LOGIC LF3321

LF3321
Horizontal Digital Image Filter
DEVICES INCORPORATED
Improved Performance
FEATURES
Supports Interleaved Data Streams
Supports Decimation up to 16:1 for Increasing Number of Filter Taps
3.3 Volt Power Supply
144 Lead PQFP
111 MHz Data Rate
12-bit Data or Coefficients (Expandable to 24-bit)
32-Tap FIR Filter, Cascadable for More Filter Taps
LF Interface™ Allows All 256 Coefficient Sets to be
Updated Within Vertical Blanking
Over 49 K-bits of on-board Memory
Various Operating Modes: Dual Filter, Single Filter,
Double Wide Data or Coefficient, Matrix Multiplication, and Accumulator Access.
Selectable 16-bit Data Output with User-Defined
Rounding and Limiting
DESCRIPTION
The LF3321 is an improved version of the LF3320 Horizontal Digital Image Filter capable of operating at
speeds of up to 111MHz. This improved speed increases flexibility and performance and enables the user
to utilize this device in more applications. For example, four interleaved data streams of 27MHz can now
be processed within one device. The part is functionally identical to the LF3320 with the exception that the
filter data path is specified to operate faster than the LF Control Interface. When operating the filter at
speeds in excess of 90MHz, loading of coefficients via the LF Interface must be throttled to a maximum
of 90MHz by asserting the PAUSE pin as required to allow sufficient setup time for the configuration data
provided to the chip via the LF Interface.
Figure 1. Switching Waveforms: LF InterfaceTM
CLK
tPWH
tS0
LDA
LDB
tPS
PAUSEA
PAUSEB
CFA 11-0
CFB 11-0
tPWL
tLH
tCYC
tPH
tCFH
tCFS
ADDRESS
CF0
CF1
Figure 1 demonstrates the switching waveforms, while the switching characteristics are given in Table
1. The LF3321 filters digital images in the horizontal dimension at real-time video rates. The input and
coefficient data are both 12 bits and in two’s complement format. The output is also in two’s complement
format and may be rounded to 16 bits.
The LF3321 is designed to take advantage of symmetric coefficient sets. When symmetric coefficient sets
are used, the device can be configured as a single 32-tap FIR filter or as two separate 16-tap FIR filters.
When asymmetric coefficient sets are used, the device can be configured as a single 16-tap FIR filter or as
two separate 8-tap FIR filters. Multiple LF3321s can be cascaded to create larger filters.
Interleave/Decimation Registers (I/D Registers) allow interleaved data to be fed directly into the device and
filtered without separating the data into individual data streams.
The LF3321 can handle a maximum of sixteen data sets interleaved together. The I/D Registers and
on-chip accumulators facilitate using decimation to increase the number of filter taps. Decimation of up
to 16:1 is supported.
The LF3321 contains enough on-board memory to store 256 coefficient sets. Two separate LF InterfacesTM
allow all 256 coefficient sets to be updated within vertical blanking.
LOGIC Devices Incorporated
1
Video Imaging Products
Feb 5, 2003 LDS.3321-A
LF3321
Horizontal Digital Image Filter
DEVICES INCORPORATED
Improved Performance
DESCRIPTION
Table 1. Switching Characteristics
Commercial Operating Range (0ºC to + 70ºC)
9* (ns) speed grade
Symbol
Parameter
MIN
MAX
tCYC
Cycle Time
9
tPWL
Clock Pulse Width Low
4
tPWH
Clock Pulse Width High
4
tS0
Input Setup Time
4
tH0
Input Hold Time
0
tSCT
Setup Time Control Inputs
4
tHCT
Hold Time Control Inputs
0
tSCC
Setup Time Coefficient Control Inputs
4
tHCC
Hold Time Coefficient Control Inputs
0
tD
Output Delay
tDCC
Cascade Output Delay
7.5
tDIS
Three State Output Display Delay
10
tDIS
Three State Output Enable Delay
10
8
Figure 2. LF3321 Block Diagram
ROUT11-0
DIN11-0
CAA7-0
12
12
INTERLEAVE / DECIMATION
REGISTERS
12
12
8
8
12
12
CENA
CFA11-0
PAUSEA
LDA
RIN11-0
COUT11-0
CAB7-0
CENB
256
COEFFICIENT
SET
STORAGE
16-TAP
FILTER A
16-TAP
FILTER B
256
COEFFICIENT
SET
STORAGE
CFB11-0
PAUSEB
LDB
ROUND
SELECT
LIMIT
CIRCUITRY
CLK
OED
16
DOUT15-0
LOGIC Devices Incorporated
2
Video Imaging Products
Feb 5, 2003 LDS.3321-A
3
FILTER A
LF
INTERFACE
Coef Bank 7
Coef Bank 6
Coef Bank 5
Coef Bank 4
Coef Bank 3
CLK
SHENB
SHENA
RSLA3-0
TXFRB
TXFRA
4
B
A
ALU
B
A
ALU
B
A
ALU
B
B
A
ALU
B
A
ALU
B
A
ALU
B
A
ALU
B
32
27
13
13
32
13
"0"
13
27
13
13
ALU
B
25
25
25
25
ACCM A
25
25
25
25
DOUT15-0
16
16
16
RSLB OUT15-0
LIMIT
FILTER B
FILTER A
25
LIMIT
ROUND
ACCM B
25
SELECT
SCALE
25
ROUND
25
SELECT
32
CONFIGURATION AND
CONTROL REGISTERS
25
25
25
25
12
13
A
12
32
ALU
12
"0"
A
12
27
B
12
27
ALU
12
13
A
12
13
B
12
13
ALU
12
13
A
12
B
12
13
ALU
12
13
A
12
13
B
12
1-16
12
13
ALU
1-16
A
12
FILTER B
LF
INTERFACE
Coef Bank 8
Coef Bank 9
Coef Bank 10
Coef Bank 11
Coef Bank 12
Coef Bank 13
Coef Bank 14
Coef Bank 15
8
4
CAB7-0
12
12
OED
RSLB3-0
ACCB
CENB
COUT11-0
OEC
RIN11-0
DEVICES INCORPORATED
ACCA
12
CENA
Coef Bank 2
ALU
1-16
1-16
Coef Bank 1
A
1-16
1-16
B
1-16
1-16
ALU
1-16
1-16
A
1-16
1-16
12
B
1-16
1-16
Coef Bank 0
ALU
1-16
13
A
S E O I
8
12
DATA
REVERSAL
1-16
FILTER B I/D REGISTERS
1-16
1-16
FILTER A I/D REGISTERS
1-16
1-16
RSLB OUT15-12
1-16
4
8
1-16
1-16
CAA7-0
DIN11-0
ROUT11-4
RSLB
OUT11-0
OEC
1-16
1-16
CFA11-0
1-16
1-16
PAUSEA
1-16
1-16
CFB11-0
1-16
LDA
R E O I
LDB
DATA
REVERSAL
PAUSEB
LOGIC Devices Incorporated
ROUT3-0
LF3321
Horizontal Digital Image Filter
Improved Performance
DESCRIPTION
Figure 3. LF3321 Functional Block Diagram
Video Imaging Products
Feb 5, 2003 LDS.3321-A
LF3321
Horizontal Digital Image Filter
DEVICES INCORPORATED
Improved Performance
Functional Description
ALU
The ALUs double the number of filter taps available, when symmetric coefficient sets are used, by preadding data values which are then multiplied by a common coefficient (see Figure 4). The ALUs can
perform two operations: A+B and B–A. Bit 0 of Configuration Register 0 determines the operation of the
ALUs in Filter A. Bit 0 of Configuration Register 2 determines the operation of the ALUs in Filter B. A+B is
used with even-symmetric coefficient sets. B–A is used with odd-symmetric coefficient sets.
Also, either the A or B operand may be set to 0. Bits 1 and 2 of Configuration Register 0 and Configuration
Register 2 control the ALU inputs in Filters A and B respectively. A+0 or B+0 are used with asymmetric
coefficient sets.
Figure 4. Symmetric Coefficient Set Examples
8 7 6 5
8 7 6 5 4 3 2 1
Even-Tap, Even-Symmetric
Coefficient Set
I/D Registers
7 6 5 4 3 2 1
Odd-Tap, Even-Symmetric
Coefficient Set
4 3 2 1
Even-Tap, Odd-Symmetric
Coefficient Set
The Interleave/Decimation Registers (I/D Registers), feed the ALU inputs. They allow the device to filter up
to sixteen data sets interleaved into the same data stream without having to separate the data sets. The I/D
Registers should be set to a length equal to the number of data sets interleaved together.
For example, if two data sets are interleaved together, the I/D Registers should be set to a length of two.
Bits 1 through 4 of Configuration Register 1 and Configuration Register 3 determine the length of the I/D
Registers in Filters A and B respectively.
The I/D Registers also facilitate using decimation to increase the number of filter taps. Decimation by N is
accomplished by reading the filter’s output once every N clock cycles. The device supports decimation up to
16:1. With no decimation, the maximum number of filter taps is sixteen. When decimating by N, the number
of filter taps becomes 16N because there are N–1 clock cycles when the filter’s output is not being read. The
extra clock cycles are used to calculate more filter taps.
When decimating, the I/D Registers should be set to a length equal to the decimation factor. For example,
when performing a 4:1 decimation, the I/D Registers should be set to a length of four. When decimation is
disabled or when only one data set (non-interleaved data) is fed into the device, the I/D Registers should
be set to a length of one.
