Data recovery circuits using oversampling for best data sample selection

A method for recovering the data in an incoming data stream with jitter and skewed with respect to a reference clock by selecting the best data sample in a set of data samples obtained by oversampling said incoming data stream comprising the steps of:

• oversampling the incoming data stream with the n phases of the reference clock to generate a set of n data samples;
• detecting the position of the data edges (or transitions) in the set of data samples and creating a map of n bits, representing the data edge information, i.e. the edge positions, by assigning a determined binary logic value to said data edge positions and the opposite value otherwise;
• memorizing the map of n bits wherein the bits are at a first binary logic level for detected data edges and at a second binary logic level otherwise and wherein the last bit map position is considered as adjacent to the first bit map position (and conversely);
• determining which data sample is the farthest from the data edges, i.e. which data sample is the best candidate for the data to recover;
• validating said selection signals to generate validated selection signals to avoid false determination due to jitter and skew;
• memorizing said validated selection signals;
• and,
• using said memorized validated selection signals and said data samples to select the best data sample to represent the recovered data.