■ Truly independent channels
❐ Each channel is able to:
• Perform reclocker function
• Operate at a different signaling rate
• Transport a different data format
■ Internal phase-locked loops (PLLs) with no external PLL
components
■ Selectable differential PECL compatible serial inputs per
channel
❐ Internal DC restoration
■ Redundant differential PECL compatible serial outputs per
channel
❐ No external bias resistors required
❐ Signaling rate controlled edge rates
❐ Source matched for 50Ω transmission lines
■ MultiFrame™ Receive Framer provides alignment options
❐ Comma or full K28.5 detect
❐ Single or multibyte Framer for byte alignment
❐ Low latency option
■ Selectable input and output clocking options
®
technology
Figure 1. HOTLink II™ System Connections
■ Synchronous LVTTL parallel interface
■ JTAG boundary scan
■ Built In Self Test (BIST) for at-speed link testing
■ Link quality indicator by channel
❐ Analog signal detect
❐ Digital signal detect
■ Low power 3W at 3.3V typical
■ Single 3.3V supply
■ 256 ball thermally enhanced BGA
■ 0.25μ BiCMOS technology
■ JTAG device ID ‘0C811069’x
Functional Description
The CYV15G0404DXB Independent Clock Quad HOTLink II™
Transceiver is a point-to-point or point-to-multipoint communications building block enabling the transfer of data over a variety of
high speed serial links including SMPTE 292, SMPTE 259, and
DVB-ASI video applications. The signaling rate can be anywhere
in the range of 195 to 1500 MBaud for each serial link. Each
channel operates independently with its own reference clock
allowing different rates. Each transmit channel accepts parallel
characters in an input register, encodes each character for
transport, and then converts it to serial data. Each receive
channel accepts serial data and converts it to parallel data,
decodes the data into characters, and presents these characters
to an output register. The received serial data can also be
reclocked and retransmitted through the serial outputs. Figure 1
illustrates typical connections between independent video
coprocessors and corresponding CYV15G0404DXB chips.
Cypress Semiconductor Corporation•198 Champion Court•San Jose, CA 95134-1709•408-943-2600
Document #: 38-02097 Rev. *B Revised December 14, 2007
[+] Feedback [+] Feedback
CYV15G0404DXB
The CYV15G0404DXB satisfies the SMPTE-259M and
CYV15G0404DXB T ransceiver Logic Block Diagram
x10
Serializer
Phase
Encoder
8B/10B
Decoder
8B/10B
x11
Framer
Deserializer
TX
RX
x10
Serializer
Encoder
8B/10B
Decoder
8B/10B
x11
Framer
Deserializer
TX
RX
x10
Serializer
Encoder
8B/10B
Decoder
8B/10B
x11
Framer
Deserializer
TX
RX
x10
Serializer
Encoder
8B/10B
Decoder
8B/10B
x11
Framer
Deserializer
TX
RX
TXDA[7:0]
RXDA[7:0]
TXDB[7:0]
RXDB[7:0]
TXDC[7:0]
RXDC[7:0]
TXDD[7:0]
RXDD[7:0]
OUTA1±
OUTA2±
INA1±
INA2±
OUTB1±
OUTB2±
INB1±
INB2±
OUTC1±
OUTC2±
INC1±
INC2±
OUTD1±
OUTD2±
IND1±
IND2±
Align
Buffer
Phase
Align
Buffer
Phase
Align
Buffer
Phase
Align
Buffer
Elasticity
Buffer
Elasticity
Buffer
Elasticity
Buffer
Elasticity
Buffer
TXCTA[1:0]
RXSTA[2:0]
TXCTB[1:0]
RXSTB[2:0]
TXCTC[1:0]
RXSTC[2:0]
TXCTD[1:0]
RXSTD[2:0]
REFCLKA±
REFCLKB±
REFCLKC±
REFCLKD±
SMPTE-292M compliance according to SMPTE EG34-1999
Pathological Test Requirements.
As a second generation HOTLink device, the CYV15G0404DXB
extends the HOTLink family with enhanced levels of integration and
faster data rates, while maintaining serial link compatibility (data,
command, and BIST) with other HOTLink devices. The transmit
(TX) section of the CYV15G0404DXB Quad HOTLink II consists of
four independent byte-wide channels. Each channel accepts
either 8-bit data characters or preencoded 10-bit transmission
characters. Data characters may be passed from the transmit
input register to an integrated 8B/10B Encoder to improve their
serial transmission characteristics. These encoded characters
are then serialized and output from dual Positive ECL (PECL)
compatible differential transmission-line drivers at a bit rate of
either 10 or 20 times the input reference clock for that channel.
The receive (RX) section of the CYV15G0404DXB Quad
HOTLink II consists of four independent byte wide channels.
Each channel accepts a serial bit stream from one of two
PECL-compatible differential line receivers, and using a
completely integrated Clock and Data Recovery PLL, recovers
the timing information necessary for data reconstruction. Each
recovered bit stream is deserialized and framed into characters,
8B/10B decoded, and checked for transmission errors.
Recovered decoded characters are then written to an internal
elasticity buffer, and presented to the destination host system.
The integrated 8B/10B encoder ordecoder may be bypassed for
systems that present externally encoded or scrambled data at
the parallel interface.
The parallel IO interface may be configured for numerous forms
of clocking to provide the highest flexibility in system architecture. In addition to clocking the transmit path with a local
reference clock, the receive interface may also be configured to
present data relative to a recovered clock or to a local reference
clock.
Each transmit and receive channel contains an independent
BIST pattern generator and checker. This BIST hardware allows
at speed testing of the high speed serial data paths in each
transmit and receive section, and across the interconnecting
links.
The CYV15G0404DXB is ideal for port applications where
different data rates and serial interface standards are necessary
for each channel. Some applications include multi-format routers
and switchers.
Pin Definitions
CYV15G0404DXB Quad HOTLink II Transceiver
NameI/O CharacteristicsSignal Description
Transmit Path Data and Status Signals
TXDA[7:0]
TXDB[7:0]
TXDC[7:0]
TXDD[7:0]
TXCTA[1:0]
TXCTB[1:0]
TXCTC[1:0]
TXCTD[1:0]
TXERRA
TXERRB
TXERRC
TXERRD
Transmit Path Clock Signals
REFCLKA±
REFCLKB±
REFCLKC±
REFCLKD±
TXCLKA
TXCLKB
TXCLKC
TXCLKD
Notes
2. When REFCLKx± is configured for half rate operation, these inputs are sampled relative to both the ri sing and falling edges of the associated REFCLKx±.
3. When REFCLKx± is configured for half rate operation, these outputs are presented relative to both the risin g and falling edges of the associated REFCLKx±.
LVTTL Input,
synchronous,
sampled by the
associated
TXCLKx↑ or
REFCLKx↑
[2]
LVTTL Input,
synchronous,
sampled by the
associated
TXCLKx↑ or
REFCLKx↑
[2]
LVTTL Output,
synchronous to
REFCLKx↑
[3]
,
synchronous to
RXCLKx when
selected as
REFCLKx,
asynchronous to
transmit channel
enable/disable,
asynchronous to loss
or return of
REFCLKx±
Differential LVPECL
or single ended
LVTTL input clock
LVTTL Clock Input,
internal pull down
Transmit Data Inputs. TXDx[7:0] data inputs are captured on the rising edge of the
transmit interface clock. The transmit interface clock is selected by the TXCKSELx
latch via the device configuration interface, and passed to the encoder or Transmit
Shifter. When the Encoder is enabled, TXDx[7:0] specifies the specific data or
command character sent.
Transm it Contr ol. TXCTx[1:0] inputs are captured on the rising edge of the transmit
interface clock. The transmit interface clock is selected by the TXCKSELx latch
through the device configuration interface, and passed to the encoder or transmit
shifter. The TXCTA[1:0] inputs identify how the associated TXDx[ 7:0 ] charac ters are
interpreted. When the encoder is bypassed, these inputs are interpreted as data bits.
When the encoder is enabled, these inputs determine if the TXDx[7:0] character is
encoded as data, a special character code, or replaced with other special character
codes. See Table 3 for details.
Transmit Path Error. TXERRx is asserted HIGH to indicate detection of a transmit
phase align buffer underflow or overflow. If an underflow or overflow condition is
detected, TXERRx, for the channel in error, is asserted HIGH and remains asserted
until either a word sync sequence is transmitted on that channel, or the transmit
phase align buffer is recentered with the PABRSTx latch through the device configuration interface. When TXBISTx = 0, the BIST progress is presented on the
associated TXERRx output. The TXERRx signal pulses HIGH for one transmit
character clock period to indicate a pass through the BIST sequence once every 511
or 527 (depending on RXCKSELx) character times. If RXCKSELx = 1, a one
character pulse occurs every 527 character times. If RXCKSELx = 0, a one character
pulse occurs every 511 character times.
TXERRx is also asserted HIGH, when any of these conditions is true:
■ The TXPLL for the associated channel is powered down. This occurs when OE2x
and OE1x for a given channel are simultaneous disabled by setting OE2x = 0 and
OE1x = 0.
■ The absence of the REFCLKx± signal.
Reference Clock. REFCLKx± clock inputs are used as the timing references for the
transmit and receive PLLs. These input clocks may also be selected to clock the
transmit and receive parallel interfaces. When driven by a single ended LVCMOS or
LVTTL clock source, connect the clock source to either the true or complement
REFCLKx input, and leave the alternate REFCLKx input open (floating). When driven
by an L VPECL clock source, the clock must be a differential clock, using both inputs.
Transmit Path Input Clock. When configuration latch TXCKSELx = 0, the
associated TXCLKx input is selected as the character-rate input clock for the
TXDx[7:0] and TXCTx[1:0] inputs. In this mode, the TXCLKx input must be
frequency-coherent to its associated TXCLKOx output clock, but may be offset in
phase by any amount. Once initialized, TXCLKx is allowed to drift in phase as much
as ±180 degrees. If the input phase of TXCLKx drifts beyond the handling capacity
of the phase align buffer, TXERRx is asserted to indicate the loss of data, and remains
asserted until the phase align buffer is initialized. The phase of the TXCLKx i nput
clock relative to its associated REFCLKx± is initialized when the configuration latch
PABRSTx is written as 0. When the associated TXERRx is deasserted, the phase
align buffer is initialized and input characters are correctly captured.
Document #: 38-02097 Rev. *BPage 8 of 44
[+] Feedback [+] Feedback
CYV15G0404DXB
Pin Definitions (continued)
CYV15G0404DXB Quad HOTLink II Transceiver
NameI/O CharacteristicsSignal Description
TXCLKOA
TXCLKOB
TXCLKOC
TXCLKOD
Receive Path Data and Status Signals
RXDA[7:0]
RXDB[7:0]
RXDC[7:0]
RXDD[7:0]
RXSTA[2:0]
RXSTB[2:0]
RXSTC[2:0]
RXSTD[2:0]
Receive Path Clock Signals
RXCLKA±
RXCLKB±
RXCLKC±
RXCLKD±
Device Control Signals
RESET
LV TTL OutputTransmit Clock Output. TXCLKOx output clock is synthesized by each channel’s
transmit PLL and operates synchronous to the internal transmit character clock.
TXCLKOx operates at either the same frequency as REFCLKx± (TXRATE = 0), or at
twice the frequency of REFCLKx± (TXRATE = 1). The transmit clock outputs have no
fixed phase relationship to REFCLKx±.
LVTTL Output,
synchronous to the
selected RXCLK±
output or REFCLKx±
input
LVTTL Output,
synchronous to the
selected RXCLK±
output or REFCLKx±
input
LVTTL Output Clock Receive Clock Output. RXCLKx± is the receive interface clock used to control timing
LVTTL Input,
asynchronous,
internal pull up
Parallel Data Output. RXDx[7:0] parallel data outputs change relative to the receive
interface clock. The receive interface clock is selected by the RXCKSELx latch. If
RXCLKx± is a full rate clock, the RXCLKx± clock outputs are complementary clocks
operating at the character rate. The RXDx[7:0] outputs for the associated receive
channels follow rising edge of RXCLKx+ or falling edge of RXCLKx–. If RXCLKx± is
a half rate clock, the RXCLKx± clock outputs are complementary clocks operating at
half the character rate. The RXDx[7:0] outputs for the associated receive channels
follow both the falling and rising edges of the associated RXCLKx± clock outputs.
Parallel Status Output. RXSTA[2:0] status outputs change relative to the receive
interface clock. The receive interface clock is selected by the RXCKSELx latch. If
RXCLKx± is a full rate clock, the RXCLKx± clock outputs are complementary clocks
operating at the character rate. The RXSTAx[2:0] outputs for the associated receive
channels follow rising edge of RXCLKx+ or falling edge of RXCLKx–. If RXCLKx± is
a half rate clock, the RXCLKx± clock outputs are complementary clocks operating at
half the character rate. The RXSTAx[2:0] outputs for the associated receive channels
follow both the falling and rising edges of the associated RXCLKx± clock outputs.
When the decoder is bypassed, RXSTx[1:0] become the two low-order bits of the
10-bit received character. RXSTx[2] = HIGH indicates the presence of a Comma
character in the Output Register. When the decoder is enabled, RXSTx[2:0] provide
status of the received signal. See Table 11 for a list of received character status.
of the RXDx[7:0] and RXSTA[2:0] parallel outputs. The source of the RXCLKx±
outputs is selected by the RXCKSELx latch via the device configuration interface.
These true and complement clocks are used to control timing of data output transfers.
These clocks are output continuously at either the dual-character rate (1/20
serial bit-rate) or character rate (1/10
as selected by RXRATEx. When configured such that the output data path is clocked
by the REFCLKx± instead of a recovered clock, the RXCLKx± output drivers present
a buffered or divided form (depending on RXRATEx) of the associated REFCLKx±
that are delayed in phase to align with the data. This phase difference allows the user
to select the optimal clock (REFCLKx± or RXCLK±) for setup orhold timing for their
specific system.
When REFCLKx± is a full rate clock, the RXCLKx± rate depends on the value of
RXRATEx.
When REFCLKx± is a half rate clock and RXCKSELx = 0, the RXCLKx± rate depends
on the value of RXRATEx.
