The ball bearings of an aero-engine are key parts that frequently fail, and it is very important to effectively carry out fault diagnosis of the ball bearings. However, in the present research work, the ball bearing faults characteristics are extracted mainly from the bearing house signals, it is well known that usually only the casing signals can be measured in practical aero-engine test, and the ball bearing faults characteristics will greatly weaken after transmitting to the casing from the bearing house, therefore, it is very important to extract the fault characteristics of ball bearings from casing vibration signals for the ball bearing fault diagnosis in the practical aero-engine. In this study, simulation experiments for ball bearing faults are conducted using two rotor experimental rigs with casings. In addition, by means of the impulse response method, the transfer characteristics from the ball bearings to casing measuring points are measured, and a sensitivity analysis is performed. Faults are created on the inner ring, outer ring, and ball of the ball bearings in the two experimental rigs. The ball bearing experiments are carried out, and the fault features are extracted by means of a wavelet envelope analysis. The experimental results indicate that, with high connection stiffness between the bearing house and the casing, there is little vibration attenuation. However, with low connection stiffness, the vibration attenuation is great. After the impulse vibrations caused by the ball bearing faults are transmitted to the casing, the casing vibration is very weak and is often submerged in other signals. However, the ball bearing fault characteristic frequencies can still be effectively extracted from the weak casing vibration signals by using a wavelet envelope analysis. The research results in this study provide an experimental basis for a ball bearing fault diagnosis based on a casing test signal from a practical aero-engine.