TABLE 19 LVDS LINK 59 BYTE CELL CONFIGURATIONS .......................... 188
TABLE 20 LVDS LINK 58 BYTE CELL CONFIGURATIONS .......................... 188
TABLE 21 LVDS LINK 57 BYTE CELL CONFIGURATIONS .......................... 190
TABLE 22 LVDS LINK 61 BYTE CELL CONFIGURATIONS .......................... 192
TABLE 23 LVDS LINK 59 BYTE CELL CONFIGURATIONS WITH CRC ....... 192
PROPRIETARY AND CONFIDENTIAL TO PMC-SIERRA, INC., AND FOR ITS CUSTOMERS’ INTERNAL USEx
RELEASED
DATA SHEET
PMC-1980581ISSUE 5DUAL SERIAL LINK PHY MULTIPLEXER
PM7350 S/UNI-DUPLEX
1 FEATURES
• Integrated analog/digital device that interfaces a high-speed parallel bus to a
high speed Low Voltage Differential Signal (LVDS) serial link with optional 1:1
protection.
• For framers or modems without Utopia bus interfaces the S/UNI-DUPLEX
provides cell delineation (I.432) across 16 clock and data (bit serial)
interfaces.
• Fault detection, redundancy, protection switching, and inserting/removing
cards while the system is running (hot swap).
• Interface to other S/UNI-DUPLEX or S/UNI-VORTEX, to satisfy a full set of
system level requirements for backplane interconnect:
• Transports user data by providing the inter-card data-path.
• Inter-processor communication by providing an integrated inter-card
control channel.
• Exchanges flow control information (back-pressure) to prevent data
loss.
• Provides embedded command and control signals across the
backplane: system reset, error indications, protection switching
commands, etc.
• Clock/timing distribution (system clocks as well as reference clocks
such as 8 kHz timing references).
• When used as a parallel bus slave device, can be configured to share the
bus with other S/UNI-DUPLEX bus slave devices.
• Can interface to another S/UNI-DUPLEX device (via a single LVDS link) to
create a simple point-to-point "Utopia bus extension" capability.
• Can interface to two S/UNI-DUPLEX devices to create a 1:1 protected bus
extension.
• Interworks with PM7351 S/UNI-VORTEX devices to implement a point-tomultipoint serial backplane architecture, with optional 1:1 protection of the
common card.
• In the LVDS receive direction: selects traffic from the LVDS link marked active
and demultiplexes the individual cell streams to the appropriate PHY device.
PROPRIETARY AND CONFIDENTIAL TO PMC-SIERRA, INC., AND FOR ITS CUSTOMERS’ INTERNAL USE1
RELEASED
DATA SHEET
PMC-1980581ISSUE 5DUAL SERIAL LINK PHY MULTIPLEXER
PM7350 S/UNI-DUPLEX
• Cell read/write to both LVDS links available through the processor port.
Provides optional hardware assisted CRC32 calculation across cells to
support an embedded inter-processor communication channel across the
LVDS links.
PROPRIETARY AND CONFIDENTIAL TO PMC-SIERRA, INC., AND FOR ITS CUSTOMERS’ INTERNAL USE3
RELEASED
DATA SHEET
PMC-1980581ISSUE 5DUAL SERIAL LINK PHY MULTIPLEXER
PM7350 S/UNI-DUPLEX
3 REFERENCES
• PMC-Sierra; “Saturn Compatible Interface For ATM PHY Layer And ATM
Layer Devices, Level 2”; PMC-940212; Dec. 8, 1995
• ATM Forum, “Universal Test & Operations PHY Interface for ATM (UTOPIA),
Level 2”, Version 1.0, af-phy-0039.000, June 1995
• PMC-Sierra; “Saturn Interface Specification And Interoperability Framework
For Packet And Cell Transfer Between Physical Layer And Link Layer
Devices”, PMC-980902, Draft
• Draft American National Standard for Telecommunications T1.413 Issue 2,
“Network and Customer Installation Interfaces - Asymmetric Digital
Subscriber Line (ADSL) Metallic Interface”, ANSI T1.413-1998, November
1998
PROPRIETARY AND CONFIDENTIAL TO PMC-SIERRA, INC., AND FOR ITS CUSTOMERS’ INTERNAL USE4
RELEASED
DATA SHEET
PMC-1980581ISSUE 5DUAL SERIAL LINK PHY MULTIPLEXER
PM7350 S/UNI-DUPLEX
4 APPLICATION EXAMPLES
When designing communication equipment such as access switches,
multiplexers, wireless base stations, and base station controllers the equipment
architect is faced with a common problem: how to efficiently connect a large
number of lower speed ports to a small number of high speed ports? Typically, a
number of line-side ports (analog modems, xDSL modems ATM PHYs, or RF
modems) are terminated on each line card. Numerous line cards are then
slotted into one or more shelves and backplane traces or inter-shelf cables are
used to connect the line cards to a centralized (often 1:1 protected) common
card, hereafter referred to as the core card. The core card normally includes one
or more high speed WAN up-link ports that transport traffic to and from a high
speed broadband network.
