MT8950 ISO-CMOS
Theory of Operation
Data Rate
Bits/Sec
0 - 8000
9600
19200
The MT8950 is an encoder/decoder which operates
on low baud rate data (up to 19.2 kbps) to convert it
to the ST-BUS format. The data can subsequently
be transparently switched or transmitted in a manner
identical to PCM encoded voice. In this respect, the
functional characteristics of the device are very
similar to many industry standard voice codecs.
Asynchronous and synchronous data from 0 to 8
kbps and at 9.6 kbps is accepted by the codec
without any restrictions. Asynchronous data at 19.2
kbps should have at least two stop bits for the device
to encode it properly. The data is encoded by the
Codec into an eight bit word which occupies one 64
kbps channel on the ST-BUS. Conversely, it accepts
an encoded 8 bit word from an incoming ST-BUS
stream and regenerates the original digital signal.
Mitel’s ST-BUS is a synchronous time division
multiplexed serial stream with a bit rate of 2048
kbps. In a telecommunications environment, it is
generally divided into 32 channels made up of 8 bits
each, with an effective bandwidth of 64 kbps per
channel. These channels may carry data or PCM
encoded voice.
Asynchronous
Restrictions
None
None
Minimum
2 Stop Bits
Synchronous
Restrictions
None
±3.2
None
±3.8
None
±7.5
Percentage
†
Distortion
Table 1. Summary of Data Codec Capabilities.
† Refers to the maximum distortion in the bit period timing of the
regenerated data. (Channel Bandwidth = 64kbps )
Percentage Distortion = |TBO - TBR| / TBO X 100
where TBO = Original Data Bit Period
TBR = Regenerated Data Bit Period
and an actual data transition be at least 125µs. The
violation pulses can be on the MARK or SPACE line.
In a communications system, these violations can be
used to carry other information when no data is
being transmitted.
Encoding/Decoding Scheme
The Data Codec uses
a
Transition Encoded
Modulation (TEM) technique to encode low speed
data onto a 56 or 64 kbps equivalent PCM voice
channel. This coding algorithm significantly reduces
data bit distortion. The timing distortion in the
regenerated data is summarized in Table 1. A simple
sampling method for encoding the data would
require a 256 kbps channel to obtain the same low
distortion figures.
Low Speed Data Format
The Data Codec can accept low speed data in either
Non Return to Zero (NRZ) or Return to Zero (RZ)
format. The NRZ format requires only one line to
carry the data. This format is suitable for interfacing
the data codec with RS-232 type terminals and
microprocessor peripherals such as UARTS, ACIAs,
etc. All signals have to be converted to TTL voltage
levels before being input to the codec.
If the encoded information is to be transmitted over
digital T1/DS1 trunks, the maximum percentage
distortion in the regenerated data is effectively
doubled. This is due to the fact that the least
significant bit in specific channels on these trunks is
used to transmit signalling information. Thus the
bandwidth per channel is reduced to 56 kbps.
The RZ format requires two separate lines to
represent the MARKs and SPACEs in the data as
illustrated in Figure 4. This format is useful when the
data terminal is located some distance from the
codec and the data is to be transmitted over a line as
a three level signal (a positive pulse for the
beginning of MARKs, negative pulse for the
beginning of SPACEs and zero level for no change in
the signal). The three level signal is converted to its
TTL-Compatible binary form as shown in Figure 4
before being applied to the codec. A pulse appears
on one line of the input indicating the beginning of
MARKs. This is followed by a pulse on the second
line indicating the beginning of SPACEs. If two or
more pulses appear consecutively on the same line
before the second line of the pair receives or
transmits another pulse, then these pulses can be
considered to be violating the normal rule of the RZ
format and are called "Violation Pulses". The data
codec will accept these violations with the restriction
that the time difference between a violation pulse
The encoder stage of the Data Codec
observes
data transitions in discrete timing windows which are
125µs wide. These timing frames are further divided
into 32 timeslots of 3.906µs duration each (see
Figure 3). The position of the first data transition, the
total number of transitions, and, the time period
between the transitions in this 125µs frame is
encoded as an 8 bit word.
The first five bits (b0 to b4) indicate the position of
the first data transition with respect to the 32
timeslots in the window. Bit 7 in the encoded word
represents the absolute value of the data in the 31st
timeslot. Bits 5 and 6 in conjunction with bit 7 are
used to identify the total number of transitions and
the time period between the transitions. Due to the
fixed bit rate restrictions above 8 kbps, a maximum
6-6
MITEL [ MITEL NETWORKS CORPORATION ]
