Noise characterization by voronoi cell modeling for perpendicular double layer media

Abstract

~~~~~~~~~~~~~~~~ disk was the same ah for our previous error rate measurements [I]. Differentiated readback waveforms for a designated position on the disk were averaged to reduce electronics noise using precise phase matching based on cross-correlation. Then, the averaged waveform containing 40 isolated pulses was cut out and accumulated so that all the isolated pulses were aligned with a fixed transition point. The standard deviation, normalized to the pulse amplitude, was calculated for the cumulative waveform. A result of the time domain measurement is depicted in fig. 1. A clear noise peak at the transition point was observed, which indicated that transition noise was dominant rather than uniformly additive noise. A positive correlation between the rise-edge and the falliug-edge of the readback pulses was also clarified, as shown in fig. 2, which showed that the transition noise wasn’t caused by a variation of pulse width or pulse amplitude, but position jitter. Direct examination of the magnetic transitions using a MFM, Magnetic Force Microscope, showed microscopic cross-track fluctuations of the magnetic transition. The fluctuated transition lines corresponded to boundaries running through magnetic ‘clusters’ whose sizes were around 80 nm [2]. This fact suggested that the noisc originated from the magnetic ‘clusters’. Noise Estimation usins Voronoi Cell Modeling Assuming that several neighboring crystallographic grains of the medium were magnetically connected together and switched in unison, we established il noise model using Voronoi cells; each cell represented the connected gains which are referred to as a ‘cluster’. We calculated transition jitter and root-mean-squared (RMS) noise voltage: a generated assembly of Voronoi cells was sliced into microtracks and the Lorentzian pulse responses for each microtracks were superposed. The calculated standard deviation of the simulated transition noise on time axis is shown in fig. 3. The results were in good quantitative agreement with the experimental result