTREQ and ABOOT are multiplexed on the same pin). After waiting for 175 ms the download should have completed. During the wait period, the host should ignore all INTREQ behavior (mask the INTREQ interrupt). The host can then verify that the code has successfully initialized itself by reading a variable from the application and checking the returned value against the default value. Any variable can be used for the verification step, but a robust design will select a variable whose value is neither all 0's nor all 1's. If the first read attempt returns an incorrect value, a 5 ms wait should be inserted and the read should be repeated. If a second invalid number is read, the entire boot process should be repeated. When the number returned matches the default value for the variable read, the host knows that the application is resident in the DSP and awaiting further instruction. Please see Section 3.2 "Autoboot" -- page 28 for more information. 39 4.4.3 RESET (LOW) 64 Kilobyte Paged Autoboot Memory Systems using fixed page sizes can implement either 64 Kilobyte or 32 Kilobyte pages when the audio decoder is one of the CS4923/4/5/7/9, but the CS4926 and CS4928 are restricted to 64 Kilobyte pages. The larger page size adds more versatility to the system design because both autoboot code and DTS tables can be included in any external ROM utilizing 64 Kilobyte pages. The next two examples show possible memory designs. The first memory image contains only autoboot code, and the second example shows an external ROM image which holds multiple pieces of autoboot code in addition to the DTS look-up tables. ABOOT (LOW) uC15 (LOW) Address AC-3 Code RESET (HIGH) ABOOT (HIGH) Wait 175 ms Ignore INTREQ READ_*(Variable) N Correct Value? Wait 5 ms Y Autoboot Complete WRITE_*(HW_CONFIG_MSG, HW_MSG_SIZE) WRITE_*(SW_CONFIG_MSG, SW_MSG_SIZE) WRITE_*(KICKSTART, MSG_SIZE) The memory image illustrated in Figure 28 is designed for a system using external ROM for autoboot purposes only. There are only two pages to choose from, which means that the host microcontroller needs to drive one memory address line. In this case, it would be bit A16. The architecture for this memory configuration is shown in Figure 23. All 16 bits of latched address are connected to the memory, and the microcontroller would provide uC16 for paging between the AC-3 Code and Crystal Original Surround. The flow diagram given in Figure 29 demonstrates the interaction required by the microcontroller when placing the DSP into autoboot mode. After placing the decoder into a reset state, the host selects the page in memory containing Crystal Original 0x00000 AC-3 Code The READ_* and WRITE_* functions are placeholders for the READ_I2C/READ_SPI and WRITE_I2C/ WRITE_SPI functions defined in the Serial Communications section. Figure 27. Autoboot Sequence for 32 Kbyte Paged Memory 0x0FFFF 0x10000 0x1FFFF Crystal Original Surround Code Address line A16 used for paging Figure 28. 64 Kbyte Paged Autoboot Memory 40 AN115REV2 Surround Code by driving uC16 to a high state. The host also drives ABOOT low and holds it in a low state until the rising edge of RESET to initiate autoboot. As noted in the autoboot section, the ABOOT pin should be connected to an open-drain output of the microcontroller so as to allow the specified pull-up resistor to generate the high value. The open-drain driver is required because the DSP will begin using the pin as an output after a successful download (INTREQ and ABOOT are multiplexed on the same pin). RESET (LOW) ABOOT (LOW) uC16 (HIGH) Address Crystal Surround Sound RESET (HIGH) ABOOT (HIGH) After waiting for 175 ms the download should have completed. During the wait period, the host should ignore all INTREQ behavior (mask the INTREQ interrupt). The host can then verify that the code has successfully initialized itself by reading a variable from the application and checking the returned value against the default value. Any variable can be used for the verification step, but a robust design will select a variable whose value is neither all 0's nor all 1's. If the first read attempt returns an incorrect value, a 5 ms wait should be inserted and the read should be repeated. If a second invalid number is read, the entire boot process should be repeated. When the number returned matches the default value for the variable read, the host knows that the application is resident in the DSP and awaiting further instruction. Please see Section 3.2 "Autoboot" -- page 28 for more information. 4.4.4 Wait 175 ms Ignore INTREQ READ_*(Variable) NO Correct Value? YES Autoboot Complete WRITE_*(HW_CONFIG_MSG, HW_MSG_SIZE) WRITE_*(SW_CONFIG_MSG, SW_MSG_SIZE) 64 Kilobyte Paged DTS & Autoboot Memory The memory image of Figure 30 provides one example of a ROM configuration for external memory which contains multiple autoboot code loads in addition to the look-up tables required for DTS decode. The four pages in memory are each 64 Kilobytes in size. The symmetry of this design simplifies the necessary address logic when compared to the CRD4923-MEM design discussed later in this section. The host can select any page in AN115REV2 Wait 5 ms WRITE_*(KICKSTART, MSG_SIZE) The READ_* and WRITE_* functions are placeholders for the READ_I2C/READ_SPI and WRITE_I2C/ WRITE_SPI functions defined in the Serial Communication section. Figure 29. Autoboot for 64 Kbyte paged Memory 41 memory by toggling the A16 and A17 address bits of the external memory. Figure 31 shows the sequence of actions required for performing an autoboot in a DTS system (CS4926 or CS4928). Note the similarities to the flow chart for a simple autoboot system. This memory arrangement requires an additional address bit, uC17, and memory must be paged to the DTS Tables before KICKSTARTing the DTS decode application. After placing the decoder into a reset state, the host selects the page in memory containing the DTS Code by driving uC16 to a low state, and uC17 to a high state. The host also drives ABOOT low and holds it in a low state until the rising edge of RESET to initiate autoboot. As noted in the autoboot section, the ABOOT pin should be connected to an open-drain output of the microcontroller so as to allow the specified pull-up resistor to generate the high value. The open-drain driver is required because the DSP will begin using the pin as an output after a successful download (INTREQ and ABOOT are multiplexed on the same pin). 