Functional Description (Continued)
BURST MODE OPERATION
For full-duplex operation over a single twisted-pair, burst
mode timing is used, with the line-card (exchange) end of
the link acting as the timing Master.
Each burst from the Master consists of the B1, B2 and D
channel data from 2 consecutive frames combined in the
format shown in
Figure 5
. During transmit bursts the Master’s receiver input is inhibited to avoid disturbing the adaptive circuits. The Slave’s receiver is enabled at this time and
it synchronizes to the start bit of the burst, which is always
an unscrambled ‘1’ (of the opposite polarity to the last ‘1’
sent in the previous burst). When the Slave detects that 36
bits following the start bit have been received, it disables the
receiver input, waits 6 line symbol periods to match the other end settling guard time, and then begins to transmit its
burst back towards the Master, which by this time has enabled its receiver input. The burst repetition rate is thus
4 kHz, which can either free-run or be locked to a synchronizing signal at the Master end by means of the MBS input
(See
Figure 10
). In the latter case, with all Master-end
transmitters in a system synchronized together, near-end
crosstalk between pairs in the same cable binder may be
eliminated, with a consequent increase in signal-to-noise ratio (SNR).
ACTIVATION AND LOOP SYNCHRONIZATION
Activation (i.e. power-up and loop synchronization) is typically completed in 50 ms and may be initiated from either
end of the loop. If the Master is activating the loop, it sends
normal bursts of scrambled ‘1’s, which are detected by the
Slave’s line-signal detect circuit, causing it to set C0
e
1in
the Status Register, and pull the INT
pin low. Pin 6, the LSD
pin, also pulls low. To proceed with Activation, the device
must be powered up by writing to the Control Register with
C6
e
1. The Slave then replies with bursts of scrambled
‘1’s synchronized to received bursts, and the flywheel circuit
at each end searches for 4 consecutive correctly formatted
receive bursts to acquire full loop synchronization. Each receiver indicates when it is correctly in sync with received
bursts by setting the C1 bit in the Status Register high and
pulling INT
low.
To activate the loop from the Slave end, bit C6 in the Control Register must be set high, which will power-up the device and begin transmission of alternate bursts i.e., the burst
repetition rate is 2 kHz, not 4 kHz. At this point the Slave is
running from its local oscillator and is not receiving any sync
information from the Master. When the Master’s line-signal
detect circuit recognizes this ‘‘wake-up’’ signal, the Master
is activated and begins to transmit bursts, synchronized, as
normal, to the MBS or FS
a
input with a 4 kHz repetition rate.
This enables the Slave’s receiver to correctly identify burst
timing from the Master and to re-synchronize its own burst
transmissions to those it receives. The flywheel circuits then
acquire full loop sync as described earlier.
Loop synchronization is considered to be lost if the flywheel
finds 4 consecutive receive burst ‘‘windows’’ (i.e. where a
receive burst should have arrived based on timing from previous bursts) do not contain valid bursts. At this point bit C1
in the Status Register is set low, the INT
output is set low
and the receiver searches to re-acquire loop sync.
DIGITAL SYSTEM INTERFACE
The digital system interface on the DASL separates B and D
channel information onto different pins to provide maximum
flexibility. On the B channel interface, phase skew between
transmit and receive directions may be accommodated at
the Master end since separate frame sync inputs, Fs
a
and
Fs
b
, are provided. Each of these synchronizes a counter
which gates the transfer of B1 and B2 channels in consecutive time-slots across the digital interface; since the counters are edge-synchronized the duration of the F
s
input signals may vary from a single-bit pulse to a square-wave. The
serial shift rate is determined by the BCLK input, and may
be any frequency from 256 kHz to 2.048 MHz, as shown in
Figure 6.
At the Slave end, both Fsaand Fsbare outputs. Fsagoes
high for 8 cycles of BCLK coincident with the 8 bits of the
B1 channel in both Transmit and Receive directions. Fs
b
goes high for the next 8 cycles of BCLK, which are coincident with the 8 bits of the B2 channel in both Transmit and
Receive directions. BCLK is also an output at 2.048 MHz,
the serial data shift rate, as shown in
Figure 7.
Data may be
exchanged between the B1 and B2 channels as it passes
through the device, by setting Control bit C0
e
1. An addi-
tional Frame Sync output, FS
c
, is provided to enable a regenerator to be built by connecting a DASL in Slave Mode
to a DASL in Master Mode. The FS
c
output from the Slave
directly drives the FS
a
and FSbinputs on the Master.
D channel information, being packet-mode, requires no synchronizing input. This interface consists of the transmit data
input, D
x
, receive data output, Dr, and 16 kHz serial shift
clock DCLK, which is an input at the Master end and an
output at the Slave end. Data shifts into D
x
on falling edges
of DCLK and out from D
r
on rising edges, as shown in
Fig-
ure 11.
DCLK should be Synchronous with BCLK.
An alternative function of the DCLK/DEN pin allows Dxand
D
r
to be clocked at the same rate as BCLK at the Master
end only. By setting bit C1 in the Control Register to a 1,
DCLK/DEN becomes an input for an enabling pulse to gate
2 cycles of BCLK for shifting the 2 D bits per frame. Thus, at
the Master end, the D channel bits can be interfaced to a
TDM bus and assigned to a time-slot (the same time-slot for
both transmit and receive), as shown in
Figure 12.
CONTROL INTERFACE
A serial interface, which can be clocked independently from
the B and D channel system interfaces, is provided for microprocessor control of various functions on the DASL device. All data transfers consist of a single byte shifted into
the Control Register via CI simultaneous with a single byte
shifted out from the Status Register via CO, see
Figure 13
.
Data shifts in to CI on rising edges of CCLK and out from
CO on falling edges when CS
is pulled low for 8 cycles of
CCLK. An Interrupt output, INT
goes low to alert the microprocessor whenever a change in one of the status bits, C1
and/or C0 has occurred. This latched output is cleared high
following the first CCLK pulse when CS
is low. No interrupt
is generated when status bit C2 (bipolar violation) goes high,
however. This bit is set whenever 1 or more violations of the
AMI coding rule is received, and cleared everytime the CS
is
pulsed. Statistics on the line bit error rate can be accumulated by regularly polling this bit.
When reading the CO pin, data is always clocked into the
Control Register; therefore the CI data word should repeat
the previous instruction if no change to the device mode is
intended.
Figure 13
shows the timing for this interface, and Table II
lists the control functions and status indicators.
5