Due to the sensitivity of the write process to temperature variations and laser position alignment, transition noise in HAMR may exhibit characteristics that are significantly different from those of perpendicular magnetic recording (PMR). It is important to have an accurate characterization of medium noise for a realistic channel model, which in turn is essential for developing reliable signal processing and coding schemes. In this work, statistical properties of transition noise in HAMR and PMR are investigated using real data taken off a spin stand. The Karhunen-Loeve (K-L) expansion method is applied to the read waveforms corresponding to a pseudorandom bit pattern written at varying locations of a disk.
A 127-bit pseudorandom pattern is written on different positions of the disk. The HAMR system under study employs a FePtL10 thin-film disk and a shielded pole head designed for PMR. The same head was used along with a CoCrPt alloy thin-film disk for the PMR measurements. Each 127-bit pattern is read repeatedly to remove additive system noise via averaging. The averaged read waveforms are interpolated with sufficient resolution and then aligned based on the least square principle. The aligned waveforms constitute an ensemble that contains meaningful statistical variations involving transitions within different local bit patterns. The K-L expansion allows decomposition of the noise associated with isolated transitions into different orthogonal directions corresponding to position jitter and width fluctuations. In this work, we apply the K-L expansion to the read waveforms corresponding to crowded transition patterns in HAMR and PMR, in order to develop insights into data-dependent transition noise characteristics.
The results indicate that for isolated transitions in PMR, the position jitter component accounts for about 80% of the total noise power whereas the width fluctuation component takes up 9-13%. This appears consistent with the results reported in pr...