I/D Registers Data
Path Control
LOGIC Devices Incorporated
The three multiplexers in the I/D Register data path control how data is routed through the forward and
reverse data paths. The forward data path contains the I/D Registers in which data flows from left to right
in the block diagram in Figure 1. The reverse data path contains the I/D Registers in which data flows from
right to left. In Single or Dual Filter Modes, data is fed from the forward data path to the reverse data path
as follows. When the filter is configured for an even number of taps, data from the last I/D Register in the
forward data path is fed into the first I/D Register in the reverse data path (see Figure 5). When the filter
is configured for an odd number of taps, the data which will appear at the output of the last I/D Register
in the forward data path on the next clock cycle is fed into the first I/D Register in the reverse data path.
Bit 5 in Configuration Register 1 and Configuration Register 3 configures Filters A and B respectively for
an even or odd number of taps.
4
Video Imaging Products
Feb 5, 2003 LDS.3321-A
LF3321
Horizontal Digital Image Filter
DEVICES INCORPORATED
Improved Performance
Functional Description
Figure 5. I/D Register Data Paths
ALU
1-16
1-16
1-16
1-16
1-16
1-16
Delay Stage N
A
A
B
DATA
REVERSAL
ALU
B
1-16
ALU
A
B
1-16
A
DATA
REVERSAL
ALU
B
1-16
1-16
DATA
REVERSAL
1-16
1-16
A
Delay Stage N 1
ALU
B
A
ALU
B
COEF 7
COEF 7
2
COEF 7
2
COEF 6
COEF 6
COEF 6
EVEN-TAP MODE
ODD-TAP MODE
ODD-TAP INTERLEAVE MODE
When interleaved data is fed through the device and an even tap filter is desired, the filter should be
configured for an even number of taps and the I/D Register length should match the number of data sets
interleaved together. When interleaved data is fed through the device and an odd tap filter is desired, the
filter should be set to Odd-Tap Interleave Mode. Bit 0 of Configuration Register 1 and Configuration Register
3 configures Filters A and B respectively for Odd-Tap Interleave Mode. When the filter is configured for
Odd-Tap Interleave Mode, data from the next to last I/D Register in the forward data path is fed into the
first I/D Register in the reverse data path.
When the filter is configured for an odd number of taps (interleaved or non-interleaved modes), the filter is
structured such that the center data value is aligned simultaneously at the A and B inputs of the last ALU in
the forward data path. In order to achieve the correct result, the user must divide the coefficient by two.
Data Reversal
Data reversal circuitry is placed after the multiplexers which route data from the forward data path to the
reverse data path (see Figure 6). When decimating, the data stream must be reversed in order for data to
be properly aligned at the inputs of the ALUs.
Figure 6. Data Reversal
LOGIC Devices Incorporated
5
Video Imaging Products
Feb 5, 2003 LDS.3321-A
LF3321
Horizontal Digital Image Filter
DEVICES INCORPORATED
Improved Performance
Functional Description
Data Reversal
Continued
When data reversal is enabled, the circuitry uses a pair of LIFOs to reverse the order of the data sent
to the reverse data path. TXFRA and TXFRB control the LIFOs in Filters A and B respectively. When
TXFRA/TXFRB goes LOW, the LIFO sending data to the reverse data path becomes the LIFO receiving
data from the forward data path, and the LIFO receiving data from the forward data path becomes the LIFO
sending data to the reverse data path.
The device must see a HIGH to LOW transition of TXFRA/TXFR B in order to switch LIFOs. If decimating by
N, TXFRA/TXFRB should go LOW once every N clock cycles. When data reversal is disabled, the circuitry
functions like an I/D Register. When feeding interleaved data through the filter, data reversal should be
disabled. Bit 6 of Configuration Register 1 and Configuration Register 3 enables or disables data reversal
for Filters A and B respectively.
Figure 7 .Filter A and Filter B Round/Limit/Select Circuitry
RSLB3-0
DATA IN
32
32
RSLA3-0
4
RB0
RA0
4
DATA IN
32
RND
32
RB15
RA15
RND
32
SELECT
SELECT
16
16
SB0
SA0
32
5
LB0
LA0
SB15
SA15
5
LIMIT
FILTER B RSL
16
DATA OUT
LOGIC Devices Incorporated
32
LA15
LIMIT
LB15
32
6
16
FILTER A RSL
DATA OUT
Video Imaging Products
Feb 5, 2003 LDS.3321-A
LF3321
Horizontal Digital Image Filter
DEVICES INCORPORATED
Improved Performance
Functional Description
Cascading
Three cascade ports are provided to allow cascading of multiple devices for more filter taps (see Figure
8). COUT11-0 of one device should be connected to DIN11-0 of another device. ROUT11-0 of one
device should be connected to RIN11-0 of another device. As many LF3321s as desired may be cascaded
together. However, the outputs of the LF3321s must be added together with external adders.
Bit 0 of Configuration Register 5 determines how the device will send data to the reverse data path when
multiple LF3321s are cascaded together. If a LF3321 is the last in the cascade chain, Bit 0 of Configuration
Register 5 should be set to a “0”. This will cause the data from the end of the forward data path to be routed
to the beginning of the reverse data path based on how the filter is configured (even/odd number of taps or
interleave mode). If a LF3321 is not the last in the cascade chain, Bit 0 of Configuration Register 5 should
be set to a “1”. This will cause RIN11-0 to feed data to the reverse data path. When not cascading, Bit 0 of
Configuration Register 5 should be set to a “0”.
Special data routing circuitry is used to feed the COUT and ROUT output registers. The data routing
circuitry is required to correctly align data in the forward and reverse data paths as data passes from one
LF3321 to another. The COUT and ROUT registers are loaded with data which is two clock cycles behind
the current output of the I/D Register just before the ROUT or COUT register. This correctly accounts for the
extra delays added to the forward and reverse data paths by the input/output cascade registers.
Output Adder
The Output Adder adds the Filter A and B outputs together when the device is in Single Filter Mode. If 24-bit
data and 12-bit coefficients or 12-bit data and 24-bit coefficients are desired, the LF3321 can facilitate this
by scaling the Filter B output by 2-12 before adding it to the Filter A output. Bit 3 in Configuration Register 5
determines if the Filter B output is scaled before being added to the Filter A output.
Figure 8. Multiple LF3321 Cascaded Together
LF3320
DIN
12
I/D
REGISTERS
I/D
REGISTERS
FILTER
A
FILTER
B
LF3320
RIN
COUT
ROUT
DIN
I/D
REGISTERS
I/D
REGISTERS
FILTER
A
FILTER
B
LF3320
RIN
ROUT
COUT
DIN
I/D
REGISTERS
I/D
REGISTERS
FILTER
A
FILTER
B
LF3320
RIN
COUT
ROUT
DIN
I/D
REGISTERS
I/D
REGISTERS
FILTER
A
FILTER
B
RSL
CIRCUIT
RSL
CIRCUIT
RSL
CIRCUIT
RSL
CIRCUIT
16
16
16
16
LF3347
25
25
RSL
CIRCUIT
16
DATA OUT
128 TAP RESULT
LOGIC Devices Incorporated
7
Video Imaging Products
Feb 5, 2003 LDS.3321-A
LF3321
Horizontal Digital Image Filter
DEVICES INCORPORATED
Improved Performance
Functional Description
Rounding
The overall filter output (Single Filter Mode) or Filter A and B outputs (Dual Filter Mode) may be rounded by
adding the contents of one of the sixteen Filter A or B round registers to the overall filter, Filter A, or Filter
B outputs (see Figure 10). The Filter A round registers are used for the overall filter (Single Filtermode)
or Filter A (Dual Filter Mode). The Filter B round registers are used for Filter B (Dual Filter Mode). Each
round register is 32-bits wide and user-programmable. This allows the filter’s output to be rounded to any
precision required. Since any 32-bit value may be programmed into the round registers, the device can
support complex rounding algorithms as well as standard Half-LSB rounding. RSLA3-0 determines which
of the sixteen Filter A round registers are used in the Filter A rounding circuitry. RSLB3-0 determines
which of the sixteen Filter B round registers are used in the Filter B rounding circuitry. A value of 0 on
RSLA/RSLB3-0 selects Filter A/B round register 0. A value of 1 selects Filter A/B round register 1 and so
on. RSLA/RSLB3-0 may be changed every clock cycle if desired. This allows the rounding algorithm to be
changed every clock cycle. This is useful when filtering interleaved data. If rounding is not desired, a round
register should be loaded with 0 and selected as the register used for rounding. Round register loading is
discussed in the LF InterfaceTM section.
Output Select
The word width of the overall filter, Filter A, and Filter B outputs is 32-bits. However, only 16-bits may be
sent to DOUT15-0 (Single or Dual Filter Modes) and COUT11-0/ROUT3-0 (Dual Filter Mode). The Filter
A/B select circuitry determines which 16-bits are passed (see Table 20). The Filter A/B select registers
control the Filter A/B select circuitry. There are sixteen Filter A and B select registers. The Filter A select
registers are used for the overall filter (Single Filter Mode) or Filter A (Dual Filter Mode). The Filter B select
registers are used for Filter B (Dual Filter Mode). Each select register is 5 bits wide and user-programmable.
RSLA3-0 determines which of the sixteen Filter A select registers are used in the Filter A select circuitry.
RSLB3-0 determines which of the sixteen Filter B select registers are used in the Filter B select circuitry.