0x00000 AC-3 Code 0x0FFFF 0x10000 0x1FFFF 0x20000 Crystal Original Surround Code DTS, unlike the other applications for the CS492X family, performs run-time accesses to the external memory. In order to ensure that the decoding process begins properly, the host should page to the DTS tables before sending a KICKSTART to the CS4926 or CS4928. The memory image in Figure 30 shows the DTS tables located at address 0x30000, so the host drives both uC16 and uC17 to a high state. After the address has been set, the host can move on to the configuration messages required for the final decoder setup. Please see Section 3.2 "Autoboot" -- page 28 for more information. 4.5 DTS Code 0x2FFFF 0x30000 DTS Tables 0x3FFFF Address lines A16 and A17 used for paging Figure 30. 64 Kbyte Paged DTS/Autoboot Memory 42 After waiting for 175 ms the download should have completed. During the wait period, the host should ignore all INTREQ behavior (mask the INTREQ interrupt). The host can then verify that the code has successfully initialized itself by reading a variable from the application and checking the returned value against the default value. Any variable can be used for the verification step, but a robust design will select a variable whose value is neither all 0's nor all 1's. If the first read attempt returns an incorrect value, a 5 ms wait should be inserted and the read should be repeated. If a second invalid number is read, the entire boot process should be repeated. When the number returned matches the default value for the variable read, the host knows that the application is resident in the DSP and awaiting further instruction. CRD4923-MEM The CRD4923-MEM is an external memory adapter card designed for use with the CRD4923 and CDB4923. The schematic for the CRD4923MEM is shown in Figure 32. In order to reduce the number of microcontroller lines needed for paging the memory and controlling the autoboot sequence, some `glue logic' was added to the board. A consequence of this logic was a hybrid paging scheme. Both 32 Kilobyte and 64 Kilobyte pages are used in the external ROM found on the AN115REV2 expander card. One example CRD4923-MEM memory image can be seen in Figure 33. RESET (LOW) The aforementioned `glue logic' was implemented with four tri-state buffers and the most significant address bit as shown in Figure 32. The high order address bit, uC17, is used to page between the Autoboot Sector and DTS Table Sector, as well as to initiate an autoboot sequence. The tri-state buffers are enabled and disabled according to the state of uC17, making the autoboot function address dependent. ABOOT (LOW) uC17 (HIGH), uC16 (LOW) Address DTS Code RESET (HIGH) ABOOT (HIGH) Wait 175 ms Ignore INTREQ READ_*(Variable) N Correct Value? Wait 5 ms Y Autoboot Complete uC17 (HIGH), uC16 (HIGH) Address DTS Tables WRITE_*(HW_CONFIG_MSG, HW_MSG_SIZE) WRITE_*(SW_CONFIG_MSG, SW_MSG_SIZE) WRITE_*(KICKSTART, MSG_SIZE) The READ_* and WRITE_* functions are placeholders for the READ_I2C/READ_SPI and WRITE_I2C/ WRITE_SPI functions defined in the Serial Communications section. Figure 31. Autoboot Sequence for DTS System using Symmetrical 64 Kilobyte Pages AN115REV2 When uC17 is high, the uC15 buffer and REQ23 buffer outputs are in a high impedance state, the DSP15 buffer is enabled, and the microcontroller is given control of only A16 and A17. This allows the CS4926 (or CS4928) to access all 64 Kbytes (A[15:0]) of the DTS Tables located in the upper half of external memory. In the DTS Sector of the memory there are only two 64 Kilobyte pages which are selected by uC16. An autoboot sequence is activated when uC17 is driven low during reset. The uC15 and REQ23 buffers are enabled, and the REQ23 (ABOOT) line is driven low. Recall that holding ABOOT low during the rising edge of RESET signals the DSP to begin an autoboot sequence. Using uC17 in this fashion frees the microcontroller from dedicating an open-drain output for driving the ABOOT pin, but it also splits the memory into the Autoboot and DTS Sectors shown in Figure 33. The Autoboot Sector has four 32 Kilobyte pages which are selected with A15 and A16. While in the DTS Sector of memory, the DSP can access two different 64 Kilobyte pages with microcontroller address line uC16. Obviously, the most complicated process involving external memory is autobooting the CS4926 or CS4928 into DTS mode. The microcontroller must first load the CS4926 with the DTS application code, and then page to the DTS Tables to ensure proper decode of DTS streams. The flow diagram 43 shown in Figure 34 illustrates the DTS autoboot process (CS4926 and CS4928 systems). Note that the control of the ABOOT line and the uC17 line are synonymous because of the glue logic used on the CRD4923-MEM. After placing the decoder into a reset state, the host selects the page in memory containing the DTS Code by driving uC15 low, uC16 high, and uC17 to a low state. In the act of addressing autoboot code, the host also drives ABOOT low because of the relationship between uC17 and ABOOT. The ABOOT pin is held in a low state until the rising edge of RESET to initiate autoboot. In fact, for the CRD4923-MEM, it is important that ABOOT (i.e. uC17) is not driven high until autoboot has completed. Placing uC17 into a high state would prevent the DSP from accessing the correct code. The tri-state buffer on the ABOOT line acts as an open-drain driver. The open-drain driver is required because the DSP will