When REFCLKx± is a half rate clock and RXCKSELx=1, the RXCLKx± rate does not
depend on the value of RXRATEx and operates at the same rate as REFCLKx±.
Asynchronous Device Reset. RESET
configuration latches in the device to a known state. RESET
for a minimum pulse width. When the reset is removed, all state machines, counters,
and configuration latches are at an initial state. As per the JTAG specifications, the
device RESET
to be reset separately. Refer to JTAG Support on page 23 for the methods to reset
the JTAG state machine. See Table 9 for the initialize values of the device configu-
ration latches.
cannot reset the JTAG controller. Therefore, the JTAG controller has
th
th
the serial bit-rate) of the data being received,
initializes all state machines, counters, and
must be asserted LOW
the
Document #: 38-02097 Rev. *BPage 9 of 44
[+] Feedback [+] Feedback
CYV15G0404DXB
Pin Definitions (continued)
CYV15G0404DXB Quad HOTLink II Transceiver
NameI/O CharacteristicsSignal Description
LDTDENLVTTL Input,
internal pull up
RCLKENA
RCLKENB
LVTTL Input,
internal pull down
RCLKENC
RCLKEND
ULCA
ULCB
LVTTL Input,
internal pull up
ULCC
ULCD
SPDSELA
SPDSELB
3-Level Select
static control input
[4]
SPDSELC
SPDSELD
INSELA
INSELB
LVTTL Input,
asynchronous
INSELC
INSELD
LPENA
LPENB
LPENC
LVTTL Input,
asynchronous,
internal pull down
LPEND
Notes
4. 3-Level Select inputs are used for static configuration. These are ternary inpu ts that make use of logic levels of LOW, MID, and HIGH. The LOW level is usually
implemented by direct connection to V
implemented by not connecting the input (left floating), which allows it to self bias to the proper level.
5. See Device Configuration and Control Interface for detailed information on the operation of the Configuration Interface.
SS
Level Detect Transition Density Enable. When LDTDEN is HIGH, the signal level
detector, range controller , and transition density detector are all enabled to determine
if the RXPLL tracks REFCLKx± or the selected input serial data stream. If the signal
level detector, range controller, or transition density detector are out of their
respective limits while LDTDEN is HIGH, the RXPLL locks to REFCLK± until such a
time they become valid. The (SDASEL[A..D][1:0]) configure the trip level of the signal
level detector. The transition density detector limit is one transition in every 60
consecutive bits. When LDTDEN is LOW, only the range controller determines if the
RXPLL tracks REFCLKx± or the selected input serial data stream. For the cases
when RXCKSELx = 0 (recovered clock), it is recommended to set LDTDEN = HIGH.
Reclocker Enable. When RCLKENx is HIGH, the RXPLL performs clock and data
recovery functions on the input serial data stream and routes the deserialized data
to the RXDx[7:0] and RXSTA[2:0] parallel data outputs as configured by DECBYPx.
It also presents the reclocked serial data to the enabled differential serial outputs.
When RCLKENx is LOW, the receive reclocker is disabled and the TXDx[7:0] parallel
data inputs and TXCTx[1:0] inputs are interpreted (as configured by ENCBYPx) to
generate appropriate 10-bit characters that are presented to the differential serial
outputs.
The reclocker feature is optimized to be used for SMPTE video applications.
Use Local Clock. When ULCx is LOW, the RXPLL is forced to lock to REFCLKx±
instead of the received serial data stream. While ULCx
is LOW, the LFIx for the
associated channel is LOW indicating a link fault.
When ULCx
is HIGH, the RXPLL performs Clock and Data Recovery functions on
the input data streams. This function is used in applications in which a stable
RXCLKx± is needed. In cases when there is an absence of valid data transitions for
a long period of time, or the high-gain differential serial inputs (INx±) are left floating,
there may be brief frequency excursions of the RXCLKx± outputs from REFCLKx±.
Serial Rate Select. The SPDSELx inputs specify the operating signaling rate range
of each channel’s transmit and receive PLL.
LOW = 195 – 400 MBd
MID = 400 – 800 MBd
HIGH = 800 – 1500 MBd.
Receive Input Selector. The INSELx input determines which external serial bit
stream is passed to the receiver’s clock and data recovery circuit. When INSELx is
HIGH, the primary differential serial data input, INx1±, is selected for the associated
receive channel. When INSELx is LOW, the secondary differential serial data input,
INx2±, is selected for the associated receive channel.
Loop Back Enable. The LPENx input enables the internal serial loop back for the
associated channel. When LPENx is HIGH, the transmit serial data from the
associated channel is internally routed to the associated receive Clock an d Data
Recovery (CDR) circuit. All enabled serial drivers on the channel are forced to differential logic-1, and the serial data inputs are ignored. When LPENx is LOW, the
internal serial loop back function is disabled.
(ground). The HIGH level is usually implemented by direct connection to VCC (power). The MID level is usually
Document #: 38-02097 Rev. *BPage 10 of 44
[+] Feedback [+] Feedback
CYV15G0404DXB
Pin Definitions (continued)
CYV15G0404DXB Quad HOTLink II Transceiver
6. See Device Configuration and Control Interface for detailed information on the internal latches.
LVTTL Output,
asynchronous
LVTTL input,
asynchronous,
internal pull up
asynchronous,
internal pull up
asynchronous,
internal pull up
Link Fault Indication Output. LFIx is an output status indicator signal. LFIx is the
logical OR of five internal conditions. LFIx
tions are true:
■ Received serial data rate outside expected range
■ Analog amplitude below expected levels
■ Transition density lower than expected
■ Receive channel disabled
■ ULCx is LOW
■ No REFCLKx±.
Control Write Enable. The WREN input writes the values of the DATA[7:0] bus into
the latch specified by the address location on the ADDR[3:0] bus.
Control Addressing Bus. The ADDR[3:0] bus is the input address bus used to
configure the device. The WREN input writes the values of the DAT A[7:0] bus into the
latch specified by the address location on the AD DR[3:0] bus.
configuration latches within the device, and the initialization value of the latches upon
the assertion of RESET
. Table 10 shows how the latches are mapped in the device.
Control Data Bus. The DATA[7:0] bus is the input data bus used to configure the
device. The WREN
input writes the values of the DATA[7:0] bus into the latch
specified by address location on the ADDR[3:0] bus.
latches within the device, and the initialization value of the latches upon the assertion
of RESET. Table 10 shows how the latches are mapped in the device.
Reframe Mode Select.
Framing Character Select.
Receiver Decoder Mode Select.
Receiver Decoder Bypass.
Receive Clock Select.
Receive Clock Rate Select.
Signal Detect Amplitude Select.
Transmit Encoder Bypass.
Transmit Clock Select.
Transmit PLL Clock Rate Select.
Reframe Enable.
Receive Channel Power Control.
Receive Bist Disabled.
Transmit Bist Disabled.
Differential Serial Output Driver 2 Enable.
Differential Serial Output Driver 1 Enable.
Transmit Clock Phase Alignment Buffer Reset.
Global Latch Enable.
Force Global Latch Enable.
is asserted LOW when any of these condi -
[5]
[5]
Table 9 lists the
[5]
Table 9 lists the configuration
Document #: 38-02097 Rev. *BPage 11 of 44
[+] Feedback [+] Feedback
CYV15G0404DXB
Pin Definitions (continued)
CYV15G0404DXB Quad HOTLink II Transceiver
NameI/O CharacteristicsSignal Description
Factory Test Modes
SCANEN2LVTTL input,
internal pull down
TMEN3LVTTL input,
internal pull down
Analog I/O
OUTA1±
OUTB1±
OUTC1±
OUTD1±
OUTA2±
OUTB2±
OUTC2±
OUTD2±
INA1±
INB1±
INC1±
IND1±
INA2±
INB2±
INC2±
IND2±
JTAG Interface
TMSLVTTL Input,
TCLKLVTTL Input,
TDO3-State LVTTL Output Test Data Out. JTAG data output buf fer . High-Z while JTAG test mode is not selected.
TDILVTTL Input,
TRST
Power
V
CC
GNDSignal and Power Ground for all internal circuits.
CML Differentia l
Output
CML Differentia l
Output
Differential InputPrimary Differential Serial Data Input. The INx1± input accepts the serial data
Differential InputSecondary Differentia l Serial Data Input. The INx2± input accepts the serial data
internal pull up
internal pull down
internal pull up
LVTTL Input,
internal pull up
Factory Test 2. SCANEN2 input is for factory testing only . Leave this input as a NO
CONNECT or GND only.
Factory Test 3. TMEN3 input is for factory testing only. Leave this input as a NO
CONNECT or GND only.
Primary Differential Serial Data Output. The OUTx1± PECL-compatible CML
outputs (+3.3V referenced) are capable of driving terminated transmission lines or
standard fiber-optic transmitter modules, and must be AC coupled for PECL
compatible connections.
Secondary Differential Serial Data Output. The OUTx2± PECL-comp atible CML outputs
(+3.3V referenced) are capable of driving terminated transmission lines or standard fiber
optic transmitter modules, and must be AC coupled for PECL comp atible connections.
stream for deserialization and decoding. The INx1± serial stream is passed to the
receive CDR circuit to extract the data content when INSELx = HIGH.
stream for deserialization and decoding. The INx2± serial stream is passed to the
receiver CDR circuit to extract the data content when INSELx = LOW.
Te s t M o d e S e lect. Used to control access to the JTAG Test Modes. If maintained
high for ≥5 TCLK cycles, the JTAG test controller is reset.
JTAG Test Clock.
Test Data In. JTAG data input port.
JTAG reset signal. When asserted (LOW), this input asynchronously resets the
JT AG test access port controller.
+3.3V Power.
CYV15G0404DXB HOTLink II Operation
The CYV15G0404DXB is a highly configurable, independent
clocking, quad-channel transceiver designed to support reliable
transfer of large quantities of data, using high speed serial links
from multiple sources to multiple destinations. This device
supports four single byte channels.
CYV15G0404DXB Transmit Data Path
Input Register
The bits in the Input Register for each channel support different
assignments, based on if the input data is encoded or
unencoded. These assignments are shown in Table 1.
When the ENCODER is enabled, each input register capture s
eight data bits and two control bits on each input clock cycle.
Document #: 38-02097 Rev. *BPage 12 of 44
When the encoder is bypassed, the control bits are part of the
preencoded 10-bit character.
When the encoder is enabled, the TXCTx[1:0] bits are interpreted along with the associated TXDx[7:0] character to
generate a specific 10-bit transmission character.
Phase Align Buffer
Data from each input register is passed to the associated phase
align buffer, when the TXDx[7:0] and TXCTx[1:0] input registers
are clocked using TXCLKx¦ (TXCKSELx = 0 and TXRA TEx = 0).
When the TXDx[7:0] and TXCTx[1:0] input registers are clocked
using REFCLKx± (TXCKSELx = 1) and REFCLKx± is a full rate
clock, the associated phase alignment buffer in the transmit path
is bypassed. These buffers are used to absorb clock phase
differences between the TXCLKx input clock and the internal
character clock for that channel.
[+] Feedback [+] Feedback
CYV15G0404DXB
Once initialized, TXCLKx is allowed to drift in phase as much as
±180 degrees. If the input phase of TXCLKx drifts beyond the
handling capacity of the phase align buffer, TXERRx is asserted
to indicate the loss of data, and remains asserted until the phase
align buffer is initialized. The phase of the TXCLKx relative to its
associated internal character rate clock is initialized when the
configuration latch PABRSTx is written as 0. When the
associated TXERRx is deasserted, the phase align buffer is
initialized and input characters are correctly captured.
If the phase offset between the initialized location of the input
clock and REFCLKx exceeds the skew handling capabilities of
the phase align buffer, an error is reported on that channel’s
TXERRx output. This output indicates an error continuously until
the phase align buffer for that channel is reset. While the error
remains active, the transmitter for that channel outputs a
continuous C0.7 character to indicate to the remote receiver that
an error condition is present in the link.
Each phase align buffer may be individually reset w ith minimal
disruption of the serial data stream. When a phase align buffer
error is present, the transmission of a word sync sequence
recenters the phase align buffer and clears the error indication.
Note. K28.5 characters may be added or removed from the data
stream during the phase align buffer reset operation. When used
with non-Cypress devices that require a complete 16-character
word sync sequence for proper receive elasticity buffer
operation, follow the phase alignment buffer reset by a word sync
sequence to ensure proper operation.
Encoder
Each character received from the Input register or phase align
buffer is passed to the encoder logic. This block interprets each
character and any associated control bits, and outputs a 10-bit
transmission character.
Depending on the operational mode, the generated transmission
character may be
■ The 10-bit preencoded character accepted in the input register.
■ The 10-bit equivalent of the 8-bit Data character accepted in
the input register
■ The 10-bit equivalent of the 8-bit Special Character code
accepted in the input register
■ The 10-bit equivalent of the C0.7 violation character if a phase
align buffer overflow or underflow error is present
■ A character that is part of the 511-character BIST sequence
■ A K28.5 character generated as an individual character or as
part of the 16-character Word Sync Sequence
Data Encoding
Raw data, as received directly from the transmit input register, is
seldom in a form suitable for transmission across a serial link.
The characters must usually be processed or transformed to
guarantee
■ a minimum transition density (to allow the receive PLL to extract
a clock from the serial data stream)
■ A DC-balance in the signaling (to prevent baseline wander)
■ Run length limits in the serial data (to limit the bandwidth
requirements of the serial link)
■ the remote receiver a way of determining the correct character
boundaries (framing)
When the encoder is enabled (ENCBYPx = 1), th e characters
transmitted are converted from data or special character codes
to 10-bit transmission characters, using an integrated 8B/10B
encoder. When directed to encode the character as a special
character code, the encoder uses the special character encoding
rules listed in Table 15. When directed to encode the character
as a data character, it is encoded using the data character
encoding rules in Table 14.
The 8B/10B encoder is standards compliant with ANSI/NCITS
ASC X3.230-1994 Fibre Channel, IEEE 802.3z Gigabit Ethernet,
the IBM® ESCON® and FICON™ channels, ETSI DVB-ASI, and
ATM Forum st andards for data transport.
