Breakthrough detection is a crucial task to reduce the risks of damaging soft tissue bone drilling operations during orthopedic surgery. Conventional drills are not equipped with this function while the recent literature has offered this capability with high cost and complex modification needs. In this study, a new breakthrough detection approach based on closed-loop control characteristics of the drilling operation is proposed. A feature set containing closed-loop signals and force sensor data is created to train K-Nearest and Ensemble Classifier for breakthrough detection tasks with drilling the synthetic bone model and animal bone with a robot manipulator. The best accuracy of breakthrough detection with only closed-loop control signal attributes is achieved as 96.9 +/- 0.8% for the synthetic bone model and 98.1 +/- 0.2% for sheep femur bone. Breakthrough detection delay which included sampling and operation time of the method guarantees that the drill bit would stop with acceptable breakthrough range of 1.0413 mm. The proposed method can be used to detect breakthrough and also to estimate the state of the drill bit in robotic orthopedic bone drilling processes using only closed-loop signals so that it would be no need to use extra high-cost sensors.