begin using the pin as an output after a successful download (INTREQ and ABOOT are multiplexed on the same pin). After waiting for 175 ms the download should have completed. During the wait period, the host should ignore all INTREQ behavior (mask the INTREQ interrupt). The host can then verify that the code has successfully initialized itself by reading a variable from the application and checking the 44 returned value against the default value. Any variable can be used for the verification step, but a robust design will select a variable whose value is neither all 0's nor all 1's. If the first read attempt returns an incorrect value, a 5 ms wait should be inserted and the read should be repeated. If a second invalid number is read, the entire boot process should be repeated. When the number returned matches the default value for the variable read, the host knows that the application is resident in the DSP and awaiting further instruction. DTS, unlike the other applications for the CS492X family, performs run-time accesses to the external memory. In order to ensure that the decoding process begins properly, the host should page to the DTS tables before sending a KICKSTART to the CS4926 or CS4928. The memory image in Figure 33 shows the DTS tables located at address 0x20000, so the host drives uC17 to a high state (this was required for releasing ABOOT), and uC16 to a low state. In this state, with uC17 high, the DSP15 buffer has been enabled giving the decoder full access to all 64 Kbytes of the DTS look-up tables. After the address has been set, the host can move on to the configuration messages required for the final decoder setup. Please see Section 3.2 "Autoboot" -- page 28 for more information. AN115REV2 AN115REV2 uC17 U3 uC16 U1 D0 D1 D2 D3 D4 D5 D6 D7 2 3 4 5 6 7 8 9 11 EMOE 1 D0 D1 D2 D3 D4 D5 D6 D7 U2 19 18 17 16 15 14 13 12 Q0 Q1 Q2 Q3 Q4 Q5 Q6 Q7 A0 A1 A2 A3 A4 A5 A6 A7 A0 A1 A2 A3 A4 A5 A6 A7 +3.3V CK EMOE 20 10 VCC OE GND 2 3 4 5 6 7 8 9 11 1 C1 TC74VHC574FW D0 D1 D2 D3 D4 D5 D6 D7 Q0 Q1 Q2 Q3 Q4 Q5 Q6 Q7 CK OE VCC GND A15 A8 A9 A10 A11 A12 A13 A14 19 18 17 16 15 14 13 12 DSP15 20 10 +3.3V TC74VHC574FW .1uF A14 A13 A12 A11 A10 A9 A8 A7 A6 A5 A4 A3 A2 A1 A0 30 2 3 29 28 4 25 23 26 27 5 6 7 8 9 10 11 12 C2 .1uF A17 A16 A15 A14 A13 A12 A11 A10 A09 A08 A7 A6 A5 A4 A3 A2 A1 A0 O0 O1 O2 O3 O4 O5 O6 O7 CE OE PGM VPP VCC GND 13 14 15 17 18 19 20 21 D0 D1 D2 D3 D4 D5 D6 D7 +3.3V EXTMEM 22 24 31 EMOE R13 10k 1 R14 10k 32 16 C3 .1uF AT27LV02 0A-90JC +3.3V 47uF C6 .1uF + C7 A[0:14] +3.3V EXTMEM EMOE D6 D4 D2 D0 2 4 6 8 10 12 14 16 18 20 uC16 REQ23 D7 D5 D3 D1 R10 0 uC17 U5A 1 2 3 3X1HDR 1 P1 1 3 5 7 9 11 13 15 17 19 MRESET23 2 uC15 3 74LVC125 A15 uC16 +3.3V 4 U5B 5 DSP15 +3.3V 6 R9 0 C5 10 U5C U5E .1uF 9 8 74LVC125 uC17 +3.3V U5D 12 13 uC15 +3.3V 11 uC17 0 1 2 3 74LVC125 R E Q 23 7 3X1HDR R8 14 74LVC125 uC15 1 2 3 74LVC125 3X1HDR Figure 32. CRD4923-MEM R1 10k 45 5. .HARDWARE CONFIGURATION RESET (LOW) After download or soft reset, and before kickstarting the application (please see the Audio Manager in the Application Messaging Section of any Application Code User's Guide for more information on kickstarting), the host has the option of changing the default hardware configuration. Hardware configuration messages are used to physically reconfigure the hardware of the audio decoder, as in enabling or disabling address checking for the serial communication port. Hardware configuration messages are also used to initialize the data type (i.e., PCM or compressed) and format (e.g., I2S, left justified, etc.) for digital data inputs, as well as the data format and clocking options for the digital output port. ABOOT (LOW) Equivalent to uC17(LOW) uC17 (LOW), uC16 (HIGH), uC15 (LOW) Address DTS Code RESET (HIGH) Wait 175 ms, Ignore INTREQ ABOOT (HIGH) Equivalent to uC17(HIGH) READ_*(Variable) Autoboot Sector 0x00000 N Correct Value? AC-3 Code 0x07FFF 0x08000 0x0FFFF 0x10000 Wait 5 ms Y Crystal Original Surround Code DTS Autoboot Complete! DTS Code uC17 (HIGH), uC16 (LOW), uC15 (DON'T CARE) Address DTS Tables 0x17FFF 0x18000 MPEG Code 0x1FFFF WRITE_*(HW_CONFIG_MSG, HW_MSG_SIZE) DTS Table Sector 0x20000 DTS Tables WRITE_*(SW_CONFIG_MSG, SW_MSG_SIZE) Unused WRITE_*(KICKSTART, MSG_SIZE) 0x2FFFF 0x30000 0x3FFFF Address lines A15 and A16 page Autoboot Sector Address line A17 selects DTS Tables or Autoboot Figure 33. Memory Map for CRD4923 Daughter Board 46 The READ_* and WRITE_* functions are placeholders for the READ_I2C/READ_SPI and WRITE_I2C/ WRITE_SPI functions defined in the Serial Communications section. Figure 34. DTS Autoboot Flow Diagrams AN115REV2 In general, the hardware configuration can only be changed immediately after download or after soft reset. However, some applications provide the capability to change the input ports without affecting other hardware configurations after sending a special Application Restart message (please see the Audio Manager in any Application Code User's Guide to determine whether the Application Restart message is supported). Section 5.3 at the end of this chapter will describe how to construct a hardware configuration message. 