A block diagram of a 1:1 redundant system is shown in Fig. 1.
Fig. 1 Typical Target Application
Modem
or PHY
Modem
or PHY
Modem
or PHY
Modem
or PHY
Modem
or PHY
Modem
or PHY
Line Card #1
S/UNI-
DUPLEX
Line Card #N
S/UNI-
DUPLEX
S/UNI-
VORTEX
WAN Card
S/UNI-
VORTEX
WAN Card
Policing
OA&M
Policing
OA&M
Buffering
Discard
Scheduling
Buffering
Discard
Scheduling
OA&M
OA&M
WAN
up-link
WAN
up-link
In this type of equipment the majority (perhaps all) user traffic goes from WAN
port to line port, or from line port to WAN port. Although the individual ports on
the line cards are often relatively low speed interfaces such as T1, E1, or xDSL,
there may be many ports per line card and many line cards per system, resulting
in hundreds or even thousands of lines terminating on a single WAN up-link. In
the upstream direction (from line card to WAN up-link), the equipment must have
capacity to buffer and intelligently manage bursts of upstream traffic
simultaneously from numerous line cards.
PROPRIETARY AND CONFIDENTIAL TO PMC-SIERRA, INC., AND FOR ITS CUSTOMERS’ INTERNAL USE5
RELEASED
DATA SHEET
PMC-1980581ISSUE 5DUAL SERIAL LINK PHY MULTIPLEXER
PM7350 S/UNI-DUPLEX
In the downstream direction the equipment must handle a similar issue, the “big
pipe feeding little pipe” problem. When a large burst of traffic destined for a
single line port is received at the high speed WAN port it must be buffered and
managed as it queues up waiting for the much lower speed line port to clear.
The line cards are always the most numerous cards in this type of equipment.
An individual line card, even if it terminates a few dozen low speed ports, does
not generate or receive enough traffic to justify putting complex buffering and
traffic management devices on it. The ideal architecture has low cost “dumb”
line cards and a feature rich, “smart” core card. In order to enhance fault
tolerance, the architecture should also inherently support 1:1 protection using a
redundant core card and WAN up-link without significantly increasing line card
complexity.
A system architecture that keeps buffering and traffic management off the line
card with typically exhibit the following features:
1. Connection setup is simpler both in terms of programming and during
execution because there is minimal or no requirement for line
intervention during the connection setup process.
2. In-service feature upgrades are simpler because feature complexity is
limited to the common equipment.
3. Component costs are reduced, while system reliability increases due
to reduced component count.
In this type of architecture there are often three stages of signal concentration or
multiplexing, as shown in Fig 2.