A value of 0 on RSLA/RSLB3-0 selects Filter A/B select register 0. A value of 1 selects Filter A/B select
register 1 and so on. RSLA/RSLB3-0 may be changed every clock cycle if desired. This allows the 16-bit
window to be changed every clock cycle. This is useful when filtering interleaved data. Select register
loading is discussed in the LF InterfaceTM section.
Output Limiting
An output limiting function is provided for the overall filter, Filter A, and Filter B outputs. The Filter A limiting
circuitry is used to limit the overall filter output (Single Filter Mode) and the Filter A output (Dual Filter
Mode). The Filter B limiting circuitry is used to limit the Filter B output (Dual Filter Mode). The Filter
A and B limit registers determine the valid range of output values for the Filter A and B limiting circuitry
respectively. There are sixteen 32-bit user-programmable limit registers for both Filters A and B. The Filter
A limit registers are used for the overall filter (Single Filter Mode) or Filter A (Dual Filter Mode). The Filter
B limit registers are used for Filter B (Dual Filter Mode). RSLA3-0 determines which of the sixteen Filter
A limit registers are used in the Filter A limit circuitry. RSLB3-0 determines which of the sixteen Filter B
limit registers are used in the Filter B limit circuitry. A value of 0 on RSLA/RSLB3-0 selects Filter A/B limit
register 0. A value of 1 selects Filter A/B limit register 1 and so on. Each limit register contains an upper
and lower limit value. If the value fed to the limiting circuitry is less than the lower limit, the lower limit value
is passed as the filter output. If the value fed to the limiting circuitry is greater than the upper limit, the upper
limit value is passed as the filter output. Bit 1 and 0 in Configuration Register 4 enable and disable Filter A
and B limiting respectively. RSLA/RSLB3-0 may be changed every clock cycle if desired. This allows the
limit range to be changed every clock cycle. This is useful when filtering interleaved data. When loading
limit values into the device, the upper limit must be greater than the lower limit. Limit register loading is
discussed in the LF InterfaceTM section.
Coefficient
Banks
The coefficient banks store the coefficients which feed into the multipliers in Filters A and B. There is a
separate bank for each multiplier. Each bank can hold 256 12-bit coefficients. The banks are loaded using
an LF InterfaceTM. There is a separate LF InterfaceTM for the Filter A and B banks. Coefficient bank loading
is discussed in the LF InterfaceTM section.
LOGIC Devices Incorporated
8
Video Imaging Products
Feb 5, 2003 LDS.3321-A
LF3321
Horizontal Digital Image Filter
DEVICES INCORPORATED
Improved Performance
Functional Description
Configuration
and Control
Registers
The configuration registers determine how the LF3321 operates. Tables 2 through 7 show the formats of
the six configuration registers. There are three types of control registers: round, select, and limit. There
are sixteen round registers for Filter A and sixteen for Filter B. Each register is 32 bits wide. RSLA3-0 and
RSLB3-0 determine which Filter A and B round registers respectively are used for rounding.
There are sixteen select registers for Filter A and sixteen for Filter B. Each register is 5 bits wide. RSLA3-0
and RSLB3-0 determine which Filter A and B select registers respectively are used in the select circuitry.
There are sixteen limit registers for Filter A and sixteen for Filter B. Each register is 32-bits wide and
stores both an upper and lower limit value. The lower limit is stored in bits 15-0 and the upper limit is
stored in bits 31-16. RSLA3-0 and RSLB3-0 determine which Filter A and B limit registers respectively
are used for limiting when limiting is enabled. Configuration and control register loading is discussed in
the LF InterfaceTM section.
Table 2. Configuration register 0 - Address 200H
BITS
FUNCTION
DESCRIPTION
0
ALU Mode
1
Pass A
2
Pass B
0:A+B
1:B–A
0 : ALU Input A = 0
1 : ALU Input A = Forward Register Path
0 : ALU Input B = 0
1 : ALU Input B = Reverse Register Path
Must be set to “0”
11-3
Reserved
Table 3. Configuration Register 1 - Address 201H
BITS
0
4-1
5
Filter B Tap Number
6
Filter B Data Reversal
11-7
LOGIC Devices Incorporated
FUNCTION
Filter B Odd-Tap
Interleave Mode
Filter B I/D Register Length
Reserved
9
DESCRIPTION
0 : Odd-Tap Interleave Mode Disabled
1 : Odd-Tap Interleave Mode Enabled
0000 : 1 Register
0001 : 2 Registers
0010 : 3 Registers
0011 : 4 Registers
0100 : 5 Registers
0101 : 6 Registers
0110 : 7 Registers
0111 : 8 Registers
1000 : 9 Registers
1001 : 10 Registers
1010 : 11 Registers
1011 : 12 Registers
1100 : 13 Registers
1101 : 14 Registers
1110 : 15 Registers
1111 : 16 Registers
0 : Even Number of Taps
1 : Odd Number of Taps
0 : Data Reversal Enabled
1 : Data Reversal Disabled
Must be set to “0”
Video Imaging Products
Feb 5, 2003 LDS.3321-A
LF3321
Horizontal Digital Image Filter
DEVICES INCORPORATED
Improved Performance
Functional Description
Table 4. Configuration Register 2 - Address 202H
BITS
0
FUNCTION
ALU Mode Filter B
DESCRIPTION
1
Pass A Filter B
2
Pass B Filter B
0 : ALU Input A=0
1 : ALU Input A=Forward Register Path
0 : ALU Input A=0
1 : ALU Input A=Reverse Register Path
Must be set to “0“
11-3
Reserved
0:A+B
1:B–A
Table 5. Configuration Register 3 - Address 203H
BITS
0
4-1
FUNCTION
Filter A Odd-Tap
Interleave Mode
Filter A I/D Register Length
5
Filter A Tap Number
6
Filter A Data Reversal
11-7
Reserved
DESCRIPTION
0 : Odd-Tap Interleave Mode Disabled
1 : Odd-Tap Interleave Mode Enabled
0000 : 1 Register
0001 : 2 Registers
0010 : 3 Registers
0011 : 4 Registers
0100 : 5 Registers
0101 : 6 Registers
0110 : 7 Registers
0111 : 8 Registers
1000 : 9 Registers
1001 : 10 Registers
1010 : 11 Registers
1011 : 12 Registers
1100 : 13 Registers
1101 : 14 Registers
1110 : 15 Registers
1111 : 16 Registers
0 : Even Number of Taps
1 : Odd Number of Taps
0 : Data Reversal Enabled
1 : Data Reversal Disabled
Must be set to “0”
Table 6. Configuration Register 4 - Adress 204H
BITS
0
FUNCTION
Filter B Limit Enable
1
Filter A Limit Enable
11-2
LOGIC Devices Incorporated
Reserved
10
DESCRIPTION
0 : Limiting Disabled
1 : Limiting Enabled
0 : Limiting Disabled
1 : Limiting Enabled
Must be set to “0”
Video Imaging Products
Feb 5, 2003 LDS.3321-A
LF3321
Horizontal Digital Image Filter
DEVICES INCORPORATED
Improved Performance
Functional Description
Table 7. Configuration Register 5 - Adress 205H
BITS
0
FUNCTION
Vertical Limit Enable
1
Horizontal Limit Enable
11-2
Reserved
DESCRIPTION
0 : Vertical Limiting Disabled
1 : Vertical Limiting Enabled
0 : Horizontal Limiting Disabled
1 : Horizontal Limiting Enabled
Must be set to “0”
Table 8. CFA/CFB11-9 Decode
11
0
0
0
0
1
1
1
1
LF Interface ™
10
0
0
1
1
0
0
1
1
9
0
1
0
1
0
1
0
1
DESCRIPTION
Coefficient Banks
Configuration Registers
Filter A Select Registers
Filter B Select Registers
Filter A Round Registers
Filter B Round Registers
Filter A Limit Registers
Filter B Limit Registers
The Filter A and B LF InterfacesTM are used to load data into the Filter A and B coefficient banks
respectively. They are also used to load data into the configuration and control registers.
The following section describes how the Filter A LF InterfaceTM works. The Filter A and B LF InterfacesTM
are identical in function. If LDA and CFA11-0 are replaced with LDB and CFB11-0, the following section will
describe how the Filter B LF InterfaceTM works.
LDA is used to enable and disable the Filter A LF InterfaceTM. When LDA goes LOW, the Filter A LF
InterfaceTM is enabled for data input. The first value fed into the interface on CFA11-0 is an address which
determines what the interface is going to load. The three most significant bits (CFA11-9) determine if the
LF InterfaceTM will load coefficient banks or Configuration/control registers (see Table 8). The nine least
significant bits (CFA8-0) are the address for whatever is to be loaded (see Tables 9 through 14). For
example, to load address 15 of the Filter A coefficient banks, the first data value into the LF InterfaceTM
should be 00FH.
Table 9 Fltr. A Round Regs.
Table 10 Fltr. A Select Regs
Table 11 Fltr. A Limit Regs.
REGISTER ADDRESS (HEX)
0
800
1
801
REGISTER ADDRESS (HEX)
0
400
1
401
REGISTER ADDRESS (HEX)
0
C00
1
C01
14
15
14
15
40E
40F
14
15
C0E
C0F
Table 12 Fltr. B Round Regs
Table 13 Fltr. B Select Regs
Table 14 Fltr. B Limit Regs
REGISTER ADDRESS (HEX)
0
A00
1
A01
REGISTER ADDRESS (HEX)
0
600
1
601
REGISTER ADDRESS (HEX)
0
E00
1
E01
14
15
LOGIC Devices Incorporated
80E
80F
A0E
A0F
14
15
11
60E
60F
14
15
E0E
E0F
Video Imaging Products
Feb 5, 2003 LDS.3321-A
LF3321
Horizontal Digital Image Filter
DEVICES INCORPORATED
Improved Performance
Functional Description
LF Interface ™
Continued
To load Filter A limit register 10, the first data value should be C0AH. The first address value should be
loaded into the interface on the same clock cycle that latches the HIGH to LOW transition of LDA (see
Figures 9 and 10).