5.1 Address Checking When using one of the serial communication modes, I2C or SPI, as discussed in Section 2.1, it is necessary to send a 7-bit address along with a read/write bit at the start of any serial transaction. By default, address checking is enabled in the CS492X with an address of 0000000b. What this means is that all transactions starting with this address will be accepted by the CS492X communication port and all other communication will be ignored. The address checking portion of the hardware configuration message allows the host to enable or disable address checking as well as assign a unique address to the CS492X. It should be noted that systems with multiple devices on the same bus require special consideration. Since the unique address can not be assigned until after download or reset, every device but one should be held in reset. That single device should then be brought out of reset, downloaded and assigned a unique address. The next device should then be brought out of reset and so on. This will insure that there is no contention on the bus and that the communication integrity is upheld. It should also be noted that performing a Soft Reset, as described in Section 3.4, will cause address checking to be re-enabled and the address will return to its default of 0000000b. 5.2 Input and Output The CS492X has two input ports and one output port. This section will describe the digital audio formats supported by the ports and give a description of the ports themselves. The full capabilities of each port will be presented, although all configurations may not be currently supported by the software. When configuring the input ports, both data format and data type must be considered. Data format is defined as the bit level presentation of the data, such as left justified or I2S. Data type refers to what the bits actually represent, such as AC-3 or PCM. It is the combination of these parameters that fully define the hardware configuration for the input ports. To allow for real-time data type changes, the hardware configuration of the input ports can be changed after a special run-time restart message (please see the Audio Manager in any Application Code User's Guide to determine whether the Application Restart message is supported). All other hardware configurations can change only immediately following download or a soft reset. 5.2.1 Digital Audio Formats This subsection will describe some common audio formats that the CS492X supports. It should be noted that the input ports always use 24-bit PCM resolution and 16-bit compressed data word lengths. The output port of the CS492X provides 20-bit PCM resolution. I2S: Figure 35 shows the I2S format. For I2S, data is presented most significant bit first, one SCLK delay after the transition of LRCLK and is valid on the rising edge of SCLK. For the I2S format, the left subframe is presented when LRCLK is low and the right subframe is presented when LRCLK is high. SCLK is expected to run at a frequency of 48Fs or greater on the input ports. Left Justified: Figure 36 shows the left justified format. Data is presented most significant bit first AN115REV2 47 on the first SCLK after an LRCLK transition and is valid on the rising edge of SCLK. For the left justified format, the left subframe is presented when LRCLK is high and the right subframe is presented when LRCLK is low. SCLK is expected to run at a frequency of 48Fs or greater on the input ports. Multi-Channel: Figure 37 shows the multichannel format. In this format up to 6 channels of audio are presented on one data line with 20 bits per channel. Channels 0, 2, and 4 are presented while the LRCLK is high and channels 1, 3, 5 are presented while the LRCLK is low. Data is valid on the rising edge of SCLK and is presented most significant bit first. LRCK Bursty: Bursty audio delivery is a special format in which only clock and data are used to deliver compressed data to the CS492X (i.e. no frame clock or LRCLK). A third line is used as a request to the host for more data. It is an indicator that the CS492X internal FIFO is low on data and can accept another block of data. Typically this mode is used for compressed data delivery where asynchronous data transfer occurs in the system, i.e. in a system such as a Set Top Box or HDTV. PCM data can not be presented in this mode since data is interpreted as a continuous stream with no word boundaries. For this reason bursty mode covers both data format and data type. Left Right SCLK SDATA MSB LSB MSB LSB Figure 35. I2S Format LRCK Left Right SCLK SDATA MSB LSB MSB LSB MSB Figure 36. Left Justified Format LRCLK SCLK SDATA MSB LSB MSB M Clocks Per Channel LSB MSB M Clocks Per Channel LSB M Clocks Per Channel MSB LSB MSB M Clocks Per Channel LSB MSB M Clocks Per Channel LSB MSB M Clocks Per Channel Figure 37. Multi-Channel Format (M == 20) 48 AN115REV2 5.2.2 Digital Input and Output Ports Digital Audio Input or DAI port: Table 13 shows the three pins associated with the DAI port. LRCLKN1 is the frame clock which frames the incoming data. It is synonymous with LRCLK in the digital audio format figures. SCLKN1 is the bit clock which clocks in the data and is synonymous with SCLK in the digital audio format figures. SDATAN1 is the serial data input and is synonymous with SDATA in the digital audio format figures. This port operates as a slave. In the slave mode both SCLKN1 and LRCLKN1 are driven from an external source. This port can take either compressed data or PCM data but it will not operate in the Bursty data format. Pin Name LRCLKN1 SCLKN1 SDATAN1 Pin Number 26 25 22 Table 13. DAI - Digital Audio Input Port Compressed Data Input or CDI port: Table 14 shows the three pins associated with the CDI port. The pins have different functions depending on how the port is configured. When configured for I2S, left justified or multi-channel, LRCLKN2 is the frame clock synonymous with LRCLK, SCLKN2 is the bit clock synonymous with SCLK, and SDATAN2 is the serial data input synonymous with SDATA. This port operates as a slave. In the slave mode both SCLKN2 and LRCLKN2 are driven from an external source. This port can take either compressed data or PCM data. Pin Name LRCLKN2, CMPREQ SCLKN2, CMPCLK SDATAN2, CMPDAT Pin Number 29 28 27 Table 14. CDI - Compressed Digital Input Port When the CDI is configured to operate in the Bursty format, CMPREQ is an output from the CS492X. When CMPREQ goes low it indicates that the internal FIFO in the CS492X can accept a block of data. The block size is defined in the hardware configuration message section. This pin could be used as an interrupt request to a host controlling data flow or as a throttle for an external FIFO. CMPCLK is the bit clock that will clock in data on its