ETC AMBE2000

Version 3.0
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
AMBE-2000™ Vocoder Chip
User’s Manual
Version 3.0
April, 2000
Copyright , 2000
Digital Voice Systems, Inc
234 Littleton Road
Westford, MA 01886
This document may not, in whole or in part be copied, photocopied, reproduced, translated, or reduced to any electronic
medium or machine readable form without prior consent in writing from Digital Voice Systems, Incorporated.
Every effort has been made to ensure the accuracy of this manual. However, Digital Voice Systems, Inc. makes no warranties
with respect to the documentation and disclaims any implied warranties of merchantability and fitness for a particular purpose.
Digital Voice Systems, Inc. shall not be liable for any errors or for incidental or consequential damages in connection with the
furnishing, performance, or use of this manual or the examples herein. This includes business interruption and/or other loss
which may arise from the use of this product. The information in this document is subject to change without notice.
Trademarks
AMBE-2000™ Vocoder Chip is a registered trademark of Digital Voice Systems, Inc. Other product names mentioned may be
trademarks or registered trademarks of their respective companies and are the sole property of their respective manufacturers.
All Rights Reserved
Data subject to change
page 2
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
AMBE-2000™ Vocoder Chip END USER License Agreement
1.0 Preliminary Statements and Definitions
7.0 Proprietary Information
1.1 “END USER” shall mean the person and/or organization to whom the AMBE2000™ Vocoder Chip was delivered or provided to as specified in the purchase order or
other documentation. In the event that the END USER transfers his rights under this
license to a third party as specified in section 2.2, then this third party shall become an
“END USER”.
1.2 Digital Voice Systems, Inc. (DVSI) has developed a voice coding method and
algorithm (the “Technology”) based on the Advanced Multi-Band Excitation
(“AMBE”) voice coder. The technology codes speech at bit rates of 2.4 to 9.6 kilobits
per second (kbps) including error correction bits.
1.3 "AMBE Voice Compression Software" shall mean the speech coding software
and/or firmware integrated into the AMBE-2000™ Vocoder chip integrated circuit.
1.4 "Voice Codec" shall mean the AMBE-2000™ Vocoder Chip integrated circuit, the
AMBE Voice Compression Software, firmware and associated documentation,
including modifications, enhancements and extensions made by or for Digital Voice
Systems, Inc. (DVSI) and including circuit diagrams, timing diagrams, logic diagrams,
layouts, operating instructions and user manuals.
1.5 DVSI represents that it owns certain “Proprietary Rights” in the Technology and the
AMBE Voice Compression Software, including patent rights in the Technology, and
patent rights, copyrights, and trade secrets in the AMBE Voice Compression Software.
7.1 The parties agree that the AMBE Voice Compression Software shall be considered Proprietary
Information.
7.2 Except as otherwise provided in this Agreement, END USER shall not use, disclose, make, or have
made any copies of the Proprietary Information, in whole or in part, without the prior written consent of
DVSI.
2.0 License Granted
2.1 Subject to the conditions herein and upon initial use of the AMBE-2000™ Vocoder
Chip, DVSI hereby grants to END USER a non-exclusive, limited license to use the
AMBE® Voice Compression Software in machine readable form solely on the AMBE2000™ Vocoder Chip. Title to the AMBE® Voice Compression Software remains with
DVSI. No license is granted for use of the AMBE® Voice Compression Software on
other than the AMBE-2000™ Vocoder Chip. No license, right or interest in any
trademark, trade name or service mark of DVSI is granted under this Agreement.
2.2 END USER shall not copy, extract, de-compile, reverse engineer or disassemble the
AMBE® Voice Compression Software contained in the AMBE-2000™ Vocoder Chip.
3.0 Transfer of License
3.1 The END USER shall have the right to transfer the AMBE-2000™ Vocoder Chip
and all rights under this Agreement to a third party by either (i) providing the third party
with a copy of this Agreement or (ii) providing the third party with an agreement written
by the END USER ( hereinafter “END USER Agreement”) so long as the END USER
Agreement is approved in writing by DVSI prior to transfer of the AMBE-2000™
Vocoder Chip. The END USER Agreement shall contain comparable provisions to those
contained herein for protecting the Proprietary Information from disclosure by such third
party. Third parties shall agree to accept all the terms and conditions under either
Agreement or the END USER Agreement.
4.0 Term and Termination
4.1 This Agreement is effective upon initial delivery of the Voice Codec and shall remain
in effect until terminated in accordance with this agreement.
4.2 This Agreement shall terminate automatically without notice from DVSI if END
USER fails to comply with any of the material terms and conditions herein. END USER
may terminate this Agreement at any time upon written notice to DVSI certifying that
END USER has complied with the provisions of Section 3.3.
4.3 Upon termination of this Agreement for any reason, END USER shall: (i) return all
AMBE-2000™ Vocoder Chip purchased or acquired, or in Licensee’s possession, to
DVSI; (ii) have no further rights to any AMBE® Voice Compression Software or the
Technology without a separate written license from DVSI; (iii) discontinue all use of the
AMBE-2000™ Vocoder Chip;
5.0 Payments
5.1 In consideration of the materials provided as part of the Voice Codec, and in
consideration of the license and rights in the AMBE Voice Compression Software
granted by DVSI, and in consideration of DVSI's performance of its obligations
hereunder, END USER agrees to pay to DVSI the fee specified in DVSI's invoice.
6.0 Proprietary Notices
6.1 END USER shall not remove any copyright or proprietary notice on the AMBE2000™ Vocoder Chip or on the AMBE Voice Compression Software.
8.0 Limited Warranty
8.1 DVSI warrants the Voice Codec to be free from defects in materials and workmanship under normal
use for a period of ninety (90) days from the date of delivery.
8.2 Except as stated in Section 7.1, the Voice Codec is provided "as is" without warranty of any kind.
DVSI does not warrant, guarantee or make any representations regarding the use, or the results of the
use, of the Voice Codec with respect to its correctness, accuracy, reliability, correctness or otherwise.
The entire risk as to the results and performance of the Voice Codec is assumed by the END USER.
After expiration of the warranty period, END USER, and not DVSI or its employees, assumes the entire
cost of any servicing, repair, replacement, or correction of the Voice Codec.
8.3 DVSI represents that, to the best of its knowledge, it has the right to enter into this Agreement and to
grant a license to use the AMBE Voice Compression Software to END USER.
8.4 Except as specifically set forth in this Section 7.0, DVSI makes no express or implied warranties
including, without limitation, the warranties of merchantability or fitness for a particular purpose or
arising from a course of dealing, usage or trade practice, with respect to the Voice Codec. Some states
do not allow the exclusion of implied warranties, so the above exclusion may not apply to END USER.
No oral or written information or advice given by DVSI or its employees shall create a warranty or in
any way increase the scope of this warranty, and END USER may not rely on any such information or
advice. The limited warranties under this section 7.0 give END USER specific legal rights, and END
USER may have other rights which vary from state to state.
9.0 Limitation of Liability
9.1 In no event shall DVSI be liable for any special, incidental, indirect or consequential damages
resulting from the use or performance of the Voice Codec whether based on an action in contract, tort
(including negligence) or otherwise (including, without limitation, damages for loss of business profits,
business interruption, and loss of business information), even if DVSI or any DVSI representative has
been advised of the possibility of such damages.
9.2 Because some states do not allow the exclusion or limitation of liability for consequential or
incidental damages, the above limitations may not apply to END USER.
9.3 DVSI's maximum liability for damages arising under this Agreement shall be limited to 20% (twenty
percent) of the fees paid by END USER for the particular Voice Codec which caused the damages or
that is the subject matter of, or is directly related to, the cause of action.
10.0 Taxes
10.1 All payments required under Section 4.0 or otherwise under this Agreement are
exclusive of taxes and END USER agrees to bear and be responsible for the payment
of all such taxes (except for taxes based upon DVSI's income) including, but not
limited to, all sales, use, rental receipt, personal property or other taxes which may be
levied or assessed in connection with this Agreement.
11.0 Export
11.1 United States export laws and regulations prohibit the exportation of certain products or technical
data received from DVSI under this Agreement to certain countries except under a special validated
license. As of November 30, 1999 the restricted countries are: Libya, Cuba, North Korea, Iraq, Serbia,
Taliban in Afghanistan, Sudan, Burma, Yugoslavia and Iran. The END USER hereby gives its
assurance to DVSI that it will not knowingly, unless prior authorization is obtained from the appropriate
U.S. export authority, export or re-export, directly or indirectly to any of the restricted countries any
products or technical data received from DVSI under this Agreement in violation of said United States
Export Laws and Regulations. DVSI neither represents that a license is not required nor that, if
required, it will be issued by the U.S. Department of Commerce. Licensee shall assume complete and
sole responsibility for obtaining any licenses required for export purposes.
12.0 Governing Law
12.1 This Agreement is made under and shall be governed by and construed in
accordance with the laws of the Commonwealth of Massachusetts, except that body of
law governing conflicts of law. If any provision of this Agreement shall be held
unenforceable by a court of competent jurisdiction, that provision shall be enforced to
the maximum extent permissible, and the remaining provisions of this Agreement
shall remain in full force and effect.
page 3
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
1. Product Introduction
2
3
4
6
1.1
General Information
6
1.2
Advantages
6
1.3
Typical Applications
7
AMBE™-2000 Application Design Overview
2.1 Basic Operation
8
2.2 Initial Design Considerations
2.2.1
A/D – D/A Overview
2.2.2
Channel Interface Overview
2.2.3
Speech and FEC Rate Selection Overview
8
8
9
9
Hardware Information
10
3.1 Special Handling Instructions
3.1.1
Storage
10
10
3.2
Pin Descriptions
11
3.3
Clock and Reset Timing
13
3.4 Crystal / Oscillator Usage
3.4.1
TTL Clock Source
3.4.2
Crystal Oscillator
14
14
15
3.5
Package Description
16
3.6
Normal Operating Conditions
17
3.7
Absolute Maximum Ratings
17
3.8
Electrical Characteristics and Requirements
17
Channel Interface
19
4.1
Overview
19
4.2
Serial Configuration Selection
19
4.3 Channel Serial Mode
4.3.1
Low Level Timing for Passive and Active Serial Mode
5
8
Channel Data Format
5.1
21
22
24
Formatted Format
24
5.2 Formatted Input Format
5.2.1
Formatted Input : Word 0 : Header
5.2.2
Formatted Input : Word 1 : Power Control
5.2.3
Formatted Input : Word 1 : Control Word 1
5.2.4
Formatted Input : Words 2-6 : Rate Info 0, Rate Info 1, Rate Info 2, Rate Info 3, Rate Info 4
5.2.5
Formatted Input : Word 7 : Unused in Input
5.2.6
Formatted Input : Word 8 : Unused in Input
5.2.7
Formatted Input : Word 9 : Unused in Input
page 4
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
24
25
25
26
26
27
27
27
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
5.2.8
5.2.9
5.2.10
6
Formatted Input : Word 10 :DTMF Control
Formatted Input : Word 11 Control Word 2
Formatted Input : Words 12-23 : Channel Data
27
28
29
5.3 Formatted Output Format
5.3.1
Formatted Output : Word 0 : Header
5.3.2
Formatted Output : Word 1 : ID (Power Control)
5.3.3
Formatted Output : Word 1 :Control Word 1
This 8-bit control word indicates the activity of various functions.
