Recent expansion of inverter-based renewable generation (RG) has introduced unprecedented stability issues. One major challenge stems from tripping RGs offline during disturbances based on predefined operation thresholds (ride through curves) on voltage magnitudes and frequency at the point of common coupling (PCC). The current practice uses a single set of thresholds for the whole year across all RGs in the system. This could result in loss of these resources during critical times, particularly endangering system voltage security, i.e a significant reduction in the load margin (LM). On the other hand, relaxing these limits could result in them staying online even during islanding scenarios, which is a safety concern to the restoration crew. Thus, an approach to find minimal relaxations to these limits to enhance voltage security is proposed here. This process first constructs an optimization model aiming to minimize adjustments on RG voltage limits to reach anticipated LM increase and then solves it with a multi-stage screening method including base power-voltage (P-V) curve acquisition, critical RGs identification, successive screenings of candidate solutions based on sensitivities, and optimal solutions validation. This largely increases the scalability of the algorithm, which is demonstrated on the IEEE 14-bus, 118-bus systems, and a 500-bus system.