8b/10b Encoder/Decoder

April 2011 Reference Design RD1012

Introduction
Many serial data transmission standards utilize 8b/10b encoding to ensure sufficient data transitions for clock recovery. This reference design describes an encoder/decoder suitable for performing 8b/10b encoding/decoding within Lattice programmable logic devices. Several generic CPLD and FPGA implementations are shown with this reference design.

Features
8b to 10b encoder and 10b to 8b decoder Previous octet disparity input and current disparity output Output to indicate when invalid control character is requested to be encoded Output to indicate when invalid data/control character is received Running disparity checking Conform to 8b/10b specified in IEEE 802.3z and ANSI X3.230-1994

Figure 1. The 8b/10b Encoder/Decoder in a System

MSB

LSB HGFEDCBA

MSB

LSB HGFEDCBA

10b to 8b Decoder

8b to 10b Encoder

Transmitter

Receiver

10
High Speed Deserializer

10
High Speed Serializer

1
1001110100101011101001100010111100000101 abcdeifghjabcdeifghjabcdeifghjabcdeifghj
D0.0 (RD-) D31.5 (RD-) D0.0 (RD+) K28.5 (RD+)

Functional Description
The 8b/10b coding scheme was initially proposed by Albert X. Widmer and Peter A. Franaszek of IBM Corporation in 1983. This coding scheme is used for high-speed serial data transmission. The encoder on the transmitter side maps the 8-bit parallel data input to 10-bit output. This 10-bit output is then loaded in and shifted out through a high-speed Serializer (Parallel-in Serial-out 10-bit Shift Register). The serial data stream will be transmitted through the transmission media to the receiver. The high-speed Deserializer (Serial-in Parallel-out 10-bit Shift Register) on the receiver side converts the received serial data stream from serial to parallel. The decoder will then remap the 10-bit data back to the original 8-bit data. When the 8b/10b coding scheme is employed, the serial data stream is DC-balanced and has a maximum run-length without transitions of 5. These characteristics aid in the recovery of the clock and data at the receiver. Figure 1 shows the 8b/10b encoder/decoder usage in a communication system.
2011 Lattice Semiconductor Corp. All Lattice trademarks, registered trademarks, patents, and disclaimers are as listed at www.latticesemi.com/legal. All other brand or product names are trademarks or registered trademarks of their respective holders. The specifications and information herein are subject to change without notice.

www.latticesemi.com

rd1012_01.2

Lattice Semiconductor DC Balance and Run Length

8b/10b Encoder/Decoder

A DC-balanced serial data stream means that it has the same number of 0s and 1s for a given length of data stream. DC-balance is important for certain media as it avoids a charge being built up in the media. The run-length is defined as the maximum numbers of contiguous 0s or 1s in the serial data stream. A small runlength data stream provides data transitions within a small length of data. Data transitions are essential for clock recovery. The PLL of the CDR generates a phase-adjustable output clock from the reference clock input. Transitions on the serial data stream provide the transmission clock phase information to the PLL and allow the PLL to recover the transmission clock with the correct phase. Note that the reference clock input is always necessary for the CDR. The serial data stream embeds the phase of the transmission clock, not the clock itself. This reference clock comes from the receiver system, not the transmitter system.

8b/10b Code Mapping

The 8b/10b encoder converts 8-bit code groups into 10-bit codes. The code groups include 256 data characters named Dx.y and 12 control characters named Kx.y. Figure 2. The 8b/10b Coding Scheme

code group
MSB

Dx.y
H G F E D

or

Kx.y
LSB
B A

8b
LSB

MSB b c d e i

LSB
f g h

MSB j

10b

The coding scheme breaks the original 8-bit data into two blocks, 3 most significant bits (y) and 5 least significant bits (x). From the most significant bit to the least significant bit, they are named as H, G, F and E, D, C, B, A. The 3bit block is encoded into 4 bits named j, h, g, f. The 5-bit block is encoded into 6 bits named i, e, d, c, b, a. As seen in Figure 2, the 4-bit and 6-bit blocks are then combined into a 10-bit encoded value.