Many of the special character codes listed in Table 15 may be
generated by more than one input character. The
CYV15G0404DXB is designed to support two independent (but
non-overlapping) special character code tables. This allows the
CYV15G0404DXB to operate in mixed environments with other
Cypress HOTLink devices using the enhanced Cypress
command code set, and the reduced command sets of other
non-Cypress devices. Even when used in an environment that
normally uses non-Cypress Special Character codes, the
selective use of Cypress command codes can permit operation
where running disparity and error handling must be managed.
Following conversion of each input character from eight bits to a
10-bit transmission character, it is passed to the transmit shifter
and is shifted out LSB first, as required by ANSI and IEEE
standards for 8B/10B coded serial data streams.
Document #: 38-02097 Rev. *BPage 13 of 44
[+] Feedback [+] Feedback
CYV15G0404DXB
Transmit Modes
Encoder Bypass
When the Encoder is bypassed, the character captured from the
TXDx[7:0] and TXCTx[1:0] input register is passed directly to the
transmit shifter without modification. With the encoder bypassed,
the TXCTx[1:0] inputs are considered part of the data character
and do not perform a control function that would otherwise
modify the interpretation of the TXDx[7:0] bits. The bit usage and
mapping of these control bits when the Encoder is bypassed is
shown in Table 2.
When the encoder is enabled, the TXCTx[1:0] data control bits
control the interpretation of the TXDx[7:0] bits and the characters
generated by them. These bits are interpreted as listed in
Table 3.
Table 3. Transmit Modes
TXCTx[1]TXCTx[0]Characters Generated
00Encoded data character
01K28.5 fill character
10Special character code
1116-character Word Sync Sequence
Word Sync Sequence
When TXCTx[1:0] = 11, a 16-character sequence of K28.5
characters, known as a word sync sequence, is generated on the
associated channel. This sequence of K28.5 characters may
start with either a positive or negative disparity K28.5 (as determined by the current running disparity and the 8B/10B coding
rules). The disparity of the second and third K28.5 characters in
this sequence are reversed from what normal 8B/10B coding
rules would generate. The remaining K28.5 characters in the
sequence follow all 8B/10B coding rules. The disparity of the
generated K28.5 characters in this sequence follow a pattern of
either ++––+–+–+–+–+–+– or ––++–+–+–+–+–+–+.
The generation of this sequence, once started, cannot be
stopped until all 16 characters have been sent. The content of
the associated input registers are ignored for the duration of this
[7]
a
b
c
d
e
i
f
g
h
j
sequence. At the end of this sequence, if the TXCTx[1:0] = 11
condition is sampled again, the sequence restarts and remains
uninterruptible for the following 15 character clocks.
Transmit BIST
Each transmit channel contains an internal pattern generator that
can be used to validate both the link and device operation. These
generators are enabled by the associated TXBISTx latch
through the device configuration interface. When enabled, a
register in the associated transmit channel becomes a signature
pattern generator by logically converting to a Linear Feedback
Shift Register (LFSR). This LFSR generates a 511-character (or
526-character) sequence that includes all data and special
character codes, including the explicit violation symbols. This
provides a predictable yet pseudo-random sequence that can be
matched to an identical LFSR in the attached Receiver(s).
A device reset (RESET sampled LOW) presets the BIST enable
latches to disable BIST on all channels.
All data and data-control information present at the associated
TXDx[7:0] and TXCTx[1:0] inputs are ignored when BIST is
active on that channel. If the receive channels are configured for
reference clock operation, each pass is preceded by a
16-character word sync sequence to allow elasticity buffer
alignment and management of clock frequency variations.
Transmit PLL Clock Multiplier
Each Transmit PLL Clock Multiplier accepts a character rate or
half character-rate external clock at the associated REFCLKx±
input, and that clock is multiplied by 10 or 20 (as selected by
TXRATEx) to generate a bit rate clock for use by the transmit
shifter. It also provides a character rate clock used by the
transmit paths, and outputs this character rate clock as
TXCLKOx.
Each clock multiplier PLL is able to accept a REFCLKx± input
between 19.5 MHz and 150 MHz, however, this clock range is
limited by the operating mode of the CYV15G0404DXB clock
multiplier (TXRATEx) and by the level on the associated
SPDSELx input.
SPDSELx are 3-level select
operating ranges for the serial data outputs and inputs of the
associated channel. The operating serial signaling rate and
allowable range of REFCLKx± frequencies are listed in Table 4.
Table 4. Operating Speed Settings
SPDSELxTXRATE
LOW
1reserved195 – 400
019.5 – 40
MID (Open)120 – 40400 – 800
040 – 80
HIGH140 – 75800 – 1500
080 – 150
[4]
inputs that select one of three
REFCLKx±
Frequency
(MHz)
Signaling
Rate (MBaud)
Document #: 38-02097 Rev. *BPage 14 of 44
[+] Feedback [+] Feedback
CYV15G0404DXB
The REFCLKx± inputs are differential inputs with each input
internally biased to 1.4V. If the REFCLKx+ input is connected to
a TTL, LVTTL, or LVCMOS clock source, the input signal is
recognized when it passes through the internally biased
reference point. When driven by a single-ended TTL, LVTTL, or
LVCMOS clock source, connect the clock source to either the
true or complement REFCLKx input, and leave the alternate
The local internal loopback (LPENx) allows the serial transmit
data outputs to be routed internally back to the clock and data
recovery circuit associated with each channel. When configured
for local loopback, the associated transmit serial driver outputs
are forced to output a differential logic-1. This prevents local
diagnostic patterns from being broadcast to attached remote
receivers.
REFCLKx input open (floating).
When both the REFCLKx+ and REFCLKx– inputs are
connected, the clock source must be a differential clock. This can
either be a differential LVPECL clock that is DC- or AC-coupled
or a differential LVTTL or LVCMOS clock.
By connecting the REFCLKx– input to an external voltage
source, it is possible to adjust the reference point of the
REFCLKx+ input for alternate logic levels. When doing so,
ensure that the input differential crossing point remains within the
parametric range supported by the input.
Serial Output Drivers
The serial output interface drivers use differential Current Mode
Logic (CML) drivers to provide source matched drivers for transmission lines. These drivers accept data from the transmit
shifters. They have signal swings equivalent to that of standard
PECL drivers, and are capable of driving AC-coupled optical
modules or transmission lines. When configured for local
loopback (LPENx = HIGH), all enabled serial drivers are
configured to drive a static differential logic 1.
Transmit Channels Enabled
Each driver can be enabled or disabled separately using the
device configuration interface.
When a driver is disabled throug h the configuration interface, it
is internally powered down to reduce device power. If both serial
drivers for a channel are in this disabled state, the associated
internal logic for that channel is also powered down. A device
reset (RESET sampled LOW) disables all output drivers.
[8]
CYV15G0404DXB Receive Data Path
Serial Line Receivers
Two differential line receivers, INx1± and INx2±, are available on
each channel for accepting serial data streams. The active serial
line receiver on a channel is selected using the associated
INSELx input. The serial line receiver inputs are differential, and
Signal Detect/Link Fault
Each selected line receiver (that is routed to the clock and data
recovery PLL) is simultaneously monitored for:
■ Analog amplitude above amplitude level selected by SDASELx
■ Transition density above the specified limit
■ Range controls report the received data stream inside normal
frequency range (±1500 ppm)
■ Receive channel enabled
■ The presence of a reference clock
■ ULCx is not asserted.
All of these conditions must be valid for the signal detect block
to indicate a valid signal is present. This status is presented on
the LFIx (Link Fault Indicator) output associated with each
receive channel, which changes synchronous to the selected
receive interface clock.
Analog Amplitude
While most signal monitors are based on fixed constants, the
analog amplitude level detection is adjustable to allow operation
with highly attenuated signals, or in high noi se environments.
The analog amplitude level detection is set by the SDASELx
latch via device configuration interface. The SDASELx latch sets
the trip point for the detection of a va lid signal at one of three
levels, as listed in Table 5. This control input affects the analog
monitors for all receive channels.
Table 5. Analog Amplitude Detect Valid Signal Levels
SDASEL Typical Signal with Peak Amplitudes Above
00Analog Signal Detector is disabled
01140 mV p-p differential
10280 mV p-p differential
11420 mV p-p differential
[9]
can accommodate wire interconnect and filtering losses or transmission line attenuation greater than 16 dB. For normal
operation, these inputs should receive a signal of at least VI
100 mV , or 200 mV peak-to-peak differential. Each Line Receiver
DIFF
can be DC- or AC-coupled to +3.3V powered fiber optic interface
modules (any ECL/PECL family, not limited to 100K PECL) or
AC-coupled to +5V powered optical modules. The common
mode tolerance of these line receivers accommodates a wide
The analog signal detect monitors are active for the line receiver
as selected by the associated INSELx input. When configured
>
for local loopback, no input receivers are selected, and the LFIx
output for each channel reports only the receive VCO frequency
out-of-range and transition density status of the associated
transmit signal. When local loopback is active, the associated
analog signal detect monitor is disabled.
range of signal termination voltages. Each receiver provides
internal DC-restoration, to the center of the receiver ’s common
mode range, for AC-coupled signals.
Notes
8. When a disabled transmit channel (i.e., both outputs disabled) is re-enabled, the dat a on the serial outputs may not meet all timing specifications for up to 250 ms.
9. The peak amplitudes listed in this table are for ty pical waveforms t hat h ave g enerally 3 – 4 t ransi tio ns for every t en bits. In a worse case environment the signals may
have a sign-wave appearance (highest transition density with repeat ing 0101...). Signal peak amplit udes levels within this en vironmen t type could increase the value s
in the table above by approximately 100 mV.
Document #: 38-02097 Rev. *BPage 15 of 44
[+] Feedback [+] Feedback
CYV15G0404DXB
Transition Density
The transition detection logic checks for the absence of transitions spanning greater than six transmission characters (60 bits).
If no transitions are present in the data received, the detection
logic for that channel asserts LFIx
Range Controls
The CDR circuit includes logic to monitor the frequency of the
PLL Voltage Controlled Oscillator (VCO) used to sample the
incoming data stream. This logic ensures that the VCO operates
at, or near the rate of the incoming data stream for two primary
cases:
■ When the incoming data stream resumes after a time in which
it has been “missing.”
■ When the incoming data stream is outside the acceptable
signaling rate range.
To perform this fu nction, the frequency of the RXPLL VCO is
periodically compared to the frequency of the REFCLKx± input.
If the VCO is running at a frequency beyond ±1500 ppm, as
defined by the REFCLKx± frequency, it is periodically forced to
the correct frequency (as defined by REFCLKx±, SPDSELx, and
TXRATEx) and then released in an attempt to lock to the input
data stream.
The sampling and relock period of the range control is calculated
in the following manner: RANGE_CONTROL_
SAMPLING_PERIOD = (RECOVERED BYTE CLOCK PERIOD)
* (4096).
During the time that the range control forces the RXPLL VCO to
track REFCLKx±, the LFIx
serial data stream is applied, it may take up to one RANGE
CONTROL SAMPLING PERIOD before the PLL locks to the
input data stream, after which LFIx
Receive Channel Enabled
The CYV15G0404DXB contains four receive channels that can
be independently enabled and disabled. Each channel can be
enabled or disabled separately through the RXPLLPDx input
latch as controlled by the device configuration interface. When
the RXPLLPDx latch = 0, the associated PLL and analog circuitry
of the channel is disabled. Any disabled channel indicates a
constant link fault condition on the LFIx
RXPLLPDx = 1, the associated PLL and receive channel is
enabled to receive and decode a serial stream.
Note. When a disabled receive channel is reenabled, the status
of the associated LFIx
the associated channel may be indeterminate for up to 2 ms.
output is asserted LOW. After a valid
output and data on the parallel outputs for
.
should be HIGH.
output. When
Clock/Data Recovery
The extraction of a bit-rate clock and recovery of bits from each
received serial stream is performed by a separate CDR block
within each receive channel. The clock extraction function is
performed by an integrated PLL that tracks the frequency of the
transitions in the incoming bit stream and align the phase of the
internal bit rate clock to the transitions in the selected serial data
stream.
Each CDR accepts a character rate (bit-rate
half-character rate (bit-rate
associated REFCLKx± input. This REFCLKx± input is used to
■ Ensure that the VCO (within the CDR) is operating at the correct
frequency (rather than a harmonic of the bit-rate)
■ Reduce PLL acquisition time
■ Limit unlocked frequency excursions of the CDR VCO when
there is no input data present at the selected serial line receiver.
Regardless of the type of signal present, the CDR attempts to
recover a data stream from it. If the signalling rate of the
recovered data stream is outside the limits set by the range
control monitors, the CDR tracks REFCLKx± instead of the data
stream. Once the CDR output (RXCLK±) frequency returns close
to REFCLKx± frequency, the CDR input is switched back to the
input data stream. If no data is present at the selected line
receiver, this switching behavior may result in brief RXCLK±
frequency excursions from REFCLKx±. However, the validity of
the input data stream is indicated by the LFIx
frequency of REFCLKx± is required to be within ±1500 ppm of
the frequency of the clock that drives the REFCLKx± input of the
remote transmitter to ensure a lock to the incoming data stream.
For systems using multiple or redundant connections, the LFIx
can be output to select an alternate data stream. When an LFIx
indication is detected, external logic can toggle selection of the
associated INx1± and INx2± input through the associated
INSELx input. When a port switch takes place, it is necessary for
the receive PLL for that channel to reacquire the new serial
stream and frame to the incoming character boundaries.
÷ 20) reference clock from the
÷ 10) or
output. The
Reclocker
The CYV15G0404DXB contains a reclocker mode on each
receive channel that can be independently enabled and
disabled. When the reclocker mode is enabled by RCLKENx, the
received serial data is reclocked and transmitted through the
enabled differential serial outputs of the selected channel. In the
reclocker mode, the RXPLL performs clock and data recovery
functions on the input serial data stream and the reclocked serial
data is routed to the enabled differential serial outputs. The serial
data is also routed to the deserializer and the deserialized data
is presented to the RXDx[7:0] and RXSTA[2:0] parallel data
outputs as configured by DECBYPx. When the reclocker is
enabled, the data on the TXDx[7:0] and TXCT[1:0] is ignored and
not transmitted through the enabled serial outputs.
