Modular and reconfigurable manipulators have gained popularity especially in the service sector, where the use of customized configurations has increased. Adaptable modular designs have come into advances in achieving required configuration of a robotic manipulator. As reported in literature, various designs of the modules mainly with conventional configurations are presented and a few are reported with unconventional adjustments. To cater the non-repetitive applications, this paper presents an optimal architectural design for unconventional parameters for customized reconfigurability. This lighter and easier-to-connect version is also applicable to n-DoF and unconventional robotic parameters. Architecture Prominent Sectioning (APS) strategy is proposed which assumes an architecture as a set of point masses and optimally relocate components with respect to the minimization of the joint torques. Modules are considered to be 3D printable using poly-lactic acid (PLA), a thermoplastic material, and thus light in weight. The new modular architecture design is validated through the assemblage of conventional/unconventional configurations using two types of modules namely Heavy (H) and Light (L). Along with that, worst torque analyses for the different configurations have been done in order to provide a strategy for assembly combinations. A comparative study is presented based upon the payload-to-weight ratio, involving other reported architectures.