PROPRIETARY AND CONFIDENTIAL TO PMC-SIERRA, INC., AND FOR ITS CUSTOMERS’ INTERNAL USE6
RELEASED
DATA SHEET
PMC-1980581ISSUE 5DUAL SERIAL LINK PHY MULTIPLEXER
PM7350 S/UNI-DUPLEX
Fig. 2 Three Stage Multiplex Architecture
Modem
Modem
Modem
Modem
Modem
Modem
Modem
Modem
Modem
Line Card #1
S/UNI-
DUPLEX
Line Card #2
S/UNI-
DUPLEX
Line Card #N
S/UNI-
DUPLEX
S/UNI-
VORTEX
S/UNI-
VORTEX
Policing
OA&M
Buffering
Discard
Scheduling
OA&M
WAN Card
WAN
up-link
The first stage resides on the line card and spans only those ports physically
terminated by that card. Since it is confined to a single card, this first stage of
multiplexing readily lends itself to the simple parallel bus based multiplex
topology implemented by the S/UNI-DUPLEX. The second stage of
concentration occurs between the core card(s) and the line cards, including line
cards that are on a separate shelf. This second stage is best served by a
redundant serial point-to-point technology. The third stage of multiplexing is
optional and resides on the core card. This third stage is used in systems with a
large number of line cards that require several devices to terminate the second
stage of aggregation. Since the third stage of aggregation is confined to the core
card, it lends itself readily to a parallel bus implementation. This three stage
approach is implemented directly by the S/UNI-DUPLEX and its sister device, the
S/UNI-VORTEX.
The S/UNI-DUPLEX acts as the line card’s bus master. It implements the first
stage of multiplexing by routing traffic from the PHYs and transmitting the traffic
simultaneously over two high speed (200 Mbps) serial 4-wire LVDS links. One
serial link attaches to the active core card, the other to the standby core card. In
the downstream direction the S/UNI-DUPLEX demultiplexes traffic from the
active core card’s LVDS serial link and routes this traffic to the appropriate PHYs.
If the active core card (or its LVDS link) should fail, protection switching
commands embedded in the spare LVDS link will direct the S/UNI-DUPLEX to
start receiving its traffic from this spare link.
PROPRIETARY AND CONFIDENTIAL TO PMC-SIERRA, INC., AND FOR ITS CUSTOMERS’ INTERNAL USE7
RELEASED
DATA SHEET
PMC-1980581ISSUE 5DUAL SERIAL LINK PHY MULTIPLEXER
PM7350 S/UNI-DUPLEX
The S/UNI-VORTEX resides on the core card and terminates up to 8 LVDS links
connected to 8 S/UNI-DUPLEX devices. The S/UNI-VORTEX implements the
second stage of multiplexing. More than one S/UNI-VORTEX will be required if
more than 8 links are required – as will be the case for a system with more than
8 line cards. The S/UNI-VORTEX device(s) share a high speed parallel bus with
the core card’s traffic management and OA&M layers, as implemented by
devices such as PMC-Sierra’s S/UNI-APEX and the S/UNI-ATLAS.
Some applications use framer or modem devices without integrated I.432
processing1 normally support a clock and data interface, and rely on external
circuitry to detect and generate ATM cell framing and overhead. To support
these applications, the S/UNI-DUPLEX provides a clock and data mode2. In this
mode, the input/output pins that normally interface to the Utopia bus are
configured to support up to 16 clock and data serial interfaces. This type of line
card is shown in Fig. 3. The I.432 processing is transparent to the far end
device, which implies that a single S/UNI-VORTEX can simultaneously interface
to line cards that implement the Utopia bus and to line cards that use clock and
data interfaces.
Fig. 3 Clock and Data PHY Interface
Line Card
Rx Clock
Rx Data
Modem #1
Tx Clock
Tx Data
S/UNI-
DUPLEX
4-wire
LVDS
Modem #16
Some PHY devices provide a 3-line interface consisting of clock, data, and
overhead indication. For these PHYs external circuitry can be used to adapt to
the S/UNI-DUPLEX’s 2-line interface.
Either Utopia mode or clock and data mode can be selected, but not both at once.