Deserializer/Framer
Each CDR circuit extracts bits from the associated serial data
stream and clocks these bits into the shifter/framer at the bit clock
rate. When enabled, the framer examines the data stream
looking for one or more COMMA or K28.5 characters at all
possible bit positions. The location of this character in the data
stream determines the character boundaries of all following
characters.
Framing Character
The CYV15G0404DXB allows selection of different framing
characters on each channel. Two combinations of framing
characters are supported to meet the requirements of different
interfaces. The selection of the framing character is made
Document #: 38-02097 Rev. *BPage 16 of 44
[+] Feedback [+] Feedback
CYV15G0404DXB
through the FRAMCHARx latches through the configuration
Note
10.The standard definition of a Comma contains only seven bits. However, since all valid Comma characters within the 8B/10B character set also have the eighth
bit as an inversion of the seventh bit, the compare p attern is extended to a full eight bits to reduce the possibility of a framing error.
interface.
The specific bit combinations of these framing characters are
listed in Table 6. When the specific bit combination of the
selected framing character is detected by the framer, the boundaries of the characters present in the received data stream are
known.
Table 6. Framing Character Selector
FRAMCHARx
0COMMA+
1–K28.5
Bits Detected in Framer
Character NameBits Detected
[10]
COMMA–
00111110XX
or 11000001XX
0011111010 or
+K28.5
1100000101
Framer
The framer on each channel operates in one of three different
modes. Each framer is enabled or disabled using the RFENx
latches using the configuration interface. When the framer is
disabled (RFENx = 0), no combination of received bits alters the
frame information.
When the low latency framer is selected (RFMODEx[1:0] = 00),
the framer operates by stretching the recovered character clock
until it aligns with the received character boundaries. In this
mode the framer starts its alignment process on the first
detection of the selected framing character. T o reduce the impact
on external circuits that use the recovered clock, the clock period
is not stretched by more than two bit periods in any one clock
cycle. When operated with a character rate output clock, the
output of properly framed characters may be delayed by up to
nine character clock cycles from the detection of the selected
framing character. When operated with a half character rate
output clock, the output of properly framed characters may be
delayed by up to 14 character clock cycles from the detection of
the framing character.
When RFMODEx[1:0] = 10, the Cypress-Mode Multi-Byte framer
is selected. The required detection of multiple framing characters
makes the associated link much more robust to incorrect framing
due to aliased SYNC characters in the data stream. In this mode,
the framer does not adjust the character clock boundary, but
instead aligns the character to the already recovered character
clock. This ensures that the recovered clock does not contain
any significant phase changes or hops during normal operation
or framing, and allows the recovered clock to be replicated and
distributed to other external circuits or components using
PLL-based clock distribution elements. In this framing mode the
character boundaries are only adjusted if the selected framing
character is detected at least twice within a span of 50 bits, with
both instances on identical 10-bit character boundaries.
When RFMODEx[1:0] = 01, the Alternate-mode Multi-Byte
Framer is enabled. Like the Cypress-mode Multi-Byte Framer,
multiple framing characters must be detected before the
character boundary is adjusted. In this mode, the data stream
must contain a minimum of four of the selected framing
characters, received as consecutive characters, on identical
10-bit boundaries, before character framing is adjusted.
10B/8B Decoder Block
The decoder logic block performs two primary functions:
■ Decoding the received transmission characters to data and
special character codes
■ Comparing generated BIST patterns with received characters
to permit at-speed link and device testing
The framed parallel output of each deserializer shifter is passed
to its associated 10B/8B Decoder where, if the decoder is
enabled, the input data is transformed from a 10-bit transmission
character back to the original data or special character code.
This block uses the 10B/8B decoder patterns in Table 14 and
Table 15. Received special code characters are decoded using
Table 15. Valid data characters are indicated by a 000b bit
combination on the associated RXSTx[2:0] status bits, and
special character codes are indicated by a 001b bit combination
of these status outputs. Framing characters, invalidpatterns,
disparity errors, and synchronization status are presented as
alternate combinations of these status bits.
When DECBYPx = 0, the 10B/8B decoder is bypassed through
the configuration interface. When bypassed, raw 10-bit
characters are passed through the receiver and presented at the
RXDx[7:0] and the RXSTA[1:0] outputs as 10-bit wide
characters.
When the decoder is enabled by setting DECBYPx = 1 through
the configuration interface, the 10-bit transmission characters
are decoded using Table 14 and Table 15. Received Special
characters are decoded using Table 15. The columns used in
Table 15 are determined by the DECMODEx latch through the
device configuration interface. When DECMODEx = 0 the
ALTERNATE table is used and when DECMODEx = 1 the
CYPRESS table is used.
Receive BIST Operation
The receiver channel contains an internal pattern checker that
can be used to validate both device and link operation. These
pattern checkers are enabled by the associated RXBISTx latch
using the device configuration interface. When enabled, a
register in the associated receive channel becomes a si gnature
pattern generator and checker by logically converting to a Linear
Feedback Shift Register (LFSR). This LFSR generates a
511-character or 526-character sequence that includes all data
and special character codes, including the explicit violation
symbols. This provides a predictable yet pseudo random
sequence that can be matched to an identical LFSR in the
attached transmitters. When synchronized with the received
data stream, the associated Receiver checks each character in
the Decoder with each character generate d by the LFSR and
indicates compare errors and BIST status at the RXSTx[2:0] bits
of the Output Register.
When BIST is first recognized as being enabled in the Receiver,
the LFSR is preset to the BIST-loop start code of D0.0. This code
D0.0 is sent only once per BIST loop. The status of the BIST
progress and any character mismatches are presented on the
RXSTx[2:0] status outputs.
Document #: 38-02097 Rev. *BPage 17 of 44
[+] Feedback [+] Feedback
CYV15G0404DXB
Code rule violations or running disparity errors that occur as part
Note
11.When the receive paths are configured for REFCLKx± operation, each pass must be preceded by a 16-character Word Sync Sequence to allow management
of clock frequency variations.
of the BIST loop do not cause an error indication. RXSTx[2:0]
indicates 010b or 100b for one character period per BIST loop to
indicate loop completion. This status can be used to check test
pattern progress. These same status values are presented when
the decoder is bypassed and BIST is enabled on a receive
channel.
The specific status reported by the BIST state machine are listed
in Table 11. These same codes are reported on the receive
status outputs.
The specific patterns checked by each receiver are described in
detail in the Cypress application note “HOTLink Built-In
Self-T est.” The sequence compared by the CYV15G0404DXB is
identical to that in the CY7B933, CY7C924DX, and
CYP(V)15G0401DXB, allowing interoperable systems to be built
when used at compatible serial signaling rates.
If the number of invalid characters received ever exceeds the
number of valid characters by 16, the receive BIST state
machine aborts the compare operations and resets the LFSR to
the D0.0 state to look for the start of the BIST sequence again.
When Receive BIST is enabled on a channel, do not enable the
low latency framer. The BIST sequence contains an aliased
K28.5 framing character, which causes the receiver to update its
character boundaries incorrectly.
The receive BIST state machine requires the characters to be
correctly framed for it to detect the BIST sequence. If the low
latency framer is enabled, the framer misaligns to an aliased
SYNC character within the BIST sequence. If the alternate
multi-byte framer is enabled and the receiver outputs are clocked
relative to a recovered clock, it is generally necessary to frame
the receiver before BIST is enabled. If the receive outputs are
clocked relative to REFCLKx±, the transmitter precedes every
511 character BIST sequence with a 16 character word sync
sequence.
[11]
A device reset (RESET sampled LOW) presets the BIST enable
latches to disable BIST on all channels.
Receive Elasticity Buffer
Each receive channel contains an elasticity buffer that is
designed to support multiple clocking modes. These buffers
allow data to be read using a clock that is asynchronous in both
frequency and phase from the elasticity buffer write clock, or to
be read using a clock that is frequency coherent but with uncontrolled phase relative to the elasticity buffe r w r it e cl ock.
If the chip is configured for operation with a recovered clock, the
elasticity buffer is byp assed.
Each elasticity buffer is 10 characters deep, and supports and an
11 bit wide data path. It is capable of supporting a decoded
character and three status bits for each character present in the
buffer. The write clock for these buffers is always the recovered
clock for the associated read channel.
Receive Modes
When the receive channel is clocked by REFCLKx±, the
RXCLKx± outputs present a buffered or divided (depending on
RXRATEx) and delayed form of REFCLKx±. In this mode, the
receive elasticity buffers are enabled. For REFCLKx± clocking,
the elasticity buffers must be able to insert K28.5 characters and
delete framing characters as appropriate.
The insertion of a K28.5 or deletion of a framing cha racter can
occur at any time on any channel. However, the actual timing of
these insertions and deletions is controlled in part by how the
transmitter sends its data. Insertion of a K28.5 character can only
occur when the receiver has a framing character in the elasticity
buffer. Likewise, to delete a framing character , one must also be
in the elasticity buffer. To prevent a buffer overflow or underflow
on a receive channel, a minimum density of framing characters
must be present in the received data streams.
When the receive channel output register is clocked by a
recovered clock, no characters are added or deleted and the
receiver elasticity buffer is bypassed.
Power Control
The CYV15G0404DXB supports user control of the powered up
or down state of each transmit and receive channel. The receive
channels are controlled by the RXPLLPDx latch through the
device configuration interface. When RXPLLPDx = 0, the
associated PLL and analog circuitry of the channel is di sabled.
The transmit channels are controlled by the OE1x and the OE2x
latches through the device configuration interface. When a driver
is disabled through the configuration interface, it is internally
powered down to reduce device power. If both serial drivers for
a channel are in this disabled state, the associated internal logic
for that channel is powered down as well.
Device Reset State
When the CYV15G0404DXB is reset by assertion of RESET , all
state machines, counters, and configuration latches in the device
are initialized to a reset state, and the elasticity buffer pointers
are set to a nominal offset. Additionally, the JTAG controller must
also be reset to ensure valid operation (even if JTAG testing is
not performed). See the JTAG Support section for JTAG state
machine initialization. See Table 9 for the initialize values of the
configuration latches.
Following a device reset, it is necessary to enable the transmit
and receive channels used for normal operation. This is done by
sequencing the appropriate values on the device configuration
interface.
[5]
Output Bus
Each receive channel presents an 11-signal output bus
consisting of
■ An 8-bit data bus
■ A 3-bit status bus.
The signals present on this output bus are modified by the
present operating mode of the CYV15G0404DXB as selected by
the DECBYPx configuration latch. This mapping is shown in
When the 10B/8B decoder is bypassed, the framed 10-bit value
is presented to the associated output register, along with a status
output signal indicating if the character in the output register is
one of the selected framing characters. The bit usage and
mapping of the external signals to the raw 10B transmission
character is shown in Table 8.
The COMDETx status output operates the same regardless of
the bit combination selected for character framing by the
FRAMCHARx latch. COMDETx is HIGH when the character in
the output register contains the selected framing character at the
proper character boundary, and LOW for all other bit combinations.
When the low-latency framer and half rate receive port clocking
are also enabled, the framer stretches the recovered clock to the
nearest 20-bit boundary such that the rising edge of RXCLKx+
occurs when COMDETx is present on the associated output bus.
When the Cypress or alternate mode framer is enabled and half
rate receive port clocking is also enabled, the output clock is not
modified when framing is detected, but a single pipeline stage
may be added or subtracted from the data stream by the framer
logic such that the rising edge of RXCLKx+ occurs when
COMDETx is present on the associated output bus.
This adjustment only occurs when the framer is enabled. When
the framer is disabled, the clock boundaries are not adjusted,
and COMDETx may be asserted during the rising edge of
RXCLKx– (if an odd number of characters were received
following the initial framing).
Receive Status Bits
When the 10B/8B decoder is enabled, each character presented
at the output register includes three associated status bits.
These bits are used to identify
■ If the contents of the data bus are valid
■ The type of character present
■ The state of receive BIST operations
■ Character violations
These conditions often overlap; for example, a valid data
character received with incorrect running disparity is not reported
as a valid data character. It is instead reported as a decoder
violation of some specific type. This implies a hierarchy or priority
level to the various status bit combinations. The hierarchy and
value of each status are listed in Table 11.
A second status mapping, listed in Table 11, is u sed when the
receive channel is configured for BIST operation. This status is
used to report receive BIST status and progress.
BIST Status State Machine
When a receive path is enabled to look for and compare the
received data stream with the BIST pattern, the RXSTx[2:0] bits
identify the present state of the BIST compare operation.
The BIST state machine has multiple states, as shown in
Figure 2 and Table 11. When the receive PLL detects an
out-of-lock condition, the BIST state is forced to the S tart-of-BIST
state, regardless of the present state of the BIST state machine.
If the number of detected errors ever exceeds the number of
valid matches by greater than 16, the state machine is forced to
the WAIT_FOR_BIST state where it monitors the receive path for
the first character of the next BIST sequence (D0.0). Also, if the
Elasticity Buffer ever hits an overflow/underflow condition, the
status is forced to the BIST_START until the buf fer is re-centered
(approximately nine character periods).
To ensure compatibility between the source and destination
systems when operating in BIST modes, the sending and
receiving ends of the link must use the same receive clock
configuration.
Device Configuration and Control Interface
The CYP(V)15G0404DX is highly configurable through the
configuration interface. The configuration interface allows the
device to be configured globally or allows each channel to be
configured independently. Table 9 lists the configuration latches
within the device including the initialization value of the latche s
upon the assertion of RESET
are mapped in the device. Each row in the Table 10 maps to a
8-bit latch bank. There are 16 such write-only latch banks. When
WREN
= 0, the logic value in the DAT A[7:0] is latched to the latch
bank specified by the values in ADDR[3:0]. The second column
of Table 10 specifies the channels associated with the corresponding latch bank. For example, the first three latch banks (0,1
and 2) consist of configuration bits for channel A. The latch banks
. Table 10 shows how the latches
Document #: 38-02097 Rev. *BPage 19 of 44
[+] Feedback [+] Feedback
CYV15G0404DXB
12, 13, and 14 consist of global configuration bits and the last
latch bank (15) is the mask latch bank that can be configured to
perform bit-by-bit configuration.
Global Enable Function
The global enable function, controlled by the GLENx bits, is a
feature that is used to reduce the number of write operations
needed to setup the latch banks. This function is beneficial in
systems that use a common configuration in multiple channels.
