Medium noise modeling, spinstand measurements, and timing recovery for magnetic recording
The digital magnetic recording for information storage industry is perhaps unparalleled in the duration and rate of its growth. For over 40 years, it has sustained at least 30% and as much as 100% compound annual growth rate in areal density. Much of this is due to overcoming tremendous hurdles in engineering over many different disciplines. In this dissertation, we explore media noise modeling including nonlinear distortion, digital magnetic recording measurements and testing, and timing recovery for digital magnetic recording.
Modern detection for digital magnetic recording uses partial response maximum likelihood (PRML) detection to combat linear intersymbol interference. Nonlinear distortions can degrade performance, so a model has been previously developed that includes nonlinear distortion effects. This model, while based on recording physics, is not a physical model, but rather a statistical one. In this work, we use spinstand measurements to evaluate the ability of the model to mimic real-life magnetic recording systems.
The magnetic recording spinstand is a valuable tool, both in research and production environments. It can be used to evaluate almost all facets of the magnetic recording system, from heads and media, to channel models, to detector designs. Here, we look at some of the tests and results obtained from a spinstand, with perpendicular recording media.
Timing recovery is an essential part of the digital magnetic recording puzzle. It takes samples with random fluctuations in their sampling phase and frequency and attempts to synchronize those samples with the read clock before they reach the detector. In this dissertation, we look at the standard timing recovery method in light of PRML channels, and we also evaluate a different timing recovery method based on cross-correlation.