5.3.4
Formatted Output : Words 2-6 : Rate Info 0, Rate Info 1, Rate Info 2, Rate Info 3, Rate Info 4.
5.3.5
Formatted Output: Word 7 : Bit Error Rate
5.3.6
Formatted Output : Word 10 : DTMF Control
5.3.7
Formatted Output : Word 11 :Control Word 2
5.3.8
Formatted Output : Words 12-23 :Channel Data
29
30
30
30
30
30
30
30
32
33
5.4 Unformatted Serial Format
5.4.1
Unformatted Serial Output Format
5.4.2
Unformatted Serial Input Format
33
33
34
A/D-D-A Interface
35
6.1
A/D-D/A Overview
35
6.2
Configuring the A/D-D/A Interface using CODEC_SEL[1-0]
35
6.3 Low Level A/D–D/A Timing
Low Level Timing for A/D-D/A in Active and Passive Modes
7
8
Special Functions
36
36
38
7.1
Hardware vs. Software Selection Note
38
7.2
Coding Rate Selection
38
7.3
Echo Cancellation
39
7.4
Voice Activation Detection (VAD), Comfort Noise Insertion (CNI)
40
7.5
Dual Tone Multiple Frequency, Detection and Generation
40
7.6 Normal Power and Power Saving Modes
7.6.1
Standard Sleep Mode
7.6.2
Power Down
40
40
41
7.7
Slip Enable
42
Appendices
45
8.1 Example A/D-D/A Usage
8.1.1
AD73311
45
45
page 5
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
1. Product Introduction
1.1 General Information
Digital Voice Systems Inc.’s AMBE-2000™ Vocoder Chip is an extremely flexible, high-performance, single chip, speech
compression coder. It provides superior voice quality at low data rates. It provides a real-time, full-duplex implementation of
the standard-setting AMBE voice compression software algorithm. DVSI’s patented AMBE voice compression technology
has been proven to outperform CELP, RELP, VSELP, MELP, ECELP, MP-MLQ, LPC-10, and other competitive technologies.
Numerous evaluations have shown its ability to provide performance equal to today’s digital cellular systems at under half the
data rate. The AMBE voice compression algorithm is used in applications throughout the world, including the next generation
of digital mobile communication systems.
The AMBE-2000™ Vocoder chip provides a high degree of flexibility in selecting the speech and FEC (Forward Error
Correction) data rates. The user can separately select these parameters in 50 bps increments for total rates from 2.4 kbps to 9.6
kbps. Typically for higher error rate channels, the user will apportion a greater percentage of the total bit rate to FEC coding.
The AMBE-2000™ voice coder maintains natural voice quality and speech intelligibility at rates as low as 2.4 kbits/sec. The
AMBE algorithm’s low complexity allows it to be fully integrated into a low cost, low power integrated circuit, the AMBE2000™ Vocoder Chip.
The AMBE-2000™ Vocoder Chip offers similar features as to DVSI’s AMBE-1000™ Vocoder Chip allowing it to be
incorporated into systems already design for the AMBE-1000™ and is interoperable with other DVSI products. The AMBE2000™ Vocoder Chip delivers improved performance and enhanced modes such as 4.0 kbps toll quality speech and
convolutional FEC coding. Along with these enhancements the AMBE-2000™ Vocoder Chip employs a control interface
along with the variable data rates and FEC selection as found in .
1.2 Advantages
•
Superior Voice Quality
•
Low Cost
•
No External Memory Required
•
Robust to Bit Errors & Background Noise
•
Variable Data Rates - 2.0 kbps to 9.6 kbps
•
Variable FEC Rates - 50 bps to 7.2 kbps
•
Very Low Power (65mW @ 3.3V, 0.11mW Deep Sleep)
•
Compact Single Chip Solution: 100 pin TQFP
•
High Quality Low Data Rate Speech Coding
•
DVSI’s Full Duplex AMBE Voice Coder
•
Supports Data Rates of 2.0 kbps to 9.6 kbps in 50 bps increments
•
User Selectable Forward Error Correction rates
•
Viterbi Decoder (rate 1/4 or more)
•
16 Level Soft Decision Decoding
•
Voice Activity Detection (VAD) / Comfort Noise Insertion
•
16 ms. Echo Cancellation
•
Single and Dual Tone (DTMF) Detection and Generation
Features
page 6
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
•
Power-Down Mode
•
Minimal algorithmic processing delay
•
DTMF detection and regeneration with North American call progress tones
1.3 Typical Applications
•
Satellite Communications
•
Digital Mobile Radio
•
Secure Communications
•
Cellular Telephony and PCS
•
Voice Multiplexing
•
Voice Mail
•
Multimedia Applications
page 7
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
2 AMBE™-2000 Application Design Overview
2.1 Basic Operation
In its simplest model, the AMBE-2000™ can be viewed as two separate components, the Encoder and the Decoder. The
Encoder receives an 8kHz. sampled stream of speech data (16-bit linear, 8-bit Alaw, or 8-bit ulaw) and outputs a stream of
channel data at the desired rate. Conversely the Decoder receives a stream of channel data and synthesizes a stream of speech
data. The timing for the interfaces for the AMBE-2000™ Encoder and Decoder are fully asynchronous.
Figure 2-A Basic Operation
AMBE-2000
AMBE-2000
8kHz Speech Data
Encoder
Compressed Data @ 2400-9600bps
Decoder
8kHz Speech Data
8kHz Speech Data
Decoder
Compressed Data @ 2400-9600bps
Encoder
8kHz Speech Data
Typically the speech interface is an external A/D-D/A chip. The format of the incoming and outgoing speech data streams are
coupled, that is to say they must be the same format (16-bit linear, 8-bit Alaw, or 8-bit µlaw). The channel interface is
commonly (but not limited to) an 8 or 16 bit microprocessor or other suitable ‘glue logic’ hardware capable of performing the
rudimentary formatting functions between the AMBE-2000™ channel format and the format of the system channel under
design.
Optional functions of the chip, such as echo cancellation, voice activation/detection, power mode control, data/FEC rate
selection, etc. are controlled either through hardware control pins (see Section 0) and/or through the decoder command
interface (see Section 4) Data sent into the decoder for function control purposes is distinguished from the data to be decoded
into speech through a channel format which is described in Section 4.
2.2
Initial Design Considerations
Some of the initial design considerations the application engineer will face are the following:
2.2.1
•
Choice of A/D-D/A chip.
•
Choice of Channel Interface.
•
Speech and FEC Rates.
A/D – D/A Overview
The choice of the A/D-D/A chip is critical to designing a system with superior voice quality. Given that Alaw and µlaw
companding chips are already incorporating some compression to reduce the number of bits per sample, it is recommended
that, when possible, a 16-bit linear device be used for maximum voice quality. When choosing a device, pay particular
attention to Signal to Noise ratios and Frequency Responses of any filters that may be present on the analog front end of these
chips. The Alaw and µlaw interfaces are also provided for the design engineer who is trying to fit to pre-existing conditions or
is under other cost type restraints.
page 8
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
2.2.2
Channel Interface Overview
The channel interface is meant to be flexible to allow for easy integration with the system under design. The basic hardware
unit of the interface is a serial port. The serial mode can run in passive or active modes. Simply stated, the control signal for
serial mode can be derived by the AMBE-2000™ chip or they can be derived externally.
Under normal operation, every 20msec the encoder outputs a frame of coded bits, and the decoder needs to be delivered a
frame of coded bits. There is some formatting of the data for both the encoder and the decoder. The primary purpose of the
formatting is to provide alignment information for the encoded bit stream. The data has two formats, Formatted and
Unformatted. Serial mode can run in either Formatted or Unformatted mode.
The Formatted and Unformatted modes are explained in full detail in Section 4, but essentially the two formats are trying to
achieve the same function, to provide positional information regarding the outgoing and incoming coded data streams. In
Formatted mode each 20msecs of output data from the encoder is preceded by a known structure. This structure also embeds
some status type flags, meant for local control purposes, within it. The only data from the Formatted format that is typically
sent across the transmission channel under design are the actual encoded bits at the desired rate.
In Formatted mode, it is the responsibility of the designed system to pass enough information along with the encoded bits such
that the Formatted format needed by the decoder can be reconstructed on the other side. This extra information, or overhead,
is going to be very specific to the system under design, but at a minimum needs to pass enough information to reliably
reconstruct the 20msec frame structure at the other end for the decoder.
In Unformatted mode the data coming out of the encoder can be thought of as a continuous stream of voice data with the
framing information embedded within the encoded bits. One advantage of this type of set-up is that the system does not have
to add any bandwidth for overhead to the channel. The disadvantage is that the decoder needs 10-12 incoming frames in order
to gain synchronization with the data stream before it can properly synthesize the speech waveform. Also, the Unformatted
mode only commits a single bit per frame to maintaining data alignment. In higher error rate channels the performance will be
improved by adding more bits per frame to the alignment information (which is more easily performed when using Formatted
mode)
Additional flexibility is given to the channel interface to the encoder and decoder by allowing the AMBE-2000™ Vocoder
Chip to run in Passive or Active modes. In Passive mode, data strobes are provided by an external source, while in Active
mode, data strobes are provided by the AMBE-2000™ Vocoder Chip. The serial interfaces can be run in Passive or Active
modes. See Section 4 for full details and timing for both Formatted and Unformatted data.
2.2.3
Speech and FEC Rate Selection Overview
The total coded bit rate is the sum of two components, the Speech Data and the Forward Error Correction (FEC) Data. The
addition of FEC data to the speech data allows the decoder to be able to correct a limited amount of errors within each frame
should they arrive corrupted. If the channel is expected to have more errors then more bits should be dedicated to FEC. At the
same time, voice quality will increase if the number of speech bits can be maximized.
page 9
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
3 Hardware Information
3.1
Special Handling Instructions
The AMBE-2000 uses the TM320-LC541-66 core. For more details on handeling, electrical characteristics, packaging, or
timing constraints please refer to the TMS320-C54x manual found at http://www-s.ti.com/sc/psheets/sprs039c/sprs039c.pdf
(Adobe Acrobat). Although the AMBE-2000™ Vocoder Chip incorporates input protection circuitry, to avoid damage from
the accumulation of a static charge, industry standard electrostatic discharge precautions and procedures must be employed
during handling and mounting.
The 100 pin TQFP package design of the AMBE-2000™ Vocoder Chip allows it to be mounted by infrared reflow, vaporphase reflow or equivalent processes. The peak package body temperature must not exceed 220°C.
