ETC STEL

STEL-2060C/CR
Data Sheet
STEL-2060C/CR
45 Mbps
Viterbi Decoder
R
Powered by ICminer.com Electronic-Library Service CopyRight 2003
FUNCTIONAL DESCRIPTION
FEATURES
■ 45 Mbps Operating Rate
■ Constraint Length K = 7
G1 = 1718
Convolutional encoding and Viterbi decoding are used to
provide forward error correction (FEC) which improves
digital communication performance over a noisy link. The
STEL-2060C is a specialized product designed to perform
this specific communications related function. At the
encoder a stream of symbols is created which introduces a
high degree of redundancy. This enables accurate decoding
of the information despite a high symbol error rate resulting
from an impaired communications link.
G2 = 1338
■
Multiple Rates: Rate 1/2 as well as
Punctured codes at Rates 2/3 through 7/8
■
Internal Depuncturing Capability at Rates
2
/3, 3/4 and 7/8
■
Multiple Devices can be Multiplexed to
Give Higher Data Rates
■
Optimized Interface to Operate with BPSK
and QPSK Demodulators
■
■
■
■
Auto Node Sync Capability
■
■
■
5.2 dB Coding Gain @10-5 BER (R = 1/2)
The STEL-2060C contains a K = 7 Viterbi Decoder. The data
inputs can be in offset binary or offset signed-magnitude
formats, with 3-bit soft decision. Auto node sync is provided
for applications where symbol uncertainty can occur. Rate
2
/3, 3/4, 4/5, 5/6, 6/7 and 7/8 punctured signals can be
decoded, as well as non-punctured, Rate 1/2, signals. The
polynomials and puncturing patterns used are industry
standards. Depuncturing logic is incorporated into the
decoder to provide automatic depuncturing of received data
at rates 2/3, 3/4 and 7/8 when the puncturing patterns
supported by the device are used. A BER monitor is also
provided in the device, along with a circuit for computing
the mean value of the BER over an extended period. These
circuits operate with punctured codes as well as
unpunctured. The STEL-2060C incorporates a descrambler
for signals scrambled with the “Invert G2” algorithm. (With
this method the G2 symbols are logically inverted at the
encoder. This provides a very effective level of scrambling
for the purpose of avoiding long strings of ones or zeroes in
the transmitted signal using BPSK modulation.)
Differential Decoder
“Invert G2” Descrambler
Internal BER Monitor and BER
Measurement Circuit
100-pin PQFP Package
0.5 Micron CMOS Technology
BLOCK DIAGRAM
DSCRAM
OBIN
PARL
LDG2
G1
G2
SYMBOL
ALIGNMENT
AND
DEPUNCTURING
CIRCUIT
3
3
2
PNCG1/G2
BRANCH
METRIC
ASSIGNMENT
VITERBI
DECODER
(ACS)
TRACEBACK
MEMORY
SYMCKIN
DCLKIN
SYNC
RATE
EXTSEL
THRES
H
COUNT
NODE SYNC
CONTROL
3
OOS
TIMING AND
CONTROL
8
8
DATO
DIFFERENTIAL
DECODER
DDIF
DATA
ADDR
WR
AUTO
ODCLK
8
3
µP
INTERFACE
BER
MONITOR
AND
COUNTER
RD
CSEL
RESET
TO ALL REGISTERS
STEL-2060C
Powered by ICminer.com Electronic-Library Service CopyRight 2003
2
BERR
G1ERR
G2ERR
INT
PIN CONFIGURATION
Package: 100-pin HQFP
Thermal coefficient, θja = 30° C/W
0.941"
± 0.010"
0.742"
±0.005"
80
81
51
50
Top View
0.487"
±0.003"
0.705"
± 0.010"
0.11" nom.
0.009"
± 0.005"
Pin 1 Identifier
31
100
1
0.031"
± 0.005"
30
Detail of pins
0.0256" ±0.002"
Notes:
(1)
(2)
(3)
(4)
0.014"
± 0.002"
0.122" max.
Tolerances on pin spacing are not cumulative
Dimensions shown are at seating plane
I.C. denotes Internal Connection. This pin must be left unconnected. Do not use for vias.
N.C. denotes No Connection. These pins can be used for vias.
PIN CONNECTIONS
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
VDD
N.C.
N.C.
G10
G11
G12
G20
G21
G22
N.C.
OBIN
N.C.
VSS
SYMCKIN
N.C.
VSS
DCLKIN
N.C.
VSS
RESET
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
VSS
N.C.
PNCG1
PNCG2
DSCRAM
SYNC
LDG2
N.C.
N.C.
VDD
VDD
COUNT0
COUNT1
COUNT2
COUNT3
COUNT4
COUNT5
COUNT6
COUNT7
VDD
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
VSS
VDD
PARL
READ
VSS
ADDR2
ADDR1
ADDR0
WRITE
CSEL
VDD
VDD
N.C.
DATA7
DATA6
DATA5
DATA4
DATA3
DATA2
DATA1
3
Powered by ICminer.com Electronic-Library Service CopyRight 2003
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
DATA0
N.C.
INT
N.C.
VSS
ODCLK
N.C.
VSS
DATO
N.C.
OOS
AUTO
N.C.
I.C.
N.C.
BERR
G1ERR
G2ERR
VDD
VDD
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
THR0
THR1
THR2
THR3
THR4
THR5
THR6
THR7
EXTSEL
VSS
VSS
VSS
VDD
VSS
DDIF
VSS
RATE2
RATE1
RATE0
VDD
STEL-2060C
INPUT SIGNALS
RESET
inputs are accepted sequentially, using the G12-0 pins for
both symbols. The sequential input is most suited for BPSK
data, and the parallel input is most suited for QPSK data. The
auto node sync sequence will operate on the assumption that
the signal is BPSK modulated when PARL is set low and
QPSK modulated when it is set high. Setting PARL low adds
two cycles of ODCLK to the pipeline delay.
Reset. A logic low on this asynchronous input will completely
reset all registers in the decoder to an initial condition within
20 nsec. Normal operation will commence after RESET goes
high. This will not affect the values stored in the decision
path memory but will reset the node sync state to the initial
condition.
DCLKIN
OBIN
Decoder clock input. It is the reference clock for all internal
synchronous functions in the decoder when operating in the
Internal Puncturing mode. It should nominally be a square
wave with a maximum frequency of 45 MHz, corresponding
to a decoded data rate of 45 Mbps. When operating at Rate
1
/2 and in the External Puncturing mode this clock will be
generated internally from SYMCKIN, and DCLKIN should
be connected to ground.