The GLENx bit is present in bit 0 of latch banks 0 through 11 only .
Its default value (1) enables the global update of the latch bank's
contents. Setting the GLENx bit to 0 disables this functionality.
Latch Banks 12, 13, and 14 load values in the related latch banks
in a global manner. A write operation to latch bank 12 could do
a global write to latch banks 0, 3, 6, and 9 depending on the value
of GLENx in these latch banks; latch bank 13 could do a global
write to latch banks 1, 4, 7, and 10; and latch banks 14 could do
a global write to latch banks 2, 5, 8, and 11. The GLENx bit
cannot be modified by a global write operation.
Force Global Enable Function
FGLENx forces the global update of the target latch banks, but
does not change the contents of the GLENx bits. If FGLENx = 1
for the associated global channel, FGLENx forces the global
update of the target latch banks.
Mask Function
An additional latch bank (15) is used as a global mask vector to
control the update of the configuration latch banks on a bit-by-bit
Table 9. Device Configuration and Control Latch Descriptions
Reframe Mode Select. The initialization value of the RFMODEx [1:0] latches = 10. RFMODEx is used to select
the operating mode of the framer. When RFMODEx[1:0] = 00, the low-latency framer is selected. This frames
on each occurrence of the selected framing character(s) in the received data stream. This mode of framing
stretches the recovered clock for one or multiple cycles to align that clock with the recovered data. When
RFMODEx[1:0] = 01, the altern ate mo de Mult i-B yte parallel framer is selected. This requires detection of the
selected framing character(s) in the received serial bit stream, on identical 10-bit boundaries, on four directly
adjacent characters. The recovered character clock remains in the same phasing regardless of character
offset. When RFMODEx[1:0] =10, the Cypress-mode Multi-Byte parallel framer is selected. This requires a
pair of the selected framing character(s), on identical 10-bit boundaries, within a span of 50 bits, before the
character boundaries are adjusted. The recovered character clock remains in the same phasing regardless of
character offset. RFMODEx[1:0] = 11 is reserved for test.
Framing Character Select. The initialization value of the FRAMCHARx latch = 1. FRAMCHARx is used to
select the character or portion of a character used for framing of each channel’s received data stream. When
FRAMCHARx = 1, the framer looks for either disparity of the K28.5 character. When FRAMCHARx = 0, the
framer looks for either disparity o f the 8-bit Comma characters. The specific bit combinations of these framing
characters are listed in Table 6.
Receiver Decoder Mode Select. The initialization value of the DECMODEx latch = 1. DECMODEx selects
the Decoder Mode used for the associated channel. When DECMODEx = 1 and decoder is enabled, the
Cypress Decoding Mode is used. When DECMODEx = 0 and decoder is enabled, the Alternate Decoding
mode is used. When the decoder is enabled (DECBYPx = 1), the 10-bit transmission characters are decoded
using Table 14 and Table 15. The column used in the Special Characters Table 15 is determined by the
DECMODEx latch.
Receiver Decoder Bypass. The initialization value of the DECBYPx latch = 1. DECBYPx selects if the
Receiver Decoder is enabled or bypassed. When DECBYPx = 1, the decoder is enabled and the Decoder
Mode is selected by DECMODEx. When DECBYPx = 0, the decoder is bypassed and raw 10-bit characters
are passed through the receiver.
basis. A logic 1 in a bit location allows for the update of that same
location of the target latch bank(s), whereas a logic 0 disables it.
The reset value of this latch bank is FFh, thereby making its use
optional by default. The mask latch bank is not maskable. The
FGLEN functionality is not affected by the bit 0 value of the mask
latch bank.
Latch Types
There are two types of latch banks: static (S) and dynamic (D).
Each channel is configured by two static and one dynamic latch
bank. The S type contain those settings that normally do not
change for a given application, while the D type controls the
settings that could change dynamically during the ap plication's
lifetime.The first row of latches for each channel (address
numbers 0, 3, 7, and 10) are the static receiver control latches.
The second row of latches for each channel (address numbers
1, 4, 8, and 11) are the static transmitter control latches. The third
row of latches for each channel (address numbers 2, 5, 9, and
12) are the dynamic control latches that are associated with
enabling dynamic functions within the device.
Latch Bank 14 is also useful for those users that do not need the
latch-based programmable feature of the device. This latch bank
could be used in those applications that do not need to modify
the default value of the static latch banks, and that can afford a
global (that is, not independent) control of the dynamic signals.
In this case, this feature becomes available when ADDR[3:0] is
left unchanged with a value of “11 10” and WREN
The signals present in DATA[7:0] effectively become global
control pins, and for the latch banks 2, 5, 8, and 11.
is left asserted.
Document #: 38-02097 Rev. *BPage 20 of 44
[+] Feedback [+] Feedback
CYV15G0404DXB
Table 9. Device Configuration and Control Latch Descriptions (continued)
Receive Clock Select. The initialization value of the RXCKSELx latch = 1. RXCKSELx selects the receive
clock source used to transfer data to the Output Registers and the clock source for the RXCLK± output. When
RXCKSELx = 1, the associated Output Registers, are clocked by REFCLKx± at the associated RXCLKx±
output buffer. When RXCKSELx = 0, the associated Output Registers, are clocked by the Recovered Byte
clock at the associated RXCLKx± output buffer . These output clocks may operate at the character-rate or half
the character-rate as selected by RXRATEx.
Receive Clock Rate Select. The initialization value of the RXRATEx latch = 1. RXRA TEx is used to select the
rate of the RXCLKx± clock output.
When RXRATEx = 1 and RXCKSELx = 0, the RXCLKx± clock outputs are complementary clocks that follow
the recovered clock operating at half the character rate. Data for the associated receive channels should be
latched alternately on the rising edge of RXCLKx+ and RXCLKx–.
When RXRATEx = 0 and RXCKSELx = 0, the RXCLKx± clock outputs are complementary clocks that follow
the recovered clock operating at the character rate. Data for the associated receive channels should be latched
on the rising edge of RXCLKx+ or falling edge of RXCLKx–.
When RXRATEx = 1 with RXCKSELx = 1 and REFCLKx± is a full rate clock, the RXCLKx± clock outputs are
complementary clocks that follow the reference clock operating at half the character rate. Data for the
associated receive channels should be latched alternately on the rising edge of RXCLKx+ and RXCLKx–.
When RXRATEx = 0 with RXCKSELx = 1 and REFCLKx± is a full rate clock, the RXCLKx± clock outputs are
complementary clocks that follow the reference clock operating at the character rate. Data for the associated
receive channels should be latched on the rising edge of RXCLKx+ or falling edge of RXCLKx–.
When RXCKSELx = 1 and REFCLKx± is a half rate clock, the value of RXRATEx is not interpreted and the
RXCLKx± clock outputs are complementary clocks that follow the reference clock operating at half the
character rate. Data for the associated receive channels should be latched alternately on the rising edge of
RXCLKx+ and RXCLKx–.
Primary Serial Data Input Signal Detector Amplitude Select. The initialization value of the SDASEL1x[1:0]
latch = 10. SDASEL1x[1:0] selects the trip point for the detection of a valid signal for the INx1± Primary
Differential Serial Data Inputs.
When SDASEL1x[1:0] = 00, the Analog Signal Detector is disabled.
When SDASEL1x[1:0] = 01, the typical p-p differential voltage threshold level is 140 mV.
When SDASEL1x[1:0] = 10, the typical p-p differential voltage threshold level is 280 mV.
When SDASEL1x[1:0] = 11, the typical p-p differential voltage threshold level is 420 mV.
Secondary Serial Data Input Signal Detector Amplitude Select. The initialization value of the
SDASEL2x[1:0] latch = 10. SDASEL2x[1:0] selects the trip point for the detection of a valid signal for the INx2±
Secondary Differential Serial Data Inputs.
When SDASEL2x[1:0] = 00, the Analog Signal Detector is disabled
When SDASEL2x[1:0] = 01, the typical p-p differential voltage threshold level is 140 mV.
When SDASEL2x[1:0] = 10, the typical p-p differential voltage threshold level is 280 mV.
When SDASEL2x[1:0] = 11, the typical p-p differential voltage threshold level is 420 mV.
Transmit Encoder Bypassed. The initialization value of the ENCBYPx latch = 1. ENCBYPx selects if the
Transmit Encoder is enabled or bypassed. When ENCBYPx = 1, the Transmit encoder is enabled. When
ENCBYPx = 0, the Transmit Encoder is bypassed and raw 10-bit characters are transmitted.
Transmit Clock Select. The initialization value of the TXCKSELx latch = 1. TXCKSELx selects the clock
source used to write data into the Transmit Input Register. When TXCKSELx = 1, the associated input register,
TXDx[7:0] and TXCTx[1:0], is clocked by REFCLKx↑. In this mode, the phase alignment buffer in the transmit
path is bypassed. When TXCKSELx = 0, the associated TXCLKx↑ is used to clock in the input registers,
TXDx[7:0] and TXCTx[1:0].
Document #: 38-02097 Rev. *BPage 21 of 44
[+] Feedback [+] Feedback
CYV15G0404DXB
Table 9. Device Configuration and Control Latch Descriptions (continued)
NameSignal Description
TXRATEA
TXRATEB
TXRATEC
TXRATED
RFENA
RFENB
RFENC
RFEND
RXPLLPDA
RXPLLPDB
RXPLLPDC
RXPLLPDD
RXBISTA
RXBISTB
RXBISTC
RXBISTD
TXBISTA
TXBISTB
TXBISTC
TXBISTD
OE2A
OE2B
OE2C
OE2D
OE1A
OE1B
OE1C
OE1D
PABRSTA
PABRSTB
PABRSTC
PABRSTD
GLEN[11..0]Global Enable. The initialization value of the GLENx la tch = 1 . Th e GLENx is used to reconfigure several
FGLEN[2..0]Force Global Enable. The initialization value of the FGLENx latch is NA. The FGLENx latch forces a GLobal
Transmit PLL Clock Rate Select. The initialization value of the TXRATEx latch = 0. TXRA TEx is used to select
the clock multiplier for the Transmit PLL. When TXRATEx = 0, each transmit PLL multiples the associated
REFCLKx± input by 10 to generate the serial bit-rate clock. When TXRA TEx = 0, the TXCLKOx output clocks
are full rate clocks and follow the frequency and duty cycle of the associated REFCLKx± input. When
TXRATEx= 1, each Transmit PLL multiplies the associated REFCLKx± input by 20 to generate the serial
bit-rate clock. When TXRATEx = 1, the TXCLKOx output clocks are twice the frequency rate of the REFCLKx±
input. When TXCKSELx = 1 and TXRATEx = 1, the Transmit Data Inputs are captured using both the rising
and falling edges of REFCLKx. TXRATEx = 1 and SPDSELx is LOW, is an invalid state and this combination
is reserved.
Reframe Enable. The initialization value of the RFENx latch = 1. RFENx selects if the receiver framer is
enabled or disabled. When RFENx = 1, the associated channel’s framer is enabled to frame per the presently
enabled framing mode and selected framing character. When RFENx = 0, the associated channel’s framer is
disabled, and no received bits alters the frame offset.
Receive Channel Enable. The initialization value of the RXPLLPDx latch = 0. RXPLLPDx selects if the
associated receive channel is enabled or powered-down. When RXPLLPDx = 0, the associated PLL and
analog circuitry is powered-down. When RXPLLPDx = 1, the associated PLL and analog circuitry is enabled.
Receive Bist Disabled. The initialization value of the RXBISTx latch = 1. RXBISTx selects if receive BIST is
disabled or enabled. When RXBISTx = 1, the receiver BIST function is disabled. When RXBISTx = 0, the
receive BIST function is enabled.
Transmit Bist Disabled. The initialization value of the TXBISTx latch = 1. TXBISTx selects if the transmit BIST
is disabled or enabled. When TXBISTx = 1, the transmit BIST function is disabled. When TXBISTx = 0, the
transmit BIST function is enabled.
Secondary Differential Serial Data Output Driver Enable. The initialization value of the OE2x latch = 0.
OE2x selects if the OUT2± secondary differential output drivers are enabled or disabled. When OE2x = 1, the
associated serial data output driver is enabled allowing data to be transmitted from the transmit shifter. When
OE2x = 0, the associated serial data output driver is disabled. When a driver is disabled via the configuration
interface, it is internally powered down to reduce device power. If both serial drivers for a channel are in this
disabled state, the associated internal logic for that channel is also powered down. A device reset (RESET
sampled LOW) disables all output drivers.
Primary Differential Serial Data Output Driver Enable. The initialization value of the OE1x latch = 0. OE1x
selects if the OUT1± primary differential output drivers are enabled or disabled. When OE1x = 1, the associated
serial data output driver is enabled allowing data to be transmitted from the transmit shifter. When OE1x = 0,
the associated serial data output driver is disabled. When a driver is disabled via the configuration interface,
it is internally powered down to reduce device power. If both serial drivers for a channel are in this disabled
state, the associated internal logic for that channel is also powered down. A device reset (RESET sampled
LOW) disables all output drivers.
Transmit Clock Phase Alignment Buffer Reset. The initialization value of the PABRSTx latch = 1. The
PA BRSTx is used to re-center the Transmit Phase Align Buffer. When the configuration latch PABRSTx is
written as a 0, the phase of the TXCLKx input clock relative to its associated REFCLKx+/- is initialized. PABRST
is an asynchronous input, but is sampled by each TXCLKx↑ to synchronize it to the internal clock domain.
PA BRSTx is a self clearing latch. This eliminates the requirement of writing a 1 to complete the initialization of
the Phase Alignment Buffer.
channels simultaneously in applications where several channels may have the same configuration. When
GLENx = 1 for a given address, that address is allowed to participate in a global configuration. When GLENx
= 0 for a given address, that address is disabled from participating in a global configuration.
ENable no matter what the setting is on the GLENx latch. If FGLENx = 1 for the associated Global channel,
FGLEN forces the global update of the target latch banks.
Document #: 38-02097 Rev. *BPage 22 of 44
[+] Feedback [+] Feedback
CYV15G0404DXB
Device Configuration Strategy
The following is a series of ordered events needed to load the
configuration latches on a per channel basis:
1. Pulse RESET
Low after device power up. This operation
resets all four channels. Initialize the JTAG state machine to
its reset state as detailed in the JTAG Support section.
