Hybrid Flowshop Scheduling Problems (HFS) are among the most realistic machine sequencing models there are. These problems deal with the scheduling of a set of jobs through a set of stages where at each stage, multiple parallel machines exist. We consider a number of extensions to this problem. First of all, we take into account the existence of sequence dependent setup times which are prevalent in practice. Second, we optimize the total weighted earliness and tardiness, but not from a due date, rather a due date window. As of late, simple local search methods have been successfully employed in scheduling problems. Their simplicity and ease of extension is remarkable. Despite being simple, state-of-the-art results are often achieved. Among these methods, Iterated Greedy (IG) has shown great promise. We elaborate on these ideas and develop new local search procedures that are much faster than existing ones, allowing many more iterations per unit of time. A parameter-less IG, which does not require calibration, is presented. In order to test it, we carry out a comprehensive computational campaign. We test both the HFS without setup times and due date windows (HFSDDW), as well as the same problem with the additional consideration of setup times (HFSDDW-SDST). A total of 14 competing methods have been carefully reimplemented, adapted for these problems, calibrated and tested against the proposed IG. The complete experiments, with more than 12,000 instances, together with the statistical analyses, indicate that the proposed IG method produces state-of-the-art results for the problems considered.