The STEL-2060CCC can accept the soft-decision input data
in either offset binary or offset signed-magnitude formats.
When the OBIN input is set high, the format expected will be
offset binary; when it is set low it will be offset signedmagnitude. The meanings of the 3-bit values for these two
codes is shown in the following table:
SYMCKIN
Symbol clock input. This is the reference clock for all
internal synchronous functions in the symbol alignment and
depuncturing circuits. It should nominally be a square wave
with a maximum frequency of 90 MHz. Its frequency should
be equal to f DCLKIN/2R in the parallel input mode
(PARL = 1) and equal to fDCLKIN/R in the sequential input
mode (PARL = 0), where R is the decoding rate when using
internal depuncturing. Please refer to the section on
Punctured Mode Operation for more detailed information.
OBIN = 1
OBIN = 0
Value
111
110
111
110
Most confident +
(Data = 1)
101
100
101
100
Least confident +
011
010
000
001
Least confident –
001
000
010
011
(Data = 0)
Most confident –
When using the STEL-2060CCC with hard-decision data, the
symbols should be loaded into the G12 and G22 pins. The
other symbol inputs should be set to a logic high level and
OBIN should be set low.
G12-0, G22-0
The G12-0 and G22-0 signals are the 3-bit soft decision input
symbols to the decoder. They are presented to the decoder
either sequentially or in parallel depending on the states of
the PARL and RATE2-0 inputs. In the parallel mode
(PARL = 1) the symbols are clocked into the device on the
rising edges of SYMCKIN when RATE2-0 = 0 (Rate 1/2 and
External Depuncturing) and on both edges of SYMCKIN
when RATE2-0 ≠ 0 (Internal Depuncturing). In the sequential
mode (PARL␣ = 0) in which theG22-0 inputs are not used, both
the G1 and G2 symbols are loaded via the G12-0 pins. The G1
symbols are then latched in on the rising edges of SYMCKIN
when LDG2 is low and the G2 symbols are latched in on the
rising edges of SYMCKIN when LDG2 is high.
RATE2-0
These signals select the decoding rate for unpunctured
operation (Rate 1/2) and internally supported depuncturing
patterns (Rates 2/3, 3/4 and 7/8). These patterns are shown
in the following table, where a 0 in the pattern indicates a
punctured symbol:
RATE2-0
Rate
Pattern
0 0 0
1
/2
N.A.
0 0 1
2
/3
G1: 10
G2: 11
1 0 0
3
/4
G1: 101
G2: 110
0 1 0
7
/8
G1: 1000101
LDG2
When this signal is high during a rising edge of SYMCKIN
the symbol loaded into the G12-0 pins will be G2. This
function is only active when PARL is set low (sequential
input mode) and RATE2-0 is set to 000 (Rate 1/2 operation or
External Puncturing mode). If auto node sync is used, the
LDG2 signal can be derived by dividing the SYMCKIN
signal by two. The auto node sync will then compensate for
the phase ambiguity.
Other puncturing patterns can be implemented externally
using the PNCG1 and PNCG2 inputs.
PARL
DDIF
When this signal is high, the input symbols are accepted in
parallel by the chip, using the G12-0 pins for the G1 symbols
and the G22-0 pins for the G2 symbols. When it is set low, the
When this input is set high, it causes the data out of the
Viterbi decoder to be differentially decoded. This adds one
cycle of ODCLK to the pipeline delay.
STEL-2060C
Powered by ICminer.com Electronic-Library Service CopyRight 2003
G2: 1111010
4
OUTPUT SIGNALS
DSCRAM
ODCLK
When this input is set high, it causes the G2 symbols to be
inverted before they enter the Viterbi decoder, thereby
reversing the effect of the G2 inversion if an "Invert G2"
scrambler is implemented at the encoder.
Output data clock. All outputs change on the rising edge of
this clock. The falling edge of ODCLK can be used as a strobe
for DATO output, which is guaranteed to be valid on this
edge.
PNCG1, PNCG2
DATO
The PNCG1 and PNCG2 signals are used to control the
STEL-2060CCC when operating in punctured modes not
supported by the internal puncturing logic. In normal
operation (Rate 1/2 and when using internal puncturing)
these pins should be set low. In the external depuncturing
mode, the PNCG1 signal must be set high to indicate that the
G1 symbol is punctured and the PNCG2 signal must be set
high to indicate that the G2 symbol is punctured. A symbol
will be depunctured when the PNCG1 or PNCG2 signals are
high during the rising edge of SYMCKIN which latches the
corresponding symbol in to the decoder. RATE2-0 should be
set to 000 when operating in external depuncturing mode.
Zero value metrics will be substituted internally for the
actual metrics corresponding to the signals present on the
G12-0 and G22-0 pins at that time. Internal depuncturing can
be selected by the use of the RATE inputs.
Decoded data output. This is the output of the Viterbi
decoder. This signal changes on the rising edges of ODCLK.
In Rate 1/2 operation there will be a pipeline delay of 526
cycles of ODCLK from the G12-0/G22-0 inputs to the DATO
output when DDIF is set low and PARL is set high. Setting
DDIF high adds one cycle to this value and setting PARL low
adds two cycles.
OOS
This output pin serves as a flag for the out-of-sync condition.
When it goes high it signifies that the renormalization count
in the internal node sync circuit has exceeded the threshold
value set by the THR7-0 signal, declaring an out-of-sync
condition. It will remain high until this condition ceases to
exist. i.e., until the next time the threshold is not exceeded
during a complete count period.
AUTO
SYNC
When the SYNC input is set high during the rising edge of
SYMCKIN the internal symbol synchronization will be
changed. When auto node sync is not desired this pin should
be set low. It should be connected to the AUTO output to use
the auto node sync capability of the STEL-2060CCC. The
state of this circuit will always be set to normal after a reset.
This is the feedback signal from the internal node sync
correction circuit. It will pulse high for one cycle of DCLKIN
each time the renormalization count in the internal node
sync circuit has exceeded the threshold value set by the
THR7-0 signal and the out-of-sync condition is declared. It
should be connected to the SYNC input when using the
internal node sync facility.
