Timing sensitivity in discrete-time equalization
Jaekyun J. Moon · [1993] Digests of International Magnetics Conference · 1993
In digital channels, pulse shaping is achieved by passing readback samples through a discrete-time filter called the equalizer, which is typically implemented using a finite impulse response (FIR) filter consisting of taps and delays. Under the finite-complexity constraint, the equalizer performance is a highly sensitive function of timing phase. In this paper, we investigate the performance of different equalization schemes suitable for high density magnetic recording - partial response linear equalization and feedback equalization - as a function of static sampling phase and of equalizer complexity. I. INTRODUCTION The impact of timing errors on the performance of ideal continuous-time equalizers has been studied thoroughly in (l) for recording channels with significant excess bandwidth. Timing sensitivities for partial response linear equalizers for both symbol-rate and fractionally-spaced digital equalizers are analyzed extensively in (2). In this paper, we consider the timing error effects on the performance of both linear equalization and feedback equalization for high density recording channels which are highly band-limited. Our focus is on adaptive discrete-time equalizers with f~te complexity. As linear density increases in magnetic recording, most of the signal energy resides below half the symbol frequency. In such regime, it is well known that the ideal continuous time equalizer or the discrete-time equalizer with infinite numbe-r of taps is completely immune to timing phase errors (3). However, when the lit is imposed on the number of taps in the finite impulse response (FIR) equalizer filters. the resulting performance is no longer immune to errors in the timing phase. In this paper. we study the effect of sampling phase on the performance of various finite-complexity equalizers considered suitable for magnetic recording applications. We consider two types of equalization: partial response linear equalization and feedback equalization. Both class IV partial response (PR4) and extended class IV (EPR4) targets are assumed in partial response equalization (4). and decision feedback equalization (DFE) (5) and fixed-delay tree search with decision feedback (FDTS/DF) (6) are considexed in feedback equalization.