rising edge and CMPDAT is the serial data input. It should be noted that CMPCLK must be gated when valid data is not present as CMPCLK will clock in data on its rising edge whether CMPREQ is high or low. Digital Audio Output or DAO port: Table 15 shows the six lines associated with the DAO port. MCLK is the master clock which can be used to synchronize data flow throughout the system. MCLK is an input. LRCLK is the frame clock which frames the outgoing data and SCLK is the bit clock which clocks out the data. AUDATA0, AUDATA1, and AUDATA2 are the PCM audio outputs from the chip. Data is valid on the rising edge of SCLK. Table 16 shows the default mapping of data on the AUDATA[2:0] lines. This default can be changed using the DAO channel messages discussed in the Audio Manager portion of the each Application Code User's guide. Pin Name MCLK SCLK LRCLK AUDATA0 AUDATA1 AUDATA2 Pin Number 44 43 42 41 40 39 Table 15. DAO - Digital Audio Output Port Data Channel 0 1 2 3 4 5 Output Name Left Right Left Surround Right Surround Center Subwoofer Subframe Left Right Left Right Left Right Signal AUDATA0 AUDATA0 AUDATA1 AUDATA1 AUDATA2 AUDATA2 Table 16. Output Channel Mapping AN115REV2 49 The DAO can operate as a master or a slave. As a master, the DAO drives both LRCLK and SCLK. LRCLK and SCLK will be divided down from MCLK. When the DAO is configured in slave mode, LRCLK and SCLK are inputs. If the DAO is a slave, then MCLK is a don't care as an input. The DAO can also be configured to drive MCLK, LRCLK, and SCLK when the internal PLL is enabled. byte at a time with the internal DSP assigning the first 8-bit write (after PCMRST) to the left channel and the second 8-bit write to the right channel. For 24-bit PCM, it assigns the first three 8-bit writes (after PCMRST) to the left channel and the next three writes to the right channel. Before starting PCM transfer, or to initiate a new PCM transfer, the PCMRST bit must be toggled as described above to insure data integrity. 5.2.3 Data must be delivered to the CS492X in blocks of data. The block size is set through a hardware configuration message. Before each block is delivered, the host should check the MFC bit. If the MFC bit is low, then the host can deliver a block of data one byte at a time. If the MFC bit is high, no more data should be sent to the CS492X. Once the MFC bit has gone low again, the host may send another block of PCM audio data. The MFC bit is FIFO level sensitive. In other words, it may change during the transfer of a block. The host should complete the block transfer and ignore the MFC bit until the block transfer is complete. Parallel Delivery of Data This section covers parallel delivery of digital audio data. The low level read and write formats are identical to those discussed in Section 2.2 "Parallel Host Communication" -- page 17. It should be noted that when switching between PCM and compressed data delivery using the parallel data delivery, a new download, soft reset or application restart must be sent along with a FIFO configuration message for the appropriate data type (along with any other required hardware configuration messages). 5.2.3.1 PCM Data Write in Parallel Host Mode Writing to the PCM audio data register entails a slightly different protocol than when writing control information. The MFC bit in the Host Control Register is an indicator of the PCM FIFO level. The MFC bit remains low until the FIFO threshold has been reached. The PCMRST bit of the CONTROL register provides absolute software/hardware synchronization by initializing the input channel to uniquely recognize the first write to the byte-wide PCMDATA port. Toggling PCMRST high and low informs the DSP that the next sample read from the PCMDATA port is the first sample of the left channel. In this fashion, the CS492X can translate successive byte writes into a variable number of channels with a variable PCM sample size. In the most simple case, the CS492X can receive stereo 8-bit PCM one 50 The generic function `Read_Byte_*()' is used in the following example as a generalized reference to either Read_Byte_MOT() or Read_Byte_INT(), and `Write_Byte_*()' is a generic reference to Write_Byte_MOT() or Write_Byte_INT(). Figure 38 shows the sequence for writing one block of PCM data when the device is in parallel host mode. The protocol presented in the flow diagram on the following page will now be described in detail. 1) The host first reads the Host Control Register (A[1:0] = 01b) in order to determine the state of the input FIFOs. 2) In order to determine whether the CS492X is ready to accept another block of data, the host must check the MFC bit of the Host Control Register (bit 4). If MFC is high, then the DSP is not prepared to accept a new block of data, and the host should poll the Host Control Register again. If MFC is low, then the host may write a AN115REV2 READ_*(HOST CONTROL REGISTER) YES MFC==1 NO WRITE_*(PCMDATA REGISTER) BLOCK WRITTEN? NO FINISHED Data must be delivered to the CS492X in blocks of data. Before each block is delivered, the host should check the MFB bit (or the CMPREQ pin). If the MFB bit (CMPREQ) is low, then the host can deliver a block of data one byte at a time. If the MFB bit (CMPREQ) is high, no more data should be sent to the CS492X. Once the MFB bit (CMPREQ) has gone low again, the host may send another block of compressed audio data. One example is given for a system using the MFB bit, and one example has been given for systems using the CMPREQ pin. (Refer to figures 39 and 40 on the following pages). The generic function 'Read_Byte_*()' is used in the following examples as a generalized reference to either Read_Byte_MOT() or Read_Byte_INT(), and 'Write_Byte_*()' is a generic reference to Write_Byte_MOT() or Write_Byte_INT(). 