Disparity
In order to create a DC-balanced data stream, the concept of disparity is employed to balance the number of 0s and 1s. The disparity of a block is calculated by the number of 1s minus the number of 0s. The value of a block that has a zero disparity is called disparity neutral. If both the 4-bit and 6-bit blocks are disparity neutral, a combined 10-bit encoded data will be disparity neutral as well. This will create a perfect DC-balanced code. However, this is not possible. Because only 6 out of the 16 possible values of the 4-bit block are disparity neutral, they are not enough for encoding the 8 values of the 3-bit block. Likewise, only 20 values of the 6-bit block are disparity neutral and they are not enough for encoding the 32 values of the 5-bit block. Because both the 4-bit and 6-bit blocks have an even number of bits, the disparity is not possible to be +1 or -1. Therefore, the values with a disparity of +2 and -2 are also used in the 8b/10b coding scheme. Table 1 and Table 2 are the values that are used for the 3-bit to 4-bit encoding and the 5-bit to 6-bit encoding respectively. Concatenating the 4-bit and 6-bit blocks together generates the 10-bit encoded value. Note that some of the encoded values in Table 1 and Table 2 have two possible values, one with a disparity value of +2 and the other with a disparity value of -2. The 8b/10b coding scheme was designed to combine the values of the 4-bit and 6-bit blocks perfectly so that the worst case disparity value of the 10-bit code group will be at most +2 or -2. For 2

Lattice Semiconductor

8b/10b Encoder/Decoder

example, the 4-bit encoded values with disparity value+2 will not be combined with the 6-bit encoded values with disparity value +2 because this will create a 10-bit value with disparity value +4. Table 1. 3-Bit to 4-Bit Encoding Values
3b Decimal 0 1 2 3 4 5 6 7 3b Binary (HGF) 000 001 010 011 100 101 110 111 4b Binary (fghi) 0100 or 1011 1001 0101 0011 or 1100 0010 or 1101 1010 0110 0001 or 1110 or 1000 or 0111

Table 2. 5-Bit to 6-Bit Encoding Values

5b Decimal 0 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 5b Binary (EDCBA) 00000 00001 00010 00011 00100 00101 00110 00111 01000 01001 01010 01011 01100 01101 01110 01111 10000 10001 10010 10011 10100 10101 10110 10111 11000 11001 11010 11011 11100 11101 11110 11111 6b Binary (abcdei) 100111 or 011000 011101 or 100010 101101 or 010010 110001 110101 or 001010 101001 011001 111000 or 000111 111001 or 000110 100101 010101 110100 001101 101100 011100 010111 or 101000 011011 or 100100 100011 010011 110010 001011 101010 011010 111010 or 000101 110011 or 001100 100110 010110 110110 or 001001 001110 101110 or 010001 011110 or 100001 101011 or 010100

Lattice Semiconductor Running Disparity

8b/10b Encoder/Decoder

Since the worst disparity of the 10-bit encoded data value is either +2 or -2, it is still possible that more 10-bit encoded data values with +2 (or -2) disparity are transmitted through the serial data stream. In this case, the data stream will no longer be DC-balanced. In order to maintain a DC-balance data stream, each code group will be converted to one of the two possible values as seen in the RD- and RD+ columns of the Table 3. The RD- disparity will be either +2 or 0 (disparity neutral) and the RD+ disparity will be either -2 or 0. The encoder will pick one of the two values based on the calculation of current Running Disparity. Table 3. Portion of the 8b/10b Encoding/Decoding Mapping Table
8-bit data Code Group D0.0 D1.0 D2.0 D3.0 : D31.0 D0.2 D1.2 D2.2 D3.2 : D31.2 D0.4 D1.4 D2.4 D3.4 : D31.4 D0.6 D1.6 D2.6 D3.6 : D31.6 K28.0 K28.1 K28.2 K28.3 K28.4 K28.5 K28.6 K28.7 K23.7 K27.7 K29.7 K30.7 0 1 1 1 1 1 1 1 1 1 1 1 1 110 11111 000 11100 001 11100 010 11100 011 11100 100 11100 101 11100 110 11100 111 11100 111 10111 111 11011 111 11101 111 11110 101011 0110 001111 0100 001111 1001 001111 0101 001111 0011 001111 0010 001111 1010 001111 0110 001111 1000 111010 1000 110110 1000 101110 1000 011110 1000 010100 0110 110000 1011 110000 0110 110000 1010 110000 1100 110000 1101 110000 0101 110000 1001 110000 0111 000101 0111 001001 0111 010001 0111 100001 0111 0 0 0 0 0 100 11111 110 00000 110 00001 110 00010 110 00011 101011 0010 100111 0110 011101 0110 101101 0110 110001 0110 010100 1101 011000 0110 100010 0110 010010 0110 110001 0110 0 0 0 0 0 010 11111 100 00000 100 00001 100 00010 100 00011 101011 0101 100111 0010 011101 0010 101101 0010 110001 1101 010100 0101 011000 1101 100010 1101 010010 1101 110001 0010 0 0 0 0 0 000 11111 010 00000 010 00001 010 00010 010 00011 101011 0100 100111 0101 011101 0101 101101 0101 110001 0101 010100 1011 011000 0101 100010 0101 010010 0101 110001 0101 kin/ kout 0 0 0 0 HGF EDCBA 000 00000 000 00001 000 00010 000 00011 10-bit data (RD-) abcdei fghj 100111 0100 011101 0100 101101 0100 110001 1011 10-bit data (RD+) abcdei fghj 011000 1011 100010 1011 010010 1011 110001 0100 Code Group D0.1 D1.1 D2.1 D3.1 : D31.1 D0.3 D1.3 D2.3 D3.3 : D31.3 D0.5 D1.5 D2.5 D3.5 : D31.5 D0.7 D1.7 D2.7 D3.7 : D31.7 0 111 11111 101011 0001 010100 1110 0 0 0 0 0 101 11111 111 00000 111 00001 111 00010 111 00011 101011 1010 100111 0001 011101 0001 101101 0001 110001 1110 010100 1010 011000 1110 100010 1110 010010 1110 110001 0001 0 0 0 0 0 011 11111 101 00000 101 00001 101 00010 101 00011 101011 0011 100111 1010 011101 1010 101101 1010 110001 1010 010100 1100 011000 1010 100010 1010 010010 1010 110001 1010 0 0 0 0 0 001 11111 011 00000 011 00001 011 00010 011 00011 101011 1001 100111 0011 011101 0011 101101 0011 110001 1100 010100 1001 011000 1100 100010 1100 010010 1100 110001 0011 kin/ kout 0 0 0 0 8-bit data HGF EDCBA 001 00000 001 00001 001 00010 001 00011 10-bit data (RD-) abcdei fghj 100111 1001 011101 1001 101101 1001 110001 1001 10-bit data (RD+) abcdei fghj 011000 1001 100010 1001 010010 1001 110001 1001

