Noise and error pattern in shingled magnetic recording

Abstract

Shingled magnetic recording (SMR) is a promising approach for achieving storage density beyond 1 Tbit/in2. The areal density increase is achieved by overlapping the tracks during writing, which results in increased impairments such as magnetic transition jitter, inter-track interference (ITI), residual and overwriting noise. Such media noises and interference becomes two-dimensional (jitter along the down track direction; ITI, overwriting and residual noise along the cross track direction) due to decreased track pitch in the SMR system, which is more challenging than the one-dimensional impairments seen in conventional magnetic recording. The goal of this effort is to better understand the effects of these 2D impairments in SMR at different channel densities by using the accurate micromagnetic writing model combined with the read channel. Recently [1], the rotated single head (RSH) has been investigated for its tolerance to ITI. To investigate the extant and effect of these impairments, this research has employed an optimized micromagnetic writer (shown in Fig.1(a)) to shingle-write an exchange coupled composite media (Voronoi grains of size 8nm) and readback with both a normally oriented single head (NSH) and rotated single head (RSH) based on the physically accurate reciprocity principle[1]. In this study, the magnetic fly height is 3nm, and the reader has 4nm×18nm×18nm free layer and 11nm shield-shield spacing. The signals are over-sampled at 2nm intervals and processed with a low-pass filter, 2D minimum mean squared error (MMSE) equalizer and detected with a Viterbi detector. In the simulation, the bit width is kept constant at 10nm, while the track width is varied from 20nm, 15nm to 10nm (corresponding to channel densities of 3.22 Tb/in2, 4.30 Tb/in2 and 6.45 Tb/in2, respectively) to mimic the increasing track density.