5.2.3.2.1 Figure 38. PCM Data Write Sequence in Parallel Host Mode Flow Diagram block of PCM audio data to the PCMDATA register (A[1:0] = 10b) one byte at a time. 5.2.3.2 Compressed Data Write in Parallel Host Mode Writing to the compressed data register is very similar to writing the PCM data register. Like PCM data transfers the host should check level of the Compressed Data FIFO before sending data, but the CS492X has two means of indicating the Compressed Data FIFO level. The MFB bit in the Host Control Register is one indicator of the Compressed Data FIFO level. The MFB bit remains low until the FIFO threshold has been reached. The alternative is to use the CMPREQ pin of the CS492X. The CMPREQ pin also remains low until the FIFO threshold has been reached. The host has the option of using either CMPREQ or the MFB bit. AN115REV2 MFB Bit Example 1) .The host first reads the Host Control Register (A[1:0] = 01b) in order to determine the state of the input FIFOs. 2) In order to determine whether the CS492X is ready to accept another block of data, the host must check the MFB bit of the Host Control Register (bit 4). If MFB is high, then the DSP is not prepared to accept a new block of data, and the host should poll the Host Control Register again. If MFB is low, then the host may write a block of compressed audio data to the CMPDATA register (A[1:0] = 11b) one byte at a time. 51 5.3.1 READ_*(HOST CONTROL REGISTER) YES MFB==1 NO Address Checking The following 4-word hex message configures the address checking circuitry of the CS492X: It should be noted that this will allow the host to disable address checking or change the address of the device. If address checking enabled with an address of 0x00 is acceptable, then these messages do not need to be sent. 0x800252 0x00FFFF WRITE_*(CMPDATA REGISTER) 0x800152 0xHH0000 NO BLOCK WRITTEN? In the last word the following bits should replace HH: bits23:17 - New Address to use for checking (if enabling address checking) FINISHED bit 16 - 1 = Address checking on 0 = Address checking off Figure 39. MFB Bit Status Polling Flow Diagram 5.2.3.2.2 CMPREQ Example 1) In order to determine whether the CS492X is ready to accept another block of data, the host can check the CMPREQ pin of the CS492X. If CMPREQ is high, then the DSP is not prepared to accept a new block of data, and the host should poll the CMPREQ again. If CMPREQ is low, then the host may write a block of compressed audio data to the CMPDATA register (A[1:0] = 11b) one byte at a time. 5.3 Configuration Messages This section discusses the actual messages to be sent to the device after soft reset or download. To assemble the entire hardware configuration message, the hex messages for each individual parameter should be concatenated, creating one large message. If the default configuration for a parameter is acceptable, then no message needs to be sent. 52 YES CMPREQ==1 NO WRITE_*(CMPDATA REGISTER) NO BLOCK WRITTEN? FINISHED Figure 40. CMPREQ Pin Status Polling Flow Diagram AN115REV2 5.3.2 Input 2 Both data format (I S, Left Justified, etc.) and data type (compressed or PCM) are required to fully define the input port's hardware configuration. The DAI and the CDI are configured by the same group of messages since their configurations are interrelated. The naming convention of the input hardware configuration is as follows: INPUT A B C D where A, B, C and D are the parameters used to fully define the input port. The parameters are defined as follows: A Value (note 1) Data Type 0 DAI - PCM (default) CDI - Compressed 1 DAI - PCM and Compressed CDI - Unused 2 DAI - Unused CDI - PCM 3 DAI - PCM CDI - Bursty Compressed (See Special Considerations Note 2) DAI - Multi-Channel PCM CDI - PCM A - Data Type B - Data Format (This is a don't care for parallel modes of data delivery) 4 C - SCLK Polarity D - FIFO Setup (only valid for parallel modes of data delivery) The following tables show the different values for each parameter as well as the hex message that needs to be sent. When creating the hardware configuration message, only one hex message should be sent per parameter. It should be noted that the entire B parameter hex message must be sent, even if one of the input ports has been defined as unused by the A parameter . 5.3.2.1 Special Considerations 1) 24-bit PCM input requires at least 24 SCLKS per sub-frame. The DSP always uses 24-bit resolution for PCM input. Systems having less than 24-bit resolution will not have a problem as the extra bits taken by the DSP will be under the noise floor of the input signal for left justified and I2S formats. For compressed input, data is always taken in 16 bit word lengths. 2) If the clocks to the audio ports are known to be corrupted, such as when an SPDIF receiver AN115REV2 5 DAI - PCM CDI - Multi-Channel PCM 6 DAI - PCM CDI - Not Used Parallel Port - Compressed (FIFO B) DAI - Not Used CDI - PCM Parallel Port - Compressed (FIFO B) DAI - Not Used CDI - Not Used Parallel Port - PCM (FIFO C) and Compressed (FIFO B) 7 8 Hex Message 0x800210 0x3FBFC0 0x800110 0x80002C 0x800210 0x3FBFC0 0x800110 0xC0002C 0x800210 0x3FBFC0 0x800110 0x800020 0x800210 0x003FC0 0x800110 0x0E002C 0x800210 0x3FBFC0 0x800110 0x80002C 0x800210 0x3FBFC0 0x800110 0x800025 0x800210 0x003FC0 0x800110 0x0E002B 0x800210 0x003FC0 0x800110 0x0E0023 0x800210 0x003FC0 0x800110 0x0E0013 Table 17. Input Data Type Configuration goes out of lock, the device should be reset and reconfigured. Failure to do so could result in corrupted data and unpredictable behavior. Any modes not listed are not supported by current software. If a certain mode is desired that is not available, please contact the factory about its availability. 