Automatic probing system with machine learning algorithm

Abstract

This paper presents a novel probe alignment system that implements machine learning methods. The developed measurement system is demonstrated at frequencies ranging from 100 MHz to 125 GHz. The measurement system measures the S-parameter with slightly shifting the probe. The S-parameter is expressed by ten trigonometric function orders using the linear least mean square method. The coefficient of each function order is used to calculate the local outlier factor (LOF). Then, the calculated LOFs are used to detect the probe touchdown, and the LOF threshold is preliminarily determined using training data. The accuracy of probe positioning was compared with that of a conventional automatic probing technique, and the difference in the probe position between the two techniques was determined to be approximately 1 $\mu$ m.