H ∞ Probabilistic Robust Controller Design for Hdd Servomechanism

Challenges in designing head positioning servo controller suitable for very high density HDD (above 1 Tbits/in) comes from the issues of external and internal disturbances and to make the controller robust in spite of parametric uncertainty. Servomechanism employing dual-stage actuator can achieve bandwidth required for such high TPI and robust control design has been proposed in the past to address the issues related to uncertainty. With robust control, the servomechanism performs well for all actuators in the mass produced hard disk drives. However, high performance and sufficient robustness are conflicting requirements; in order to maintain robustness of the controller, performance is often sacrificed. Several control techniques such as Robust H2, Mixed H2/H∞ and -synthesis have been extensively studied for HDD servo control. These approaches provide the framework required achieve the bandwidth optimally in presence of uncertainty. However, these frameworks suffer from a number of theoretical limitations in terms of complexity and conservatism due to their deterministic nature of dealing with uncertainty. With increasing number of uncertain parameters, the design problem often becomes unsolvable. We propose a probabilistic approach for designing robust controller which remains solvable in spite of increased number of uncertain parameters. The order of the obtained controller is lower than those obtained by other existing robust control frameworks and yet handles uncertainty equally good. Preliminary results show significant improvement compared to conventional robust control paradigm. Utilizing this approach we are able to design a lower order controller which have a better performance compared to a controller of similar order obtained using robust H2 design.