The AMBE-2000™ Vocoder Chip requires baking before mounting, if any of the following conditions exist:
•
Humidity indicator card (included in packaging) shows exposure to > 20 %
when read at 23°C + 5°C
•
Devices were not shipped in a package designated as “moisture
controlled.”
•
Not mounted within 168 hours of receipt, at factory conditions of <30°C
and <60% RH
•
If the device has not been stored at < 20% RH
DVSI’s recommended bake out procedures:
•
For low-temperature device containers: 192 hours at 40°C + 5°C / -0°C
and < 5% Relative Humidity
•
For high-temperature device containers : 24 hours at 125°C + 5°C.
3.1.1 Storage
To insure maximum shelf life in long term storage, AMBE-2000™ Vocoder Chips should be kept in a moisture controlled
package at <40°C and <90% Relative Humidity
page 10
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
3.2
Pin Descriptions
Pin Number
Pin Descriptive Name
Pin Direction
Notes
77
CHANN_SEL1
Input
Channel Interface Selection Pins : Use these bits to select the channel interface type
(formatted,unformatted active, passive) according to Table 4-A. See full description in
section 4.2.
75
CHANN_SEL0
Input
Used with CH_SEL1 to select channel operation mode
85
CODEC_SEL1
Input
84
CODEC_SEL0
Input
74
RATE_SEL4
Input
73
RATE_SEL3
Input
72
RATE_SEL2
Input
71
RATE_SEL1
Input
70
RATE_SEL0
Input
86
VAD_EN
Input
Voice Activation Detection Enable Pin. Active HIGH. See Section 7.4. VAD can also be
enabled/disabled using the Control Word interface as described in section 5.2.9.
78
ECHOCAN_EN
Input
Echo Canceller Enable Pin. Active HIGH. See Section 7.3. The Echo Canceller can also
be enabled/disabled using the Control Word interface as described in section 5.2.9.
83
SLEEP_EN
Input
Standard Sleep Enable Pin. Active HIGH. See Section 7.6.1.
82
SLIP_EN
Input
Slip Control Enable Pin. Active HIGH. See Section 7.7.
68
X2/CLKIN
Input
Clock Input 1. 16.384 Mhz input. See Section 3.4
67
X1
Input
Output from internal oscillator for the crystal. If the internal oscillator is not used this pin
should be unconnected.
69
RESETN
Input
AMBE-2000 Reset pin. Active LOW. See Section 3.3
20
EPR
Output
79
SOFT_EN
Input
80
BAUD_SEL0
Input
81
BAUD_SEL1
Input
32
CHAN_RX_DATA
Input
42
CHAN_TX_DATA
Output
28
CHAN_RX_CLK
Input
Channel Receive Clock
34
CHAN_TX_CLK
Input
Channel Transmit Clock
38
CHAN_TX_STRB
I/O
Channel Transmit Frame Synchronization Pulse
30
CHAN_RX_STRB
Input
Channel Receive Frame Synchronization Pulse
29
CODEC_RX_STRB
Input
Frame synchronization pulse for A/D data. Should be connected to CODEC_TX_STRB
37
CODEC_TX_STRB
Input
Frame synchronization pulse for D/A data. Should be connected to CODEC_RX_STRB
31
CODEC_RX_DATA
Input
PCM Data from A/D Conveter to AMBE-2000
A/D-D/A Select Pins see Table 6-A to select the interface.
Coding Rate Select Pins : Use these bits to select the voice and FEC rates according to
section 7.2. The coding rates are also selectable using the Control Word interface
described in section 5.2.4.
Encode Packet Ready: Following a reset, this signal will have a high to low transition to
indicate the first packet is ready. The next packet will be ready approximately 20 msec
later. See Note 1.
Soft decision decoding enable. Enables 4 bit soft decision error decoder.
Baud Rate Selector for unformatted serial mode See Table TBDL
Channel Receive Data to AMBE-2000
Channel Transmit Data from AMBE-2000
page 11
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
Pin Number
Pin Descriptive Name
Pin Direction
Output
Notes
41
CODEC_TX_DATA
PCM Data from AMBE-2000 to D/A Conveter
27
CODEC_RX_CLK
Input
A/D Serial clock. Should be connected to CODEC_TX_CLK
33
CODEC_TX_CLK
Input
D/A Serial clock Should be connected to CODEC_RX_CLK
51
CLOCK_MODE
Input
If high enables crystal oscillator option for clock source. If low then external oscillator
option is selected. See Section 3.4 for details.
8,11,12,23
36,39,44,45
46, 47,48
49,54,57,64
76,87,90
VDD
Power
Supply Voltage
1,9,10,25,26
35,40,50,52
53,56,63,65
88,89
GND
Power
Ground
2,3,4,5,6,7,
13,14,15,16
17,18,19,21
22,24,43,55
58,59,60,61
62,66,91,92
93,94,95,96
97,98,99
100
No Connection
These pins must remain unconnected
Note 1:The AMBE-2000 expects an encoder packet to be read approximately every 20 msec. Following the initial
reset, wait for EPR to go low and read the initial packet (t0). 20 msec later, the next packet (t1) should be ready.
For packet t1 and all following packets use the procedure below:
1) Wait for slightly less than 20 msec.
2) Assert CHAN_TX_STB and read word on CHAN_TX_DATA.
3) If transmitted word not 0x13EC, discard it and repeat step 2.
4) If transmitted word 0x13EC, read 23 more words (rest of packet).
A new packet should be ready every 20 msec after the initial EPR high to low transition. A packet read should take
place every 20 msec. If there is a delay in the read (i.e. a packet is missed), it is recommended that the device be
reset.
page 12
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
3.3
Clock and Reset Timing
To reset the AMBE-2000 chip, the reset signal must be held low for a minimum of six clock cycles. The recovery time from
reset is approximately 95 msec. In other words, 95 msec after the rising edge of the reset signal the AMBE-2000 starts
processing PCM samples. The first packet will be ready after 252 PCM samples are read.
Figure 3-A X2/CLKIN and CLKOUT Timing Diagram
tw(CIH)
tw(CIL)
tf(CI)
tr(CI)
tc(CI)
~
~
~
~
X2/CLKIN
td(CIH-CO)
tw(COH)
~ ~
~
~
~
~
CLKOUT
tc(CO)
tp
tf(CO)
tr(CO)
tw(COL)
Unstable
Table 3-A X2/CLKIN and CLKOUT Timing Parameters
Reference
tc(CI)
•
•
Parameter
Cycle time,
Integer PLL multiplier N (N=4)
X2/CLKIN
Fall time, X2/CLKIN
Rise time, X2/CLKIN
Pulse duration, X2/CLKIN low
Pulse duration, X2/CLKIN high
Transitory phase, PLL lock-up time
Min
Max
Units
20
400
ns
4
4
16
ns
ns
ns
ns
µs
ns
ns
H
H
ns
ns
tf(CL)
tr(CL)
6
tw(CIL)
6
tw(CIH)
tp
15
tc(CO)
Cycle time, CLKOUT (typical is tc(CI)/4
Delay time, X2/CLKIN high/low to CLKOUT high/low
4
td(CIH-CO)
tf(CO)/tr(CO) Fall/Rise time, CLKOUT (typical is 2 ns)
Pulse duration, CLKOUT low
H-4
Tw(COL)
Pulse duration, CLKOUT high
H-4
Tw(COH)
CLKOUT is shown for reference only. It is not to be connected in the circuit.
H = 7.629 ns
page 13
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
50
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
X2/CLKIN
~
tsu(RS)
tw(RSL)
RESETN
~
tsu(INT)
th(RS)
CLKOUT
~
Figure 3-B Hardware Reset Timing Diagram
Table 3-B Reset Timing Parameters
Reference
th(RS)
tw(RSL)
tsu(RS)
tsu(INT)
•
3.4
Parameter
Hold time, RS after CLKOUT low
Pulse duration, RS low
Setup time, RS before X2/CLKIN low
Setup time, INTn, NMI, RS before CLKOUT low
Min
0
50
5
10
Max
Units
ns
µs
ns
ns
CLKOUT is shown for reference only. It should not be connected in the circuit.
Crystal / Oscillator Usage
The AMBE-2000™ Vocoder Chip has an input clock frequency of 16.384 MHz. Two options are outlined below in providing
this signal. The option used is specified by the CLOCK_MODE pin 51.
The following points should be noted when designing any printed circuit board layout:
•
Keep the crystal and external capacitors as close to the CLK_I and CLK_I2
pins as possible to minimize board stray capacitance.
•
Keep X2/CLKIN and X1 away from high frequency digital traces (example
CLKOUT) to avoid coupling.
3.4.1 TTL Clock Source
If CLOCK_MODE pin is low then a TTL/CMOS source is used as the clock input. Connect X2/CLKIN and X1 as follows:
page 14
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
Figure 3-C X2/CLKIN and X1 with TTL Clock Source
16.384 MHz
TTL/CMOS Clock
Source
X2/CLKIN (pin 68)
AMBE-2000
Unconnected
X1 (pin 67)
3.4.2 Crystal Oscillator
The Crystal Oscillator option is selected with CLOCK_MODE pin set to a high level. To use the crystal oscillator, connect the
crystal across X2/CLKIN and X1 along with one external capacitor from each of these pins to ground. Recommended values
for C1 and C2 is 10 pF.
Figure 3-D X2/CLKIN and X1 with Crystal Oscillator
C1 = 10 pF
X2/CLKIN (pin 68)
16.384 MHz
AMBE-2000
X1 (pin 67)
C2 = 10 pF
page 15
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
3.5
Package Description
100 pin TQFP (Thin Quad Flat Pack)
All Dimensions are in millimeters
Figure 3-E Package Dimensions
16.20 / 15.80 SQ
14.20 / 13.80 SQ
12.00 TYP
75
51
76
50
100
26
25
1.40/1.60 mm
1
o
12 All Around
0.5 mm
Detail
0.17 / 0.27
Gage Plane
0.13NOM
1.35 /1.45
1.60 MAX
0.05
0.25
0 - 7 MAX
0.45 / 0.75
Not Drawn to Scale
page 16
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
3.6
Normal Operating Conditions
Table 3-C Normal Operating Conditions
Normal Operating Conditions
3.7
Operating Voltage
3.3V
Operating Case Temperature Range
-40°C to 100°C
Storage Temperature Range
-55°C to 150°C
Absolute Maximum Ratings
Stresses in excess of the Absolute Maximum Ratings can cause permanent damage to
the device. These are absolute stress ratings only. Functional operation of the device is
not implied at these or any other conditions in excess of those given in the operational
sections of the data sheet. Exposure to Absolute Maximum Ratings for extended periods
can aversely affect device reliability.