2. Set the static receiver latch bank for the target channel. May
be performed using a global operation, if the application
permits it. [Optional step if the default settings match the
desired configuration.]
3. Set the static transmitter latch bank for the target channel.
May be performed using a global operation, if the application
permits it. [Optional step if the default settings match the
desired configuration.]
4. Set the dynamic bank of latches for the target channel. Enable
the Receive PLLs and transmit channels. May be performed
using a global operation, if the application permits it.
[Required step.]
5. Reset the Phase Alignment Buffer for the target channel. May
be performed using a global operation, if the application
permits it. [Optional if phase align buffer is bypassed.]
When a receive channel is configured with the decoder
bypassed and the receive clock selected as recovered clock in
half rate mode (DECBYPx = 0, RXRATEx = 0, RXCKSELx = 0),
the channel cannot be dynamically reconfigured to enable the
decoder with RXCLKx selected as the REFCLKx (DECBYPx =
1, RXCKSELx = 1). If such a change is desired, a global reset
should be performed and all channels should be reconfigured to
the desired settings.
Table 10. Device Control Latch Configuration Table
The CYV15G0404DXB contains a JTAG port to allow system
level diagnosis of device interconnect. Of th e available JTAG
modes, boundary scan, and bypass are supported. This
capability is present only on the LVTTL inputs and outputs and
the REFCLKx± clock input. The high-speed serial inputs and
outputs are not part of the JTAG test chain.
To ensure valid device operation after power up (including
non-JTAG operation),
the JTAG state machine must also be
reset (using RESET)
TRST
(asserting it LOW and de-asserting it or leaving it
asserted), or by asserting TMS HIGH for at least five consecutive
TCLK cycles. This is necessary to ensure that the JTAG
controller does not enter any of the test modes after device
power up. In this JTAG reset state, the rest of the device is in
normal operation.
Note. The order of device reset (using RESET
ization does not matter.
. The JT AG state machine is initialized using
) and JT AG initial-
initialized to a reset state. This is done in addition to the device
Document #: 38-02097 Rev. *BPage 23 of 44
[+] Feedback [+] Feedback
CYV15G0404DXB
3-Level Select Inputs
Each 3-Level select inputs reports as two bits in the scan register.
These bits report the LOW, MID, and HIGH state of the
associated input as 00, 10, and 11 respectively.
JTAG ID
The JTAG device ID for the CYV15G0404DXB is ‘0C811069’x
.
Receive Character Status Bits
Description
RXSTx[2:0] Priority
0007Normal character received. The valid Data
character on the output bus meets all the
formatting requirements of Data characters
listed in Table 14.
0017Special code detected. The valid special
character on the output bus meets all the
formatting requirements of Special Code
characters listed in Table 15, but is not the
presently selected framing character or a
decoder violation indication.
0102Receive Elasticity buffer underrun/overrun
error. The receive buffer was not able to
add/drop a K28.5 or framing character
0115Framing character detected. This indicates
that a character matching the patterns identified
as a framing character (as selected by
FRAMCHARx) was detected. The decoded
value of this character is present in the
associated output bus.
1004Codeword violation. The character on the
output bus is a C0.7. This indicates that the
received character cannot be decoded into any
valid character.
1011Loss of sync. This indicates a PLL Out of Lock
condition
1106Running disparity error. The character on the
output bus is a C4.7, C1.7, or C2.7.
1113ReservedBIST Wait. The receiver is comparing characters. but
Normal Status
BIST Data Compare. Character compared correctly.
BIST Command Compare. Character compared
correctly.
BIST Last Good. Last Character of BIST sequence
detected and valid.
BIST Last Bad. Last Character of BIST sequence
detected invalid.
BIST Start. Receive BIST is enabled on this channel,
but character compares have not yet commenced. This
also indicates a PLL Out of Lock condition, and
Elasticity Buffer overflow/underflow conditions.
BIST Error. While comparing characters, a mismatch
was found in one or more of the decoded character bits.
has not yet found the start of BIST character to enable
the LFSR.
12.Tested one output at a time, output shorted for less than one second, less than 10% duty cycle.
13.This is the minimum difference in voltage between the true and complement input s required t o ensure detectio n of a logic-1 or logic-0. A logic-1 exists when the
true (+) input is more positive than the complement (-) input. A logic-0 exists when the complement (-) inp ut is more positive than true (+) input.
14.The common mode range defines the allowable range of REFCLKx+ and REFCLKx- when REFCLKx+ = REFCLKx-. This marks the zero-cro ssing between the
true and complement inputs as the signal switches between a logic-1 and a logic-0.
15.The common mode range defines the allowable range of INPUT+ and INPUT- when INPUT+ = INPUT-. This marks the zero-crossing between the true and
complement inputs as the signal switches between a logic-1 and a logic-0.
16.Maximum ICC is measured with VCC = MAX, RFENx = 0,TA = 25°C, with all channels and Serial Line Drivers enabled, sending a continuous alternating 01
pattern, and outputs unloaded.
17.Typical ICC is measured under similar conditions except with VCC = 3.3V, TA = 25°C, RFENx = 0, with all channels enabled and one Serial Line Driver per
transmit channel sending a continuous alternating 01 patte rn. The redu ndant outpu t s on each channel are power ed down and t he p arallel out put s are unl oaded.
18.Cypress uses constant current (ATE) load configurations and forcing functions. This figure is for reference only.
19.The LVTTL switching threshold is 1.4V. All timing references are made relative to where the signal edges cross the threshold voltage.
Document #: 38-02097 Rev. *BPage 27 of 44
[+] Feedback [+] Feedback
CYV15G0404DXB
CYV15G0404DXB AC Electrical Characteristics
ParameterDescriptionMin.MaxUnit
CYV15G0404DXB Transmitter LVTTL Switching Characteristics Over the Operating Range
f
TS
t
TXCLK
[20]
t
TXCLKH
[20]
t
TXCLKL
[20, 21, 22]
t
TXCLKR
[20, 21, 22]
t
TXCLKF
t
TXDS
t
TXDH
f
TOS
t
TXCLKO
t
TXCLKOD
CYV15G0404DXB Receiver LVTTL Switching Characteristics Over the Operating Range
f
RS
t
RXCLKP
t
RXCLKD
[20]
t
RXCLKR
[20]
t
RXCLKF
[23]
t
RXDv–
[23]
t
RXDv+
CYV15G0404DXB REFCLKx Switching Characteristics Over the Operating Range
f
REF
t
REFCLK
t
REFH
t
REFL
[25]
t
REFD
[20, 21, 22]
t
REFR
[20, 21, 22]
t
REFF
Notes
20.Tested initially and after any design or process changes that may affect these parameters, but not 100% tested.
21.The ratio of rise time to falling time must not vary by greater than 2:1.
22.For a given operating frequency, neither rise or fall specification can be greater than 20% of the clock-cycle period or the data sheet maximum time.
23. Parallel data output specifications are only valid if all outputs are lo aded with similar DC an d AC loads.
24.Receiver UI (Unit Interval) is c a l c u l a t e d a s 1 / ( f
25.The duty cycle specification is a simultaneous condition with the t
cannot be as large as 30%–70%.
6.6651.28ns
TXCLKx HIGH Time2.2ns
TXCLKx LOW Time2.2ns
TXCLKx Rise Time0.21.7ns
TXCLKx Fall Time0.21.7ns
Transmit Data Set-up Time to TXCLKx↑ (TXCKSELx ≠ 0)2.2ns
Transmit Data Hold Time from TXCLKx↑ (TXCKSELx ≠ 0)1.0ns
TXCLKOx Clock Frequency = 1x or 2x REFCLKx Frequency19.5150MHz
TXCLKOx Period=1/f
TOS
6.6651.28ns
TXCLKO Duty Cycle centered at 60% HIGH time–1.90ns
RXCLKx± Clock Output Frequency9.75150MHz
RXCLKx± Period = 1/f
RS
RXCLKx± Duty Cycle Centered at 50% (Full Rate and Half Rate when
6.66102.56ns
–1.0+1.0ns
RXCKSELx = 0)
RXCLKx± Rise Time0.31.2ns
RXCLKx± Fall Time0.31.2ns
Status and Data Valid Time to RXCLKx± (RXRATEx = 0, RXCKSELx = 0)
5UI–2.0
[24]
(Full Rate)
Status and Data Valid Time to RXCLKx± (RXRATEx = 1, RXCKSELx = 0)
5UI–1.3
[24]
(Half Rate)
Status and Data Valid Time to RXCLKx± (RXRATEx = 0, RXCKSELx = 0)
5UI–1.8
[24]
(Full Rate)
Status and Data Valid Time to RXCLKx± (RXRATEx = 1, RXCKSELx =0)
5UI–2.6
[24]
(Half Rate)
REFCLKx Clock Frequency19.5150MHz
REFCLKx Period = 1/f
REF
6.651.28ns
REFCLKx HIGH Time (TXRATEx = 1)(Half Rate)5.9ns
REFCLKx HIGH Time (TXRATEx = 0)(Full Rate)2.9
[20]
REFCLKx LOW Time (TXRATEx = 1)(Half Rate)5.9ns
REFCLKx LOW Time (TXRATEx = 0)(Full Rate)2.9
[20]
REFCLKx Duty Cycle3070%
REFCLKx Rise Time (2 0% – 8 0% )2ns
REFCLKx Fall Time (20%–80%)2ns
* 20) (when TXRATEx = 1) or 1/(f
REF
REFH
and t
* 10) (when TXRATEx = 0). In an operating link this is equivalent to tB.
REF
parameters. This means that at faster character rates the REFCLKx± duty cycle
REFL
ns
ns
ns
ns
ns
ns
Document #: 38-02097 Rev. *BPage 28 of 44
[+] Feedback [+] Feedback
CYV15G0404DXB
CYV15G0404DXB AC Electrical Characteristics (continued)
ParameterDescriptionMin.MaxUnit
t
TREFDS
t
TREFDH
t
RREFDA
t
RREFDW
t
REFxDV–
t
REFxDV+
[28]
t
REFRX
CYV15G0404DXB Bus Configuration Write Timing Characteristics Over the Operating Range
t
DATAH
t
DATAS
t
WRENP
CYV15G0404DXB JTAG Test Clock Characteristics Over the Operating Range
f
TCLK
t
TCLK
CYV15G0404DXB Device RESET Characteristics Over the Operating Range
t
RST
CYV15G0404DXB Transmit Serial Outputs and TX PLL Characteristics Over the Operating Range
ParameterDescriptionConditionMin.Max.Unit
t
B
Notes
26.Since this timing parameter is greater than the minimum time period of REFCLK it se t s an u pper li mit to the f requency in whi ch REFCLKx can be used t o clock
the receive data out of the output register. For predictable timing, users can use this parameter only if REFCLK period is greater than sum of t
time of the upstream device. When t his condit ion is n ot tru e, RXCLKx± (a b uf fered or divided ve rsion of REFCLK when RXCK SELx = 1) could b e used to clock
the receive data out of the device.
27.Measured using a 50% duty cycle reference clock
28.REFCLKx± has no phase or frequency relationship with the recovered clock(s) and only acts as a centering reference to reduce clock synchronization time.
REFCLKx± must be within ±1500 PPM (±0.15%) of the transmit ter PLL reference (REFCLKx±) frequency. Although transmitting to a HOTLink II receiver necessitates the frequency difference between the transmitter and receiver ref erence clocks to be within ±1500-PPM, the st ability of the cr ystal needs to be within t he
limits specified by the appropriate standard when transmitti ng to a remote receiver that is compliant to that standard. For example, to be IEEE 802.3z Gigabit
Ethernet compliant, the frequency stability of the crystal needs to be within ±100 PPM.l.
29.While sending continuous K28.5s, outputs loaded to a balanced 100Ω load, measured at the cross point of differential outputs, over the operating range.
30.While sending continuous K28.7s, after 100,000 samples measured at the cross point of differential outputs, time referenced to REFCLKx± input, over the
operating range.
31.Total jitter is calculated at an assumed BER of 1E -12. Hence: Total Jitter (tJ) = (tRJ * 14) + tDJ.
32.Also meets all Jitter Generation and Jitter Tolerance requirements as specified by SMPTE 259, SMPTE 292, ESCON, FICON, Fibre Channel, and DVB-ASI.
