The application of absolutely calibrated piezoelectric (PZT) sensors is increasingly used to help interpret the information carried by radiated seismic waves in laboratory and in situ seismology. In this paper, we revisit the methodology based on the finite element method (FEM) to characterize PZT sensors. The FEM-based modelling tool is used to numerically compute the Green’s function between a ball impact source, and an array of PZT sensors used to detect laboratory-induced elastic stress wave propagation excited by a unit step force-time function. Realistic boundary conditions that capture the experimental conditions, are adopted to physically constrain the problem of elastic wave propagation, reflection and transmission in/on the elastic medium. The modelling methodology is first validated against the reference approach (generalized ray theory) and is then extended down to 1 kHz where elastic wave reflection and transmission along different types of boundaries are explored. We find the Green’s functions calculated for realistic boundaries have distinct differences between commonly employed 2 Wu et al., idealized boundary conditions, especially around the anti-resonant and resonant frequencies. Unlike traditional methods that use singular ball drops, we find that each ball drop is only partially reliable over specific frequency bands. We demonstrate, by adding spectral constraints, that the individual instrumental responses are accurately cropped and linked together over 1 kHz to 1 MHz after which they overlap with little amplitude shift. This study finds that ball impacts with a broad range of diameters as well as the corresponding valid frequency bandwidth and equivalent seismic magnitude, are necessary to characterize broadband PZT sensors from 1 kHz to 1 MHz. This work bridges the gap between microcrack/damage mechanics and laboratory/in situ acoustic emissions (AEs) by unraveling sources in terms of the physics that generates AE signals.