I. Introduction
Many common timing recovery schemes [1] are based on an inductive architecture where the sampler is contained within a feedback loop. Decision-directed inductive schemes [2], where output decisions are used to aid the timing recovery, are often used for the magnetic recording channel, as they improve performance in the presence of high levels of channel distortion. However, as the detector is inside the timing recovery loop, unwanted interactions can occur, especially when the detector is adaptive.