Transmit Data Set-up Time to REFCLKx - Full Rate
2.4ns
(TXRATEx = 0, TXCKSELx = 1)
Transmit Data Set-up Time to REFCLKx - Half Rate
2.3ns
(TXRATEx = 1, TXCKSELx = 1)
Transmit Data Hold Time from REFCLKx - Full Rate
1.0ns
(TXRATEx = 0, TXCKSELx = 1)
Transmit Data Hold Time from REFCLKx - Half Rate
1.6ns
(TXRATEx = 1, TXCKSELx = 1)
Receive Data Access Time to REFCLKx (RXCKSELx = 1)9.7
[26]
Receive Data Valid Time Window (RXCKSELx = 1)10UI – 5.8ns
Received Data Valid Time to RXCLK when RXCKSELx = 1
10UI
[24 ]
– 6.16ns
(TXRATEx = 0, RXRATEx = 0)
Received Data Valid Time to RXCLK when RXCKSELx = 1
5UI – 2.53
[27]
(TXRATEx = 0, RXRATEx = 1)
[27]
Received Data Valid T ime to RXCLK when RXCKSELx = 1 (TXRATEx = 1) 10UI – 5.86
Received Data Valid Time from RXCLK when RXCKSELx = 1
1.4ns
ns
(TXRATEx = 0, RXRATEx = 0)
Received Data Valid Time from RXCLK when RXCKSELx = 1
5UI – 1.83
[27]
(TXRATEx = 0, RXRATEx = 1)
Received Data Valid Time from RXCLK when RXCKSELx = 1
1.0
[27]
(TXRATEx = 1)
REFCLKx Frequency Referenced to Received Clock Period–0.15+0.15%
Bus Configuration Data Hold0ns
Bus Configuration Data Setup10ns
Bus Configuration WREN Pulse Width10ns
JTAG Test Clock Frequency20MHz
JTAG Test Clock Period50ns
Device RESET Pulse Width30ns
Bit Time5128666ps
and set-up
RREFDA
ns
ns
ns
ns
Document #: 38-02097 Rev. *BPage 29 of 44
[+] Feedback [+] Feedback
CYV15G0404DXB
CYV15G0404DXB AC Electrical Characteristics (continued)
TXCLKx
TXDx[7:0],
TXCTx[1:0],
t
TXDH
t
TXDS
t
TXCLK
t
TXCLKH
t
TXCLKL
Transmit Interface
Write Timing
TXCLKx selected
REFCLKx
Transmit Interface
t
REFCLK
t
REFH
t
REFL
t
TREFDS
t
TREFDH
Write Timing
TXRATEx = 0
TXDx[7:0],
TXCTx[1:0],
REFCLKx selected
ParameterDescriptionMin.MaxUnit
[20]
t
RISE
[20]
t
FALL
[20, 29, 31]
t
DJ
[20, 30, 31]
Z
RJ
[20]
t
REFJ
t
TXLOCK
CYV15G0404DXB Receive Serial Inputs and CDR PLL Characteristics Over the Operating Range
t
RXLOCK
t
RXUNLOCK
[20]
t
JTOL
[20]
t
DJTOL
CML Output Rise Time 20−80% (CML Test Load)SPDSELx = HIGH60270ps
SPDSELx = MID100500ps
SPDSELx =LOW1801000ps
CML Output Fall Time 80−20% (CML Test Load)SPDSELx = HIGH60270ps
SPDSELx = MID100500ps
SPDSELx =LOW1801000ps
Deterministic Jitter (peak-peak)
Random Jitter (σ)
CYV15G0404DXB HOTLink II Transmitter Switching Waveforms
Document #: 38-02097 Rev. *BPage 30 of 44
[+] Feedback [+] Feedback
CYV15G0404DXB
CYV15G0404DXB HOTLink II Transmitter Switching Waveforms (continued)
t
TREFDH
Transmit Interface
Write Timing
TXRATEx = 1
REFCLKx
t
REFCLK
t
REFL
t
REFH
Note 33
TXDx[7:0],
TXCTx[1:0],
t
TREFDS
t
TREFDS
t
TREFDH
REFCLKx selected
TXCLKOx
t
TXCLKO
Transmit Interface
TXCLKOx Timing
TXRATEx = 1
(internal)
REFCLKx
t
REFCLK
t
REFL
t
REFH
Note 34
Note 35
TXCLKOx
t
TXCLKO
t
TXOH
t
TXOL
Transmit Interface
TXCLKOx Timing
REFCLKx
Note 34
Note 35
t
REFCLK
t
REFH
t
REFL
TXRATEx = 0
Note
33.When REFCLKx± is configured for half rate operation (TXRATE = 1) and data is captured using REFCLKx instead of a TXCLKx clock. Data is captured using
both the rising and falling edges of REFCLKx.
Document #: 38-02097 Rev. *BPage 31 of 44
[+] Feedback [+] Feedback
CYV15G0404DXB
Switching Waveforms for the CYV15G0404DXB HOTLink II Receiver
REFCLKx
RXDx[7:0],
RXSTx[2:0],
t
REFCLK
t
REFH
t
REFL
Receive Interface
Read Timing
full rate RXCLKx±
RXCLKx
t
REFxDV+
t
REFxDV
–
t
RREFDA
REFCLKx Selected
t
RREFDW
t
RREFDW
TXERRx
[36]
REFCLKx
RXDx[7:0],
RXSTx[2:0],
t
REFCLK
t
REFH
t
REFL
Receive Interface
Read Timing
half rate RXCLKx±
t
RREFDA
RXCLKx
t
REFxDV+
t
REFxDV
–
Note 37
t
RREFDA
REFCLKx Selected
t
RREFDW
t
RREFDW
TXERRx
[36]
RXCLKx+
RXDx[7:0],
RXSTx[2:0],
t
RXDV+
t
RXCLKP
Receive Interface
Read Timing
RXRATEx = 0
RXCLKx-
t
RXDV
–
Recovered Clock selected
Notes
34.The TXCLKOx output remains at the character rate regardless of the state of TXRATE and does not follow the duty cycle of REFCLKx±.
35.The rising edge of TXCLKOx output has no direct phase relationship to the REFCLKx± input.
36.TXERRx is synchronous to RXCLKx only when RXCLKx is selected as REFCLK.
Document #: 38-02097 Rev. *BPage 32 of 44
[+] Feedback [+] Feedback
CYV15G0404DXB
Switching Waveforms for the CYV15G0404DXB HOTLink II Receiver
RXCLKx+
RXDx[7:0],
RXSTx[2:0]
t
RXDV+
t
RXDV
–
t
RXCLKP
Receive Interface
Read Timing
RXRATEx = 1
RXCLKx-
Recovered Clock selected
ADDR[3:0]
t
DATAS
Bus Configuration
Write Timing
DATA[7:0]
WREN
t
DATAH
t
WRENP
Note
37.When operated with a half rate REFCLKx±, the setup and hold specifications for data relative to RXCLKx are relative to both rising and falling edges of th e
respective clock output
Document #: 38-02097 Rev. *BPage 33 of 44
[+] Feedback [+] Feedback
CYV15G0404DXB
Table 11. Package Coordinate Signal Allocation
Ball
Signal NameSignal Type
ID
A01INC1–CML INC07ULCCLVTTL IN PUF17RCLKENALVTTL IN PD
A02OUTC1–CML OUTC08GNDGROUNDF18RXSTB[1]LVTTL OUT
A03INC2–CML INC09DA TA[7]LVTTL IN PUF19TXCLKOBLVTTL OUT
A04OUTC2–CML OUTC10DA TA[5]LVTTL IN PUF20RXSTB[0]LVTTL OUT
A05VCCPOWERC11DATA[3]LVTTL IN PUG01TXDC[7]LVTTL IN
A06IND1–CML INC12DATA[1]LVTTL IN PUG02WRENLVTTL IN PU
A07OUTD1–CML OUTC13GNDGROUNDG03TXDC[4]LVTTL IN
A08GNDGROUNDC14RCLKENBLVTTL IN PDG04TXDC[1]LVTTL IN
A09IND2–CML INC15SPDSELD3-LEVEL SELG17SPDSELB3-LEVEL SEL
A10OUTD2–CML OUTC16VCCPOWERG18LPENCLVTTL IN PD
A1 1INA1–CML INC17LDTDENLVTTL IN PUG19SPDSELA3-LEVEL SEL
A12OUTA1–CML OUTC18TRSTLVTTL IN PUG20RXDB[1]LVTTL OUT
A13GNDGROUNDC19LPENDLVTTL IN PDH0 1G NDGROUND
A14I NA2–CML INC20TDOLVTTL 3-S OUTH02GNDGROUND
A15OUTA2–CML OUTD01TCLKLVTTL IN PDH03GNDGROUND
A16VCCPOWERD02RESET
A17INB1–CML IND03INSELDLVTTL INH17GNDGROUND
A18OUTB1–CML OUTD04INSELALVTTL INH18GNDGROUND
A19INB2–CML IND05VCCPOWERH19GNDGROUND
A20OUTB2–CML OUTD06ULCALVTTL IN PUH20GNDGROUND
B01INC1+CML IND07SPDSELC3-LEVEL SELJ01TXCTC[1]LVTTL IN
B02OUTC1+CML OUTD08GNDGROUNDJ02TXDC[5]LVTTL IN
B03INC2+CML IND09DATA[6]LVTTL IN PUJ03TXDC[2]LVTTL IN
B04OUTC2+CML OUTD10DATA[4]LVTTL IN PUJ04TXDC[3]LVTTL IN
B05VCCPOWERD11DATA[2]LVTTL IN PUJ17RXSTB[2]LVTTL OUT
B06IND1+CML IND12DATA[0]LVTTL IN PUJ18RXDB[0]LVTTL OUT
B07OUTD1+CML OUTD13GNDGROUNDJ19RXDB[5]LVTTL OUT
B08GNDGROUNDD14LPENBLVTTL IN PDJ20RXDB[2]LVTTL OUT
B09IND2+CML IND15ULCBLVTTL IN PUK01RXDC[2]LVTTL OUT
B10OUTD2+CML OUTD16VCCPOWERK02REFCLKC–PECL IN
B1 1INA1+CML IND17LPENALVTTL IN PDK03TXCTC[0]LVTTL IN
B12OUTA1+CML OUTD18VCCPOWERK04TXCLKCLVTTL IN PD
B13GNDGROUNDD19SCANEN2LVTTL IN PDK17RXDB[3]LVTTL OUT
B14INA2+CML IND20TMEN3LVTTL IN PDK18RXDB[4]LVTTL OUT
B15OUTA2+CML OUTE01VCCPOWERK19RXDB[7]LVTTL OUT
B16VCCPOWERE02VCCPOWERK20LFIBLVTTL OUT
B17INB1+CML INE03VCCPOWERL01RXDC[3]LVTTL OUT
B18OUTB1+CML OUTE04VCCPOWERL02REFCLKC+PECL IN
B19INB2+CML INE17VCCPOWERL03LFICLVTTL OUT
B20OUTB2+CML OUTE18VCCPOWERL04TXDC[6]LVTTL IN
C01TDILVTTL IN PUE19VCCPOWERL17RXDB[6]LVTTL OUT
C02TMSLVTTL IN PUE20VCCPOWERL18RXCLKB+LVTTL OUT
C03INSELCLVTTL INF01RXDC[6]LVTTL OUTL19RXCLKB–LVTTL OUT
Ball
Signal NameSignal Type
ID
Ball
Signal NameSignal Type
ID
LVTTL IN PUH04GNDGROUND
Document #: 38-02097 Rev. *BPage 34 of 44
[+] Feedback [+] Feedback
CYV15G0404DXB
Table 11. Package Coordinate Signal Allocation (continued)
Ball
Signal NameSignal Type
ID
C04INSELBLVTTL INF02RXDC[7]LVTTL OUTL20TXDB[6]LVTTL IN
C05VCCPOWERF03TXDC[0]LVTTL INM01RXDC[4]LVTTL OUT
C06ULCDLVTTL IN PU F04RCLKENDLVTTL IN PDM02RXDC[5]LVTTL OUT
M03RCLKENCLVTTL IN PDU03TXDD[2]LVTTL INW03LFID
M04TXERRCLVTTL OUTU04TXCTD[1]LVTTL INW04RXCLKD–LVTTL OUT
M17REFCLKB+PECL INU05VCCPOWERW05VCCPOWER
M18REFCLKB–PECL INU06RXDD[2]LVTTL OUTW06RXDD[4]LVTTL OUT
M19TXERRBLVTTL OUTU07RXDD[1]LVTTL OUTW07RXSTD[1]LVTTL OUT
M20TXCLKBLVTTL IN PD U08GNDGROUNDW08GNDGROUND
N01GNDGROUNDU09TXCTA[1]LVTTL INW09ADDR [3]LVTTL IN PU
N02GNDGROUNDU10ADDR [0]LVTTL IN PUW10ADDR [1]LVTTL IN PU
N03GNDGROUNDU11REFCLKD–PECL INW11RXCLKA+LVTTL OUT
N04GNDGROUNDU12TXDA[1]LVTTL INW12TXERRALVTTL OUT
N17GNDGROUNDU13GNDGROUNDW13GNDGROUND
N18GNDGROUNDU14TXDA[4]LVTTL INW14TXDA[2]LVTTL IN
N19GNDGROUNDU15TXCTA[0]LVTTL INW15TXDA[6]LVTTL IN
N20GNDGROUNDU16VCCPOWERW16VCCPOWER
P01RXDC[1]LVTTL OUTU17RXDA[2]LVTTL OUTW17LFIALVTTL OUT
P02RXDC[0]L VTTL OUTU18TXCTB[0]LVTTL INW18REFCLKA+PECL IN
P03RXSTC[0]LVTTL OUTU19RXSTA[2]LVTTL OUTW19RXDA[4]LVTTL OUT
P04RXSTC[1]LVTTL OUTU20RXSTA[1]LVTTL OUTW20RXDA[1]LVTTL OUT
P17TXDB[5]LVTTL INV01TXDD[3]LVTTL INY01TXDD[6]LVTTL IN
P18TXDB[4]LVTTL INV02TXDD[4]LVTTL INY02TXCLKDLVTTL IN PD
P19TXDB[3]LVTTL INV03TXCTD[0]LVTTL INY03RXDD[7]LVTTL OUT
P20TXDB[2]LVTTL INV04RXDD[6]LVTTL OUTY04RXCLKD+LVTTL OUT
R01RXSTC[2]LVTTL OUTV05VCCPOWERY05VCCPOWER
R02TXCLKOCLVTTL OUTV06RXDD[3]LVTTL OUTY06RXDD[5]LVTTL OUT
R03RXCLKC+LVTTL OUTV07RXSTD[0]LVTTL OUTY07RXDD[0]LVTTL OUT
R04RXCLKC–LVTTL OUTV08GNDGROUNDY08GNDGROUND
R17TXDB[1]LVTTL INV09RXSTD[2]LVTTL OUTY09TXCLKODLVTTL OUT
R18TXDB[0]LVTTL INV10ADDR [2]LVTTL IN PUY10NCNO CONNECT
R19TXCTB[1]LVTTL IN V11REFCLKD+PECL INY11TXCLKALVTTL IN PD
R20TXDB[7]LVTTL INV12TXCLKOALVTTL OUTY12RXCLKA–LVTTL OUT
T01VCCPOWERV13GNDGROUNDY13GNDGROUND
T02VCCPOWERV14TXDA[3]LVTTL INY14TXDA[0]LVTTL IN
T03VCCPOWERV15TXDA[7]LVTTL INY15TXDA[5]LVTTL IN
T04VCCPOWERV16VCCPOWERY16VCCPOWER
T17VCCPOWERV17RXDA[7]LVTTL OUTY17TXERRDLVTTL OUT
T18VCCPOWERV18RXDA[3]LVTTL OUTY18REFCLKA–PECL IN
T19VCCPOWERV19RXDA[0]LVTTL OUTY19RXDA[6]LVTTL OUT
T20VCCPOWERV20RXSTA[0]LVTTL OUTY20RXDA[5]LVTTL OUT
U01TXDD[0]LVTTL INW01TXDD[5]LVTTL IN
U02TXDD[1]LVTTL INW02TXDD[7]LVTTL IN
Ball
Signal NameSignal Type
ID
Ball
Signal NameSignal Type
ID
LVTTL OUT
Document #: 38-02097 Rev. *BPage 35 of 44
[+] Feedback [+] Feedback
CYV15G0404DXB
X3.230 Codes and Notation Conventions
Information transmitted over a serial link is encoded eight bits at
a time into a 10-bit Transmission Character and then sent
serially, bit-by-bit. Information received over a serial link is
collected ten bits at a time, and those transmission characters
that are used for data characters are decoded into the correct
8-bit codes. The 10-bit transmission code supports all 256 8-bit
combinations. Some of the remaining transmission characters
(special characters) are used for functions other than data transmission.
