Backchannel Modeling and Simulation Using Recent Enhancements to the IBIS Standard

摘要

Backchannel training is a technique used in modern serial link interfaces like 10GBASEKR and PCI Express (PCIe) to optimize transmitter and receiver equalization settings for the characteristics of a particular channel. In this scheme, a known training pattern of bits is sent from the transmitter to the receiver. The receiver evaluates the signal quality of the pattern, takes judgement on it, and communicates desired equalization adjustments through the "backchannel" to the transmitter. The transmitter adjusts its settings, and retransmits the training pattern. This process is repeated until the receiver is satisfied with the signal quality, or the limit of the number of training patterns has been reached. Then the transmitter equalization settings are locked in, and the intended data is transmitted. Performed automatically by SerDes hardware, backchannel training is an essential part of signal quality in high performance serial link interfaces. Until recently, there has been no standard way to model this behavior in serial link simulations with commercial tools. But recent enhancements to the upcoming IBIS standard now support backchannel training, enabling IBIS-AMI models to emulate this real-world SerDes behavior. AMI modelers will now have the ability to incorporate backchannel algorithms into their IBIS-AMI models, automating the optimization of transmitter and receiver equalization settings in the same manner as their actual SerDes hardware devices. This will save system designers significant time by avoiding a multitude of computationally intensive sweeping in order to determine optimum equalization settings for their link, while at the same time yielding more realistic and higher quality results that are more consistent with the hardware they seek to model. This paper will review backchannel basics as well as the upcoming IBIS enhancements. It will also show specifically how to incorporate backchannel capability into standard IBIS-AMI models, and present simulation results using these new capabilities.