Discrete Element Method (DEM) simulations suffer from two key limitations: the absence of a standardized environment for comparing DEM models, and the reliance on potentially flawed particle packing generators (PPGs) that mask calibration errors. To overcome these issues, we introduce a principle-based PPG framework for representative volume elements (RVEs), employing radius expansion with servo control. This approach guarantees packings with rigorous isotropy, homogeneity, and precisely controlled porosity and stress, thereby circumventing the need for extensive validation. Implemented in our open-source software, DEM-Gen, this framework provides a reliable foundation for material modeling. We showcase the methodology through numerical tests on sand and glass beads, and discuss how it enables fair DEM model comparisons, enhances parameter reusability, reduces calibration workloads, and minimizes error masking, ultimately fostering more accurate and repeatable DEM simulations.