Table 3-D Absolute Maximum Ratings
Absolute Maximum Ratings
Voltage Range on any Pin with Respect to Ground
3.8
-0.3V to 4.6V
Electrical Characteristics and Requirements
Table 3-E Recommended Operating Conditions
Parameter
Min
Nom
Max
Unit
DVDD
Device Supply Voltage
3
3.3
3.6
V
VSS
Supply Voltage, GND
-
0
-
V
VIH
High-level input voltage, I/O
2.5
-
DVDD + 0.3
V
VIL
Low-level input voltage
-0.3
-
0.8
V
IOH
High-level output current
-
-
-300
µA
IOL
Low-level output current
-
-
1.5
mA
page 17
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
Table 3-F Electrical Characteristics over Recommended Operating Case
Temperature Range (Unless Otherwise Noted)
Parameter
Test Conditions
VOH
High-level output voltage‡
VDD = 3.3 ±V, IOH = MAX
VOL
Low-level output voltage‡
IOL = MAX
II
Input current in high impedance
(VI = VSS to VDD)
VDD = MAX, VI = VSS to VDD
MIN
TYP†
MAX
2.4
Unit
V
-10
0.4
V
10
µA
CI
Input capacitance
10
pF
CO
Output capacitance
10
pF
† All values are typical unless otherwise specified.
page 18
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
4
4.1
Channel Interface
Overview
The Channel Interface is the general term used for the interface for the compressed bits coming from the encoder and the
compressed bits going to the decoder. This same interface is also used to output status information from the encoder and
decoder such as whether a DTMF tone has just been detected in the speech input, or whether the decoder has detected and
synthesized a frame of silence. Additionally, this interface is used to perform more complex control operations on both the
encoder and decoder (usually at start-up). These control functions include speech data and FEC rate control as well as enabling
features such as echo cancellation.
It is important to realize that not all data being output from the AMBE-2000 is intended for transmission over the channel.
Status type of data is typically only useful at the ‘local’ end. In most voice transmission systems, the actual encoded bits are
extracted from the channel formatting, combined into the systems transmission stream, sent over the transmission path,
extracted from the transmission path at the receiving end, and reassembled into the AMBE-2000’s channel format for synthesis
by the decoder.
Figure 4-A Channel Interface Overview
Typical Voice Frame Output From Encoder
Overhead Data
Voice Data (48 to 192 bits)
(Header, Status)
Transmission Channel
System extracts
relevant Voice Data bits
and formats them for
transmission over
Channel
System Overhead
Voice Data (48 to 192 bits)
System extracts
relevant Voice Data bits
and formats them for
Input into the Decoder
adding Header and
Control Information
Typical Voice Frame Input to the Decoder
Overhead Data
Voice Data (48 to 192 bits)
(Header, Control)
4.2
Serial Configuration Selection
The hardware interface to the Channel Interface is configured as a serial interface based exclusively on the hardware settings of
CHANN_SEL[1-0]. See Table 4-A.
page 19
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
Table 4-A Channel Interface Selection Table
Port Type
CH_SEL1
(pin 77)
CH_SEL0
(pin 75)
Interface
Select Pins
Active Formatted
0
0
Active
Unformatted
0
1
Passive
Formatted
1
0
Passive
Unformatted
1
1
Table 4-B Unformatted Bit per Word Selection Table
BAUD_SEL0
(pin 80)
Number
of Voice
Data
Bits per
Word
BAUD_SEL1
(pin 81)
Interface
Select Pins
1
0
0
2
0
1
3
1
0
4
1
1
All transfers occur through a serial port. The serial port inputs and outputs a 16 bit word for every write and read strobe signal
respectively. Serial mode can be formatted or unformatted. Within the unformatted mode, the data is input and output in
16 bits words still but with only 1 to 4 voice data bits carried within each word. These four configurations can be seen in Table
4-B. See section 4.3 for all the details on the serial interface.
page 20
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
4.3
Channel Serial Mode
The signals in Table 4-C make up the serial channel interface. The serial channel mode transfers data in and out of the AMBE2000™ using 16 bit words on the two data lines CHAN_RX_DATA and CHAN_TX_DATA. The selection of the
formatted or unformatted format of this data is made using information in Table 4-A.
Table 4-C Channel Serial Interface Pin Descriptions
Pin
Symbol
Pin
Direction
Pin
Number
Description
EPR
Out
20
Encoder Packet Ready: Following a reset, this output signal will have a
high to low transition to indicate that the encoder has a frame of data to
output. The next packet will be ready approximately 20 msec later. See
Note 1.
CHAN_RX_DATA
In
32
Serial Data Input: 16 bits of channel data are input on CHAN_RX_DATA,
synchronous to CHAN_RX_CLK, with each CHAN_RX_STRB pulse.
CHAN_RX_CLK
In
28
Serial
Input
Clock:
In
coordination
with
CHAN_RX_STRB,
CHAN_RX_DATA is latched by the AMBE-2000™ on the falling edges of
CHAN_RX_CLK.
CHAN_RX_STRB
In
30
Input (Write) Data Strobe: This signal indicates to the AMBE-2000™
when the data on CHAN_RX_DATA will be latched by CHAN_RX_CLK.
See Figure 4-B.
CHAN_TX_DATA
Out
42
Serial Data Output: 16 bits
CHAN_TX_DATA,
synchronous
CHAN_TX_STRB pulse.
CHAN_TX_CLK
In
34
Serial Output Clock: In coordination with CHAN_TX_STRB, the data on
CHAN_TX_DATA is output by the AMBE-2000™ on the falling edges of
CHAN_TX_CLK.
CHAN_TX_STRB
Active
Output/
Passive
Input
38
Output (Read) Data Strobe: This signal indicates to the AMBE-2000™
when to bring the data to the CHAN_TX_DATA pin. See Figure 4-B.
of
to
channel data are output on
CHAN_TX_CLK,
with
each
Note 1:The AMBE-2000 expects an encoder packet to be read approximately every 20 msec. Following the initial
reset, wait for EPR to go low and read the initial packet (t0). 20 msec later, the next packet (t1) should be ready.
For packet t1 and all following packets use the procedure below:
1) Wait for slightly less than 20 msec.
2) Assert CHAN_TX_STB and read word on CHAN_TX_DATA.
3) If transmitted word not 0x13EC, discard it and repeat step 2.
4) If transmitted word 0x13EC, read 23 more words (rest of packet).
A new packet should be ready every 20 msec after the initial EPR high to low transition. A packet read should take place
every 20 msec. If there is a delay in the read (i.e. a packet is missed), it is recommended that the device be reset.
page 21
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
4.3.1 Low Level Timing for Passive and Active Serial Mode
Figure 4-B Low Level Timing for Passive and Active Serial Mode
tc(SCK)
tf(SCK)
~
tw(SCK)
CHAN_RX_CLK
th(FSR)
tsu(FSR)
tw(SCK)
tr(SCK)
tsu(DR)
~
CHAN_RX_STRB
CHAN_RX_DATA
MSB
1
~ ~
th(DR)
2
LSB
8/16
7/15
tc(SCK)
tf(SCK)
~
tw(SCK)
CHAN_TX_CLK
td(FSX)
th(FSX)
th(FSX)H
td(DX)
tdis(DX)
~
CHAN_TX_STRB
tr(SCK)
tw(SCK)
CHAN_TX_DATA
MSB
1
2
~ ~
th(DX)
7/15
LSB
8/16
Table 4-D Switching Characteristics Over Recommended Operating Conditions
for Serial Port Receive
Reference
th(FSR)
th(DR)
Parameter
Hold time, FSR after CLKR falling edge
3.3 Volts
Min
Max
6
Hold time, DR after CLKR falling edge
6
page 22
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
Units
ns
ns
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
Table 4-E Switching Characteristics Over Recommended Operating Conditions
for Serial Port Receive
Reference
td(SCK)
tc(SCK)
tf(SCK)
tr(SCK)
tw(SCK)
tsu(FSR)
tsu(DR)
td(FSX)
td(DX)
th(FSX)
th(FSX)H
tdis(DX)
th(DX)
3.3 Volts
Min
Max
Parameter
Delay time, DX valid after CLKX rising
(Passive Mode)
Cycle time, serial port clock
25
6H
Units
ns
ns
Fall time, serial port clock
6
Rise time, serial port clock
6
ns
ns
Pulse duration, serial port clock low/ high
3H
ns
Setup time, FSR before CLKR falling edge
6
ns
Setup time, DR before CLKR falling edge
6
Delay time, CLKX rising to FSX
Delay time, CLKX rising to DX (Active
Mode)
Hold time, FSX after CLKX falling edge
ns
15
ns
6
Hold time, FSX after CLKX rising edge
Disable time, CLKX rising to DX
Hold time, DX valid after CLKX rising
edge
ns
15
ns
2H-5
ns
20
ns
-5
ns
•
Note: H = 7.629 ns; however, do not operate serial clocks any faster than 2.048 MHz. Thus tc(SCK) should be a
minimum of 488.3 nsec.
page 23
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
5
Channel Data Format
The channel interface is responsible for outputting the compressed data from the encoder and inputting compressed data to the
decoder. In addition to these most basic functions the channel interface is also capable of reporting certain events, such as the
detection of a DTMF tone. The channel interface can also control certain selectable functions of the AMBE-2000™, such as
the voice coding rate. This chapter will describe how the AMBE-2000™ uses the channel interface to multiplex these
capabilities.
There are two formats to the data, Formatted and Unformatted, both of which operate in serial mode. Generally speaking the
Unformatted mode is used only when the connection between the AMBE-2000™ and the channel under design is relatively
direct, and the designer wants to simplify the extraction of the relevant voice data. In this mode configuration is accomplished
using hardwired pins. In most cases, when a controller is present between the AMBE-2000™ and the channel, the system
designer will find that using the Formatted format is more flexible.
5.1
Formatted Format
The Formatted format is a 24 by sixteen-bit word format for a total of 48 bytes or 384 bits. Every 20 milliseconds the encoder
outputs 24 sixteen-bit words, and likewise the decoder expects to receive 24 words. The format of the input and output frames
are detailed below. The first 12 sixteen bit words are made up of header, ID and status or control information. The remaining
12 sixteen bit words make up the encoded data bit field. These 12 words, or 192 bits, will be fully populated with relevant
voice data only when the AMBE-2000 is operating in a 9600bps mode (9600 bits/sec ÷ 50 frames/sec = 192 bits/frame).
Otherwise, when the data rate is less than 9600bps, the coded voice bits are filled starting from the MSB of the first word in the
field, leaving any unused bits as zeros. It is important to note here that even when the AMBE-2000 is operating at less than
9600bps, all 384 bits of the Formatted format (including any unused trailing zeros) must be transferred out of the encoder and
into the decoder.
5.2
Formatted Input Format
The format of the Formatted input is shown in Figure 5-A. Keep in mind that even though the channel data in this Formatted
input format is closely associated with the decoder, the control information will apply to both encoder and decoder functions.