COUNT7-0
BERR
The 8-bit COUNT7-0 input defines the period (number of
bits) used in the node synchronization circuit when EXTSEL
is set high. The 8-bit number N is used to set up a period of
(256N + 256) bits internally, where N is the value of
COUNT7-0. An out-of-sync condition is declared (i.e., the
output pin OOS is set high and AUTO pulses high) if the
renormalization count exceeds the threshold value during a
period of this number of bits.
The Bit Error output indicates that an error has been detected
in either the G1 or G2 symbols corresponding to the current
output bit.
G1ERR
The G1 Error output indicates that an error has been detected
in the G1 symbol corresponding to the current output bit.
G2ERR
The G2 Error output indicates that an error has been detected
in the G2 symbol corresponding to the current output bit.
Note: The BERR, G1ERR and G2ERR signals are in NRZ
format, i.e., the signals will not return to zero between two
consecutive errors. To generate pulsed outputs, the signals
can be gated with the ODCLK signal.
THR7-0
The 8-bit THR7-0 input defines the threshold for node
synchronization when EXTSEL is set high. The 8-bit number
N is used to set up a threshold value of (8N + 6) internally,
where N is the value of THR7-0. An out-of-sync condition is
declared (i.e., the output pin OOS is set high and AUTO
pulses high) if the renormalization count exceeds this
threshold value.
MICROPROCESSOR INTERFACE
DATA7-0
EXTSEL
All I/O and control functions can be accessed via the
DATA7-0 bus with the associated control signals. The
STEL-2060CCC is used as a memory or I/O mapped
peripheral to the host processor.
When the EXTSEL input is set high, the COUNT7-0 and
THR7-0 information is derived from the COUNT7-0 and
THR7-0 input pins. When it is set low, this information is
derived from the data written into addresses 0 and 1.
5
Powered by ICminer.com Electronic-Library Service CopyRight 2003
STEL-2060C
ADDR2-0
Since the default (reset) value of the BPER23-0 register is
FF FF FFH a potential problem occurs if the desired value is
not written into this register before the value of the counter
has already incremented past this value. If this is not done
the equality will not be detected until after the counter
overflows and increments to the desired value once again.
Even at the maximum rate of 45 Mbps this will take over 6
minutes and, at a more modest data rate, such as 1 Mbps, it
will take over 41/2 hours! In any case, the user can easily be
misled into believing that the circuit is not operating correctly
since the interrupts will not be generated as expected. It is
therefore imperative that the BPER23-0 value be written into
the STEL-2060CCC as soon as possible after a reset to ensure
that this condition does not take place. The maximum time
allowable is just less than the desired interrupt period itself,
since the counter begins counting right after the reset is
released.
e.g., if the desired interrupt period is one second, the
BPER23-0 value must be written within one second of the
reset. At a data rate of 1 Mbps the period would correspond
to 106 bits and the correct BPER23-0 value would be 103, or
00 03 E8H.
If, for some reason, it is not possible to do this, a dummy
value should first be written into the STEL-2060CCC. This
should be large enough so that, at the time of writing, the bit
counter will not have exceeded the dummy value. In this
way the first interrupt will be generated within a reasonable
period of time and the dummy value can then be overwritten
with the desired value. Again, care must be taken to ensure
that the BPER 23-0 value written is greater than the
instantaneous counter value, otherwise the same problem
will occur.
e.g., in the above example, if it is not possible to write the
BPER23-0 value until 5 seconds after the reset, then a dummy
BPER23-0 value corresponding to >5 seconds, e.g., 6 seconds,
or 00 17 70H should first be written. The desired value of
00 03 E8H must then be written within one second of an
interrupt generated by the STEL-2060CCC, thereby ensuring
that the counter has not exceeded the new value at that time.
The 3-bit address bus is used to access the various I/O
functions, as shown in the Memory Map table, below. Note
that some addresses contain both Read and Write registers.
These read and write mode registers are separate and contain
different data.
WRITE
The Write input is used to write data to the microprocessor
data bus. It is active low and is normally connected to the
write line of the host processor.
READ
The Read input is used to read data from the microprocessor
data bus. It is active low and is normally connected to the
read line of the host processor.
CSEL
The Chip Select input can be used to selectively enable the
microprocessor data bus. It is active low.
INT
The Interrupt output indicates when the Period Counter in
the BER Monitor has completed a count period, and that a
new value of BERCT is ready to be read from addresses 0H
and 1H, when INT will go high for one symbol period.
INPUT (WRITE) FUNCTIONS
COUNT7-0
The 8-bit COUNT7-0 data defines the period (i.e., the number
of bits) used in the node synchronization circuit. The 8-bit
number N is used to set up a period of (256N + 256) internally,
where N is the value of COUNT7-0. If the renormalization
count exceeds the threshold value during a period of this
number of bits then an out-of-sync condition is declared (i.e.,
the output pin OOS is set high and AUTO pulses high).
Reset value 00H.
THR7-0
The 8-bit THR7-0 data defines the threshold for node
synchronization when EXTSEL is set low. The function is
identical to that of the THR7-0 input signal. Reset value 00H.
BPER23-0
OUTPUT (READ) FUNCTIONS
The 24-bit BER Period data is used to set the period (number
of data bits) over which the mean BER is measured by the
BER Monitor. The period used is 1000 times the value of
BPER23-0. Reset value FFFFFFH.
Note: The BER Count function incorporated in the
STEL-2060CCC uses a counter to count the number of
thousands of bits received. When the value of this counter is
equal to the value written into BPER23-0 the number of errors
counted is dumped into the BERCT15-0 output register and
can be read from read addresses 0-1H. Simultaneously, both
the error and bit counters are reset and the process is restarted,
and an interrupt (INT) is generated to indicate that the new
value is ready to be read.
STEL-2060C
Powered by ICminer.com Electronic-Library Service CopyRight 2003
BERCT15-0
The 16-bit Bit Error Count data represents the mean Bit Error
Rate over the period determined by the BER Period data
BPER23-0. The actual BER is given by:
BER =
8 x BERCT15-0
1000 x BPER23-0
The value will be updated each time the period counter
completes its count. Completion is indicated by the INT
output going high for one clock cycle. If the accumulator
overflows during a measurement period its output will be
caused to saturate at a value of FFFFH.