53 B Value (note 1) Data Format 2 0 PCM - I S 24 Bit (default) Compressed - I2S 16 Bit 1 2 3 4-6 7 8 Hex Message 0x800217 0x8080FF 0x80021A 0x8080FF 0x800117 0x011100 0x80011A 0x011900 PCM - Left Justified 24 Bit 0x800217 Compressed - Left Justified 16 0x8080FF Bit 0x80021A 0x8080FF 0x800117 0x001000 0x80011A 0x001800 PCM - I2S 24 Bit 0x800217 Multi-channel PCM - Left Justi- 0x8080FF fied 24 bit PCM 0x80021A 0x8080FF 0x800117 0x0048C0 0x80011A 0x0119C0 PCM - Left Justified 24 Bit 0x800217 Multi-channel PCM - Left Justi- 0x8080FF fied 24 bit 0x80021A 0x8080FF 0x800117 0x0048C0 0x80011A 0x0018C0 Not Used PCM - I2S 24 Bit 0x800217 Multi-channel PCM - Left Justi- 0x8080FF fied 20 bit 0x80021A 0x8080FF 0x800117 0x003CC0 0x80011A 0x0119C0 PCM - Left Justified 24 Bit 0x800217 Multi-channel PCM - Left Justi- 0x8080FF fied 20 bit 0x80021A 0x8080FF 0x800117 0x003CC0 0x80011A 0x0018C0 C Value SCLK Polarity (Both CDI & Hex (note 1) DAI Port) Message 0 Data Clocked in on Rising 0x800217 (default) Edge 0xFFFFDF 0x80021A 0xFFFFDF 1 Data Clocked in on Falling 0x800117 Edge 0x000020 0x80011A 0x000020 Table 19. Input SCLK Polarity Configuration FIFO Size & Blocksize (no default - only applicable to D Value parallel delivery modes) 1 Compressed FIFO B Size 6kbyte Blocksize - 2kbyte 2 PCM FIFO C Size - 6kbyte Blocksize - 2kbyte Hex Message 0x800014 0x280D00 0x800014 0x820300 Table 20. FIFO Setup Configuration Table 18. Input Data Format Configuration 54 AN115REV2 5.3.3 Output The naming convention for the DAO configuration is as follows: B Value (Note 3) DAO Data Format 2 0 I S 20-bit (default) OUTPUT A B C D E where the parameters are defined as: A - DAO Mode (Master/Slave for LRCLK and SCLK) B - Data Format C - MCLK Frequency D - SCLK Frequency E - SCLK Polarity The following tables show the different values for each parameter as well as the hex message that needs to be sent. When creating the hardware configuration message, only one hex message should be sent per parameter. DAO Modes (LRCLK & A Value SCLK) 0 MCLK - Slave (default) SCLK - Slave LRCLK - Slave 1 MCLK - Slave SCLK - Master LRCLK - Master 2 MCLK - Master SCLK - Master LRCLK - Master Hex Message 0x80017F 0x400000 0x80027F 0xBFDFFF Special Considerations 1) All PCM output is 20-bit resolution 2) An SCLK frequency of at least 128Fs must be selected for the 20-bit multi-channel mode. 3) An SCLK frequency of at least 256Fs must be selected for the 24-bit multi-channel mode. 4) If the clocks to the audio ports are known to be corrupted, such as when an SPDIF receiver goes out of lock, the device should be reset and AN115REV2 2 0x80027F 0xBFFFFF Table 21. Output Clock Configuration 5.3.3.1 1 3 Hex Message 0x80027F 0xFC7FFF 0x80027C 0xF01F00 0x80027D 0xF01F00 0x80027E 0xF01F00 0x80017F 0x038000 0x80017C 0x000001 0x80017D 0x000001 0x80017E 0x000001 Left Justified 20-bit 0x80027F 0xFC7FFF 0x80027C 0xF01F00 0x80027D 0xF01F00 0x80027E 0xF01F00 0x80017F 0x018000 Multi-Channel 0x80027F 20 bit Left Justified 0xFC7FFF (SCLK must be at least 128Fs 0x80027C for this mode) 0xF01F00 0x80027D 0xF01F00 0x80027E 0xF01F00 0x80027F Multi-Channel 0xFC7FFF 24 bit Left Justified (SCLK must be at least 256Fs 0x80027C 0xF01F00 for this mode) 0x80027D 0xF01F00 0x80027E 0xF01F00 0x80017C 0x008000 Table 22. Output Data Format Configuration reconfigured. Failure to do so could result in corrupted data and unpredictable behavior. 55 C Value Output MCLK Frequency 0 256Fs (default) 1 512Fs 2 128Fs 3 384Fs (SCLK must be 64Fs in this mode) Hex Message 0x80027F 0xFFE7FF 0x80027F 0xFFE7FF 0x80017F 0x001000 0x80027F 0xFFE7FF 0x80017F 0x001800 0x80027F 0xFFE7FF 0x80017F 0x000800 Table 23. Output MCLK Configuration D Value Output SCLK Frequency 0 64Fs (default) 1 128Fs 2 256Fs Hex Message 0x80027F 0xFFF8FF 0x80017F 0x000100 0x80027F 0xFFF8FF 0x80017F 0x000200 0x80027F 0xFFF8FF 0x80017F 0x000300 Table 24. Output SCLK Configuration E Value Output SCLK Polarity 0 Data Valid on Rising Edge (default) (clocked out on falling) 1 Data Valid on Falling Edge (clocked out on rising) Hex Message 0x80027F 0xF7FFFF 0x80017F 0x080000 Table 25. Output SCLK Polarity Configuration 5.3.4 Creating Hardware Configuration Messages The single hardware configuration message that must be sent to the CS492X after download or soft reset should be a concatenation of the messages in the Section 5.3.1 through Section 5.3.3. The complete hardware configuration message should be created by taking a message for each parameter 56 (where the default is not acceptable) and concatenating the messages together. No messages need to be sent if the default configuration for a particular parameter is acceptable. This example can be easily expanded to fit other system requirements. For example if the host system has the following configuration: Address Checking: Enabled with an address of 0000000b The above configuration is default so no configuration message is required. DAI: Left Justified Slave Mode PCM and Compressed data CDI: Not used The above configuration corresponds to INPUT 1 1 which corresponds to a configuration message of: 0x800210 0x3FBFC0 0x800110 0xC0002C 0x800217 0x8080FF 0x80021A 0x8080FF 0x800117 0x001000 0x80011A 0x001800 DAO: Left Justified slave mode (LRCLK, SCLK inputs) MCLK @ 256Fs SCLK @ 64Fs The above configuration corresponds to OUTPUT 0 1 0 0 which has a configuration message of: AN115REV2 0x80027F 0xFC7FFF 0x80027C 0xF01F00 0x80027D 0xF01F00 0x80027E 0xF01F00 0x80017F 0x018000 Concatenating the messages together gives the following hardware configuration message that should be sent after download or soft reset: WORD# 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 VALUE 0x800210 0x3FBFC0 0x800110 0xC0002C 0x800217 0x8080FF 0x80021A 0x8080FF 0x800117 0x001000 0x80011A 0x001800 0x80027F 0xFC7FFF 0x80027C 0xF01F00 0x80027D 0xF01F00 0x80027E 0xF01F00 0x80017F 0x018000 Table 26. Example Values to be Sent to CS492X After Download or Soft Reset AN115REV2 57 6. APPENDIX A - PSEUDOCODE FOR THE CS4923/4/5/6/7/8/9 FAMILY In the pseudocode, it is assumed a function call to a signal name forces the signal HIGH or LOW (i.e. CS(LOW) forces the chip select line of the CS492X LOW). Also, the function Read(SIGNAL) returns the value of SIGNAL. The pseudocode that is not communication specific uses function calls to Write_* and Read_*. The * should be replaced by the communication mode for the system. 