
MT9072
Data Sheet
55
Zarlink Semiconductor Inc.
will be phase-locked to the CKi clock by an external phase locked loop (PLL). Therefore, in a single trunk system,
the receive data is in phase with the EXCLi clock, the CKi clock is phase-locked to the EXCLi clock, and the read
and write positions of the elastic buffer will remain fixed with respect to each other.
In a multi-trunk slave or loop-timed system (i.e., PABX application) a single trunk will be chosen as a network
synchronizer, which will function as described in the previous paragraph. The remaining trunks will use the system
timing derived from the synchronizer to clock data out of their slip buffers. Even though the signals from the network
are synchronous to each other, due to multiplexing, transmission impairments and route diversity, these signals
may jitter or wander with respect to the synchronizing trunk signal. Therefore, the EXCLi clocks of non-synchronizer
trunks may wander with respect to the EXCLi clock of the synchronizer and the system bus.
Network standards state that, within limits, trunk interfaces must be able to receive error-free data in the presence
of jitter and wander (refer to network requirements for jitter and wander tolerance). The MT9072 will allow a
maximum of 26 channels (208 UI, unit intervals) of wander and low frequency jitter before a frame slip will occur.
The minimum delay through the receive slip buffer is approximately two channels and the maximum delay is
approximately 60 channels (see Figure 8).
When the CKi and the EXCLi clocks are not phase-locked, the rate at which data is being written into the slip buffer
from the line side may differ from the rate at which it is being read out onto the ST-BUS. If this situation persists, the
delay limits stated in the previous paragraph will be violated and the slip buffer will perform a controlled frame slip.
That is, the buffer pointers will be automatically adjusted so that a full frame is either repeated or lost. All frame slips
occur on frame boundaries.
Two status bits, RSLP and RSLPD (register address Y10 for E1 and Y13 for T1), give indication of a slip occurrence
and direction. RSLP changes state in the event of a slip. If RSLPD=0, the slip buffer has overflowed and a frame
was lost; if RSLPD=1, a underflow condition occurred and a frame was repeated. A maskable interrupt status bit
RSLIPI (register address Y34 for E1 and Y36 for T1) is also provided.
Figure 8 illustrates the relationship between the read and write pointers of the receive slip buffer. Measuring
clockwise from the write pointer, if the read pointer comes within two channels of the write pointer a frame slip will
occur, which will put the read pointer 34 channels from the write pointer. Conversely, if the read pointer moves more
than 60 channels from the write pointer, a slip will occur, which will put the read pointer 28 channels from the write
pointer. This provides a worst case hysteresis of 13 channels peak (26 channels peak-to-peak) or a wander
tolerance of 208 UI. The registers relevant to controlling and observing the elastic buffer in T1 mode are shown in
Table 14. The registers related to elastic buffer for E1 are shown in Table 15.
Figure 8 - Read and Write Pointers in the Slip Buffers
Write Pointer
60 CH
2 CH
47 CH
15 CH
34 CH
28 CH
512 Bit
Elastic
Store
13 CH
-13 CH
Wander Tolerance
Read Pointer
Read Pointer
Read Pointer
Read Pointer
2