The next value(s) loaded into the interface are the data value(s) which will be stored in the bank or register
defined by the address value. When loading coefficient banks, the interface will expect eight values to be
loaded into the device after the address value. The eight values are coefficients 0 through 7. When loading
configuration or select registers, the interface will expect one value after the address value. When loading
round or limit registers, the interface will expect four values after the address value. Figures 9 and 10 show
the data loading sequences for the coefficient banks and Configuration/control registers.
Both PAUSEA and PAUSEB allow the user to effectively slow the rate of data loading through the LF
InterfaceTM. When PAUSEA is HIGH, the LF InterfaceTM affecting the data used for Filter A is held until
PAUSEA is returned to a LOW. When PAUSEB is HIGH, the LF InterfaceTM affecting the data used for Filter
B is held until PAUSEB is returned to a LOW. Figures 11 through 14 display the effects of both PAUSEA
and PAUSEB while loading coefficient and control data.
Table 15 shows an example of loading data into the coefficient banks. The following data values are written
into address 10 of coefficient banks 0 through 7: 210H, 543H, C76H, 9E3H, 701H, 832H, F20H, 143H.
Table 16 shows an example of loading data into a Configuration Register. Data value 003H is written into
Configuration Register 4. Table 17 shows an example of loading data into a round register. Data value
7683F4A2H is written into Filter A round register 12. Table 18 shows an example of loading data into a
select register. Data value 00FH is loaded into Filter A select register 2.
Figure 9. Coefficient Bank Loading Sequence
COEFFICIENT SET 1
COEFFICIENT SET 2
COEFFICIENT SET 3
CLK
W1
W2
W3
LDA/LDB
CFA/CFB11-0
ADDR1
COEF0
COEF7
ADDR2
COEF0
COEF7
ADDR3
COEF0
COEF7
W1: Coefficient Set 1 written to coefficient banks during this clock cycle.
W2: Coefficient Set 2 written to coefficient banks during this clock cycle.
W3: Coefficient Set 3 written to coefficient banks during this clock cycle.
Figure 10. Configurational/Control Register Loading Sequence
CONFIG REG
ROUND REGISTER
SELECT REG
LIMIT REGISTER
CLK
W1
W2
W3
W4
LDA/LDB
CFA/CFB11-0
ADDR1
DATA1 ADDR2
DATA1 ADDR3
DATA1
DATA2
DATA3
DATA4 ADDR4
DATA1
DATA2
DATA3
DATA4
W1: Configuration Register loaded with new data on this rising clock edge.
W2: Select Register loaded with new data on this rising clock edge.
W3: Round Register loaded with new data on this rising clock edge.
W4: Limit Register loaded with new data on this rising clock edge.
LOGIC Devices Incorporated
12
Video Imaging Products
Feb 5, 2003 LDS.3321-A
LF3321
Horizontal Digital Image Filter
DEVICES INCORPORATED
Improved Performance
Functional Description
Figure 11. Coefficient Bank Loading Sequence with PAUSE Implementation
COEFFICIENT SET 1
CLK
W1
PAUSEA/PAUSEB
LDA/LDB
CFA/CFB11-0
ADDR1
COEF0
COEF1
COEF7
W1: Coefficient Set 1 written to coefficient banks during this clock cycle.
Figure 12. Config and Select Register Loading Sequence w/ PAUSE Implementation
SELECT REGISTER
CONFIGURATION REGISTER
CLK
W1
W2
PAUSEA/PAUSEB
LDA/LDB
CFA/CFB11-0
ADDR1
DATA1
ADDR2
DATA1
W1: Configuration Register loaded with new data on this rising clock edge.
W2: Select Register loaded with new data on this rising clock edge.
Figure 13. Round Register Loading Sequence with PAUSE Implementation
ROUND REGISTER
CLK
W1
PAUSEA/PAUSEB
LDA/LDB
CFA/CFB11-0
ADDR1
DATA1
DATA2
DATA3
DATA4
W1: Round Register loaded with new data on this rising clock edge.
LOGIC Devices Incorporated
13
Video Imaging Products
Feb 5, 2003 LDS.3321-A
LF3321
Horizontal Digital Image Filter
DEVICES INCORPORATED
Improved Performance
Functional Description
Figure 14. Limit Register Loading Sequence with PAUSE Implementation
LIMIT REGISTER
CLK
W1
PAUSEA/PAUSEB
LDA/LDB
CFA/CFB11-0
ADDR1
DATA1
DATA2
DATA3
DATA4
W1: Limit Register loaded with new data on this rising clock edge.
LF Interface TM
Continued
Table 19 shows an example of loading data into Filter B limit register 7. Data value 3B60H is loaded as
the lower limit and 72A4H is loaded as the upper limit. It takes 9S clock cycles to load S coefficient sets
into the device. Therefore, it takes 2304 clock cycles to load all 256 coefficient sets. Assuming an 83 MHz
clock rate, all 256 coefficient sets can be updated in less than 27.7 µs, which is well within vertical blanking
time. It takes 5S clock cycles to load S round or limit registers. Therefore, it takes 320 clock cycles to
update all round and limit registers (both Filters A and B). Assuming an 83 MHz clock rate, all Filter A and B
round/limit registers can be updated in 3.84 µs.
The coefficient banks and Configuration/Control registers are not loaded with data until all data values for
the specified address are loaded into the LF InterfaceTM. In other words, the coefficient banks are not written
to until all eight coefficients have been loaded into the LF InterfaceTM. A round register is not written to
until all four data values are loaded.
After the last data value is loaded, the interface will expect a new address value on the next clock cycle.
After the next address value is loaded, data loading will begin again as previously discussed. As long
as data is loaded into the interface, LDA must remain LOW. After all desired coefficient banks and
Configuration/Control registers are loaded with data, the LF InterfaceTM must be disabled. This is done by
setting LDA HIGH on the clock cycle after the clock cycle which latches the last data value. It is important
that the LF InterfaceTM remain disabled when not loading data into it.
The Filter A coefficient banks may only be loaded with the Filter A LF InterfaceTM and the Filter B coefficient
banks may only be loaded with the Filter B LF InterfaceTM. The Configuration and Control registers may be
loaded with either the Filter A or B LF InterfacesTM.
Since both LF InterfacesTM operate independently of each other, both LF InterfacesTM can load data into their
respective coefficient banks at the same time. Or, one LF InterfaceTM can load the Configuration/Control
registers while the other loads it’s respective coefficient banks. If both LF InterfacesTM are used to load a
Configuration or Control register at the same time, the Filter B LF InterfaceTM will be given priority over the
Filter A LF InterfaceTM. For example, if the Filter A LF InterfaceTM attempts to load data into a configuration
register at the same time that the Filter B LF InterfaceTM attempts to load a Filter A round register, the
Filter B LF InterfaceTM will be allowed to load the round register while the Filter A LF InterfaceTM will not
be allowed to load the configuration register. However, the Filter A LF InterfaceTM will continue to function
as if the write occurred.
LOGIC Devices Incorporated
14
Video Imaging Products
Feb 5, 2003 LDS.3321-A
LF3321
Horizontal Digital Image Filter
DEVICES INCORPORATED
Improved Performance
Functional Description
Table 15. Coefficient Bank Loading Format
1st Word - Address
2nd Word - Bank 0
3rd Word - Bank 1
4th Word - Bank 2
5th Word - Bank 3
6th Word - Bank 4
7th Word - Bank 5
8th Word - Bank 6
9th Word - Bank 7
CFA/B11 CFA/B10 CFA/B 9 CFA/B 8 CFA/B 7 CFA/B 6 CFA/B 5 CFA/B 4 CFA/B 3 CFA/B 2 CFA/B 1 CFA/B 0
0
0
0
0
0
0
0
0
1
0
1
0
0
0
1
0
0
0
0
1
0
0
0
0
0
1
0
1
0
1
0
0
0
0
1
1
1
1
0
0
0
1
1
1
0
1
1
0
1
0
0
1
1
1
1
0
0
0
1
1
0
1
1
1
0
0
0
0
0
0
0
1
1
0
0
0
0
0
1
1
0
0
1
0
1
1
1
1
0
0
1
0
0
0
0
0
0
0
0
1
0
1
0
0
0
0
1
1
Table 16. Configuration Register Loading Format
1st Word - Address
2nd Word - Data
CFA/B11 CFA/B10 CFA/B 9 CFA/B 8 CFA/B 7 CFA/B 6 CFA/B 5 CFA/B 4 CFA/B 3 CFA/B 2 CFA/B 1 CFA/B 0
0
0
1
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
1
1
Table 17. Round Register Loading Format
1st Word - Address
2nd Word - Data
3rd Word - Data
4th Word - Data
5th Word - Data
CFA/B11 CFA/B10 CFA/B 9 CFA/B 8 CFA/B 7 CFA/B 6 CFA/B 5 CFA/B 4 CFA/B 3 CFA/B 2 CFA/B 1 CFA/B 0
1
0
0
0
0
0
0
0
1
1
0
0
R
R
R
R
0
1
0
1
0
0
1
0*
R
R
R
R
1
1
1
1
0
1
0
0
R
R
R
R
1
0
0
0
0
0
1
1
R
R
R
R
0 **
1
1
1
0
1
1
0
Table 18. Select Register Loading Format
1st Word - Address
2nd Word - Data
CFA/B11 CFA/B10 CFA/B 9 CFA/B 8 CFA/B 7 CFA/B 6 CFA/B 5 CFA/B 4 CFA/B 3 CFA/B 2 CFA/B 1 CFA/B 0
0
1
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
1
1
1
1
Table 19. Limit Register Loading Format
1st Word - Address
2nd Word - Data
3rd Word - Data
4th Word - Data
5th Word - Data
CFA/B11 CFA/B10 CFA/B 9 CFA/B 8 CFA/B 7 CFA/B 6 CFA/B 5 CFA/B 4 CFA/B 3 CFA/B 2 CFA/B 1 CFA/B 0
1
1
1
0
0
0
0
0
0
1
1
1
R
R
R
R
0
1
1
0
0
0
0
0
R
R
R
R
0*
0
1
1
1
0
1
1
R
R
R
R
1
0
1
0
0
1
0
0
R
R
R
R
0 **
1
1
1
0
0
1
0
R = Reserved. Must be set to “0“.
