Abstract: A method of generating an encoded signal from a sequential stream of digital data, where the encoded signal has a non-power carrying null state and a power carrying active state with two opposing polarities. Logical one bits are distinguished from logical zero bits by inverting the encoded signal's polarity at the start of only the logical one bits. The encoded signal is set to the active state during a bit set-up period before, and held in the active state during a bit hold period after each polarity inversion. At other times the encoded signal is set to the null state. The method may include the addition of equalization pulses during strings of consecutive logical zero bits to keep the encoded signal from remaining in the null state for extended periods. Each equalization pulse may be preceded by an equalization set-up period and followed by an equalization hold period where the encoded signal is in the active state.

Abstract: A method of write equalization in digital data recording systems comprising the steps of: providing a digital data signal to be recorded on recordable media, wherein the digital data signal includes 1s and 0s spaced at time interval T with 1s represented by a signal transition between first and second signal levels and 0s represented by no signal transition, and in which there is a minimum of d+1 intervals T between transitions; providing write equalization pulses of width WT for each 0 starting with the (d+1)th 0 following a 1; and causing a transition of a write equalization pulse adjacent to a 1 to occupy the same position as the transition of the 1 so that there is no transition for the 1. In one embodiment, the initial transition of the first added 0 pulse following a 1 is advanced in time so that it occupies the same position as the transition for the 1.

Abstract: An input string of binary data bits which is a random sequence of 0s and 1s is selectively spaced encoded into an output string of binary channel bits which is a sequence of 0s and 1s. An input string of binary data bits are recursively encoded according to run-length selection means having rate p/q where p is the number of input data bits that are encoded into q-channel bits. The channel bits obey constraints such that the encoding specifies an ordered n element list of integers that are the lengths of the selected runs of consecutive zeros, denoted b.sub.0 -1, b.sub.1 -1, . . . , b.sub.n-1 -1, and such that the element list of integers does not contain all the integers between the minimum element, b.sub.0, and the maximum element, b.sub.n-1.