A high-speed adaptive equalizer for magnetic disk drives

Abstract

The design of a high-speed adaptive decision feedback equalizer (DFE) is described for detection of magnetic disk data. A RAM is used in the feedback path instead of the usual transversal filter, and the equalizer is referred to as the RAM-DFE. The RAM compensates for trailing nonlinear distortion in the disk data. However, use of the RAM makes the equalizer more sensitive to latency in the adaptive update and can slow training considerably when using LMS adaptation. A technique is presented for training the RAM which reduces convergence time to near that of a transversal feedback filter. The RAM-DFE design is targeted for a VLSI implementation using BiCMOS technology. Simulation predict clock speeds in excess of 50 MHz, corresponding to data rates in excess of 50 Mb/s. The design also includes circuits to perform digital timing recovery and initial synchronization.