A new digital signal processing channel for data storage products

The author presents a run-length coded channel with digital signal processing and a sequence detection algorithm. A novel look-ahead decoder provides maximum-likelihood performance without Viterbi computations by focusing, on the minimum-distance error events. The channel, called (1,7) ML, is characterized by the extended PR4 (EPR) polynomial (1-D)(1+D)(1+D). This channel retains the write path and the gain control and timing recovery circuits of the read path from the conventional peak-detection channel; the read signal, however, is equalized, sampled, and digitized. The decoder processes these sample values iteratively, receiving one new sample per clock with a look-ahead of at least five sample values. A set of linear functions of the look-ahead sample values and a set of threshold constants, called decision boundaries, are at the heart of the decoder architecture. The linear functions and the decision boundaries are chosen so as to minimize the overall mean-square error in the sequence detection process. The author introduces a number of concepts and tools that are novel to signal processing technology and its application to data storage products