* This bit represents the MSB of the Lower Limit.
** This bit represents the MSB of the Upper Limit.
LOGIC Devices Incorporated
15
Video Imaging Products
Feb 5, 2003 LDS.3321-A
LF3321
Horizontal Digital Image Filter
DEVICES INCORPORATED
Improved Performance
Operating Modes
Single Filter
Mode
In this mode, the device operates as a single FIR filter (see Figure 15). It can be configured to have
as many as 32 taps if symmetric coefficient sets are used. If asymmetric coefficient sets are used, the
device can be configured to have as many as 16 taps. Cascade ports are provided to facilitate cascading
multiple devices to increase the number of filter taps. Bit 1 in Configuration Register 5 determines the
filter mode. In Single Filter Mode, DIN11-0 is the data input for the filter and DOUT15-0 is the data
output for the filter.
Figure 15. Single Filter Mode
ROUT11-0
DIN11-0
12
12
12
I/D
REGISTERS
I/D
REGISTERS
FILTER
A
FILTER
B
12
RIN11-0
COUT11-0
RSL
CIRCUIT
16
DOUT15-0
Dual Filter
Mode
In this mode, the device operates as two separate FIR filters (see Figure 16). Each filter can be configured
to have as many as 16 taps if symmetric coefficient sets are used. If asymmetric coefficient sets are used,
each filter can be configured to have as many as 8 taps. In Dual Filter Mode, DIN11-0 is the data input for
Filter A. Either RIN11-0 or DIN11-0 can be the data input for Filter B. The Filter B input is determined by Bit
2 in Configuration Register 5. DOUT15-0 is the data output for Filter A. COUT11-0 and ROUT3-0 together
form the data output for Filter B. COUT11-0 is the twelve least significant bits and ROUT3-0 is the four most
significant bits of the 16-bit Filter B output.
Figure 16. Dual Filter Mode
12
DIN11-0
12
I/D
REGISTERS
I/D
REGISTERS
FILTER
A
FILTER
B
R.S.L.
CIRCUIT
16
DOUT15-0
Matrix-vector
Multiply Mode
RIN11-0
R.S.L.
CIRCUIT
16
ROUT3-0 / COUT11-0
In this mode, the LF3321 can be configured to multiply a square matrix of maximum size N (N = 8 or
16), multiplied by a matrix-vector of maximum size [8,1] or [16,1]. The mathematical representation for
this operation is in Figure 18. When configured in the dual filter mode, the LF3321 can process two matrixvector multipliers simultaneously (i.e. [8x8][8x1]). In the single filter mode, the LF3321 can process a single
matrix-vector multiply (i.e. [16x16][16x1]). This mode of operation allows the user to organize data values
(e.g. pixels) into an array (e.g. blocks). This function is useful for any application requiring the operation of
matrix multiplication; a function that is used when generating Discrete Cosine Transform coefficients (DCT)
for the purpose of further processing.
When configuring the LF3321 for an [8x8][8x1] matrix-vector operation, the coefficient banks will require
8 coefficient sets to be loaded into the coefficient memory banks; each coefficient set will have 8, 12-bit
coefficients. The input data, [8x1] column-vector, will be loaded through DIN11-0 for Filter A; either RIN11-0
or DIN11-0 can be the data input for Filter B. Conversely, when configured for a [16x16][16x1] matrix-vector
operation, the coefficient banks will require 16 coefficient sets to be loaded into the coefficient memory
banks; each coefficient set will have 16, 12-bit coefficients. The input data, [16x1] column-vector, will be
loaded through DIN11-0.
LOGIC Devices Incorporated
16
Video Imaging Products
Feb 5, 2003 LDS.3321-A
LF3321
Horizontal Digital Image Filter
DEVICES INCORPORATED
Improved Performance
Operating Modes
Matrix-vector
Multiply Mode
Continued
To configure the LF3321 for matrix-vector multiplication, bit 4 of Configuration Register 5 must be set to
1 (Table 7). The configuration for single filter mode or dual filter mode will still apply. Writing 012H or
016H to Configuration Register 5 will configure the device for dual filter mode, [8x8][8x1] matrix-vector
multiplication. Subsequently, writing 014H to Configuration Register 5 will configure the device for single
filter mode, [16x16][16x1] matrix-vector multiplication.
Some functions of the LF3321 must be disabled when configured for matrix-vector multiplication. This
will apply to both the single filter mode and the dual filter mode; these functions are data reversal and
interleave/decimation. The LF3321 can be cascaded to realize larger matrices.
Figure 17. Matrix-vector Multiply Mode
N
DIN11-0
RIN11-0
12
0
A
ALU
B
0
A
ALU
0
0
B
A
ALU
B
A
ALU
B
TXFRA
TXFRB
COEF (N-1)
N
COEF 2
COEF 1
COEF 0
12
12
12
12
Dual Filter Mode, N=8
Single FIlter Mode, N=16
32
Data reversal can be disabled by setting bit 6, of Configuration Register 1 (Filter A) and Configuration
Register 3 (Filter B), both to 1. The Odd-Tap, interleave mode will need to be disabled. Writing a 0 to bit 0
of Configuration Register 1 and Configuration Register 3 will disable the odd-tap interleave mode for Filter
A and Filter B. When data is not being interleaved or decimated, the I/D Register length should be set to a
length of one (Table 3 and Table 5). Therefore, writing 040H to Configuration Register 1 and 3 will disable
the data reversal and set the corresponding inherent characteristics for the desired matrix function.
The Filter A ALU and Filter B ALU are to be configured for A+B (Table 2 and Table 4); so that condition
A+0 is satisfied. To accomplish this, bit 0 is to be reset to 0, bit 1 is to be set to 1, and bit 2 is to be reset
to 0. Writing 002H to Configuration Register 0 (Filter A) and Configuration Register 2 (Filter B) will set the
corresponding registers to satisfy the A+0 condition.
The timing diagrams in Figure 19 and 20 will assume that the Configuration Registers, the coefficient sets,
and the first set of data values (data set) have been loaded. Loading input data for an [8x8][8x1] matrix
operation requires 9 clock cycles and loading input data for a [16x16][16x1] matrix operation requires 17
clock cycles. When configured for an [8x8][8x1] matrix-vector operation, 8 data values are required for
loading. When configured for a [16x16][16x1] matrix-vector operation, 16 data values are required for
loading. Each data value is fed through the I/D Registers, using the corresponding input. Once the final
data value, of the data set, has been loaded TXFRA/TXFRB should be brought LOW for one clock cycle to
complete the loading. Once this occurs, the data set is then bank loaded into the respective registers ready
to begin the matrix-vector multiplication operation. The current data set will not change until TXFRA/TXFRB
is brought LOW again. To satisfy the matrix equation (see Figure 18), the current data set is held for the
duration of the required matrix dimension while cycling through each coefficient set (CENA/CENB must
be held LOW). During this time new data values can be loaded serially, ready for the next activation of
TXFRA/TXFRB. To insure the correct evaluation of the matrix-vector multiplication equation, it is imperative
that the coefficient values are paired with their corresponding data values.
For the [8x8][8x1] matrix-vector configuration (dual filter mode), the first result will appear 19 clock cycles
from the first data input, DIN15-0 (Filter A) and RIN15-0 (Filter B); device latency for the first result is 10
clock cycles (10+9 = 19).
LOGIC Devices Incorporated
17
Video Imaging Products
Feb 5, 2003 LDS.3321-A
LF3321
Horizontal Digital Image Filter
DEVICES INCORPORATED
Improved Performance
Operating Modes
Matrix-vector
Multiply Mode
Continued
The result will appear at the corresponding filter output, DOUT15-0 (Filter A) and ROUT3-0/COUT11-0
(Filter B). For the [16x16][16x1] matrix-vector configuration (single filter mode), the first result will appear 28
clock cycles from the first data input, DIN15-0; device latency for the first result is 11 clock cycles (11+17 =
28). The result will appear at the corresponding filter output, DOUT15-0. Subsequently, for both dual and
single filter mode configurations, the sum of products will continue to appear every clock cycle thereafter
until the matrix dimension has been realized. The total pipeline latency for a complete [8x8][8x1] matrixvector operation is 26 clock cycles and the total pipeline latency for a complete [16x16][16x1] matrix-vector
operation is 43 clock cycles. Therefore, to process two square matrices simultaneoulsy, of size N=8, a total
of 73 clock cycles are all that is required. Similarly, to process a single square matrix, of size N=16, a
total of 283 clock cycles are required.