6.1 SPI Pseudocode 6.1.1 SPI Write Operation #define ADDR23 0x00 #define WRITE 0xFE unsigned char message[size]; void Write_SPI(unsigned char *message, int size) { int x; /* size=message length in bytes */ CS(LOW); Write_Byte_SPI(ADDR4923 & WRITE); for(x=0;x<size;x++) Write_Byte_SPI(message[x]); CS(HIGH); } /* Write_SPI */ void Write_Byte_SPI(unsigned char data_byte) { char bit_number; for (bit_number=7;bit_number>=0;bit_number--) { if((data_byte>>bit_number)&0x01) /* check each bit to write */ CDIN(HIGH); else CDIN(LOW); SCL(HIGH); /* clock in the bit */ SCL(LOW); /* insure data byte is clocked in to CS4923/4/5/6/7/8/9 */ } } /*Write_Byte_SPI */ 58 AN115REV2 6.1.2 SPI Read Operation #define ADDR23 0x00 #define READ 0x01 unsigned char data_byte_array[size]; void Read_SPI(unsigned char *data_byte_array) { int i = 0; char not_end_of_read = 1; CS23(LOW); Write_Byte_SPI(ADDR4923 | READ); while (not_end_of_read) { not_end_of_read = Read_Byte_SPI(data_byte_array + i); printf("%02x ", data_byte_array[i]); i++; } CS23(HIGH); } /* SPI_Read */ unsigned char Read_Byte_SPI(unsigned char *DataIn) { int bit_number; int end_of_data = 1; *DataIn=0; for (bit_number=0; bit_number<8; bit_number++) { SCL(HIGH); /*clock out data */ if (bit_number == 6) if (Read(/INTREQ) == 1)/*check request */ end_of_data = 0; /*if HIGH -> no more data, if LOW -> read again */ *DataIn |= Read(CDOUT) << bit_number; SCL(LOW); } return(end_of_data); } /*Read_Byte_SPI */ AN115REV2 59 6.2 I2C Pseudocode I2C Write Operation 6.2.1 #define ADDR23 0x00 #define WRITE 0xFE unsigned char message[size]; void Write_I2C(unsigned char *message, int size) /* size=message length in bytes */ { int x; Send_I2C_Start(); if(Write_Byte_I2C(ADDR4923 & WRITE)) { /* Received NACK so resend address */ if(Write_Byte_I2C(ADDR4923 & WRITE)) { /*Second NACK so send stop condition and return ERROR*/ printf("\n error sending address byte"); Send_I2C_Stop(); return ERROR; } } for(x=0;x<size;x++) { if(Write_Byte_I2C(message[x])) { /*Received NACK so resend data byte */ if(Write_Byte_I2C(message[x])) { /*Second NACK so send stop condition and return ERROR*/ printf("\n error sending byte %d", x); Send_I2C_Stop(); return ERROR; } } } Send_I2C_Stop(); return 0; }/*Write_I2C*/ void Send_I2C_Start() { /* This function assumes that SCCLK and SCDIO are both HIGH when called */ SCDIO(LOW);/* drive SCDIO LOW while SCCLK is HIGH for start condition */ SCCLK(LOW);/* drive SCCLK LOW to prepare for data transfer */ } void Send_I2C_Stop() { SCDIO(LOW); /* make sure SCDIO is LOW */ SCCLK(HIGH);/* drive SCCLK HIGH */ SCDIO(HIGH);/* drive SCDIO for the STOP condition */ } unsigned char Write_Byte_I2C(unsigned char data) { 60 AN115REV2 int bit_number; for (bit_number=7;bit_number>=0;bit_number--) { if((data>>bit_number)&0x01) /* check each bit to write and drive data line */ SCDIO(HIGH); else SCDIO(LOW); SCCLK(HIGH); /*clock in data */ SCCLK(LOW); } SCDIO(HIGH); /*release bus so 4923 can ACK*/ return Get_ACK(); /* return value of ACK*/ } unsigned char Get_ACK() { unsigned char ack; SCCLK(HIGH);/* latch the ACK */ ack = Read(SCDIO); /*Read ACK*/ SCCLK(LOW); return(ack); } AN115REV2 61 6.2.2 I2C Read Operation #define ADDR23 0x00 #define READ 0x01 unsigned char data_byte_array[size]; char Read_I2C(unsigned char *data_byte_array) { int i = 0; char not_end_of_read = 1; Send_I2C_Start(); if(Write_Byte_I2C(ADDR4923 | READ)) { /*Received NACK so send again */ if(Write_Byte_I2C(ADDR4923 | READ)) { /*Second NACK so send stop condition and return ERROR*/ printf("\n error sending address byte for read condition"); Send_I2C_Stop(); return ERROR; } } while (not_end_of_read) { not_end_of_read = Read_Byte_I2C(data_byte_array + i); printf("%02x ", data_byte_array[i]); i++; } Send_I2C_Stop(); return(0); } unsigned char Read_Byte_I2C(unsigned char *DataIn) { int bit_number; int end_of_data = 0; *DataIn=0; for (bit_number=0; bit_number<8; bit_number++) { SCL(HIGH); if (bit_number == 7) if (Read(/INTREQ)) /* check request */ end_of_data = 1; /* if HIGH -> no more data, if LOW -> read again */ *DataIn |= Read(CDOUT) << bit_number SCL(LOW); } if (end_of_data) { Send_NACK(); return 0; } else 62 AN115REV2 { Send_ACK(); return 1; } } void Send_ACK() { SCDIO(LOW);/* force SCDIO LOW to ACK */ SCCLK(HIGH);/* clock the ACK */ SCCLK(LOW); } void Send_NACK() { SCDIO(HIGH);/* release SCDIO HIGH to NACK */ SCCLK(HIGH);/* clock the NACK */ SCCLK(LOW); } AN115REV2 63 6.3 Typical Download Session with the CS4923/4/5/6/7/8/9 #define BOOT_MSG_SIZE 3 void Boot_CS4923() { unsigned char error = 0; unsigned char message_bytes[3]; RESET(LOW); /* hard reset the CS4923/4/5/6/7/8/9 */ RESET(HIGH); Write_*(DOWNLOAD_BOOT, BOOT_MSG_SIZE); while(Read(/INTREQ));/* wait for /INTREQ to fall */ Read_*(message_bytes);/* read CS4923/4/5/6/7/8/9 response */ switch(message_bytes[0]) { case 0x01: printf("\n BOOT_START "); error = 0; break; case 0xfa: case 0xfc: printf("\n BOOT_ERROR "); error = 1; break; case 0xfb: printf("\n INVALID_MSG "); error = 1; break; case 0xfd: case 0xfe: printf("\n INIT_FAILURE "); error = 1; break; default: printf("\n UNRECOGNIZED BYTE "); error = 1; break; } if(error) exit(); Write_*(.LD_FILE_IMAGE_POINTER, .LD_FILE_IMAGE_SIZE); while(Read(/INTREQ));/* wait for /INTREQ to fall */ Read_*(message_bytes);/* read CS4923/4/5/6/7/8/9 response */ switch(message_bytes[0]) { case 0x02: printf("\n BOOT_SUCCESS "); error=0; break; case 0xff: printf("\n BAD_CHECKSUM "); error = 1; break; default: printf("\n UNRECOGNIZED BYTE AFTER DOWNLOAD"); error = 1; break; } 64 AN115REV2 if(error) exit(); Write_*(BOOT_SUCCESS_RECEIVED, BOOT_MSG_SIZE); }/* Boot_CS4923/4/5/6/7/8/9 */ 6.4 Typical Reset Sequence for the CS4923/4/5/6/7/8/9 void Reset_CS492X() { RESET(LOW); RESET(HIGH); Write_*(SOFTRESET, BOOT_MSG_SIZE); Delay(1); /* Insure 1 ms pause before sending configuration messages */ Write_*(Configuration_Messages, Message Size) } AN115REV2 65