The primary use of a transmission code is to improve the transmission characteristics of a serial link. The encoding defined b y
the transmission code ensures that sufficient transitions are
present in the serial bit stream to make clock recovery possible
at the receiver. Such encoding also greatly increases the
likelihood of detecting any single or multiple bit errors that may
occur during transmission and reception of information. In
addition, some special characters of the transmission code
selected by Fibre Channel Standard contain a distinct and easily
recognizable bit pattern that assists the receiver in achieving
character alignment on the incoming bi t stre a m .
Notation Conventions
The documentation for the 8B/10B Transmission Code uses
letter notation for the bits in an 8-bit byte. Fibre Channel Standard
notation uses a bit notation of A, B, C, D, E, F, G, H for the 8-bit
byte for the raw 8-bit data, and the letters a, b, c, d, e, i, f, g, h, j
for encoded 10-bit data. There is a correspondence between bit
A and bit a, B and b, C and c, D and d, E and e, F and f, G and
g, and H and h. Bits i and j are derived, respectively, from
(A,B,C,D,E) and (F,G,H).
The bit labeled A in the description of the 8B/10B Transmission
Code corresponds to bit 0 in the numbering scheme of the FC-2
specification, B corresponds to bit 1, as shown below.
FC-2 bit designation—76543210
HOTLink D/Q designation—76543210
8B/10B bit designation—HGFEDCBA
To clarify this correspondence, the following example shows the
conversion from an FC-2 Valid Data Byte to a Transmission
Character.
FC-2 45H
Bits: 7654
Converted to 8B/10B notation, note that the order of bits has
been reversed):
Data Byte Name D5.2
Bits: ABCDE
10100 010
Translated to a transmission Character in the 8B/10B Transmission Code:
Bits: abcdei fghj
101001 0101
Each valid transmission character of the 8B/10B Transmission
Code has been given a name using the following convention:
cxx.y, where c is used to show whether the Transmission
Character is a Data Character (c is set to D, and SC/D = LOW)
or a special character (c is set to K, and SC/D = HIGH). When c
3210
0100 0101
FGH
is set to D, xx is the decimal value of the binary number
composed of the bits E, D, C, B, and A in that order, and the y is
the decimal value of the binary number composed of the bits H,
G, and F in that order. When c is set to K, xx and y are derived
by comparing the encoded bit patterns of the Special Character
to those patterns derived from encoded valid data bytes and
selecting the names of the patterns most similar to the encoded
bit patterns of the special character.
Using these conventions, the transmission character used for
the examples above, is referred to by the name D5.2. The special
character K29.7 is so named because the first six bits (abcdei)
of this character make up a bit pattern similar to that resulting
from the encoding of the unencoded 11101 pattern (29), and
because the second four bits (fg hj) make up a bit pattern similar
to that resulting from the encoding of the unencoded 111 pattern
(7).
Note. This definition of the 10-bit transmission code is based on
the following references, which describe the same 10-bi t transmission code.
■ A.X. Widmer and P.A. Franaszek. “A DC-Balanced, Parti-
tioned-Block, 8B/10B Transmission Code” IBM Journal of
Research and Development, 27, No. 5: 440-451 (September, 1983).
■ U.S. Patent 4,486,739. Peter A. Franaszek and Albert X.
■ Fibre Channel Physical and Signaling Interface (ANS
X3.230-1994 ANSI FC-PH Standard).
■ IBM Enterprise Systems Architecture/390 ESCON I/O
Interface (document number SA22-7202).
8B/10B Transmission Code
The following information describes how the tables areused for
both generating valid transmission characters (encoding) and
checking the validity of received transmission characters
(decoding). It also specifies the ordering rules followed when
transmitting the bits within a character and the characters within
any higher level constructs specified by a standard.
Transmission Order
Within the definition of the 8B/10B transmission code, the bit
positions of the transmission characters are labeled a, b, c, d, e,
i, f, g, h, j. Bit “a” is transmitted first followed by bits b, c, d, e, i,
f, g, h, and j in that order.
Note that bit i is transmitted between bit e and bit f, rather than
in alphabetical order.
Valid and Invalid Transmission Characters
The following tables define the valid data characters and valid
special characters (K characters), respectively. The tables are
used for both generating valid transmission characters and
checking the validity of received transmission characters. In the
tables, each valid-data-byte or special-character-code entry has
two columns that represent two transmission characters. The
two columns correspond to the current value of the running
disparity. Running disparity is a binary parameter with either a
negative (–) or positive (+) value.
After powering on, the transmitter may assume either a positive
or negative value for its initial running disparity. Upon trans-
Document #: 38-02097 Rev. *BPage 36 of 44
[+] Feedback [+] Feedback
CYV15G0404DXB
mission of any transmission character, the transmitter selects the
proper version of the transmission character based on the
current running disparity value, and the transmitter calculates a
new value for its running disparity based on the contents of the
transmitted character. Special character cod es C1.7 and C2.7
can be used to force the transmission of a specific special
character with a specific running disparity as required for some
special sequences in X3.230.
After powering on, the receiver may assume either a positive or
negative value for its initial running disparity. Upon reception of
any transmission character, the receiver decides whether the
transmission character is valid or invalid according to the
following rules and tables and calculates a new value for its
running disparity based on the contents of the received
character.
The following rules for running disparity are used to calculate the
new running disparity value for transmission characters that
have been transmitted and received.
Running disparity for a transmission character is calculated from
subblocks, where the first six bits (abcdei) form one subblock and
the second four bits (fghj) form the other subblock. Running
disparity at the beginning of the 6-bit subblock is the running
disparity at the end of the previous transmission character.
running disparity at the beginning of the 4-bit subblock is the
running disparity at the end of the 6-bit subblock. Running
disparity at the end of the transmission character is the running
disparity at the end of the 4-bit subblock.
Running disparity for the subblocks is calculated as follows:
1. Running disparity at the end of any subblock is positive if the
subblock contains more ones than zeros. It is also positive at
the end of the 6-bit subblock if the 6-bit subblock is 000111,
and it is positive at the end of the 4-bit subblock if the 4-bit
subblock is 0011.
2. Running disparity at the end of any subblock is negative if the
subblock contains more zeros than ones. It is also negative
at the end of the 6-bit subblock if the 6-bit subblock is 11 1000,
and it is negative at the end of the 4-bit subblock if the 4-bit
subblock is 1100.
3. Otherwise, running disparity at the end of the subblock is the
same as at the beginning of the subblock.
transmitted, a new value of the running disparity is calculated.
This new value is used as the transmitter’s current running
disparity for the next valid data byte or Special Character byte
encoded and transmitted. Table 12 shows naming notations and
examples of valid transmission characters.
Use of the T ables for Checking the V alidity of Received
Transmission Characters
The column corresponding to the current value of the receiver’s
running disparity is searched for the received transmission
character. If the received transmission character is found in the
proper column, then the transmission character is valid and the
associated data byte or special character code is determined
(decoded). If the received transmission character is not found in
that column, then the transmission character is invalid. This is a
code violation. Independent of the transmission character’s
validity, the received transmission character is used to calculate
a new value of running disparity. The new value is used as the
receiver’s current running disparity for the next received transmission character.
Table 12. Valid Transmission Characters
Data
Byte Name
D0.00000000000
D1.00000000101
D2.00000001002
.
.
D5.20100010145
.
.
D30.711111110FE
D31.711111111FF
DIN or Q
76543210
.
.
.
.
OUT
.
.
.
.
Hex Value
.
.
.
.
Use of the Tables for Generating Transmission
Characters
The appropriate entry in Table 14 for the valid data byte or
Table 15 for Special Character byte identify which transmission
character is generated. The current value of the transmitter’s
running disparity is used to select the transmission character
from its corresponding column. For each transmission character
Table 13. Code Violations Resulting from Prior Errors
RDCharacterRDCharacterRDCharacterRD
Transmitted data character–D21.1–D10.2–D23.5+
Transmitted bit stream–101010 1001–010101 0101–111010 1010+
Bit stream after error–101010 1011+010101 0101+111010 1010+
Decoded data character–D21.0+D10.2+Code Violation+
Detection of a code violation does not necessarily show that the
transmission character in which the code violation was detected
is in error. Code violations may result from a prior error that
altered the running disparity of the bit stream which did not result
in a detectable error at the transmission character in which the
error occurred. Table 12 shows an example of this behavior.
39.Notation for Special Character Code Name is consistent with Fibre Channel and ESCON naming conventions. Special Chara cter Code Name is intended to
40.Both the Cypress and alternate encodings may be used for data transmission to generate specific Special Character Codes. The decoding process for received
41.These characters are used for control of ESCON interfaces. They can be sent as embedded commands or other markers when not operating using ESCON
42.The K28.5 character is used for framing operations by the receiver. It is also the pad or fill cha racter transmitte d to main t ain the se rial link when no u ser data is
43.Care must be taken when using this Special Character code. When a C7.0 or a C0.7 is followed by a D11.x or D20.x, an alias K28.5 sync character is created.
44.C2.1 = Transmit either –K28.5+ or +K28.5– as determined by Current RD and modi fy t he Transmission Character that follows, by setting its least significant bit
45.C0.7 = Transmit a deliberate code rule violation. The code chosen for this function follows the normal Running Disparity rules. Transmission of this Special
46.C1.7 = Transmit Negative K28.5 (−K28.5+) disregarding Current RD. The receiver only outputs t his Special Character if K28.5 is received with the wrong running
47.C2.7 = Transmit Positive K28.5 (+K28.5−) disregarding Current RD. The receiver on ly output s this Special Character if K28.5 is received with the wrong running
48.C4.7 = Transmit a deliberate code rule violation to indicate a Running Disparity violation. The receiver only outputs this Special Character if the Transmission
[48]
describe binary information present on I/O pins. Common usage for the name can either be in the form used for describing Data patterns (i.e., C0.0 through
C31.7), or in hex notation (i.e., Cnn where nn = the specified value between 00 and FF).
characters generates Cypress codes or Alternate codes as selected by the BOE[7:0] configur ation inputs.
protocols.
available.
These sequences can cause erroneous framing and should be avoided while RFENx = 1.
to 1 or 0. If Current RD at the start of t he following character is plus (+) t he LSB is set to 0, a nd if Curren t RD is minus ( –) t he LSB become s 1. This mod ification
allows construction of X3.230 “EOF” frame delimiters wherein the second data byte is determined by the Current RD.
For example, to send “EOFdt” the controller could issue the sequence C2.1–D21.4– D2 1.4–D21.4, and the HOTLink Transmitter sends either
K28.5–D21.4–D21.4–D21.4 or K28.5–D21.5– D21.4–D21.4 based on Current RD. Likewise to send “EOFdti” the controller could issue the sequence
C2.1–D10.4–D21.4–D21.4, and the HOTLink Transmitter sends either K28.5–D10.4–D21.4– D21.4 or K28.5–D10.5–D21.4–D21.4 based on Current RD.
The receiver never outputs this Special Character, since K28.5 is decoded as C5.0, C1.7, or C2.7, and the subsequent bytes are decoded as data.
Character has the same effect as asserting TXSVS = HIGH. The receiver only outputs this Special Character if the Transmission Character being decoded is
not found in the tables.
disparity. The receiver outputs C1.7 if −K28.5 is received with RD+, otherwise K28.5 is decoded as C5.0 or C2.7.
disparity. The receiver outputs C2.7 if +K28.5 is received with RD−, otherwise K28.5 is decoded as C5.0 or C1.7.
Character being decoded is found in the tables, but Running Disparity does not match. This might indicate that an error occurred in a prior byte.
Any Source Code (software and/or firmware) is owned by Cypress Semiconductor Corporation (Cypress) and is protected by and subject to worldwide patent protection (United States and foreign),
United States copyright laws and international treaty provisions. Cypress hereby gr ant s to l icense e a pers onal, no n-exclu sive , non-tr ansfer able license to copy, use, modify, create derivative works of,
and compile the Cypress Source Code and derivative works for the sole purpose of creating custom software and or firmware in support of licensee product to be used only in conjunction w ith a Cypress
integrated circuit as specified in the ap plicable agreem ent. Any reprod uction, modificatio n, translation, co mpilation, or repr esentation of this Source Co de except as speci fied above is pro hibited with out
the express written permission of Cypress.
Disclaimer: CYPRESS MAKES NO WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, WITH REGARD TO THIS MATERIAL, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. Cypress reserves the right to make changes without further notice to the materials described herein. Cypress does n ot
assume any liability arising out of the applic ation or use o f any pr oduct or circ uit de scribed herein . Cypr ess does n ot author ize its p roducts fo r use as critical compon ents in life-su pport systems whe re
a malfunction or failure may reason ably be expected to result in significant injury to the user. The inclusion of Cypress’ product in a life-support systems application implies that the manufacturer
assumes all risk of such use and in doing so indemnifies Cypress against all charges.
Use may be limited by and subject to the applicable Cypress software license agreement.
Document #: 38-02097 Rev. *BRevised December 14, 2007Page 44 of 44
IBM and ESCON are registered trademarks, and FICON is a trademark, of International Business Machines. HOTLink is a registered trademark and HOTLink II and MultiFrame are trademarks of
Cypress Semiconductor. All product and company names mentioned in this document may be the trademarks of their respective holders.
[+] Feedback [+] Feedback
Loading...
+ 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.