This paper addresses the optimization problem of printed circuit board (PCB) assembly scheduling for surface mount placement (SMP) machines with multiple heads. Most of the related studies have oversimplified the practical problems and limited their focus to three sub-problems: the feeder assignment problem (FAP), the nozzle setup problem, and the component pick-and-place sequence problem (PAPSP). To extend the range of problems that can be addressed, in this study, we propose a hybrid algorithm to consider the component height, the pick-and-place restrictions, and simultaneous pickup restrictions. A roulette wheel (RW) is designed to generate an automatic nozzle changer (ANC) sequence, with the nearest neighbor search (NNS) used to find an initial placement path. Furthermore, we apply the 2-optimization (2-opt) method to improve the placing sequence. Practical PCB datasets provided by an EVEST EM-780 multi-head SMP machine are adopted as experimental examples. Experimental results show that the proposed algorithm effectively enhances the productivity of the multi-head SMP machine and reduces the cycle time in PCB assembly.