PROPRIETARY AND CONFIDENTIAL TO PMC-SIERRA, INC., AND FOR ITS CUSTOMERS’ INTERNAL USE8
RELEASED
DATA SHEET
PMC-1980581ISSUE 5DUAL SERIAL LINK PHY MULTIPLEXER
PM7350 S/UNI-DUPLEX
In the clock and data receive direction the S/UNI-DUPLEX performs bit level ATM
cell delineation function. In the transmit direction the S/UNI-DUPLEX can
operate in either bit or frame aligned mode. In frame mode (also called byte
aligned mode) the two wire transmit interface continuously monitors for gaps in
the transmit clock to determine where the frame or byte alignment should occur.
The circuitry assumes that when a gap in the transmit clock is detected this is
either the framing bit position (e.g. the DS-1 framing bit) or an overhead byte
(e.g. ADSL modem). In either case the next clock period after the gap is
assumed to represent the byte alignment position.
In the multiplexer application discussed previously the S/UNI-DUPLEX’s LVDS
interfaces are connected to one or two S/UNI-VORTEX devices. It is also
possible to interface S/UNI-DUPLEX to S/UNI-DUPLEX via the LVDS link. Since
the S/UNI-DUPLEX bus interface can be configured in several ways (clock and
data, 8/16 bits bus master, 8/16 bit bus slave) there are various applications
where two S/UNI-DUPLEX devices can be used “back-to-back” in order to
perform one or more of the following functions:
1. Interfacing a bus master device to another bus master.
2. Interfacing a bus slave device to another bus slave.
3. Converting between 8 bit and 16 bit buses.
4. Off card or off shelf bus extension.
5. Cell delineation (I.432 processing).
6. Protection switching.
Examples of these types of configurations are shown in Fig. 4 and Fig. 5.
PROPRIETARY AND CONFIDENTIAL TO PMC-SIERRA, INC., AND FOR ITS CUSTOMERS’ INTERNAL USE9
RELEASED
DATA SHEET
PMC-1980581ISSUE 5DUAL SERIAL LINK PHY MULTIPLEXER
PM7350 S/UNI-DUPLEX
Fig. 4 S/UNI-DUPLEX to S/UNI-DUPLEX Applications
Bus
master
8 bit bus
S/UNI-
DUPLEX
LVDS
S/UNI-
DUPLEX
16 bit bus
Example of on-card bus conversion:
8 bit bus master to 16 bit bus master
Bus
master
S/UNI-
DUPLEX
LVDS
S/UNI-
DUPLEX
Example of basic bus extension between cards
Framer #1
LVDS
DUPLEX
Framer #2
S/UNI-
DUPLEX
...
S/UNI-
Bus
master
Bus
slave
Bus
master
Framer # 16
Clock + data
8/16 bit bus
Example of on-card I.432 processing
Fig. 5 S/UNI-DUPLEX to S/UNI-DUPLEX Protection Switching
Bus
master
Bus
master
S/UNI-
DUPLEX
S/UNI-
DUPLEX
S/UNI-
DUPLEX
S/UNI-
DUPLEX
Bus
slave
Bus
slave
Example of protection switching between cards
PROPRIETARY AND CONFIDENTIAL TO PMC-SIERRA, INC., AND FOR ITS CUSTOMERS’ INTERNAL USE10
RELEASED
DATA SHEET
PMC-1980581ISSUE 5DUAL SERIAL LINK PHY MULTIPLEXER
PM7350 S/UNI-DUPLEX
5 BLOCK DIAGRAM
Although separated to improve clarity, many signals in the following diagram
share physical package pins. The use of the SCI-PHY/Any-PHY interfaces and
the clocked serial data interfaces is mutually exclusive.
