AMBE-2000™ Vocoder Chip Users Manual
Version 4.92, June, 08
DVSI Confidential Proprietary, Subject to Change
Page 46
Visit us at
www.dvsinc.com
7.6
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-2020™ to react to small slips in
the encoder channel signals. When the AMBE-2020™ is in active mode, the chip produces the signals internally for the
transfer of data. Because this timing will then likely drive the transmission channel, the necessity of controlling slip
becomes a moot point.
Any time the AMBE-2020™ 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-2020™ Vocoder chip process 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-2020™ Vocoder Chip 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-2020™ in passive mode
In order to help understand the Slip control feature here is a brief description on reading encoder packets from the AMBE-
2020™ in passive framed mode.
When transmitting a packet, the AMBE-2020™ 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-2020™
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-2020™ 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 ( 23 data words) is read in full. If the full 24 words of the packet are not
read from the AMBE-2020™, 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 hence:
It is recommended that in passive framed mode the system always read packets by requesting
words from the AMBE-2020™ until a packet Header is received and then continuing to request
23 additional output words from the AMBE-2020™ until a total of 24 words beginning with the
Header word = 0x13ec are received. Any words output by the AMBE-2020™ prior to the Header
should be ignored by the system (except for monitoring as discussed in Method 3 below).