The transmitter assumes a negative Running Disparity (RD-) at start up. When an 8-bit data is encoding, the encoder will use the RD- column for encoding. If the 10-bit data been encoded is disparity neutral, the Running Disparity will not be changed and the RD- column will still be used. Otherwise, the Running Disparity will be changed and the RD+ column will be used instead. Similarly, if the current Running Disparity is positive (RD+) and a disparity neutral 10-bit data is encoded, the Running Disparity will still be RD+. Otherwise, it will be changed from RD+ 4

Lattice Semiconductor

8b/10b Encoder/Decoder

back to RD- and the RD- column will be used again. The state diagram in Figure 3 describes how the current Running Disparity is calculated. Figure 3. Running Disparity State Machine
Power Up otherwise

if the 10-bit encoded data of the current transmitting code group is disparity neutral

RDuse RD- column for encoding otherwise

RD+
use RD+ column for encoding

if the 10-bit encoded data of the current transmitting code group is disparity neutral

Encoder and Decoder Pin Descriptions

The 8b/10b encoder and 10b/8b decoder are implemented in HDL language. These designs can be combined to create a full duplex channel. The pin descriptions of the designs are shown in the following tables. Table 4. Encoder Pin Descriptions
Name clk reset_n datain_8b_[7:0] kin rdispin dataout_10b_[9:0] rdispout k_err Type I I I I I O O O Description Encoder Clock. This pin is the main clock of the encoder. All registered inputs and outputs of the encoder are based on the rising of this clock. Master Reset. This low active asynchronous reset will reset all internal registers of the encoder to their initial states. 8-bit Data Input. This is the 8-bit raw data to be encoded. Character Type Control. This high active signal indicates that the 8-bit data on datain_8b bus is going to be encoded to a control character instead of a data character. Running Disparity Input. This pin provides to the encoder the running disparity before the encoding of current 8-bit data on datain_8b bus. 10-bit Data Output. This is the 10-bit encoder output. Running Disparity Output. This is the running disparity of the present dataout_10b bus. Invalid Control Character Requested. This high active signal indicates that a invalid control character is requested.