Figure 18. Matrix Equation
C = COEFFICIENTS
D = DATA INPUT
R = DATA OUTPUT
R0
R1
R2
Ri
=
C00 C01 C02
C10 C11 C12
C20 C21 C22
C0j
C1j
C2j
D0
D1
D2
Ci0
Cij
Di
Ci1
Ci2
(N-1)
=
Ri
Cij
Di
i=0
For j=0,1,2,...,(N-1)
N=8 or 16
Figure 19. Dual Filter, Matrix Multiply Timing Sequence
Data Set 1 with 8 Coefficient Sets
1
2
3
8*
Data Set 2 with 8 Coefficient Sets
9
10**
11
17***
18
CLK
DIN11-0
RIN11-0
DATA SET 1
DATA SET 0
CAA7-0
CAB7-0
CF00
CF01
CF02
CF07
CF10
CF11
CF12
CF17
CF20
CF21
TXFRA/ TXFRB
DOUT15-0
ROUT3-0/COUT15-0
OUT0
OUT1
OUT7
OUT0
OUT1
CENA / CENB
SHENA / SHENB
*
**
***
8 Clocks - End of First Data/Coefficient Set
10 Clocks - First Output of First Data/Coefficient Set
17 Clocks - Final Output of First Data/Coefficient Set
Once again, the timing diagrams (see Figure 19 and 20) will assume that the Configuration Registers,
the coefficient sets, and the data values have been loaded. The corresponding timing diagram loading
sequence for the coefficient banks and Configuration/Control registers are included in the LF3321 data
sheets (Figure 9 and Figure 10 respectively). Further reference to timing diagram loading sequence for the
RSL registers are also included in the device data sheet (Figure 14, Figure 13, and Figure 12). The Filter
A and Filter B LF InterfaceTM are used to load data into the Filter A and Filter B Configuration Registers
and coefficient banks.
The Matrix Multiplication Mode is valid in the Double Wide Data/Coefficient Mode. However, there are some
special considerations when this mode is desired. The LF3321 must be configured for single filter mode
only, for a maximum (8x8) matrix. The user must disable the cascaded filter mode, the accumulator access
mode, and the data reversal (see Table 7).
LOGIC Devices Incorporated
18
Video Imaging Products
Feb 5, 2003 LDS.3321-A
LF3321
Horizontal Digital Image Filter
DEVICES INCORPORATED
Improved Performance
Operating Modes
Figure 20. Single Filter, Matrix Multiply Timing Sequence
1 Data Set with 16 Coefficient Sets
1
2
11*
3
12
13
14
15
CF0D
CF0E
17
16**
26***
CLK
DIN11-0
RIN11-0
DATA SET 0
DATA SET 0
CAA7-0
CAB7-0
CF00
CF01
CF02
CF0A
CF0B
CF0C
CF10
CF0F
TXFRA/ TXFRB
DOUT15-0
OUT0
OUT1
OUT2
OUT3
OUT4
OUT5
OUT6
OUT15
CENA / CENB
SHENA / SHENB
*
**
***
Double Wide
Data/Coefficient
Mode
11 Clocks - First Output of First Data/Coefficient Set
16 Clocks - End of First Data/Coefficient Set
26 Clocks - Final Output of First Data/Coefficient Set
The LF3321 is capable of supporting 24-bit data and 12-bit coefficients or 12-bit data and 24-bit coefficients.
When configured for this mode of operation, the Filter B output is scaled by 2-12 before adding it to the Filter
A output. This mode of operation is only valid in single filter mode.
To configure the LF3321 for this mode, bit 3 of Configuration Register 5 must be set to 1; this will account
for the scaling function (Table 7). For 24-bit data, DIN11-0 becomes the MSB (Filter A) and RIN11-0
becomes the LSB (Filter B), bit 2 of Configuration Register 5 must be set to 0. To insure correct results,
the coefficient sets must be aligned appropriately; that is to say, the coefficient set used for the MSB must
be the same for the LSB. For 24-bit coefficients, the coefficient banks for Filter A will correspond to the
coefficient MSB and the coefficient banks for Filter B will correspond to the coefficient LSB; therefore, bit
2 must be set to 1.
Once again, to insure correct results, the coefficient sets must be aligned appropriately; that is to say, the
MSB coefficients must correspond to their LSB coefficients. The output data will appear at DOUT15-0;
output appearing at ROUT3-0/COUT11-0 will not be of any value. Bit 1 is set to 0 (for single filter mode)
and bit 0 is set to 0 (cascade mode must be disabled). Therefore, to realize 24-bit data/12-bit coefficients
the user must write 008H to Configuration Register 5; conversely, for 12-bit data/24-bit coefficients the user
must write 00CH to Configuration Register 5.
The Double Wide Data/Coefficient Mode is valid in the Matrix Multiplication Mode; however, special
considerations must be observed when these two modes are combined. The LF3321 must be configured
for single filter mode only, for a maximum (8x8) matrix. In addition, the user must disable the cascaded filter
mode, the accumulator access mode, and the data reversal (Table 7). For additional considerations, refer to
the corresponding mode of operation section.
LOGIC Devices Incorporated
19
Video Imaging Products
Feb 5, 2003 LDS.3321-A
LF3321
Horizontal Digital Image Filter
DEVICES INCORPORATED
Improved Performance
Operating Modes
Figure 21. Double Wide Data Coefficient Mode
DIN11-0
RIN11-0
12
I/D
REGISTERS
I/D
REGISTERS
FILTER
A
FILTER
B
12
SCALE
R.S.L.
CIRCUIT
16
DOUT15-0
Accumulator
Access Mode
Accumulator access allows the user to accumulate the Filter A output with the Filter B output. Therefore,
this mode is only valid when the device has been configured for dual filter operation. To configure the device
for this mode, bit 1 and bit 5 must be set to 1; bit 5 is the corresponding accumulator access bit (Table
7). Writing 022H to Configuration Register 5 configures the device to accumulate the Filter A output with
the Filter B output. All remaining Configuration Registers, 0 through 4 inclusive, will depend on specific
application requirements (see Tables 2 through 4). In this mode of operation, the accumulated output is
realized at DOUT15-0. The output data at ROUT15-0/COUT15-0 is the Filter B output that is normally
expected; however, the accumulated output data (DOUT15-0) will be delayed by one clock cycle, compared
to the Filter B output data.
This type of operation is useful when two filtered data streams (i.e. I+jQ) need to be accumulated. Such
is the requirement to satisfy the equation, y(t) = cos(vt)+jsin(vt). The complex data can be streamed and
filtered using a respective ‘I’ filter and ‘Q’ filter. To convert the complex result into a real result, as seen at the
LF3321 output, two multiplies and one accumulation is required. The cosine and sine functions are realized
through the coefficient sets; consequently, multiplied by the corresponding ‘I’ and ‘Q’ data streams. To satisfy
the remainder of the equation, Filter A and Filter B must be accumulated.
As previously stated, this mode of operation is only valid with the dual filter mode configuration. All modes of
operation, that are valid in the dual filter mode, are valid with the accumulator access mode. For additional
considerations, refer to the corresponding mode of operation section.
Figure 22. Accumulator Access Mode
DIN11-0
12
I/D
REGISTERS
I/D
REGISTERS
FILTER
A
FILTER
B
R.S.L.
CIRCUIT
R.S.L.
CIRCUIT
16
DOUT15-0
LOGIC Devices Incorporated
20
12
RIN11-0
16
ROUT3-0 / COUT11-0
Video Imaging Products
Feb 5, 2003 LDS.3321-A
LF3321
Horizontal Digital Image Filter
DEVICES INCORPORATED
Improved Performance
Signal Definitions
Power
Vcc and GND
+3.3 V power supply. All pins must be connected.
Clock
CLK — Master Clock
The rising edge of CLK strobes all enabled registers.
Inputs
DIN11-0 — Data Input
DIN11-0 is the 12-bit data input port to Filter A. In Dual Filter Mode, DIN11-0 can also be the 12-bit input port
to Filter B. Data is latched on the rising edge of clock.
RIN11-0 — Reverse Cascade Input
In Single Filter Mode, RIN11-0 is the 12-bit reverse cascade input port. This port is connected to ROUT11-0
of another LF3321. In Dual Filter Mode, RIN11-0 can be the 12-bit input port to Filter B. Data is latched
on the rising edge of CLK.
CFA11-0 — Coefficient A Input
CFA11-0 is used to load data into the Filter A coefficient banks (banks 0 through 7) and the configuration/
control registers. Data present on CFA11-0 is latched into the Filter A LF InterfaceTM on the rising edge of
CLK when LDA is LOW (see the LF InterfaceTM section for a full discussion).
CAA7-0 — Coefficient Address A
CAA7-0 determines which row of data in coefficient banks 0 through 7 is fed to the multipliers. CAA7-0 is
latched into Coefficient Address Register A on the rising edge of CLK when CENA is LOW.
CFB11-0 — Coefficient B Input
CFB11-0 is used to load data into the Filter B coefficient banks (banks 8 through 15) and the configuration/
control registers. Data present on CFB11-0 is latched into the Filter B LF InterfaceTM on the rising edge of
CLK when LDB is LOW (see the LF InterfaceTM section for a full discussion).
CAB7-0 — Coefficient Address B
CAB7-0 determines which row of data in coefficient banks 8 through 15 is fed to the multipliers. CAB7-0 is
latched into Coefficient Address Register B on the rising edge of CLK when CENB is LOW.