6
MICROPROCESSOR INTERFACE MEMORY MAP
WRITE MODE REGISTERS
ADDR2-0
DATA7
DATA6
DATA5
DATA4
DATA3
DATA2
DATA1
DATA0
0
1
COUNT7
THR7
COUNT6
THR6
COUNT5
THR5
COUNT4
THR4
COUNT3
THR3
COUNT2
THR2
COUNT1
THR1
COUNT0
THR0
2
3
BPER7
BPER6
BPER5
BPER4
BPER3
BPER2
BPER1
BPER0
4
5
BPER15
BPER23
BPER14
BPER22
BPER13
BPER21
BPER12
BPER20
BPER11
BPER19
BPER10
BPER18
BPER9
BPER17
BPER8
BPER16
READ MODE REGISTERS
ADDR2-0
0
1
DATA7
DATA6
DATA5
DATA4
DATA3
DATA2
DATA1
DATA0
BERCT7
BERCT15
BERCT6
BERCT14
BERCT5
BERCT13
BERCT4
BERCT12
BERCT3
BERCT11
BERCT2
BERCT10
BERCT1
BERCT9
BERCT0
BERCT8
PUNCTURED MODE OPERATION
CONCEPT OF PUNCTURING
USING INTERNAL DEPUNCTURING
In punctured codes some of the symbols generated by the
convolutional encoder are deleted, or punctured, from the
transmitted sequence. For example, in a rate 1 / 2
(unpunctured) sequence, four symbols are transmitted for
every two data bits. If one symbol out of every group of
four was punctured from the sequence then only three
symbols would be transmitted for every two data bits. This
would result in a rate 2/3 code. The coding gain is
significantly less than that for unpunctured operation, but
this is the trade-off for the reduced bandwidth required to
transmit the information. The STEL-2060CC decoder is
designed to operate with punctured codes as well as rate
1
/2 code. Two methods are provided for depuncturing the
symbols. The external method can be used for all rates and
patterns up to rate 7/8 by indicating which symbols were
punctured at the encoder (and then reinserted prior to
decoding) with the PNCG1 and PNCG2 signals. When one
of these signals is set high the input data at the corresponding
symbol input is ignored, and the internally generated metric
for the symbol pair (G1 and G2) assigns a zero weight to the
punctured symbol in the pair. This is done because the
Viterbi decoder has no way of knowing what the punctured
symbol should have been. The recommended puncturing
sequences for the various (N-1)/ N rates of punctured
operation are shown in the tables. The portions shown in
boldface are the basic sequences, which are then repeated.
The use of the PNCG1 and PNCG2 signals for rate 3/4 is
shown in the External Puncturing timing diagrams. The
sequence for rate 3/4 is G1 G2 P G2 G1 P, and the punctured
symbols are marked with asterisks in the timing diagrams.
The internal depuncturing method supports Rates 2/3, 3/4
and 7/8 using the puncturing sequences specified in the
tables, and the node sync process will automatically find the
correct places to reinsert the punctured symbols. The
puncturing sequences used for the rates supported are shown
in the tables, along with the speeds of the two input clocks,
SYMCKIN and DCLKIN. The portions shown in boldface
are the basic sequences, which are then repeated. The
symbol sequence for rate 2/3 is shown in the Internal
Puncturing timing diagrams. The sequence is G1 G2 P G2;
only G1 symbols are punctured in this particular sequence.
The SYMCKIN and DCLKIN clocks are used to clock in the
symbols and to clock out the data bits. The speeds of these
two clocks vary according to the modes and rate in use.
When operating in the External Depuncturing modes the
DCLKIN signal is not used and the SYMCKIN speed will be
equal to the data rate (D) in the parallel mode (PARL = 1) and
be equal to twice the data rate (2D) in the sequential mode
(PARL = 0). In the Internal Depuncturing modes, however,
DCLKIN is used and the speed of SYMCKIN will be as
shown in the tables.
The depuncturing circuit in the STEL-2060CC takes symbols
synchronized to the SYMCKIN signal and depunctures them,
i.e., inserts dummy symbols into the signal stream at the
appropriate positions. To do this the circuit handles the
symbols as a group of up to four pairs, depending on the code
rate. This circuit operates at the SYMCKIN rate and, at this
point, the group of depunctured symbols is transferred into
another set of registers clocked with DCLKIN. There is an
internal timing requirement that the falling edge of DCLKIN
7
Powered by ICminer.com Electronic-Library Service CopyRight 2003
STEL-2060C
must follow the rising edge of SYMCKIN by a minimum of
8 nsec. at this instant, otherwise the handover will not occur
correctly. This occurs once every 4 cycles of DCLKIN at rate
7
/8, every 2 cycles at rate 3/4 and every 3 cycles at rate 2/3, the
number of cycles of SYMCKIN depending on whether the
parallel or sequential input mode is used, as well as the code
rate. Two examples of clock phasing for rate 3/4 parallel
operation are shown below.
Here, the timing is such that the falling edges of DCLKIN
only coincide with falling edges of SYMCKIN, never with
rising edges. In this case the timing violation never occurs at
any speed, since the non-coincident falling edges of DCLKIN
will trail the rising edges of SYMCKIN by approximately 8
nsec. at a speed of 45 Mbps. Again, the same caveat regarding
jitter must be observed. However, the symbol signal setup
and hold requirements, shown in page 11 of the data sheet
make it necessary for SYMCKIN to have a minimum low
time of 12 nsec. to satisfy these requirements, so that it is not
possible for this signal to be a square wave above 40 MHz for
this reason. Since the method for eliminating the clock
timing violation presented here relies on the use of square
waves (50% mark-space ratio), it cannot be used above 40
MHz because of the setup and hold time requirements.
Again, a similar condition exists for rate 7/8 operation, as
shown below.
SYMCKIN
Good
Bad
DCLKIN
In the first example the falling edges of DCLKIN never
coincide with the rising edges of SYMCKIN. Thus the timing
violation will never occur provided that the delay from the
non-coincident rising edges of SYMCKIN to the following
falling edge of DCLKIN (shown by the arrows) exceeds 8
nsec. In the second case the falling edge of DCLKIN coincides
with the rising edge of SYMCKIN once every two cycles of
DCLKIN, resulting in a 50% probability that this may be the
point at which the handover occurs, creating the problem
discussed above.
There are several ways to satisfy the timing requirement,
depending on the code rate. One effective way which works
at all rates is to generate SYMCKIN from DCLKIN by
puncturing the clock to reduce its frequency while keeping
all the edges synchronized; this will require the use of a small
FIFO to buffer the input symbols to cope with the punctured
clock. The rate 3/4 timing is shown below as an example.
SYMCKIN
DCLKIN
However, in this case the timing violation will begin to occur
at speeds over 15 Mbps, so that this method of solving the
timing problem is less useful for rate 7/8 operation.