B
K
O
K
K
L
8
SCIANY
OBUS8
OANYPHY
OMASTER
OENB
OADDR[4:0]
OAVALID
ODAT[15:0]
OPRTY
OSOC
OSX
OFCLK
OCA
LTXD[15:0]
LTXC[15:0]
LRXD[15:0]
LRXC[15:0]
IBU S8
IAN YP HY
IMAS TE R
IEN B
IADDR[4:0]
IAV AL ID
IDAT[15 :0]
IPR TY
ISOC
ISX
IFC LK
ICA
A[7:0]
RDB
WRB
CSB
ALE
IN TB
RSTB
D[7:0]
SCI-PHY
Transmit
Master/
Receive
Slave
Elastic
Store
SCI-PHY
Receive
Master/
Transmit
Slave
Micro-
Processor
Interface
Time-Sliced
ATM
Transmission
Convergence
2 Cell
Buffer
4 Cell
FIFO
to all
blocks
per-PHY
buffers
per-PHY
buffers
T
S
R
Cell
Processor
X
C
R
R
Clock
Synthesis
JTAG
Test Access
Port
8
X
T
RXD1+
RXD1-
TXD1+
TXD1-
RXD2+
RXD2-
TXD2+
TXD2-
REFCLK
TDO
TDI
TCK
TMS
TRS TB
PROPRIETARY AND CONFIDENTIAL TO PMC-SIERRA, INC., AND FOR ITS CUSTOMERS’ INTERNAL USE11
RELEASED
DATA SHEET
PMC-1980581ISSUE 5DUAL SERIAL LINK PHY MULTIPLEXER
PM7350 S/UNI-DUPLEX
6 DESCRIPTION
The PM7350 S/UNI-DUPLEX is a monolithic integrated circuit typically used with
its sister device, the S/UNI-VORTEX, to implement a point-to-point serial
backplane interconnect architecture. The primary role of the S/UNI-DUPLEX is
to interface to up to 32 devices (typically framers or PHYs) and transfer 52-56
byte data cells in serial format to/from a backplane. Devices interface to the
S/UNI-DUPLEX via an 8 or 16-bit SCI-PHY/Utopia/Any-PHY bus, or optionally
via a 16 port clock and data interface.
Each S/UNI-DUPLEX can connect to two 100 to 200 Mb/s Low Voltage
Differential Signal (LVDS) serial links. A microprocessor port provides access to
internal configuration and monitoring registers. The microprocessor port may
also be used to insert and extract cells in support of an embedded
microprocessor communication channel.
BUS INTERFACE:
One of four modes can be selected:
• 8 or 16 bit, Utopia L2 bus master operating at up to 33 MHz bus clock
frequency. Also supports PMC-Sierra’s SCI-PHY bus standard which is
compatible with Utopia L2 but allows extended length cells and supports
an additional bus address signal in order to support 32 PHY devices
rather than Utopia’s 31. See Table 1 for a comparison of these bus
standards.
• 16 port, 4 pin clocked serial data interface (Tx, TxClk, Rx, RxClk), with
integrated I.432 ATM cell delineation operating at up to 52 MHz serial
clock frequency.
• 8 or 16 bit, SCI-PHY/Utopia bus slave operating at up to 52 MHz bus
clock frequency. The slave input port presents itself as 32 addressable
logical channels. The slave output port appears as a single addressable
channel carrying the multiplexed traffic from up to 32 logical channels
where each cell’s channel number can optionally be embedded in the H5
header field (Utopia bus mode) or indicated in a cell prepend (SCI-PHY
bus mode).
• 8 or 16 bit, Any-PHY bus slave (bus protocol compatible with the PM7351
S/UNI-VORTEX) operating at up to 52 MHz bus clock frequency. The
slave input port presents itself as 32 addressable logical channels. The
slave output port appears as a single addressable channel carrying the
multiplexed traffic from up to 32 logical channels. In both directions each
cell’s logical channel number is indicated in a cell prepend.
PROPRIETARY AND CONFIDENTIAL TO PMC-SIERRA, INC., AND FOR ITS CUSTOMERS’ INTERNAL USE12
RELEASED
DATA SHEET
PMC-1980581ISSUE 5DUAL SERIAL LINK PHY MULTIPLEXER
PM7350 S/UNI-DUPLEX
LVDS INTERFACES (both directions):
• Dual 4-wire LVDS serial transceivers each operating at 100 to 200 Mbps
across PCB or backplane traces, or across up to 10 meters of 4-wire
twisted pair cabling for inter-shelf communications.
• Full integrated LVDS clock synthesis and recovery. No external analog
components are required.
• Usable bandwidth (excludes system overhead) of 186 Mbps.
