Mapping DSP algorithms to FPGA/VLSI circuits is an important issue in Application-Specific Array Processor design. Since a DSP algorithm can be abstracted as a graph where each node is a shift-invariant DG (Dependence Graph) and the edges denote the data flow, it is possible to map a DSP algorithm to a set of processor arrays with some interface circuits. The interface design depends on the projection/scheduling vectors used on the two corresponding shift-invariant DGs and the interfacing cost is very significant when a lot of delays are necessary or when the processor operations are relatively inexpensive in terms of area. Therefore, when selecting these vectors in a design environment, the effect on the interface circuit must be accurately computed. In this paper, various interface circuit designs are presented and categorized based on the data conversion requirement. An algorithm to select a design from many design options to minimize the cost is also described.