Note that for sequential mode operation (PARL = 0) the
frequency of the SYMCKIN signal will be doubled in every
case. This presents a problem with the synchronized clock
method presented here since it will not be possible to generate
the necessary waveforms with the correct mutual phasing
guaranteed because of the phase ambiguity of the SYMCKIN
signal itself relative to the internal handover process. In this
case it will be necessary to use either the punctured clock
approach or the synchronized reset approach.
The third method, which, while having the disadvantage
that it is susceptible to loss of sync from disturbances, is
easier to implement than clock puncturing and provides a lot
more margin than simple clock phase synchronization. It
consists of a synchronized reset generator used in conjunction
with clock phase synchronization. This is shown below for
rate 7/8.
SYMCKIN
DCLKIN
For rate 3/4 operation it is possible to generate the two clocks
with a mutual phase relationship that can exclude the timing
violation, as was shown in the first figure. As previously
stated, provided the two clocks can be generated as shown in
the first example, where only the rising edges of DCLKIN
coincide with the edges of SYMCKIN, the timing violation
will never occur provided that the delay from the noncoincident rising edges of SYMCKIN to the following falling
edge of DCLKIN (shown by the arrows) exceeds 8 nsec. In
the example shown this will be true at data rates up to 30
Mbps. Care must be taken to ensure that jitter between the
clocks is kept low enough to avoid the timing violation
condition. A similar condition exists for rate 2/3 operation,
as shown below.
SYMCKIN
➀
➁
➂
➃
➄
➅
➆
➇
➀
DCLKIN
15 nsec. min.
RESET
SYMCKIN
This ensures that the STEL-2060CC starts up during the
optimum phase of the SYMCKIN/DCLKIN repetition cycle,
i.e., the phase with the maximum separation between the
rising edge of SYMCKIN and the next falling edge of DCLKIN;
DCLKIN
STEL-2060C
Powered by ICminer.com Electronic-Library Service CopyRight 2003
8
separation between the rising edge of SYMCKIN and the
following falling edge of DCLKIN the same situation will
occur again every four cycles of DCLKIN, which is precisely
where the timing sensitivity occurs, as shown in the figure.
This method ensures good timing margins for stable operation
at all data rates up to 40 Mbps, this limit again being set by
the symbol setup and hold time requirements and need to
use square waves for the clocks.
the decoder will then continue to operate correctly provided
that nothing occurs to upset this cycle. This method exploits
the fact that not all falling edges of DCLKIN have a timing
sensitivity relative to the rising edges of SYMCKIN, as was
discussed earlier. For example, in rate 7/8 operation only
every fourth falling edge of DCLKIN is critical, starting with
the fourth one after the reset is released. By timing the rising
edge of the reset at a point in the cycle where there is good
ELECTRICAL CHARACTERISTICS
ABSOLUTE MAXIMUM RATINGS
Warning:
Stresses greater than those shown below may cause permanent damage to the
device. Exposure of the device to these conditions for extended periods may also affect device
reliability.
Symbol
Parameter
Range
Units
Tstg
Storage Temperature
–65 to +150
°C
Ta
Operating Temperature (Ambient)
–40 to +85
°C
VDDmax
Max. voltage between VDD and VSS
+7 to –0.7
volts
VI/O(max)
Max. voltage on any input or output pin
VDD + 0.3
volts
VI/O(min)
Min. voltage on any input or output pin
VSS – 0.3
volts
RECOMMENDED OPERATING CONDITIONS
Symbol
Parameter
Range
Units
VDD
Supply Voltage
+5 ± 10%
volts
Ta
Operating Temperature (Ambient)
0 to +70
°C
D.C. CHARACTERISTICS
Symbol
(Operating Conditions: VDD = 5.0 ± 5% volts, Ta = 0° to 70° C)
Parameter
Min. Typ. Max.
1.0
Units
Conditions
mA
Static, no clock
mA/Mbps
@ 45 Mbps (fDATA)
volts
Guaranteed Logic '1'
volts
Guaranteed Logic '0'
volts
IO = –4.0 mA
IDD(Q)
Supply Current, Quiescent
IDD
Supply Current, Operational
VIH(min)
Min. High Level Input Voltage
VIL(max)
Max. Low Level Input Voltage
VOH(min)
Min. High Level Output Voltage
VOL(max)
Max. Low Level Output Voltage
0.4
volts
IO = +4.0 mA
IIH(max)
Max. High Level Input Current
10
µA
VIN = +5.0 volts
IIL(max)
Max. Low Level Input Current
–10
µA
VIN = 0 volts
8
2.0
0.8
2.4
9
Powered by ICminer.com Electronic-Library Service CopyRight 2003
STEL-2060C
RATE 1/2 AND EXTERNAL DEPUNCTURING
TIMING.
(a)
PARALLEL INPUT MODE (PARL = 1)
SYMCKIN
tSU
tSHPE
tHD
SYMB. N+1
SYMB. N
G1/2
tSLPE
SYMB. N+2
PNCG1/G2
tSO
ODCLK
tOD
DATO
(b)
SEQUENTIAL INPUT MODE (PARL = 0)
SYMCKIN
tSHS
tHD
tSU
tSLS
LDG2
G1/2
G1
G2
G1
G2
G1
PNCG1
PNCG2
tSO
ODCLK
tOD
DATO
STEL-2060C
Powered by ICminer.com Electronic-Library Service CopyRight 2003
10
G2
G1
G2
A.C. CHARACTERISTICS
Symbol
fDAT
(Operating Conditions: VDD = 5.0 ±5% volts, Ta = 0° to 70° C)
Parameter
Min.
Data speed, Parallel input mode (PARL = 1),
Max.
45
Units
Conditions
Mbps
PARL = 1 and
all rates when not using Internal Depuncturing
fDAT
RATE2-0 = 0
Data speed, Sequential input mode (PARL = 0), Rate 1/2
45
Mbps
PARL = 0 or
RATE2-0 ≠ 0
and all rates and modes when using Internal Depuncturing
tSU
G1, G2, PNCG1 or PCNG2, LDG2 to SYMCKIN Setup
3
nsecs.
PARL = 0 or
tHD
G1, G2, PNCG1 or PCNG2, LDG2 to SYMCKIN Hold
5
nsecs.
RATE2-0 = 0
tSUI
G1, G2, PNCG1 or PCNG2, LDG2 to SYMCKIN Setup
6
nsecs.
