As airborne, correlation-based, ultrasonic sensors are becoming more widely used for robotic applications, it becomes increasingly important to have a good understanding of how the transducer’s filtering properties affect the performance of the sensor system. Using well-known results from acoustics, a simple, yet accurate, filtermodel for the Polaroid transducer is described. In the course of this derivation the reason for the accuracy of the often used “moving piston” approximation is also clarified. This filtermodel is then used to analyze the characteristic “peak-doubling” from which the correlation based range sensor used in the tri-aural sensor suffers. The predicted errors are compared with actual measurements by the real sensor system and are found to be in good agreement. Finally, the use of broad beamwidth transducers to avoid these errors is proposed, and it is and argued that this approach has some distinct advantages compared to alternative solutions.