LVDS TRANSMIT DIRECTION
• Simple round robin multiplex of up to 32 PHYs (or 16 clock and data
interfaces) plus the microprocessor port’s cell transfer buffer.
• Multiplexed cell stream broadcast to both LVDS simultaneously.
• 6 bit port ID prepended to each cell for use by ATM layer to identify cell
source (1 of 32 PHYs or processor).
• Back-pressure provided by far end (active link only) to prevent overflow of
far end receiver.
LVDS RECEIVE DIRECTION
• Cells received from the active LVDS link are forwarded to the appropriate
PHY, bit serial interface, or the microprocessor port as specified by a 6 bit
port ID added to each cell at the far end device.
• The LVDS link marked as "spare" is monitored for errors, PHY cells are
discarded, microprocessor port cells are accepted.
• Individual PHY and microprocessor FIFO back-pressure indications are
sent to the far end to prevent FIFO overflows. Per stream back-pressure
prevents head-of-line blocking.
MICROPROCESSOR INTERFACE
• 8 bit data bus, 8 bit address bus.
• Provides read/write access to all configuration and status registers.
• Provides CRC32 calculation and cell transfer registers to support an
embedded microprocessor to microprocessor communication channel
over the LVDS link.
PROPRIETARY AND CONFIDENTIAL TO PMC-SIERRA, INC., AND FOR ITS CUSTOMERS’ INTERNAL USE13
RELEASED
PM7350 S/UNI-DUPLEX
DATA SHEET
PMC-1980581ISSUE 5DUAL SERIAL LINK PHY MULTIPLEXER
7 PIN DIAGRAM
The S/UNI-DUPLEX is packaged in a 160-pin plastic ball grid array (PBGA)
package having a body size of 15 mm by 15 mm and a ball pitch of 1.00 mm.
1413121110987654321
A
B
C
D
E
F
RXD1+RXD1-RESTAVDGNDGND
G
TXD1+TXD1-TAVSRESKGNDGNDISX LTXC[6]
H
RXD2+RXD2-RAVSRAVD
J
TXD2+TXD2-TAVSTAVDVSS
K
TRSTBTDIA[2]RSTBA[6]RDBBIAS
TX8KTCKTMSA[0]A[4]A[5]ALEINTB
D[7]VDDVSSA[1]VSSA[7]VSSSCIANY
D[4]D[6]D[5]A[3]VDDWRBCSBVSS
VDDRX8KV SSRCLK
D[0]D[2]D[1]D[3]
IMASTER
LTXD[15]
IAVALID
LTXC[15]
IBUS8
LTXD[14]
IADDR[4]
LRXD[15]
IADDR[3]
LRXC[15]
VSS
VDD
IADDR[2]
LTXC[14]
IADDR[1]
LRXD[14]
IENB
LRXD[13]
IADDR[0]
LRXC[14]
VSS
IDAT[7]
LRXC[11]
IFCLK
LTXC[9]
IDAT[3]
LTXD[8]
ODAT[15]
LRXC[10]
ICA
LRXC[13]
VDD
VSS
IDAT[11]
LRXC[12]
IDAT[8]
LRXD[11]
VDD
IDAT[1]
LTXD[7]
IDAT[2]
LTXC[8]
VSS
ODAT[14]
LRXD[9]
RSTOB
IPRTY
LTXD[13]
IDAT[14]
LTXD[12]
IDAT[12]
LRXD[12]
VDD
IDAT[6]
LTXD[10]
VSS
ISOC
LTXD[6]
OPRTY
LRXD[10]
ODAT[13]
LRXC[9]
IDAT[15]
LTXC[13]
IDAT[13]
LTXC[12]
IDAT[10]
LTXD[11]
IDAT[9]
LTXC[11]
IDAT[5]
LTXC[10]
IDAT[4]
LTXD[9]
IDAT[0]
LTXC[7]
IANYPHY
LTXD[5]
VDD
ATP0ATP1CAVSCAVD
L
QAVSQAVDVDD
M
VSSV DD
N
P
TDOREFCLK
Unused
LTXD[0]
OMASTER
LTXD[1]
VSS
OADDR[0]
LRXC[0]
OADDR[1]
LRXD[0]
OADDR[3]
LTXC[1]
OADDR[2]
LTXC[0]
VSS
VDDVSS
OAVALID
LRXD[1]
OADDR[4]
LRXC[1]
OENB
LRXD[2]
OSOC
LRXD[3]
VSSVDDVSS
OCA LRXC[ 2]