PARL = 1 and
tHDI
G1, G2, PNCG1 or PCNG2, LDG2 to SYMCKIN Hold
6
nsecs.
RATE2-0 ≠ 0
10
nsecs.
PARL = 1 and
tSHPE, tSLPE SYMCKIN Pulse width (high or low), Parallel input mode,
RATE2-0 = 0
all rates when not using Internal Depuncturing
tSHPI, tSLPI
SYMCKIN Pulse width (high or low), Parallel input mode,
12
nsecs.
RATE2-0 ≠ 0
all rates when using Internal Depuncturing
tSHS, tSLS
PARL = 1 and
SYMCKIN Pulse width (high or low), Serial input mode
6
tSO
SYMCKIN to ODCLK stable delay (RATE2-0 = 000)
3
8
nsecs.
tDO
DCLKIN to ODCLK stable delay (RATE2-0 ≠ 000)
3
8
nsecs.
Load = 15 pF
tOD
ODCLK to output stable delay, all other outputs
1
3
nsecs.
Load = 15 pF
11
Powered by ICminer.com Electronic-Library Service CopyRight 2003
nsecs.
}
}
}
}
}
}
PARL = 0
STEL-2060C
PUNCTURED SYMBOL SEQUENCES
RATE 1/2 AND EXTERNAL DEPUNCTURING - SEQUENTIAL INPUTS (PARL = 0)
1.
Rate
Symbol sequence (suffix is symbol number in sequence, all at G12-0 input)
1
/2
G11
G21
G12
G22
G13
G23
2
/3
G11
G21
P2
G22
G13
G23
P4
G24
G15
G25
P6
G26
3
/4
G11
G21
P2
G22
G13
P3
G14
G24
P5
G25
G16
P6
G17
4
/5
G11
G21
P2
G22
P3
G23
P4
G24
G15
G25
P6
P6
5
/6
G11
G21
P2
G22
G13
P3
P4
G24
G15
P5
G16
6
/7
G11
G21
P2
G22
P3
G23
G14
P4
P5
G25
7
/8
G11
G21
P2
G22
P3
G23
P4
G24
G15
P5
SYMCKIN DCLKIN
(Not Punctured)
2D
0
2D
0
G27
2D
0
G27
P7
2D
0
G26
P7
G27
2D
0
G16
P6
G17
G27
2D
0
P6
G26
G17
P7
2D
0
2. RATE 1/2 AND EXTERNAL DEPUNCTURING - PARALLEL INPUTS (PARL = 1)
Rate
Input
Symbol sequence (suffix is symbol number in sequence)
1
/2
G12-0 G11
G22-0 G21
G12
G22
G13
G23
G14
G24
G15
G25
G16
G26
2
/3
G12-0 G11
G22-0 G21
P2
G22
G13
G23
P4
G24
G15
G25
P6
G26
G17
G27
P8
G28
G19
G29
3
/4
G12-0 G11
G22-0 G21
P2
G22
G13
P3
G14
G24
P5
G25
G16
P6
G17
G27
P8
G28
4
/5
G12-0 G11
G22-0 G21
P2
G22
P3
G23
P4
G24
G15
G25
P6
G26
P7
G27
5
/6
G12-0 G11
G22-0 G21
P2
G22
G13
P3
P4
G24
G15
P5
G16
G26
6
/7
G12-0 G11
G22-0 G21
P2
G22
P3
G23
G14
P4
P5
G25
7
/8
G12-0 G11
G22-0 G21
P2
G22
P3
G23
P4
G24
G15
P5
3.
SYMCKIN DCLKIN
(Not Punctured)
D
0
P10
G210
D
0
G19
P9
G110
G210
D
0
P8
G28
G19
G29
P10
G210
D
0
P7
G27
G18
P8
P9
G29
G110
P10
D
0
G16
P6
G17
G17
P8
G28
P9
P9
G110
G210
D
0
P6
G26
G17
P7
G18
G28
P9
G29
P10
G210
D
0
INTERNAL DEPUNCTURING - SEQUENTIAL INPUTS (PARL = 0)
Rate
1
2
3
7
4.
/2
/3
/4
/8
Symbol sequence (suffix is symbol number in sequence, all at G12-0 input)
G11
G11
G11
G11
G21
G21
G21
G21
G12
G22
G22
G22
G22
G13
G13
G23
G13
G23
G14
G24
G23
G24
G24
G15
G14
G24
G15
G25
G25
G16
G26
G17
G15
G26
G17
G18
SYMCKIN DCLKIN
(Not Punctured)
G17
G27
G28
G27
G28
G29
2D
0
G28
3
/2 D
D
G19
4
/3 D
D
G210
8
/7 D
D
INTERNAL DEPUNCTURING - PARALLEL INPUTS (PARL = 1)
Rate
Input
Symbol sequence (suffix is symbol number in sequence)
1
/2
G12-0
G22-0
G11
G21
G12
G22
G13
G23
G14
G24
G15
G25
G16
G26
G17
G27
2
/3
G12-0
G22-0
G11
G21
G22
G13
G23
G24
G15
G25
G26
G17
G27
G28
G19
G29
G210
G111
G211
G212
3
/4
G12-0
G22-0
G11
G21
G22
G13
G14
G24
G25
G16
G17
G27
G28
G19
G110
G210
G211
G112
G113
G213
7
/8
G12-0
G22-0
G11
G21
G22
G23
G24
G15
G26
G17
G18
G28
G29
G210
G211
G112
G113
G214
G115
G215
STEL-2060C
Powered by ICminer.com Electronic-Library Service CopyRight 2003
12
SYMCKIN DCLKIN
(Not Punctured)
D
0
3
/4 D
D
G214
G115
2
/3 D
D
G216
G117
4
/7 D
D
INTERNAL DEPUNCTURING TIMING.
(a)
RATE 2/3 SHOWN
PARALLEL INPUT MODE (PARL = 1)
SYMCKIN *
t SUI
tSHPI
t HDI
G2 N+1**
G1 N
G1
tSLPI
G2 N+2
G1 N+4
G2 N+3***
G2 N+4
Basic Pattern
G2 N
G2
G1 N+2
DCLKIN
tDO
ODCLK
tOD
DATO
** G1 N+1 Punctured
*
(b)
*** G1 N+3 Punctured
Within each symbol period, rising edge of SYMCLKIN must precede falling edge, as shown
SEQUENTIAL INPUT MODE (PARL = 0)
SYMCKIN
tSUI
tSHS
tHDI
tSLS
Basic Pattern
G1/2
G1 N
G2 N
G2 N+1 **
G1 N+2
G2 N+2
G2 N+3 ***
G1 N+4
G2 N+4
DCLKIN
tDO
ODCLK
tOD
DATO
**
G1 N+1 Punctured
13
Powered by ICminer.com Electronic-Library Service CopyRight 2003
***
G1 N+3 Punctured
STEL-2060C
RATE 3/4 SHOWN
EXTERNAL DEPUNCTURING TIMING.