Lattice Semiconductor
Table 5. Decoder Pin Descriptions
Name clk reset_n datain_10b_[9:0] rdispin dataout_8b_[7:0] kout rdispout disp_err code_err Type I I I I O O O O O Description

8b/10b Encoder/Decoder

Decoder Clock. This pin is the main clock of the decoder. All registered inputs and outputs of the decoder are based on the rising of this clock. Master Reset. This low active asynchronous reset will reset all internal registers of the decoder to their initial states. 10-bit Data Input. This is the 10-bit raw data to be decoded. Running Disparity Input. This pin provides to the decoder the running disparity before the decoding of current 10-bit data on datain_10b bus. 8-bit Data Output. This is the 8-bit decoder output. Character Type Output. This high active signal indicates that the 8-bit data on dataout_8b bus is a control character instead of a data character. Running Disparity Output. This is the running disparity of the present dataout_8b bus. Disparity Error. This high active signal indicates that a running disparity error has occurred. Invalid Code Group Error. This high active signal indicates that an invalid 10-bit code group has been received.

Block Diagrams and Implementation Guidelines

The block diagrams of the encoder and decoder are shown in Figure 4 and Figure 5. Notice that the encoder contains a clock latency of 2 and the decoder contains a clock latency of 3. The clock latency allows the throughput to be increased. All signals in the designs are based on the rising edge of the main clock. Once the data inputs are provided and the setup/hold times are satisfied, the data inputs will be sampled at the rising edge of the clock. As seen in Figure 6, the data outputs will be available after another one or two rising edges and a clock-to-output delay. For both the encoder and decoder designs, the rdispout output should be connected back to the rdispin input. Except clk, reset_n, and rdispin, the other inputs are pipelined to generate the outputs. However, since the running disparity of the data currently being transmitted is required for the encoding of the following data, the rdispin-to-rdispout loopback path can only contain one register level and cannot be pipelined. Figure 4. Encoder Block Diagram

datain_8b_[7:0]

SET

5
CLR

5b/6b Encoding Table 3b/4b Encoding Table Controls Disarity Generation

SET

10
CLR

10
Q

dataout_10b_[9:0]

6 4
D
SET

kin

8
D
SET

k_err

CLR

Q
CLR

rdispin clk

SET

rdispout

CLR

reset_n

Lattice Semiconductor
Figure 5. Decoder Block Diagram

8b/10b Encoder/Decoder

SET

datain_10b_[9:0]

8
D
SET

8
CLR

dataout_8b_[7:0]

6
CLR

6b/5b Decoding Table 4b/3b Decoding Table Controls

5
D
SET

kout

CLR

3
D
SET

disp_err

CLR

10

SET

code_err

CLR

rdispin clk

Disparity Generation
D
SET

rdispout

CLR

reset_n

Figure 6. Timing Diagrams

clk datain_8b_[7:0], kin dataout_10b_[9:0], rdispout, k_err
in(1) in(2) in(3) in(4)

clk datain_10b_[9:0] dataout_8b_[7:0], rdispout, kout, disp_err, code_err

in(1) in(2) in(3) in(4) in(5)

out(1)

out(2)

out(1)

out(2)

Encoder Timing

Decoder Timing

Encoder and Decoder Behavior

This 8b/10b encoder/decoder is fully tested to cover all possible cases. The encoder will generate the correctly encoded 10-bit codes based on the kin, datain_8b_[7:0], and rdispin inputs. The total number of combination cases is 2 x 256 x 2 = 1024. In these 1024 cases, only (256 + 12) x 2 = 536 cases are valid and the k_err will be generated for the rest of the 488 cases. The decoder will convert 10-bit codes into 8-bit outputs. The code_err will be generated for all invalid 10-bit codes the decoder received. However, in some cases even though the received 10bit input data is one of the 536 valid codes, the disp_err will be generated if the running disparity rule is violated. The behavior of all possible input combinations of the 8b/10b encoder/decoder is shown in Table 6 and Table 7.

Lattice Semiconductor
Table 6. 8b/10b Encoder Behavior
Inputs kin datain_8b_[7:0] Any of the 256 Dx.y 8-bit Data Characters rdispin 0 1 0 1 0 1 dataout_10b_[9:0]

8b/10b Encoder/Decoder

Outputs rdispout k_err 10-bit dataout from the (RD-) column of 0, If dataout has 5 ones the corresponding Dx.y row in Table 3 1, If dataout has 6 ones 10-bit dataout from the (RD+) column of 0, If dataout has 4 ones the corresponding Dx.y row in Table 3 1, If dataout has 5 ones 10-bit dataout from the (RD-) column of 0, If dataout has 5 ones the corresponding Kx.y row in Table 3 1, If dataout has 6 ones 10-bit dataout from the (RD+) column of 0, If dataout has 4 ones the corresponding Kx.y row in Table 3 1, If dataout has 5 ones 10-bit dataout from the (RD-) column of 0, If dataout has 5 ones the corresponding Dx.y row in Table 3 1, If dataout has 6 ones 10-bit dataout from the (RD+) column of 0, If dataout has 4 ones the corresponding Dx.y row in Table 3 1, If dataout has 5 ones