OSX LRXC[3]
ODAT[1]
LRXC[4]
ODAT[0]
LTXC[2]
ODAT[2]
LRXD[4]
ODAT[3]
LRXD[5]
VDDVSS
ODAT[4]
LRXC[6]
OANYPHY
LTXD[2]
OFCLK
LRXC[5]
VDD
ODAT[5]
LRXD[6]
ODAT[6]
LRXC[7]
OBUS8
LTXD[4]
ODAT[12]
ODAT[9]
LTXC[5]
LTXC[4]
VDD
VSS
ODAT[11]
LRXD[8]
VSS
ODAT[8]
LTXC[3]
ODAT[7]
LRXD[7]
VDD
ODAT[10]
LRXC[8]
Unused
LTXD[3]
BOTTOM VIEW
PROPRIETARY AND CONFIDENTIAL TO PMC-SIERRA, INC., AND FOR ITS CUSTOMERS’ INTERNAL USE14
RELEASED
DATA SHEET
PMC-1980581ISSUE 5DUAL SERIAL LINK PHY MULTIPLEXER
PM7350 S/UNI-DUPLEX
8 PIN DESCRIPTION
Ball
Name
RXD1+
RXD1RXD2+
RXD2-
TXD1+
TXD1TXD2+
TXD2-
Type
Diff.
LVDS
Input
Diff.
LVDS
Output
Ball
No.Function
High Speed LVDS Links
G14
G13
J14
J13
The high-speed receive data (RXD1+/-, RXD2+/-)
inputs present NRZ data from a serial backplane.
Two pairs are provided for redundancy. The active
link may be chosen by the local microprocessor or
determined by a simple handshake.
RXD1+/- and RXD2+/- are truly differential inputs
offering superior common-mode noise rejection.
They have sufficient sensitivity and common-mode
range to support LVDS signals.
H14
H13
K14
K13
The transmit differential data (TXD1+/-, TXD2+/-)
outputs present NRZ encoded data to a serial
backplane. These outputs are open drain current
sinks which interface directly with twisted-pair
cabling or board interconnect. Edge rates are
controlled to minimize radiated emissions.
Both differential links carry identical traffic except
the exact phase relationship is not guaranteed.
REFCLKInputP12The reference clock input (REFCLK) must provide a
jitter-free reference clock. It is used as the
reference clock by both clock recovery and clock
synthesis circuits. Any jitter below 1 MHz is
transferred directly to the TXD1+/- and TXD2+/outputs. The high-speed serial interface bit rate is
eight times the REFCLK frequency.
RES
RESK
AnalogG12
H11
A 4.75k
Ω ±1% resistor must be connected between
these two pins to achieve the correct LVDS output
signal levels.
ATP0
ATP1
AnalogL14
L13
The Analog Test Points (ATP) are provided for
production test purposes. In mission mode they
are high impedance and should be connected to
ground.
PROPRIETARY AND CONFIDENTIAL TO PMC-SIERRA, INC., AND FOR ITS CUSTOMERS’ INTERNAL USE15
RELEASED
DATA SHEET
PMC-1980581ISSUE 5DUAL SERIAL LINK PHY MULTIPLEXER
allows a traceable signal to be transmitted to the far
end of the high-speed serial links via TXD1+/- and
TXD2+/-. A rising edge on TX8K is encoded in the
next cell transmitted.
Although TX8K is targeted at a typical need of
transporting an 8 kHz signal, its frequency is not
constrained to 8 kHz. Any frequency less than the
cell rate is permissible.
presents the timing extracted from one of the
receive high-speed serial links, RXD1+/- or
RXD2+/-.