Figure 23. Input Formats
Input Data
11 10 9
–211 210 29
(Sign)
Coefficient Data
2 1 0
22 21 20
11 10 9
–20 2–1 2–2
(Sign)
2 1 0
2–9 2–10 2–11
Figure 24. Accumulator Output Formats
Accumulator A Output
31 30 29
–220 219 218
(Sign)
LOGIC Devices Incorporated
2 1 0
2–9 2–10 2–11
21
Accumulator B Output
31 30 29
–220 219 218
(Sign)
2 1 0
2–9 2–10 2–11
Video Imaging Products
Feb 5, 2003 LDS.3321-A
LF3321
Horizontal Digital Image Filter
DEVICES INCORPORATED
Improved Performance
Signal Definitions
Table 20. Output Formats
SLCT4-0
00000
00001
00010
·
·
·
01110
01111
10000
Outputs
S15
F15
F16
F17
·
·
·
F29
F30
F31
S14
F14
F15
F16
·
·
·
F28
F29
F30
S13
F13
F14
F15
·
·
·
F27
F28
F29
···
···
···
···
···
···
···
S8
F8
F9
F10
·
·
·
F22
F23
F24
S7
F7
F8
F9
·
·
·
F21
F22
F23
···
···
···
···
···
···
···
S2
F2
F3
F4
·
·
·
F16
F17
F18
S1
F1
F2
F3
·
·
·
F15
F16
F17
S0
F0
F1
F2
·
·
·
F14
F15
F16
DOUT15-0 — Data Output
DOUT15-0 is the 16-bit registered data output port for the overall filter (Single Filter Mode) or Filter A
(Dual Filter Mode).
COUT11-0 — Cascade Output
In Single Filter Mode, COUT11-0 is a 12-bit registered cascade output port. COUT11-0 should be
connected to DIN11-0 of another LF3321. In Dual Filter Mode, COUT11-0 is a 12-bit registered output port
for the lower twelve bits of the 16-bit Filter B output.
ROUT11-0 — Reverse Cascade Output
In Single Filter Mode, ROUT11-0 is a 12-bit registered cascade output port. ROUT11-0 on one device
should be connected to RIN11-0 of another LF3321. In Dual Filter Mode, ROUT3-0 is a 4-bit registered
output port for the upper four bits of the 16-bit Filter B output. In this mode, ROUT11-4 is disabled.
Controls
LDA — Coefficient A Load
When LDA is LOW, data on CFA11-0 is latched into the Filter A LF InterfaceTM on the rising edge of CLK.
When LDA is HIGH, data is not loaded into the Filter A LF InterfaceTM. When enabling the LF InterfaceTM for
data input, a HIGH to LOW transition of LDA is required in order for the input circuitry to function properly.
Therefore, LDA must be set HIGH immediately after power up to ensure proper operation of the input
circuitry (see the LF InterfaceTM section for a full discussion).
CENA — Coefficient Address Enable A
When CENA is LOW, data on CAA7-0 is latched into Coefficient Address Register A on the rising edge
of CLK. When CENA is HIGH, data on CAA7-0 is not latched and the register’s contents will not be
changed.
LDB — Coefficient B Load
When LDB is LOW, data on CFB11-0 is latched into the Filter B LF InterfaceTM on the rising edge of CLK.
When LDB is HIGH, data is not loaded into the Filter B LF InterfaceTM. When enabling the LF InterfaceTM for
data input, a HIGH to LOW transition of LDB is required in order for the input circuitry to function properly.
Therefore, LDB must be set HIGH immediately after power up to ensure proper operation of the input
circuitry (see the LF InterfaceTM section for a full discussion).
LOGIC Devices Incorporated
22
Video Imaging Products
Feb 5, 2003 LDS.3321-A
LF3321
Horizontal Digital Image Filter
DEVICES INCORPORATED
Improved Performance
Signal Definitions
Controls Cont’d
CENB — Coefficient Address Enable B
When CENB is LOW, data on CAB7-0 is latched into Coefficient Address Register B on the rising edge
of CLK. When CENB is HIGH, data on CAB7-0 is not latched and the register’s contents will not be
changed.
TXFRA — Filter A LIFO Transfer Control
TXFRA is used to change which LIFO in the data reversal circuitry sends data to the reverse data path
and which LIFO receives data from the forward data path in Filter A. When TXFRA goes LOW, the LIFO
sending data to the reverse data path becomes the LIFO receiving data from the forward data path, and
the LIFO receiving data from the forward data path becomes the LIFO sending data to the reverse data
path. The device must see a HIGH to LOW transition of TXFRA in order to switch LIFOs. TXFRA is
latched on the rising edge of CLK.
TXFRB — Filter B LIFO Transfer Control
TXFRB is used to change which LIFO in the data reversal circuitry sends data to the reverse data path
and which LIFO receives data from the forward data path in Filter B. When TXFRB goes LOW, the LIFO
sending data to the reverse data path becomes the LIFO receiving data from the forward data path, and
the LIFO receiving data from the forward data path becomes the LIFO sending data to the reverse data
path. The device must see a HIGH to LOW transition of TXFRB in order to switch LIFOs. TXFRB is
latched on the rising edge of CLK.
ACCA — Accumulator A Control
When ACCA is HIGH, Accumulator A is enabled for accumulation and the Accumulator A Output Register
is disabled for loading. When ACCA is LOW, no accumulation is performed and the Accumulator A Output
Register is enabled for loading. ACCA is latched on the rising edge of CLK.
ACCB — Accumulator B Control
When ACCB is HIGH, Accumulator B is enabled for accumulation and the Accumulator B Output Register
is disabled for loading. When ACCB is LOW, no accumulation is performed and the Accumulator B Output
Register is enabled for loading. ACCB is latched on the rising edge of CLK.
SHENA — Filter A Shift Enable
In Dual Filter Mode, SHENA enables or disables the loading of data into the Input (DIN11-0) and Filter A
I/D Registers. When SHENA is LOW, data is latched into the Input/Cascade Registers and shifted through
the I/D Registers on the rising edge of CLK. When SHENA is HIGH, data can not be loaded into the
Input/Cascade Registers or shifted through the I/D Registers and their contents will not be changed.
In Single Filter Mode, SHENA also enables or disables the loading of data into the Reverse Cascade
Input (RIN11-0), Cascade Output (COUT11-0), Reverse Cascade Output (ROUT11-0) and Filter B I/D
Registers. It is important to note that in Single Filter Mode, both SHENA and SHENB should be connected
together. Both must be active to enable data loading in Single Filter Mode. SHENA is latched on the
rising edge of CLK.
LOGIC Devices Incorporated
23
Video Imaging Products
Feb 5, 2003 LDS.3321-A
LF3321
Horizontal Digital Image Filter
DEVICES INCORPORATED
Improved Performance
Signal Definition
Controls Cont’d
SHENB — Filter B Shift Enable
In Dual Filter Mode, SHENB enables or disables the loading of data into the Reverse Cascade Input
(RIN11-0), Cascade Output (COUT11-0), Reverse Cascade Output (ROUT3-0) and Filter B I/D Registers.
When SHENB is LOW, data is latched into the Cascade Registers and shifted through the I/D Registers
on the rising edge of CLK. When SHENB is HIGH, data can not be loaded into the Cascade Registers or
shifted through the I/D Registers and their contents will not be changed.
In Single Filter Mode, SHENB also enables or disables the loading of data into the Input (DIN11-0), Reverse
Cascade Output (ROUT11-0) and Filter A I/D Registers. It is important to note that in Single Filter Mode,
both SHENA and SHENB should be connected together. Both must be active to enable data loading in
Single Filter Mode. SHENB is latched on the rising edge of CLK.
RSLA3-0 — Filter A Round/Select/Limit Control
RSLA3-0 determines which of the sixteen user-programmable Round/Select/Limit registers (RSL registers)
are used in the Filter A RSL circuitry. A value of 0 on RSLA3-0 selects RSL register 0. A value of 1 selects
RSL register 1 and so on. RSLA3-0 is latched on the rising edge of CLK (see the round, select, and limit
sections for a complete discussion).
RSLB3-0 — Filter B Round/Select/Limit Control
RSLB3-0 determines which of the sixteen user-programmable RSL registers are used in the Filter B RSL
circuitry. A value of 0 on RSLB3-0 selects RSL register 0. A value of 1 selects RSL register 1 and
so on. RSLB3-0 is latched on the rising edge of CLK (see the round, select, and limit sections for a
complete discussion).
OED — DOUT Output Enable
When OED is LOW, DOUT15-0 is enabled for output. When OED is HIGH, DOUT15-0 is placed in a
high-impedance state.
OEC — COUT/ROUT Output Enable
When OEC is LOW, COUT11-0 and ROUT3-0 are enabled for output. When OEC is HIGH, COUT11-0 and
ROUT3-0 are placed in a high-impedance state.
PAUSEA — LF InterfaceTM Pause
When PAUSEA is HIGH, the Filter A LF InterfaceTM loading sequence is halted until PAUSEA is returned to
a LOW state. This effectively allows the user to load coefficients and control registers at a slower rate than
the master clock (see the LF InterfaceTM section for a full discussion).
PAUSEB — LF InterfaceTM Pause
When PAUSEB is HIGH, the Filter B LF InterfaceTM loading sequence is halted until PAUSEB is returned to
a LOW state. This effectively allows the user to load coefficients and control registers at a slower rate than
the master clock (see the LF InterfaceTM section for a full discussion).