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
(12) 16 bit words of
overhead (192 bits)
(12) 16 bit words of data
(192 bits)
24 sixteen-bit words = 48 bytes = 384 bits
20 ms frame
Figure 5-A Basic Formatted Input Format
Word #
Description
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Header always set to 0x13EC
Power Control (8bits)
Control Word 1 (8 bits) – see Table 5-B
Rate info 0
Rate info 1
See Tables 5-C and 5-D
Rate info 2
Rate info 3
Rate info 4
Unused in Input
Unused in Input
Unused in Input
DTMF Control – see Tables 5-E and 5-F
Control Word 2 – see Table 5-G
Channel Data
Channel Data
Channel Data
Channel Data
Channel Data
Channel Data
Channel Data
Channel Data
Channel Data
Channel Data
Channel Data
Channel Data
5.2.1 Formatted Input : Word 0 : Header
The decoder uses the header information to synchronize with the beginning of each 20 millisecond frame. this 16 bit word
MUST be 0x13EC.
5.2.2 Formatted Input : Word 1 : Power Control
Set the encoder to 0x00 in the 8-bit Power Control field of an output frame for normal use. For Power Down Mode, set this
value to 0x55. This causes the AMBE-2000 to enter low power mode. To exit low power mode, the device must be reset
through the hardware.
Table 5-A Formatted Input : ID Values Summary
ID
Type
Description
0x00
Voice Data
0x55
Low Power Mode
This ID value instructs AMBE-2000™ to operate in normal fashion
When this mode is activated the AMBE-2000™ Vocoder Chip will go into
a mode which conserves power, where no voice packets are being
processed.
page 25
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
5.2.3 Formatted Input : Word 1 : Control Word 1
Use the 8-bit control word to set various functions.
Table 5-B Control Word 1 Format
7: MSB
Lost
Frame
Indicator
6
Unused
Control Word 1 – 8-bits (See Table 1-F)
5
4
3
2
Unused
Unused
Unused
Unused
1
0 : LSB
CNI
Unused
Lost Frame Indicator : Setting the Lost Frame Indicator bit to a 1 will cause the AMBE-2000™ decoder to construct the
voice frame using the parameters from the previous frame. This is an effective way to mask the effects of short periods of data
loss. This bit should be set by the user when channel data is lost or corrupted. It works by replacing the frame of corrupted
data with the previous, intact, frame.
Comfort Noise Insertion (CNI) : Setting the CNI bit will cause the decoder to output a frame of comfort noise. This bit is
used with systems that are capable of discontinuous transmission (DTX).
5.2.4 Formatted Input : Words 2-6 : Rate Info 0, Rate Info 1, Rate Info 2, Rate Info 3,
Rate Info 4
The initial rate of the AMBE-2000 is set through the hardware pins RATE_SEL[4-0] (see Section 7.2 and Tables 7-A and 7-B)
after resetting the device. The coding rate can then be modified for both the encoder and the decoder through the Unformatted
channel interface. The Rate Selector control field (describe in section) determines where to apply the rate change.
The AMBE-2000 used these five words to set the source and FEC coding rates. Tables 5-C and 5-D list predefined value for
various source and FEC rates. The software configurations in Table 4-C are compatible for use with the AMBE-1000™. If
compatibility is not an issue, use the software codes in Table 4-D to select speech and FEC rates to optimize use of the AMBE2000™.
Table 5-C Rate Selection Using Rate Info 0-4, compatible w/ AMBE-1000™
Rate Info 0
Rate Info 1
Rate Info 2
Rate Info 3
Rate Info 4
0x9030
0x902f
0x9348
0x9243
0xab50
0x934b
0xab60
0xab5b
0x9348
0x923e
0xab53
0xab58
0xbf9b
0xab5d
0xbfc0
0xab16
0x0000
0x0000
0x0000
0x0080
0x0000
0x0080
0x0000
0x0080
0x2030
0x2800
0x2c00
0x3000
0x0080
0x3400
0x0000
0xe400
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
0x4330
0x6930
0x6f48
0x5348
0x3950
0x3950
0x7960
0x6860
0x7060
0x7460
0x5680
0x4490
0x49a0
0x31a0
0x72c0
0x67c0
Speech Rate
(bps)
2400
2350
3600
3350
4000
3750
4800
4550
3600
3100
4150
4400
7750
4650
9600
4850
page 26
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
FEC Rate
(bps)
0
50
0
250
0
250
0
250
1200
1700
2250
2800
250
3350
0
4750
Total Rate
(bps)
2400
3600
4000
4800
6400
7200
8000
9600
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
Table 5-D Rate Selection Using Rate Info 0-4, AMBE-2000™ only
Rate Info 0
Rate Info 1
Rate Info 2
Rate Info 3
Rate Info 4
Speech Rate
(bps)
FEC Rate
(bps)
Total Rate
(bps)
0x0028
0x5048
0x5250
0x1030
0x5360
0x5250
0x5048
0x1030
0x6b80
0x5250
0x5258
0x7fa0
0x5250
0x7fc0
0x5048
0x1030
0x0000
0x0000
0x0000
0x0001
0x0000
0x2010
0x0001
0x0005
0x0000
0x0001
0x0009
0x0000
0x0005
0x0000
0x000e
0x000e
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
0x180c
0x0000
0x0000
0x1e0c
0x0000
0x2010
0x0000
0x4010
0x681a
0x0000
0x0000
0x0000
0x341a
0x0000
0x0000
0x2412
0x3018
0x0000
0x542a
0x4127
0x0000
0x6834
0x0000
0x6a2e
0x511b
0x6428
0x3948
0x4150
0x6750
0x6c60
0x7460
0x6860
0x7360
0x6c80
0x5280
0x7390
0x52a0
0x72a0
0x69c0
0x65c0
0x76c0
2000
3600
4000
2400
4800
4000
3600
2400
6400
4000
4400
8000
4000
9600
3600
2400
0
0
0
1600
0
800
1200
2400
0
2400
2800
0
4000
0
6000
7200
2000
3600
4000
4800
6400
7200
8000
9600
5.2.5 Formatted Input : Word 7 : Unused in Input
Should be set to 0x0000
5.2.6 Formatted Input : Word 8 : Unused in Input
Should be set to 0x0000
5.2.7 Formatted Input : Word 9 : Unused in Input
Should be set to 0x0000
5.2.8 Formatted Input : Word 10 :DTMF Control
Use this word to set DTMF tones. See Table 5-F for a list of tones and their corresponding values.
Table 5-E DTMF Control Format
DTMF Control – 16-bits
15:
MSB
14
13
12
11
10
9
8
7
DTMF Amplitude
6
5
4
3
2
DTMF Digit Detect/ Generate
page 27
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
1
0:
LSB
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
Table 5-F DTMF Codes for Digit Detect/ Generate
DTMF Code
0x80
0x84
0x88
0x81
0x85
0x89
0x82
0x86
0x8A
0x87
0x83
0x8B
0x8C
0x8D
0x8E
0x8F
0xff
DTMF Digit
1
2
3
4
5
6
7
8
9
0
*
#
A
B
C
D
deactivate
To deactivate the Digit Detect/Genetrate function, set the DTMF Code to 0xff. Dial, ring, and busy tones are standard North
American call progress tones.
DTMF Amplitude
The DTMF Amplitude runs from 3 to –60 dBm0. This value is a signed bite (example: 0x03 = 3, 0x00 = 0, 0xC4 = -60).
5.2.9 Formatted Input : Word 11 Control Word 2
Table 5-G Control Word 2 Format
Control_Word 2 – 16-bits (See Table 1-F)
Bit 15
14
13
12
11
10
Decoder Output Volume Control
9
8
7
6
Unused Unused
0
0
5
4
3
2
VAD
Unused
0
SL
EC
Rate Information Selector (RIS) :
Use these 2 bits to select which part(s) of the vocoder will be affected by the rate selection words.
Table 5-H Rate Information Selection Codes
value
0x0
0x1
0x2
0x3
Area Controlled
Encoder and Decoder
Encoder only
Decoder only
Neither Encoder nor Decoder
page 28
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
1
0
RIS
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
Echo Canceller (EC) :
Set this bit to 1 to enable the echo cancel function on a frame by frame basis. This must be set to 1 in every packet sent to
continue use.
Sleep (SL) :
Set this bit to 1 to enter Sleep mode on a frame by frame basis. Sleep is a low power mode, not to be confused with Power
Down Mode. The sleep function must be enabled in every packet to continue in sleep mode. Set this bit to 0 to exit Sleep
mode.
VAD :
In order to set the Voice Activity Detector on, set the VAD bit to 1. To disable Voice Activity Detection, set the VAD bit to 0.
5.2.10 Formatted Input : Words 12-23 : Channel Data
This is the field that contains the actual coded bits. Input of the data begins with the MSB of the first word in this field and
continues through with the final bit output being the LSB of the final word. If the data rate selected is less than 9600bps then
the unused bits in each frame are zero and populate the end of the field. As is noted in the Channel Interface definitions, these
unused bits must still be clocked out of the AMBE-2000™. The packet must always consist of 24 words.
5.3
Formatted Output Format
The format for Formatted output data is shown in Figure 5-B. It is only the bits in the Voice Data Bits field, which are
transmitted along with framing information (data used to locate the start of each frame for proper reconstruction at the decoder)
over the channel. The first 192 bits provide overhead information, which is sometimes useful to the host but is generally not
transmitted over the channel.
(12) 16 bit words of
overhead (192 bits)
(12) 16 bit words of data
(192 bits)
24 sixteen-bit words = 48 bytes = 384 bits
20 ms frame
Figure 5-B Basic Formatted Output Format
Word #
Description
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Header always set to 0x13ec
Power Control (8bits)
Control Word 1 (8 bits) – see Table 5-J
Rate info 0
Rate info 1
Rate info 2
Rate info 3
Rate info 4
Bit Error Rate
Unused
Unused
DTMF Control – see Table 5-L
Control Word 2 – see Table 5-N
Channel Data
Channel Data
Channel Data
Channel Data
Channel Data
Channel Data
Channel Data
Channel Data
Channel Data
Channel Data
Channel Data
Channel Data
page 29
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
5.3.1 Formatted Output : Word 0 : Header
The header is a 16 bit word that begins each valid frame corresponding to 20 milliseconds of speech. This field will always be
0x13EC.
5.3.2 Formatted Output : Word 1 : ID (Power Control)
The encoder will always use 0x00 in the 8-bit field of an output frame.
5.3.3 Formatted Output : Word 1 :Control Word 1
This 8-bit control word indicates the activity of various functions.
Table 5-J Control Word 1 Format
7: MSB
Unused
6
Unused
Control Word 1 – 8-bits (See Table 1-A)
5
4
3
2
Decoder Decoder
Frame
Silence
Unused
Unused
Repeat
Detect
1
Encoder
Silence
Detect
0 : LSB
Encoder
DTMF
Detect
Decoder Frame Repeat: When the Decoder Frame Repeat flag is set to 1, the decoder is reporting that the last frame decoded
was a repeat of the previous frame.
