Improved Transmitter Jitter Modeling for Accurate Bit Error Rate (BER) Eye Contours Using Transient Simulation of Short Bit Patterns

Increase in the signaling speeds has led to decrease in jitter budget available for the channel to pass the product specifications. The current serial link specifications are based on a bit-error-rate (BER) metric and hence there is a need to account for jitter statistically. Accurately capturing jitter through the channel is very important to have confidence in the product specifications. Current statistical eye approaches are based on either single pulse/step response or multiple edge responses, which do not account for transmitter (Tx) driver jitter accurately since jitter interaction between the edges is ignored. There is an inherent assumption in pulse/step response methods that there is zero correlation between independent responses created from 2
ⁿ
combinations for n cursor pulse response where n is number where the tail settles, hence one zero crossing does not impact an adjacent zero crossing. Transient simulation accurately accounts for Tx jitter but is time consuming and requires large memory when simulating very large bit sequences. A new methodology is proposed to generate BER contours that capture the Tx driver jitter and inter symbol interference (ISI) through the channel accurately using unique waveforms created from truth table bit combinations. It utilizes 2
^N
short N bit patterns as waveforms with jitter correlation from current bit patterns into adjacent bit patterns to get equivalent transient simulation of a desired pseudo random bit sequence. The method accounts for the edge interactions from any current bit pattern to next successive bit pattern to account for ISI accurately. The statistical eye diagram generated with the above approach includes nonideal channel characteristics, including ISI, XTK from nearby aggressor channels, and jitter through the channel.