基于DB-PAM4的高速SerDes自适应均衡器设计

Design of high-speed SerDes adaptive equalizer based on DB-PAM4

PAM4编码作为如今高速串口的主流编码方式,其均衡存在多种方式,但是传统的均衡方式所需均衡强度大,抽头数量多,算法自适应调节复杂。基于此,提出了一种基于背板信道的联合自适应均衡算法——Co-LMS算法,旨在降低均衡强度的同时,减少抽头数量,提高算法的收敛速度。算法对原始PAM4信号进行预编码处理,各种均衡器结合信道特性在接收端生成DB-PAM4信号。该算法在MATLAB中进行了行为级仿真验证。结果表明,在112 Gb/s的数据传输速率下,信道衰减达到-35 dB,16抽头的FFE和1抽头的DFE就能完成均衡处理,且能在28 ns内达到稳定,相较于传统的均衡器算法抽头数量大大减少,收敛更快。

PAM4 encoding,the mainstream encoding method for high-speed serial interface port today,could be equalized by a variety of methods.However,the traditional equalization method requires large equalization intensity,large number of taps and complex adaptive adjustment.Because of that,a joint adaptive equalization algorithm was proposed based on backplane channel-Co-LMS algorithm,which aims to reduce the equalization strength,reduce the number of taps and improve the convergence speed of the algorithm.The algorithm principle was that the original PAM4 signal was first pre-coded,and then various equalizers combined the channel characteristics to generate the DB-PAM4 signal at the receiver.The algorithm was verified by behavior level simulation in MATLAB.And it turns out that at 112 Gb/s data rate,channel attenuation reaches-35 dB.16 taps FFE and 1 tap DFE can complete the equalization process,and can be stable within 28 ns.Compared with the traditional equalizer algorithm,the number of taps is greatly reduced and the convergence is faster.