Decoder Silence Detect: When the Decoder Silence Detect flag is set to 1, the decoder is reporting that the last frame decoded
was a comfort noise frame.
Encoder DTMF Detect: The Encoder DTMF Detected Flag will be set to a 1 when the encoder detects a DTMF tone. The
DTMF Detect option is controlled by the DTMF Detect/Generate bits in control Word 10 described in section 5.2.8 enable the
DTMF Detect option.
Silence Detect: The Encoder Silence Detected Flag will be set to 1 when no voice activity is detected. The Silence Detect
option is controlled by and can be disabled by the VAD as described in section 5.2.9.
5.3.4 Formatted Output : Words 2-6 : Rate Info 0, Rate Info 1, Rate Info 2, Rate Info 3,
Rate Info 4.
Words 2-6 in the packet indicate the rate at which the AMBE-2000™ encoder is operating . These words are output. See
tables 5-C and 5-D for corresponding values.
5.3.5 Formatted Output: Word 7 : Bit Error Rate
This status field is used for the decoder to report bit error information. The 16 bit number output is used to compute the BER
using the following calculation
% BER = (Word 7) / (32768)
5.3.6 Formatted Output : Word 10 : DTMF Control
This word corresponds to the DTMF Detection capabilities of the vocoder. It uniquely identifies specific tones recognized by
the encoder. See table 5-J for a list of tones and their corresponding values.
page 30
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
Table 5-K DTMF Tone Detection Parameters
DTMF Tone
Detection
Requirement
Value
Minimum
Input Level
-25 dBm0
An input signal shall not be rejected as a DTMF tone if its amplitude is greater
than -25 dBm0 (maximum sinusoid dBm0 is defined as+3.17 dBm0).
Minimum
Signal to Noise
Distortion ratio
15 dB
In order for an input signal to correspond to a valid DTMF tone, the ratio of
inband to out-of-band energy must be greater than 15dB. Inband energy is
defined to be the energy in frequency components within ±3.5% of the two
frequencies defined by the DTMF frequencies. Out-of-band energy is defined to
be the total energy minus in the inband energy.
Minimum
Frequency
Tolerance
±1.5%
An input signal shall not be rejected as a DTMF tone if both of its principal
frequency components are within ±1.5% of the frequencies needed for the DTMF
tone.
Maximum
Frequency
Tolerance
±3.5%
An input signal shall not be rejected as a DTMF tone if either of its principal
frequency components are outside ±3.5% of the frequencies needed for the DTMF
tone.
8-10 dB
An input signal does not correspond to a valid DTMF tone if the energy contained
within the low frequency band is more than 10 dB greater than the energy
contained in the high frequency band. An input signal shall not be rejected as a
DTMF if energy contained within the low frequency band is less than 8 dB greater
than the energy contained in the high frequency band. Each low and high
frequency band is limited to ±3.5% of the frequencies needed for the DTMF tone.
4-10 dB
An input signal does not correspond to a valid DTMF tone if the energy contained
within the high frequency band is more than 10 dB greater than the energy
contained in the low frequency band. An input signal shall not be rejected as a
DTMF if energy contained within the high frequency band is less than 4 dB
greater than the energy contained in the low frequency band. Each low and high
frequency band is limited to ±3.5% of the frequencies needed for the DTMF tone.
45 mS
An input signal shall not be rejected as a DTMF tone as long as its time duration
is greater than 45 mS. In addition a minimum of two frames will be transmitted
of the DTMF tone if a valid tone is detected. The duration of a tone is defined by
the points at which the envelope is 20 dB below its peak value.
Normal Twist
Range
Reverse Twist
Range
Minimum
Tone Duration
Description
Table 5-L DTMF Control Format
DTMF Control – 8-bits (See Table 1-F)
15:
MSB
14
13
12
11
10
9
8
7
DTMF Amplitude
6
5
4
3
2
DTMF Digit Detect/ Generate
page 31
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
1
0:
LSB
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
Table 5-M DTMF Codes for Digit Detect/ Generate
DTMF Code
0x80
0x84
0x88
0x81
0x85
0x89
0x82
0x86
0x8A
0x87
0x83
0x8B
0x8C
0x8D
0x8E
0x8F
0xff
DTMF Digit
1
2
3
4
5
6
7
8
9
0
*
#
A
B
C
D
deactivate
To deactivate the Digit Detect/Genetrate function, set the DTMF Code to 0xff. Dial, ring and busy tones are standard North
American call progress tones.
DTMF Amplitude
The DTMF Amplitude runs from 3 to –60 dBm0. This value is a signed bite (example: 0x03 = 3, 0x00 = 0, 0xC4 = -60).
5.3.7 Formatted Output : Word 11 :Control Word 2
Table 5-N Control Word 2 Format
Control_Word 2 – 16-bits (See Table 1-A)
Bit 15
14
13
12
11
10
Decoder Output Volume Control
9
8
7
6
5
4
Unused Unused Unused Unused
0
0
0
0
3
2
SL
EC
Rate Information Selector (RIS) :
This field is used to indicate which part of the device is affected by words 2-6, the rate control words.
Table 5-O Rate Information Selection Codes
value
0x0
0x1
0x2
0x3
Area Controlled
Encoder and Decoder
Encoder only
Decoder only
Neither Encoder nor Decoder
page 32
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
1
0
RIS
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
Echo Canceller (EC) :
A 1 in this output field indicates that the echo canceller has been enabled.
Sleep (SL) :
A 1 in the this field indicates the device has been put into sleep mode.
Decoder Output Volume Control : Indicates the current decoder volume.
5.3.8
Formatted Output : Words 12-23 :Channel Data
This is the field that contains the actual coded bits. Output of the data begins with the MSB of the first word in this field and
continues through with the final bit output being the LSB of the final word. If the data rate selected is less than 9600bps then
the unused bits in each frame are zero and populate the end of the field. As is noted in the Channel Interface definitions, these
unused bits must still be clocked out of the AMBE-2000™. The packet must always consist of 24 words.
5.4
Unformatted Serial Format
The Unformatted Format for the channel data is useful for applications which desire minimal glue logic between the AMBE2000™ and the channel hardware. The use of minimal hardware in place of a microcontroller can be realized using this data
format. Another distinct difference in this data format is that framing information (data which carries the positional
information relating to the coded bits) is embedded into the data stream itself. Using this data format, the system designer need
only transfer the coded data itself. A single bit each frame is ‘borrowed’ from the voice data to embed the framing information.
Keep in mind that this ‘borrowed’ bit reduces the effective voice coding rate quality by 50 bits per second. For example, a
system with no FEC running at 2450 bps in Unformatted mode will sound equivalent to one running at 2400 bps in Formatted
mode.
The designer should also be aware that it takes approximately 15 frames (300 milliseconds) for the decoder to attain
synchronization with the incoming stream before it can output synthesized speech. Systems which are attempting to save
power by shutting down transmission during periods of silence, and then resuming during periods of speech can not handle this
300 millisecond delay for each synchronization, and thus should use Formatted mode with a more sophisticated framing
method.
The 16 bit per word format, pictured in Section 5.2, is maintained in this mode but only a fraction of the full 16 bits is used to
transfer the coded data. The user selects whether 1, 2, 3 or 4 bits will be transferred in each word based on pins
BAUD_SEL[0:1] and Table 4-A. IMPORTANT : The voice coding data rate selected must be evenly divisible by the number
of voice data bits per word selected.
5.4.1 Unformatted Serial Output Format
The Unformatted output format contains 1 to 4 bits within each 16 bits serial output word. The formats which contain more
than one bit each word the MSB of the data bits is considered first in the transmission. In Unformatted mode, only the coded
voice data bits are output. None of the superfluous information that exists in formatted mode is available in this mode. The
number of words that need to be transferred out of the encoder for each 20 millisecond frame will be the number of bits per
frame divided by the number of bits per word. So a system coding at 4800 bps with 3 bits per word will need to read 32 (
[4800 ÷50] ÷3 = 32) words each frame.
Table 5-P Unformatted Serial Output Data Format
Bits per Word
See Table 4-A
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
1 bit per Word Format
2 bits per Word Format
3 bits per Word Format
4 bits per Word Format
D msb
D msb
D msb
D msb
0
D
D
D
0
0
D
D
0
0
0
D
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Data
Unused
page 33
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
5.4.2 Unformatted Serial Input Format
The Unformatted Input format contains 1 to 4 bits within each 16 bits serial output word. For the formats which contain more
than one bit each word the MSB of the data bits is considered first in the transmission. In Unformatted mode, the header data
from Formatted mode is dropped and each 16 bit write contains 1 to 4 coded voice data bits. The number of words that need to
be transferred into the decoder for each 20 millisecond frame will be the number of bits per frame divided by the number of bits
per word. So a system coding at 4800 bps with 3 bits per word will need to write exactly 32 ( [4800 ÷50] ÷3 = 32) words each
frame.
Table 5-Q Unformatted Serial Input Data Format
Bits per Word
See Table 4-A
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
1 bit per Word Format
2 bits per Word Format
3 bits per Word Format
4 bits per Word Format
D msb
D msb
D msb
D msb
0
D
D
D
0
0
D
D
0
0
0
D
0
0
0
0
0
0
0
0
0
0
0
0
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Data
Control Offset
Control Data
page 34
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
6
6.1
A/D-D-A Interface
A/D-D/A Overview
The interface from the analog world of speech and the AMBE-2000™ is typically an A/D-D/A chip. Selection of the A/D-D/A
chip should be made carefully, with a preference given to 16 bit linear devices. Additionally, consideration should be given for
signal to noise ratios and filtering characteristics typically built into many such devices. Generally speaking, the flatter the
frequency response over the voice spectrum (20-4000Hz) the better the overall system will sound.
The AMBE-2000™ Vocoder Chip operates with a speech data sample rate of 8kHz for both the A/D and D/A interfaces. This
8kHz data is input and output using a serial port on the AMBE-2000.
In order to simplify the process of configuring the interface to the A/D-D/A chip, a number of preset configurations can be
chosen through the CODEC_SEL[1-0] pins shown in Table 6-A. These preset configurations control signal directions for the
interface as well as the sequence of programming words for the programmable devices, specifically the AD7331AR.
6.2
Configuring the A/D-D/A Interface using CODEC_SEL[1-0]
In order to simplify the process of configuring the A/D-D/A interface certain preset configurations are available to the user.
Selection of these preset modes is made through the 3 hardware pins CODEC_SEL[1-0]. In Table 6-A, the 2 digit binary value
for CODEC_SEL[1-0] corresponds to the levels present on the hardware pins, with a 0 corresponding to GND, and a 1
corresponding to VCC.
