Most of the current day computer applications require Pseudo Random Number Generators (PRNGs) that generate sequence of pseudo-random numbers efficiently inside CPUs or GPUs and with good theoretical quality merits. Uniformity is the most crucial quality criterion of PRNG; in the case of linear PRNG, it is measured using a theoretical quality measure called equidistribution. This work targets to design light-weight PRNGs with cellular automata (CAs) having maximal equidistribution and close to maximal period. For that, first the theoretical quality merit of linear cellular automaton (CA)-based PRNGs whose sizes are close to the computer word size (32 and 64) are studied and it is observed that they lack in uniformity. To address this problem, this work introduces light-weight two-component combined CAs-based PRNGs with time spacing that achieve maximal equidistribution. We show that, the proposed combined CA-based PRNGs also pass almost all tests in the standard statistical testbeds, and the results are comparable to the existing state-of-the-art linear PRNGs.