(a)
PARALLEL INPUT MODE (PARL = 1)
SYMCKIN
tSU
G1
tHD
*
*
G2
*
*
*
PNCG1
PNCG2
ODCLK
DATO
* Indicates Punctured Symbols
(b)
SEQUENTIAL INPUT MODE (PARL = 0)
SYMCKIN
tSU
tHD
LDG2
G1/2
G2
G1
*
G2
G1
G2
*
G1
G2
G1
*
G2
G1
G2
*
G1
G2
G1
PNCG1
PNCG2
ODCLK
DATO
* Indicates Punctured Symbols
STEL-2060C
Powered by ICminer.com Electronic-Library Service CopyRight 2003
14
*
G2
MICROPROCESSOR INTERFACE TIMING
1. WRITE MODE
CSEL
ADDR2-0
DON'T CARE
DATA 7-0
DON'T CARE
tSM
tHM
tWW
WRITE
2. READ MODE
CSEL
ADDR2-0 DON'T CARE
DON'T CARE
tSM
tHM
READ
tVZ
tZV
DATA 7-0
A.C. CHARACTERISTICS
(Operating Conditions: VDD = 5.0 ±5% volts, Ta = 0° to 70° C)
Symbol
Parameter
Min.
Max.
Units
tSM
CSEL, ADDR2-0 or DATA7-0 to WRITE or READ Setup
10
nsecs.
tHM
CSEL, ADDR2-0 or DATA7-0 to WRITE or READ Hold
5
nsecs.
tWW
WRITE Pulse width
5
nsecs.
tZV
READ (low) to DATA7-0 Valid
10
nsecs.
tVZ
READ (high) to DATA7-0 High-Impedance
10
nsecs.
15
Powered by ICminer.com Electronic-Library Service CopyRight 2003
STEL-2060C
NODE SYNCHRONIZATION
assumed that the data was modulated using QPSK. The
appropriate synchronization sequences will be invoked, as
shown in the node sync sequence tables. Note that the
pipeline delay through the device will be affected by the
node sync state. If multiple devices are used in parallel to
achieve higher data rates, it is necessary for the all devices to
have the same node sync state to equalize their pipeline
delays. It will be necessary to reset the devices together to
achieve this state
In a communication system using Viterbi decoding the decoder will only operate correctly when the symbols G1 and
G2 are loaded into the decoder in the correct order. Identifying which symbol is G1 and which one is G2 is referred to
as node synchronization. The STEL-2060C contains a circuit
designed to carry out the node synchronization function
automatically. It uses the internally generated metrics of the
received sequence to do this. These constantly changing
parameters are periodically renormalized to keep them
within bounds. If renormalization occurs too frequently it is
a good indication that the system is not converging, most
likely due to lack of node synchronization. The renormalization rate at which the system will decide to change the
node sync is determined by the threshold parameter. This is
an 8-bit number which is set by the THR7-0 inputs. When the
renormalization count exceeds this value, the OOS output
will go high and the AUTO output will pulse high for one
clock cycle, as shown during Count Window N in the timing
diagram below. The counter is reset after a number of bits
determined by the number set by the COUNT7-0 inputs, so
that the threshold must be exceeded somewhere in that
period for resynchronization to take place. OOS will be
reset if the counter then counts through an entire window
and the threshold is not exceeded, as shown during Count
Window N + 1 in the timing diagram below. The most
suitable threshold setting will depend on the value of
Eb/N0, the coding rate, and the signal level at the G1 and G2
inputs. For full scale inputs, i.e., the peak signal values
almost saturate the digital inputs, suitable starting values
for the threshold will be 1% for Rate 1/2, 0.5% for Rates 2/3 to
6
/7, and 0.1% for Rate 7/8. e.g., for Rate 1/2, if the number of
bits over which the measure is made is set to 512
(COUNT7-0 = 01H) the threshold should be set to 5. Setting
THR7-0 = 0 gives a value of 6, which is adequately close.
More reliable results will be obtained by counting over a
longer period to improve the averaging process, but this
increases the time taken to make a decision and hence to
acquire node sync. Thus, starting with a low count period
and then increasing it (and adjusting the threshold accordingly to maintain a value of 1%) when OOS goes low will
result in a faster acquisition of correct node sync with a
lower probability of accidental loss of node sync once correct sync has been achieved. To use the internal node sync
the AUTO output must be connected to the SYNC input.
The synchronization sequence depends on the setting of the
PARL input. When PARL is set low it is assumed that the
data was modulated using BPSK, and when it is set high it is
When internal depuncturing is used, additional node sync
states exist because of the uncertainty of the current symbol
position in the puncture sequence. In this case the node sync
circuit will also search through the sequence by adding
delays in the depuncturing process to precess through the
sequence. In the sequential input mode (PARL = 0) this is
simply an extension of the node sync process, since the
alternate state is achieved by delaying the symbols. In the
parallel input mode, however, this is different from the
"invert G2 and swap" process, and in this sync sequence
"invert G2 and swap" precedes the delay addition, so that
the system goes through both the initial and alternate states
for each delay addition tried. This is shown for the Rate
2
/3 case. In each case the symbols are read into the depuncturing circuit in groups of three (in the BPSK mode) or six (in
the QPSK mode) and attempts are made to reinsert the
punctured symbol in all of the possible insertion positions.
The positions of the punctured symbols in the sequences are
shown by the asterisks (*). The resulting groups of four or
eight symbols are then decoded in pairs, resulting in two
decoded bits in the BPSK mode and four bits in the QPSK
mode. For higher rates the sequences will be extensions of
this procedure.
When external depuncturing is used, the determination of
which symbols were punctured, and need to be reinserted
into the symbol sequence, is part of the node sync process.
This is because the acquisition of correct node sync cannot be
completed until the punctured symbols are reinserted correctly. The AUTO and OOS outputs of the STEL-2060C can
be used as indicators of the operation of the internal node
sync process; OOS will remain high as long as node sync has
not been achieved and AUTO will pulse each time a new
node sync state is being tried. Since there are only two
possible internal node sync states, alternate pulses on the
AUTO output can be used as an indication that the depuncturing is incorrect and a new depuncturing sequence should
be tried externally.