Table 6-A CODEC_SEL[1-0] : A/D-D/A Hardware Configuration Values
CODEC_SEL[1-0]
pins
85,84
A/D-D/A Type
Generic 16 bit Linear
8Khz
Analog Devices 7331
32kHz
Generic µlaw 8kHz
Generic Alaw 8kHz
00b
01b
10b
11b
page 35
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
6.3
Low Level A/D–D/A Timing
Low Level Timing for A/D-D/A in Active and Passive Modes
Figure 6-A Low Level Timing for A/D-D/A in Active and Passive Modes
tc(SCK)
tf(SCK)
~
tw(SCK)
CODEC_RX_CLK
th(FSR)
tsu(FSR)
tw(SCK)
tr(SCK)
tsu(DR)
~
CODEC_RX_STRB
CODEC_RX_DATA
MSB
1
~ ~
th(DR)
2
LSB
8/16
7/15
tc(SCK)
tf(SCK)
~
tw(SCK)
CODEC_TX_CLK
td(FSX)
th(FSX)
th(FSX)H
td(DX)
tdis(DX)
~
CODEC_TX_STRB
tr(SCK)
tw(SCK)
CODEC_TX_DATA
MSB
1
2
~ ~
th(DX)
7/15
LSB
8/16
Table 6-B Switching Characteristics Over Recommended Operating Conditions
for Serial Port Receive
Reference
th(FSR)
th(DR)
Parameter
Hold time, FSR after CLKR falling edge
3.3 Volts
Min
Max
6
Hold time, DR after CLKR falling edge
6
page 36
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
Units
ns
ns
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
Table 6-E Switching Characteristics Over Recommended Operating Conditions
for Serial Port Receive
Reference
td(SCK)
tc(SCK)
tf(SCK)
tr(SCK)
tw(SCK)
tsu(FSR)
tsu(DR)
td(FSX)
th(FSX)
th(FSX)H
tdis(DX)
th(DX)
Delay time, DX valid after CLKX rising
3.3 Volts
Min
Max
25
Cycle time, serial port clock
6H
Parameter
Units
ns
ns
Fall time, serial port clock
6
Rise time, serial port clock
6
ns
ns
Pulse duration, serial port clock low/ high
3H
ns
Setup time, FSR before CLKR falling edge
6
ns
Setup time, DR before CLKR falling edge
6
Delay time, CLKX rising to FSX
Hold time, FSX after CLKX falling edge
6
Hold time, FSX after CLKX rising edge
Disable time, CLKX rising to DX
Hold time, DX valid after CLKX rising
edge
ns
15
ns
ns
2H-5
ns
20
ns
-5
ns
•
Note: H = 7.629 ns; however, do not operate serial clocks any faster than 2.048 MHz. Thus tc(SCK) should be a
minimum of 488.3 nsec.
page 37
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
7
7.1
Special Functions
Hardware vs. Software Selection Note
Many of the functions of the AMBE-2000™ can be accessed through both a hardware and software interfaces to the device.
The following hardware inputs, CHANN_SEL[1-0], RATE_SEL[4-0], CODEC_SEL[1-0], VAD_EN, ECHOCAN_EN, and
SLEEP_EN, are only accessed for input during the first 200 microseconds after a hardware reset on RESETN. For predictable
operation these signals must remain stable over this time period. After this initialization period the functions that these pins
access can only be reconfigured through the Control Word interface described in section 5.3.4. Changes on these pins after the
200 microseconds initialization period after reset are ignored, unless another reset is performed.
7.2
Coding Rate Selection
The Voice coding rate as well as the FEC coding rate can be selected individually on the AMBE-2000™. These rates are
selected by using a Control Word as described in section 5.3.4, or through hardware pins RATE_SEL[4-0] subject to the
restrictions in section 7.1. The five input pins RATE_SEL[4-0] give 16 preconfigured voice/FEC rates. The voice and FEC
rates are individually configurable in 50 bit per second intervals. If rates other than these are desired then the Control Word
method of configuring the rates must be used.
Table 7-A Rate Selection Using Rate Info 0-4, compatible w/ AMBE-1000™
RATE_SEL4
Pin
RATE_SEL3
Pin
RATE_SEL2
Pin
RATE_SEL1
Pin
RATE_SEL0
Pin
Speech Rate
(bps)
FEC Rate
(bps)
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
1
1
0
0
0
1
1
1
1
1
0
0
0
1
0
0
1
1
0
1
0
0
0
0
1
1
1
1
0
0
0
1
1
1
1
1
1
0
1
0
0
0
0
1
0
1
1
1
1
0
1
1
0
0
0
1
0
1
0
0
2400
2350
3600
3350
4000
3750
4800
4550
3600
3100
4150
4400
7750
4650
9600
4850
0
50
0
250
0
250
0
250
1200
1700
2250
2800
250
3350
0
4750
page 38
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
Total Rate
(bps)
2400
3600
4000
4800
6400
7200
8000
9600
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
Table 7-B Rate Selection Using Rate Info 0-4, AMBE-2000™ only
7.3
RATE_SEL4
Pin
RATE_SEL3
Pin
RATE_SEL2
Pin
RATE_SEL1
Pin
RATE_SEL0
Pin
Speech Rate
(bps)
FEC Rate
(bps)
Total Rate
(bps)
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
0
0
0
0
1
0
1
0
1
1
0
1
0
1
1
1
0
0
1
0
0
1
0
0
0
0
1
1
1
1
1
1
0
0
1
1
0
1
0
1
1
1
0
0
0
0
1
1
0
1
0
0
0
1
1
1
0
1
0
0
1
1
0
2000
3600
4000
2400
4800
4000
3600
2400
6400
4000
4400
8000
4000
9600
3600
2400
0
0
0
1600
0
800
1200
2400
0
2400
2800
0
4000
0
6000
7200
2000
3600
4000
4800
6400
7200
8000
9600
Echo Cancellation
The AMBE-2000™ Vocoder Chip provides a 16 millisecond echo canceller that is suitable for canceling the local echo caused
by a 2-to-4 wire hybrid and can achieve echo cancellation of approximately 30dB or more. Only the linear portion of the echo
is cancelled, so circuits should be designed to minimize non-linearities.
The AMBE-2000™ Vocoder Chip echo canceller operates by sending a 240 millisecond audible training sequence to the A/DD/A immediately following a reset. Best results will be achieved if the analog circuit causing any echo is stable at this time. If
the analog circuit changes substantially following this training , the echo canceller must be re-initialized, by resetting the
AMBE-2000™ Vocoder Chip for optimum performance.
Figure 7-A Typical Echo Path
AMBE-2000
Decoder
8kHz Speech Data
Echo Path A
A/D-D/A
Encoder
4 to 2 wire
Converter
Echo Path B
8kHz Speech Data
The Echo Return Loss (ERL) of the analog circuit must be 6dB or more (in diagram ERL = Echo Path A – Echo Path B) for
proper echo canceller operation. Linear A/D-D/A chips will generally provide better echo cancellation performance than µlaw
or Alaw chips due to lower quantization noise.
The echo canceller can be activated either through the hardware pin 78, ECHOCAN_EN, or through the Control Word
interface described in section 5.2.9. See section 7.1 for important note about the ECHOCAN_EN pin.
page 39
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
7.4
Voice Activation Detection (VAD), Comfort Noise Insertion (CNI)
The Voice Activation Detection (VAD) algorithm along with the Comfort Noise Insertion (CNI) feature of the AMBE-2000™
chip performs useful functions in systems trying to convert periods of silence, that exist in normal conversation, to savings in
system bandwidth or power.
With the VAD functions enabled, periods of silence will be denoted by the encoder in two ways. First, the encoder will output
a silence frame (in-band). This silence frame contains information regarding the level of background noise which allows the
corresponding decoder to synthesize a “Comfort Noise” signal at the other end. The comfort noise is intended to give the
listener the feeling that the call is still connected, as opposed to producing absolute silence which can give the impression that
the call has been “dropped”. Second, the Encoder Silence Detected flag is set in Control Word 1 of the Formatted Output
format described in section 5.3.3.
VAD can be enabled in one of two ways. A high signal on the hardware pin VAD_EN (pin 86), subject to the restrictions of
section 7.1, enables VAD. The Control Frame described in section 5.2.9 describes how to enable/disable the VAD algorithm
once the AMBE-2000™ has begun operating.
If the VAD features are being used to reduce transmit power during times of conversational silence, DVSI recommends that a
silence frame be transmitted at the start of the period and approximately each 500-1000 milliseconds thereafter. This is to
ensure that the parameters regarding the levels of background noise are transmitted to the decoder for the smoothest audible
transitions between synthesized speech and synthesized silence.
There is a silence threshold value is –25 dB in the VAD algorithm.
The synthesis of a Comfort Noise frame by the decoder is not dependant on VAD being enabled. The decoder will produce a
comfort noise frame if it receives an in-band silence frame (produced only by an encoder with VAD enabled).
7.5
Dual Tone Multiple Frequency, Detection and Generation
The AMBE-2000™ Vocoder Chip is capable of detecting, transmitting, and synthesizing DTMF tones. DTMF features are
always enabled. Detection of a DTMF tone by the encoder sets the DTMF Digit Detect in the DTMF Control Format found in
section 5.4.6. Which DTMF tone is detected along with amplitude information is placed in the bits 0-7 described in section
5.3.6. Additionally, the encoder passes the DTMF data in-band (within the regular voice data bits) so that normal DTMF tones
pass seamlessly from the encoder to the decoder for synthesis.
The decoder synthesizes a DTMF tone in response to reception of an in-band DTMF tone frame.
7.6
Normal Power and Power Saving Modes
Power savings can be achieved during times of longer inactivity of the AMBE-2000™ chip by placing it into one of three
available Low Power Modes. The chip can be placed into low-power and stand-by modes via hardware or software Control
Words. In low power modes the A/D-D/A port will be disabled, concurrently halting any processing of voice frames in either
direction. Depending on the low power state selected, either a Wake Up Control Word or a hardware reset on RESETN is
necessary to return the AMBE-2000™ to normal operation.
7.6.1
Standard Sleep Mode
The standard sleep is the only low power mode that can be entered into either through hardware or software. The AMBE2000™ Chip can be placed into Standard Sleep mode either by setting SLEEP_EN (pin 83) high, subject to the restrictions of
section 7.1, or through software by using Control Word 2 with bit 3 set to 1 as described in section 5.2.9.
SLEEP_EN should be tied high if you plan to configure the A/D-D/A chip from Standard Sleep mode upon power-up or reset.
When using software SLEEP_EN with A-law or µ-law codecs, it is important to note that if packets are sent to the decoder
while it is in sleep mode, noise will be heard at the output. It is recommended that no packets be sent to the decoder until it is
commanded to wake up.
page 40
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
7.6.2
Power Down
Power Down provides the lowest power usage of the sleep modes, the only drawback to this mode is the necessity of a
hardware reset on RESETN (pin 69) to resume normal operation.
Table 7-C Summary of Power Saving Modes
Power Consumption
Sleep Mode
Enter State via
Return to Normal
Operation via
Wake Up
Time
Normal Operation
N/A
N/A
N/A
Standard Sleep
SLEEP_EN pin at
reset OR Control
Word
Control Word
N/A
Power Down
Control Word
RESETN
200 µ secs.