LOGIC Devices Incorporated
24
Video Imaging Products
Feb 5, 2003 LDS.3321-A
LF3321
Horizontal Digital Image Filter
DEVICES INCORPORATED
Improved Performance
Specifications
Maximum Ratings - Above which useful life may be impaired (Notes 1,2,3,8)
Storage temperature .......................................................................................... –65°C to +150°C
Operating ambient temperature ......................................................................... –55°C to +125°C
VCCO supply voltage with respect to ground ........................................................ –0.5 V to +4.5 V
Input signal with respect to ground ........................................................................ –0.5 V to 5.5 V
Signal applied to high impedance output ............................................................... –0.5 V to 5.5 V
Output current into low outputs ........................................................................................... 25 mA
Latchup current .............................................................................................................. > 400 mA
ESD Classification (MIL-STD-883E METHOD 3015.7) ..................................................... Class 3
Operating Conditions - To meet specified electrical and switching characteristics
Mode
Active Operation, Commercial
Active Operation, Military
Temperature Range (Ambient)
0°C to +70°C
-55°C to +125°C
Supply Voltage
3.00 V < VCCO < 3.60 V
3.00 V < VCCO < 3.60 V
Electrical Characteristics - Over Operating Conditions (Note 4)
Symbol
Parameter
VOH
VOL
VIH
VIL
I IX
I OZ
I CC1
I CC2
C IN
C OUT
Output High Voltage
Output Low Voltage
Input High Voltage
Input Low Voltage
Input Current
Output Leakage Current
VCC Current, Dynamic
VCC Current, Quiescent
Input Capacitance
Output Capacitance
Test Condition
Switching Waveforms:
1
Min Typ
VCC = Min., IOH = -4 mA
VCC = Min., IOL = 8.0 mA
Max Unit
2.4
0.4
5.5
0.8
+10
+10
250
2
10
10
2.0
0.0
(Note 3)
Ground < VIN < VCC (Note 12)
Ground < VOUT < VCC (Note 12)
(Notes 5, 6)
(Note 7)
TA = 25°C, f = 1 MHz
TA = 25°C, f = 1 MHz
V
V
V
V
µA
µA
mA
mA
pF
pF
Data I/O
2
3
4
5
6
7
CLK
DIN11-0
RIN11-0
ACA8-0
BCA8-0
CONTROLS
(Except OED & OEC)
OED
OEC
tS
tH
DIN/RINN
tSCC
tHCC
CAA/CABN
tSCT
tPWH
DIN/RINN+1
CAA/CABN+1
tHCT
tDIS
tENA
HIGH IMPEDANCE
DOUT15-0
tDIS
25
tD
OUTPUTN-1
tENA
HIGH IMPEDANCE
ROUT11-0/COUT11-0
LOGIC Devices Incorporated
tPWL
tCYC
OUTPUTN
tDCC
OUTPUTN-1
OUTPUTN
Video Imaging Products
Feb 5, 2003 LDS.3321-A
LF3321
Horizontal Digital Image Filter
DEVICES INCORPORATED
Improved Performance
Notes
1. Maximum Ratings indicate stress specifications only. Functional operation of these products at values
beyond those indicated in the Operating Conditions table is not implied. Exposure to maximum rating
conditions for extended periods may affect reliability.
2. The products described by this specification include internal circuitry designed to protect the chip from
damaging substrate injection currents and accumulations of static charge. Nevertheless, conventional
precautions should be observed during storage, handling, and use of these circuits in order to avoid
exposure to excessive electrical stress values.
3. This device provides hard clamping of transient undershoot. Input levels below ground will be clamped
beginning at –0.6 V. The device can withstand indefinite operation with inputs or outputs in the range of
–0.5V to +5.5V. Device operation will not be adversely affected, however, input current levels will be well
in excess of 100 mA.
4. Actual test conditions may vary from those designated but operation is guaranteed as specified.
5. Supply current for a given application can be accurately approximated by:
where
NCV2 F
4
N
C
V
F
= total number of device outputs
= capacitive load per output
= supply voltage
= clock frequency
6. Tested with outputs changing every cycle and no load, at a 40 MHz clock rate.
7. Tested with all inputs within 0.1 V of VCC or Ground, no load.
8. These parameters are guaranteed but not 100% tested.
9. AC specifications are tested with input transition times less than 3ns, output reference levels of 1.5V
(except tDIS test), and input levels of nominally 0 to 3.0V. Output loading may be a resistive divider which
provides for specified IOH and IOL at an output voltage of VOH min and VOL max respectively. Alternatively,
a diode bridge with upper and lower current sources of IOH and IOL respectively, and a balancing voltage of
1.5V may be used. Parasitic capacitance is 30 pF minimum, and may be distributed.
This device has high-speed outputs capable of large instantaneous current pulses and fast turn-on/turn-off
times. As a result, care must be exercised in the testing of this device. The following measures are
recommended:
a. A 0.1 µF ceramic capacitor should be installed between VCC and Ground leads as close to the Device
Under Test (DUT) as possible. Similar capacitors should be installed between device VCC and the tester
common, and device ground and tester common.
b. Ground and VCC supply planes must be brought directly to the DUT socket or contactor fingers.
c. Input voltages on a test fixture should be adjusted to compensate for inductive ground and VCC noise to
maintain required DUT input levels relative to the DUT ground pin.
10. Each parameter is shown as a minimum or maximum value. Input requirements are specified from the
point of view of the external system driving the chip. Setup time, for example, is specified as a minimum
since the external system must supply at least that much time to meet the worst-case requirements of all
parts. Responses from the internal circuitry are specified from the point of view of the device. Output
delay, for example, is specified as a maximum since worst-case operation of any device always provides
data within that time.
LOGIC Devices Incorporated
26
Video Imaging Products
Feb 5, 2003 LDS.3321-A
LF3321
Horizontal Digital Image Filter
DEVICES INCORPORATED
Improved Performance
Notes
11. For the tENA test, the transition is measured to the 1.5V crossing point with datasheet loads. For
the tDIS test, the transition is measured to the ±200mV level from the measured steady-state output
voltage with ±10mA loads. The balancing voltage, VTH, is set at 3.0 V for Z-to-0 and 0-to-Z tests, and
set at 0 V for Z-to-1 and 1-to-Z tests.
12. These parameters are only tested at the high temperature extreme, which is the worst case
for leakage current.
Figure A. Output Loading Circuit
Figure B. Threshold Levels
tENA
OE
S1
DUT
IOL
Z
0
tDIS
1.5 V
1.5 V
1.5 V
VTH
CL
IOH
Z
1
1.5 V
3.0V Vth
VOL*
VOH*
0.2 V
0.2 V
0
Z
1
Z
0V Vth
VOL* Measured VOL with IOH = –10mA and IOL = 10mA
VOH* Measured VOH with IOH = –10mA and IOL = 10mA
LOGIC Devices Incorporated
27
Video Imaging Products
Feb 5, 2003 LDS.3321-A
LF3321
Horizontal Digital Image Filter
DEVICES INCORPORATED
Improved Performance
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Top
View
144-pin
108
107
106
105
104
103
102
101
100
99
98
97
96
95
94
93
92
91
90
89
88
87
86
85
84
83
82
81
80
79
78
77
76
75
74
73
GND
COUT11
COUT10
COUT9
COUT8
COUT7
COUT6
COUT5
COUT4
COUT3
COUT2
COUT1
COUT0
GND
VCC
CENB
CAB7
CAB6
CAB5
CAB4
CAB3
CAB2
CAB1
CAB0
CFB11
CFB10
CFB9
CFB8
CFB7
CFB6
CFB5
CFB4
CFB3
CFB2
CFB1
CFB0
LDA
PAUSEA
ACCA
RSLA3
RSLA2
RSLA1
RSLA0
VCC
GND
DOUT15
DOUT14
DOUT13
DOUT12
DOUT11
DOUT10
DOUT9
DOUT8
GND
OED
DOUT7
DOUT6
DOUT5
DOUT4
DOUT3
DOUT2
DOUT1
DOUT0
GND
VCC
RSLB0
RSLB1
RSLB2
RSLB3
ACCB
PAUSEB
LDB
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
VCC
DIN11
DIN10
DIN9
DIN8
DIN7
DIN6
DIN5
DIN4
DIN3
DIN2
DIN1
DIN0
GND
VCC
CENA
CAA7
CAA6
CAA5
CAA4
CAA3
CAA2
CAA1
CAA0
CFA11
CFA10
CFA9
CFA8
CFA7
CFA6
CFA5
CFA4
CFA3
CFA2
CFA1
CFA0
144
143
142
141
140
139
138
137
136
135
134
133
132
131
130
129
128
127
126
125
124
123
122
121
120
119
118
117
116
115
114
113
112
111
110
109
GND
ROUT11
ROUT10
ROUT9
ROUT8
ROUT7
ROUT6
ROUT5
ROUT4
ROUT3
ROUT2
ROUT1
ROUT0
GND
VCC
SHENA
TXFRA
GND
CLK
VCC
TXFRB
SHENB
RIN0
RIN1
RIN2
RIN3
RIN4
RIN5
RIN6
RIN7
RIN8
RIN9
RIN10
RIN11
OEC
VCC
Package and Ordering Information
Plastic Quad Flatpack (Q5)
0°C to 70°C--Commercial Screening
Speed
9 ns
LOGIC Devices Incorporated
LF3321QC9
28
Video Imaging Products
Feb 5, 2003 LDS.3321-A