Partial response maximum likelihood (PRML) detection trellis method adaptable to signal dropouts

A tape drive having a maximum likelihood detector comprising:

• an adaptive trellis adaptable to read a signal having signal amplitude dropouts, the adaptive trellis including a standard partial response maximum likelihood (PRML) trellis, the standard PRML trellis associated with a standard state diagram having standard branches between states, wherein a first set of the standard branches produce a non-zero amplitude read signal sample and a second set of the standard branches produce a zero amplitude read signal sample, wherein each state of the standard state diagram has two standard branches feeding into that state, wherein each standard branch has an associated error metric;
• the adaptive trellis further including an estimating partial response maximum likelihood (PRML) trellis, the estimating PRML trellis associated with an estimating state diagram having estimating branches between states, the estimating branches produce estimated read signal samples that adapt to the amplitude of the signal, wherein each state of the estimating state diagram has two estimating branches feeding into that state, wherein each estimating branch has an associated error metric;
• wherein each state of the standard state diagram has two estimating branches feeding into that state;
• wherein each state of the estimating state diagram has two standard branches feeding into that state;
• wherein the states are connected by paths formed by branches between the states, wherein a path error metric is the sum of all branch error metrics for all of the preceding branches that form the path, wherein the path arriving at one state having the lowest path error metric survives to be identified as the expected path to the next state for reading the signal.

An adaptive trellis detection method adaptable to read a signal having signal amplitude dropouts, the adaptive trellis detection method comprising:

• providing a standard partial response maximum likelihood (PRML) trellis;
• associating the standard PRML trellis with a standard state diagram having standard branches between states, wherein a first set of the standard branches produce a non-zero amplitude read signal sample and a second set of the standard branches produce a zero amplitude read signal sample, wherein each state of the standard state diagram has two standard branches feeding into that state, wherein each standard branch has an associated error metric;
• and;
• providing an estimating partial response maximum likelihood (PRML) trellis;
• associating the estimating PRML trellis with an estimating state diagram having estimating branches between states, wherein the estimating branches produce estimated read signal samples that adapt to the amplitude of the signal, wherein each state of the estimating state diagram has two estimating branches feeding into that state, wherein each estimating branch has an associated error metric;
• wherein each state of the standard state diagram has two estimating branches feeding into that state;
• wherein each state of the estimating state diagram has two standard branches feeding into that state;
• connecting the states by paths formed by branches between the states, wherein a path error metric is the sum of all branch error metrics for all of the branches that form the path;
• and
• identifying a path arriving at one state having a lowest path error metric as the expected path to the next state for reading the signal.