NODE SYNC TIMING
Count Window N
Count Window N+1
ODCLK
AUTO
OOS
STEL-2060C
Powered by ICminer.com Electronic-Library Service CopyRight 2003
16
NODE SYNC SEQUENCES
1. RATE 1/2
PARL
0
Input
Initial State
2. RATE 2/3, PARL = 0 (BPSK MODE)
Alternate State
G1
G1n
G2
N.A.
N.A.
G1
G1n
G2n
G2
G2n
G1n
G2n
G2n
Input
Initial State
G1n+1
G1
2
G1n G2n* G2n+1 G2n* G2n+1 G1n+2 G2n+1 G1n+2 G2n+2*
(No delay)
1
Alternate States
1
(One symbol
delay)
(Two symbol
delay)
3. RATE 2/3, PARL = 1 (QPSK MODE)
Input
Initial State
1
Alternate States
3
2
4
5
G1
G1n G2n+1 G2n+2*
G2n* G1n+2 G2n+3
G2
G2n* G1n+2 G2n+3
G1n G2n+1 G2n+2* G2n+1 G2n+2* G1n+4 G2n* G1n+2 G2n+3 G1n+2 G2n+3 G2n+4 G2n+1 G2n+2*G1n+4
(No Invert/Swap (Invert/Swap
No delay)
No delay)
G2n* G1n+2 G2n+3 G2n+1 G2n+2* G1n+4 *G2n+1 G2n+2*G1n+4 G1n+2 G2n+3 G2n+4*
(No Invert/Swap (Invert/Swap
(No Invert/Swap
One symb. delay) One symb. delay) Two symb. delay)
(Invert/Swap
Two symb. delay)
BER PERFORMANCE
10–1
10–2
R = 1/2
–3
BER 10
R = 2/3
R = 3/4
10–4
R = 7/8
10–5
Uncoded
10–6
10–7
2
3
4
5
6
7
Eb/No
8
The coding gain obtained by the use of Convolutional coding
and Viterbi decoding is extremely dependent on many
parameters. Not surprisingly, the code rate is a primary
factor, but so are the bit error rate (BER) and amplitude of the
input signal. The BER affects the coding gain because the
error correction capability of the Viterbi decoder is
dependent on the statistics of the errors, specifically the
clustering of errors. As the BER of the input signal increases,
so does the clustering, causing a reduction in the error
correcting capability of the device, along with the coding
gain. The signal amplitude is important because of the
weighting given to the signal amplitude as an indication of
the likelihood of an error in a given symbol pair.
10
11
Consequently it is important to maintain the signal
amplitude at an optimum level in order to maximize the
performance. The performance curves shown above were
measured using a digital link simulator with the signal level
set at one half of full scale; i.e., the signal amplitude without
noise ranged from 101 to 001 in signed magnitude format, or
101 to 010 in offset binary format. The coding gain under
these conditions is about 0.2 dB less than that under
optimum signal level conditions. The performance of the
STEL-2060C is shown here for unpunctured operation
(Rate 1/2) as well as punctured operation at the rates for
which internal depuncturing is supported (Rates 2/3, 3/4
and 7/8). The error rate for uncoded data is shown for
comparison.
17
Powered by ICminer.com Electronic-Library Service CopyRight 2003
9
STEL-2060C
APPLICATION INFORMATION
The STEL-2060C can be used in a variety of different
environments. One example of a system using a convolutional
encoder with the STEL-2060C Viterbi decoder is illustrated
here. The STEL-2060C cannot be used as a common decoder
in multi-channel applications because of the memory
incorporated on the chip which is dedicated to a single
channel.
The system modulates a data stream of rate 45 Mbps using
binary PSK (BPSK) or quaternary PSK (QPSK). To be able to
use convolutional coding, the system must either have
available the additional bandwidth needed to transmit
symbols at a higher rate or must be able to make use of higher
levels of modulation. e.g., by changing from BPSK to QPSK
modulation, the data can be encoded at Rate 1/2 without
requiring any additional bandwidth. The performance
improvement that can be expected is shown in the graph
below.
The STEL-2060C is designed to accept symbols
synchronously. SYMCKIN is supplied by the user to clock
in the symbols. The maximum data rate is 45 Mbps, using a
SYMCKIN frequency of 45 MHz (when PARL is set high) or
90 MHz (when PARL is set low) at rate 1/2, corresponding to
90 MSymbols per sec.
10–1
6
3
2
10–2
I
Tx DATA
45 Mbps
RATE 1/2
CONV.
ENCODER
6
QPSK
MODULATOR
Q
3
2
CHANNEL
BW=90 MHz
CODED DATA @ 90 Mbps
45 Mbps
RATE 1/2
VITERBI
DECODER
Q
6
BER
3
2
Coded
Uncoded
10–4
6
3
2
I
Rx DATA
10–3
Coding Gain
10–5
QPSK
DEMOD.
6
3
2
10–6
6
CODED DATA @ 90 Mbps
3
2
10–7 2
3
4
5
6
7
8
Eb/N 0 dB
QPSK Communication System Using Convolutional
Encoding and Viterbi Decoding. Rate = 1/2
STEL-2060C
Powered by ICminer.com Electronic-Library Service CopyRight 2003
18
9
10
11
12
Information in this document is provided in connection with
Intel® products. No license, express or implied, by estoppel
or otherwise, to any intellectual property rights is granted by
this document. Except as provided in Intel’s Terms and Conditions of Sale for such products, Intel assumes no liability
whatsoever, and Intel disclaims any express or implied
warranty, relating to sale and/or use of Intel® products including liability or warranties relating to fitness for a particular purpose, merchantability, or infringement of any patent,
copyright or other intellectual property right. Intel products
are not intended for use in medical, life saving, or life sustaining applications.
Intel may make changes to specifications and product descriptions at any time, without notice.
For Further Information Call or Write
INTEL CORPORATION
Cable Network Operation
350 E. Plumeria Drive, San Jose, CA 95134
Customer Service Telephone: (408) 545-9700
Technical Support Telephone: (408) 545-9799
FAX: (408) 545-9888
Copyright © Intel Corporation, December 15, 1999. All rights reserved
Powered by ICminer.com Electronic-Library Service CopyRight 2003