Differential Photoacoustic (PA) sensors are frequently mentioned in the literature for their high sensitive and favorable signal-to-background ratio. Nevertheless, their operation principles are usually described briefly. In this article we provide a comprehensive analysis of the acoustic behavior of these sensors through both computational simulation, by using the Finite Element Method (FEM) and experimental analysis. Through two independent analyses, we identified acoustic modes supported by the sensor and those effectively excited by the incidence of a Gaussian laser beam through the PA sensor. Our findings revealed that the longitudinal (in-phase) mode is efficiently excited by the optical windows heating and strongly depends on their distance, as experimentally verified. Conversely, the ring (out-of-phase) mode is efficiently excited by the CH4 sample heating. Finally, based on the simulated sensor model, two differential PA sensors were constructed using precision machining in metal and 3D printing in photosensitive liquid resin. The performance of both sensors was evaluated in detecting traces of CH4. While the metal PA sensor exhibited slightly better precision (σCmetal=±0.160ppmV and σC,printed=±0.541ppmV) and detection limit (LODmetal = 0.358ppmV and LODprinted = 1.05ppmV), both sensors demonstrated comparable sensitivity (Smetal = 1.07 ± 0.02μV/ppmV and Sprinted = 0.97 ± 0.06μV/ppmV) and can be employed for CH4 trace detection. Our study contributed to a better comprehension of the physical mechanisms of this sensor type and highlights the potential of the proposed computational modeling, in conjunction with the 3D printing technique, to design and develop highly sensitive and selective differential PA sensors for detecting trace gases.