The rising edge of RX8K is accurate to the nearest
byte boundary of the high-speed serial link;
therefore, a small amount of jitter is present. At a
link rate of 155.52 Mb/s, the jitter is 63ns peak-topeak.
Pulses on RX8K are always 16 high-speed serial
link bit periods wide (two REFCLK periods).
RCLKOutputE11The Recovered Clock (RCLK) output presents the
byte clock for the active receive high-speed serial
link. The RCLK frequency shall be 0.125 of the
RXD1+/- or RXD2+/- bit rate.
Due to the digital clock recovery technique
employed, jitter is introduced as 12.8 ns phase
steps. If the active link is changed, RCLK can not
be guaranteed to be glitch free. Because of these
two factors, RCLK must be cleaned up by a PLL
before it is suitable for use as a timing reference.
Clocked Data Serial Interface
PROPRIETARY AND CONFIDENTIAL TO PMC-SIERRA, INC., AND FOR ITS CUSTOMERS’ INTERNAL USE16
RELEASED
DATA SHEET
PMC-1980581ISSUE 5DUAL SERIAL LINK PHY MULTIPLEXER
selects the type of PHY device interface. If
SCIANY is a logic high, the S/UNI-DUPLEX will be
configured to communicate to the PHY devices via
a shared SCI-PHY Level 2, Utopia L2, or Any-PHY
cell bus. If SCIANY is a logic low, each PHY device
has a dedicated clocked bit serial interface.
The two types of interfaces share common package
pins. Failure to present the correct logic level on
this signal for the application may result in damage
to the S/UNI-DUPLEX or the PHY devices.
When SCIANY is logic high, LTXD[3] and LTXD[0]
become inputs and need to be tied to VDD or VSS
through a pull up or a pull down.
The low-speed receive data (LRXD[15:0]) inputs
provide data from individual modem channels. The
data streams must carry contiguous ATM cells with
valid HCS (Header Check Sequence) bytes.
LRXD[n] can be clocked either by the rising or
falling edge of the corresponding LRXC[n] input,
depending on the value of the LRXCINV bit of the
Master Configuration register. By default, the rising
edge is used.
These inputs are only active if the SCIANY input is
a logic low.
PROPRIETARY AND CONFIDENTIAL TO PMC-SIERRA, INC., AND FOR ITS CUSTOMERS’ INTERNAL USE17
RELEASED
DATA SHEET
PMC-1980581ISSUE 5DUAL SERIAL LINK PHY MULTIPLEXER
The low-speed receive clock (LRXC[15:0]) inputs
provide timing for the receive links. Each LRXC
signal is independent of the others.
Each signal in LRXD[15:0] is sampled either on the
rising or the falling edge of the corresponding
LRXC[15:0] clock, depending the value of the
LRXCINV bit of the Master Configuration register.
By default, the rising edge is used.
The active edge on each LRXC must only occur
during those bit periods containing ATM cell data. It
must be suppressed during bit periods containing
transmission overhead. These inputs are only
active if the SCIANY input is a logic low.
Maximum clock rate is 50 MHz.
P11
The low-speed transmit data signals (LTXD[15:0])
C6
B2
C2
D1
F2
G1
G4
G3
H2
J1
P4
N1
M5
M11
N12
carry the outgoing link data in bit serial format.
Each LTXD signal is independent of the other
signals. The most significant bit of each data byte is
transmitted first.
Each signal in LTXD[15:0] can be updated either on
the rising or falling edge of the corresponding
LTXC[15:0] clock, depending the value of the
LTXCINV bit of the Master Configuration register.
By default, the rising edge is used.
These outputs are only active if the SCIANY input is
a logic low.
When SCIANY is logic high, LTXD[3] and LTXD[0]
become inputs and need to be tied to VDD or VSS
through a pull up or a pull down.
PROPRIETARY AND CONFIDENTIAL TO PMC-SIERRA, INC., AND FOR ITS CUSTOMERS’ INTERNAL USE18
Loading...
+ 211 hidden pages
You need points to download manuals.
1 point = 1 manual.
You can buy points or you can get point for every manual you upload.