Any of the 12 Kx.y 8-bit Control Characters 1 Other undefined 244 Kx.y 8-bit Control Characters

Table 7. 10b/8b Decoder Behavior

Inputs datain_10b_[9:0] Any 10-bit data from either the (RD-) or (RD+) column of the Dx.y rows in Table 3 Any 10-bit data from either the (RD-) or (RD+) column of the Kx.y rows in Table 3 Others rdispin code_err 0 1 0 0 1 0 1 1 disp_err 0, if data is from (RD-) column 1, if data is from (RD+) column 0, if data is from (RD+) column 1, if data is from (RD-) column 0, if data is from (RD-) column 1, if data is from (RD+) column 0, if data is from (RD+) column 1, if data is from (RD-) column X Outputs kout dataout_8b_[7:0] rdispout 0, If dataout has 5 ones 1, If dataout has 6 ones Corresponding 8-bit Data 1 0, If dataout has 4 ones 1, If dataout has 5 ones 0, If dataout has 5 ones 1, If dataout has 6 ones 0, If dataout has 4 ones 1, If dataout has 5 ones X 0, If rdispin is 0 1, If rdispin is 1

Note: X means that the outputs may be 0 or 1 and should be ignored because of the code_err.

Lattice Semiconductor

8b/10b Encoder/Decoder

HDL Simulation and Verification

Figure 7 shows the timing simulation of the 8b/10b encoder/decoder implemented in a LC4128V-5T100C device. After reset, D3.7 (RD-), D0.6 (RD+), D0.0 (RD-), D0.0 (RD-), ..., are transmitted. Figure 7. Timing Simulation Diagrams

Lattice Semiconductor Implementation

8b/10b Encoder/Decoder

This design is implemented in Verilog and VHDL. When using this design in a different device, density, speed, or grade, performance and utilization may vary. Default settings are used during the fitting of the design. Table 8. Performance and Resource Utilization
Device Family MachXO2 2 MachXO 3 LatticeECP3 4 LatticeECP2M 5 LatticeECP 6 LatticeXP2 7 ispMACH 40008 Language Verilog VHDL Verilog VHDL Verilog VHDL Verilog VHDL Verilog VHDL Verilog VHDL Verilog VHDL Speed Grade -6 -6 -3 -3 -7 -7 -6 -6 -5 -5 -5 -5 -3 (ns) -3 (ns) Utilization1 184 LUT4 184 LUT4 152 LUTs 152 LUTs 184 LUT4 184 LUT4 184 LUT4 184 LUT4 184 LUTs 184 LUTs 184 LUTs 184 LUTs 74 Macrocells 74 Macrocells fMAX (MHz) 100 100 100 100 200 200 200 200 100 100 100 100 90 90 I/Os 43 43 43 43 43 43 43 43 43 43 43 43 43 43 Architecture Resources N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A

1. Utilization is the total resources used for the Encoder and Decoder. The Encoder occupies about 30% of the total resource used. 2. Performance and utilization characteristics are generated using LCMXO2-1200HC-6MG132C with Lattice Diamond 1.2 design software. 3. Performance and utilization characteristics are generated using LCMXO1200C-3T100C with Lattice Diamond 1.2 design software. 4. Performance and utilization characteristics are generated using LFE3-150EA-7FN1156C with Lattice Diamond 1.2 design software. 5. Performance and utilization characteristics are generated using LFE2M-50E-6F672C with Lattice Diamond 1.2 design software. 6. Performance and utilization characteristics are generated using LFECP-6E-5T144C with Lattice Diamond 1.2 design software. 7. Performance and utilization characteristics are generated using LFXP2-5E-5M132C with Lattice Diamond 1.2 design software. 8. Performance and utilization characteristics are generated using LC4256B-3T100C with ispLEVER Classic 1.3 design software.

Technical Support Assistance

Hotline: 1-800-LATTICE (North America) +1-503-268-8001 (Outside North America) e-mail: [email protected] Internet: www.latticesemi.com

Revision History
Date November 2002 June 2010 April 2011 Version 01.0 01.1 01.2 Initial release. Updated for LatticeXP2, LatticeECP and MachXO device support. Updated Timing Simulation diagram. Added support for LatticeECP2M and LatticeECP3 device families. Added support for Lattice Diamond design software. Removed ispXPLD 5000MX from Implementation table. Change Summary

10