Smart grids aim at evolving the traditional electrical grid system by making increasing use of sophisticated control and communication network technology, to properly deal with the high penetration of controllable assets, such as distributed generators and flexible loads, and their associated challenges. Since electrical grids are critical infrastructures, control strategies regulating their operation need to face both efficiency and cost aspects, as well as resilience related ones in order to assure reliable service. Keeping the focus on the medium voltage control functionality, in this paper we present a study of the event-triggered voltage control algorithm to satisfy voltage control requirements. In particular, the opportunity to introduce soft bounds to improve preserving voltage values in bounds notwithstanding the effect of control delay is explored in a variety of scenarios, including fault presence due to attacks to the communication network. Through a developed stochastic model-based framework, quantitative analyses are performed on a realistic MV testbed grid, to demonstrate the feasibility and utility of the proposed contribution.