3V
CMOS
TTL
Crystal
Approx. 65mW
24 mW
36 mW
0.11 mW
0.11 mW
page 41
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
7.7
Slip Enable
In any real time communication system, clock skew issues must be anticipated to keep the flow of data smooth from one end of
the system to the other. The SLIP_EN (pin 82) signal allows the encoder of the AMBE-2000™ to react to small slips in the
encoder channel signals. When the AMBE-2000™ is in active (parallel or serial) mode, the chip produces the signals
internally for the transfer of data. Because the transmission channel will then likely be driven by this timing, the necessity of
controlling slip becomes a moot point.
Any time the AMBE-2000™ encoder channel is in one of the passive modes and the channel timing is asynchronous to the
A/D-D/A clock (very rarely are these two interfaces coupled) then the SLIP_EN pin should be set active high.
The AMBE-2000™ Vocoder chip processes speech in voice frames that are approximately 20 ms in duration. When
configured appropriately the chip provides a slip control feature that automatically adjusts the frame size to either 160 or 161
speech samples per frame. This slip control feature allows the vocoder chip to compensate for drift between the frame and
sample rate clocks on the order of approximately 0.6% (6,000 ppm.) The vocoder chip also accepts Slip Control Packets that
extend the range of allowable frame sizes to be between 159 and 161 samples per voice frame. When properly used these Slip
Control Packets provide the designer with additional flexibility in dealing with clock drift.
There are three recommended methods for using slip control on the AMBE-2000™ Vocoder Chips which are described below.
The system designer should select the method that best meets the needs of their system configuration. Also included is some
background information on the operation of the AMBE-2000 in passive mode
In order to help understand Slip control feature here is a brief description on reading encoder packets from the AMBE-2000 in
passive formatted mode.
When transmitting a packet, the AMBE-2000 writes a Header = 0x13ec followed by 23 words of data, followed by 0xfffe into
the transmit (i.e. output) buffer. The terminating word 0xfffe is written into the transmit buffer by the AMBE-2000 at the end
of each encoder packet. Normally in passive mode this terminating word is in the transmit buffer at the beginning of each
transmission cycle (from the previous frame) and so it is the first word output whenever a packet is transmitted. If the encoder
packet is ready, then the second output word will be the packet Header=0x13ec followed by 23 words of data. However if the
packet is not ready then the AMBE-2000 will continue to output the terminating word (0xfffe) until the packet is ready and
placed in the transmit buffer. At this point the full 24 word packet beginning with the Header will be output on subsequent
transmissions. This process continues for each packet transmission which occurs nominally every 20 ms, provided that each 24
word packet (Header + 23 data words) is read in full. If the full 24 words of the packet are not read from the AMBE-2000, then
the chip’s transmit buffer will contain some words left over from a previous packet. Generally these words would be 0x0000
for lower data rates which don't use the last words of the packet. This case should be avoided:
It is recommended that in passive formatted mode the system always read packets by requesting
words from the AMBE-2000 until a packet Header is received and then continuing to request 23
additional output words from the AMBE-2000 until a total of 24 words beginning with the Header
word = 0x13ec are received. Any words output by the AMBE-2000 prior to the Header should be
ignored by the system (except for monitoring as discussed in Method 3 below).
page 42
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
Word #
0
1
2
3
4
5
6
7
8
9
10
11
12-23
Value
0x13ec
0x07xx
0x1355
0x009f
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
0x0000
Description
Header
Slip Control Packet ID, xx=CWD1 (See table 5-B) Typical value xx=00
Slip Control indicator for AMBE-2000
Slip Control data
Must be 0x0000
Must be 0x0000
Must be 0x0000
Must be 0x0000
Must be 0x0000
Must be 0x0000
Must be 0x0000
Must be 0x0000
Channel Data for AMBE® voice decoder
Table 7D: Slip Control Packet
Method 1 - Internal Slip Control
This is the simplest method of slip control and is the default method provided slip control is enabled (SLIP_EN is high) on the
chip. In this method the vocoder chip’s internal slip adjustment of 160 to 161 sample per frame is used. In order to work
properly, the system designer must set-up the sample rate and packet timing so that the following constraint is met:
160 < Average-Frame-Interval * (1 +/- Frame-Drift) * Sample-Rate * (1 +/- Sample-Drift) < 161
For example using an Average-Frame-Interval = 20 ms, and assuming 100 parts per million oscillator accuracy (i.e Frame-Drift
= Sample-Drift = .0001), then the above constraint requires:
8002 Hz < Sample-Rate < 8048 Hz.
In practice Sample-Rate=8002 Hz would be preferred since it is closer to the nominal value of 8000 Hz.
In another example the Sample-Rate = 8000 Hz. as provided by a PCM source. Again assuming 100 parts per million
oscillator accuracy for both clocks (i.e. Frame-Drift = Sample-Drift = .0001), then the above constraint requires:
20.0041 ms. < Average-Frame–Interval < 20.1209 ms.
This can be achieved by slightly decreasing the channel bit rate or adding an extra bit into the channel bit stream every several
voice frames.
Method 2 - Extended Slip Control with Periodic Slip Control Packets
The AMBE-2000 can provide extended slip compensation through the insertion of Slip Control Packets. One method of using
this capability is for the system to periodically insert these Slip Control Packets into the data stream sent to the AMBE-2000.
Note that for this method slip control must be enabled (SLIP_EN is high) on the chip. This approach gives the designer a way
to accommodate clock drift while providing very flexible frame-interval and sample-rate timing. Furthermore minimal system
overhead is required. In this method a Slip Control Packet is generated by the system by setting the Control Words as shown in
Table 7D above, where the Channel Data is the compressed voice data being sent to the AMBE decoder. In the Periodic Slip
Control method such a Slip Control Packet is input into the vocoder chip every N frames. The value of N must be selected by
the design engineer to meet the following constraint:
page 43
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
0.25 > (1/N) > Average-Frame-Interval * Sample-Rate * (Frame-Drift + Sample-Drift)
For example with an Average-Frame-Interval = .02 (i.e. 20 ms) and a Sample-Rate = 8000, then with 100 parts per million
oscillator accuracy (i.e. Frame-Drift = Sample-Drift = .0001), then above constraint equates to 4 < N < 31.25, and N=30 would
be a reasonable selection. In this case the system would input the specified Slip Control Packet into the AMBE-20X0 vocoder
chip every 30’th frame enabling the vocoder chip to adjust for the actual clock drift.
***RECOMMENDED***
Method 3 - Extended Slip Control with as needed Slip Control Packets
The preferred method for using Slip Control Packets is to monitor the availability of data from the AMBE-2000 vocoder chip
and to only input Slip Control Packets into the data stream sent to the AMBE-2000 as needed. Note that for this method slip
control must be enabled (SLIP_EN is high) on the chip. This method provides compensation for the widest range of clock drift
(1.2% or 12000 ppm), with the greatest flexibility in frame-interval and sample-rate timing. In this method the same Slip
Control Packets shown in Table 7D are inserted into the data stream going to the AMBE-2000. However, unlike in the
Periodic method, the packets are not input at regular intervals but are instead only input to the chip when needed. The
recommended procedure for this Method of Slip Control is for the system to read a packet from the AMBE-2000 at regular
fixed frame intervals where the fixed interval must be within the range [19.875 – 20.125] ms for a sample rate = 8000 Hz. The
system application should check each word output by the AMBE-2000 and should continue requesting words from the chip
until the packet Header followed by 23 data words are received. If the words received before the Header word consist of only a
single termination word (0xfffe) then no further action is required. However if two or more termination words are received
prior to the Header then the system should input a Slip Control Packet to the AMBE-2000 on the next available transmission
into the chip (i.e. the next packet going into the AMBE-2000 decoder should be a Slip Control Packet). Once this Slip Control
Packet is input into the AMBE-2000 it will respond within 1-2 frames by advancing the time when packets are ready for
transmission by 125 microseconds. Note that this procedure also may require a small amount of buffering in the system to
account for the fact that the packet my not be ready for some small time (< 125 microseconds) after it is first requested.
page 44
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
8 Appendices
8.1
Example A/D-D/A Usage
The following examples of A/D-D/A chips have been included to show connections necessary for interfacing to a number of
popular chips.
8.1.1
AD73311
U1
CODEC_RX_CLCK
CODEC_TX_CLCK
CODEC_TX_STRB
CODEC_RX_STRB
CODEC_RX_DATA
CODEC_TX_DATA
27
U2
33
14
37
17
29
18
31
16
41
19
SCLCK
SE
SDOFS
MCLK
SDIFS
AVDD1
SDO
AVDD2
SDI
DVDD
20
15
SERIAL PORT ENABLE (HIGH)
16.384 MHz
3
9
5V
12
3.3 V
AD73311AR
AMBE-2000
page 45
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
Revision Number
1.0
History of Revisions
Date of Revision
Description
November 1999
Initial Version
Pin descriptions
Serial Configuration Selection
1.1
April 2000
1.2
May 2000
1.3
August 2000
1.4
October 2000
Channel Serial Interface Pin
Descriptions
Table 6-A CODEC_SEL[1-0]:
A/D- D/A Hardware Configuration
Values
Deleted the word parallel
Table 5-N Control Word Format:
Removed VAD in bit 5. This is not
used as an output.
Added Application note for AD
73311AR
Clarification on VAD
1.5
November 2000
2.0
January 2001
Pages
11-12
19
20
35
9
32
43
28
Expanded Description of Control
Word 1
29
Table 5-M added DTMF Code 0xff
32
Channel and Codec Timing
Diagrams and Tables
Added Note describing H
21, 22,
36, 37
22, 37
10, 11,
13, 14,
16,17, 18,
20, 21,
22, 23,
27, 28,
29, 32,
34, 35,
36, 40
Clarified/corrected the following
pages.
page 46
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
AMBE-2000™ Vocoder Chip
User’s Manual Version 3.0
Revision Number
2.1
2.2
3.0
History of Revisions
Date of Revision
Description
February 2001
Updated Timing Diagrams and
Tables for Channel and Codec
February 2001
Changed Pin Description CLK_I to
X2/CLKIN and CLK_I2 to X1
Updated timing diagram and table
for X2/CLKIN and RESETN
Replaced …set the DTMF Code to
0x00 to …set the DTMF Code to
0xff
August 2001
Company address updated
Removed description of Decoder
Output Volume Control
Modified SLEEP_EN Section 7.6.1
Modified description of EPR
Added Detailed Explination of Slip
Enable Control
Added description of Decoder
Silence Detect, Decoder Frame
Repeat and Encoder DTMF Detect
bits
page 47
Visit us at www.dvsinc.com
Confidential Proprietary, Subject to Change
Pages
21, 22,
36, 37
11
13,14
27
2
28
40
11,12,21
43,44
30