A Data Mining Approach for Diagnosis of Nonlinear Stochastic Dynamics of TFC Active Slider in Nano-Meter Spacing

Abstract

Storage of 10 Tb/in2 in hard disk drives requires a physical spacing in the level of 0.25 nm at the read-write transducer location. A lot of physical and tribology issues exist to such a low flying height. At such a small spacing intermittent contact of solid-to-solid and solid-to-liquid between the slider and disk surface becomes inevitable, which motivates the improvement needs for the present MEMS-based thermal fly-height control (TFC) technology to satisfy the future storage needs. How to control slider to reduce touchdown instability and eventually eliminate bouncing has been a pressing and challenging research topic. There are many researches dedicated in addressing this challenging issue. For examples, the hysteresis, the influences of surface roughness and waviness, the lubricant modulation and uncertainty, and the nonlinear properties and identification have been investigated and examined in the literatures [1–9]. Knowledge obtained from the literatures concludes that the contact dynamics of slider involve strong nonlinearity and stochastic properties. By using FFT spectral analysis, existing research has clarified many nonlinear dynamics phenomena during touchdown. However, many more complicated phenomena such as narrowband and wideband frequency spectrum and the stochastic features of system response have yet clarified. The existing analytical and numerical researches have been centered on deterministic model analysis. The widely observed random properties of roughness and waviness of solid surfaces and modulation of lubricant have not been fully reflected in the analysis. Inherently not applicable to strong nonlinear and stochastic dynamics problems by classical FFT approach, this paper then proposes to develop a data mining-based approach to diagnosis the complex nonlinear stochastic dynamics of the TFC active slider at nano-meter spacing to the surface during touchdown event.