Online Quality Evaluation of Ultrasonic Wire Bonding Using Input Electrical Signal of Piezoelectric Transducer

Current bond quality evaluation technologies mainly depend on offline methods such as visual inspection of the appearance of bonds and batch destructive testing after bonding. In this paper we propose a new method for online evaluation of bonding quality. The method directly exploits the input electrical signal of piezoelectric transducer in ultrasonic wire bonder to evaluate the bonding quality. We present a feature extraction method in which the signal envelope is firstly obtained and then the envelope is separated into three sections, namely signal raise, smooth and attenuation, to characterize the transient property of the bonding process. The back-propagation artificial neural network (ANN) is used to establish the nonlinear relationships between the features and the bonding conditions. The experimental result shows that the proposed method can accurately identify four bonding conditions including wire missing, good bond, peeled off bond and non-stick bond.