Wind is an abundant renewable energy source. One form of utilizing wind energy is by using wind turbines. However, components in wind turbines are often damaged, and one of the most-frequently damaged components is bearing. The role of bearing is very important in maintaining the performance of the entire engine. Damaged bearings will have an impact on the performance of the windmill. Broken bearing detection research has been carried out on rotary machines, while the detection of damaged bearings in windmill is still minimally conducted. Therefore, this study is aimed at detecting early bearing damage in windmill using vibration analysis by applying the frequency spectrum feature and envelope analysis. This research used a bearing in normal condition and multi-fault bearings that were intentionally damaged with a size of 2 mm deep bearing and damaged at a 0.7 mm wide bearing. Multi-faults were damaged in the outer trajectory and the inner trajectory simultaneously and the detection of broken bearings was carried out using spectrum envelope analysis. The Measurements were carried out using a motor drive as a wind simulation with a shaft speed of 1200 RPM. The bearings used in this research were Self Aligning Double Row, TAM Brand, and 1208K series. The frequency spectrum does not show damaged frequency amplitude on the bearing, but there is high frequency amplitude of the shaft rotating speed. The frequency spectrum on the damaged bearing shows a broken frequency of the bearing of the outer trajectory and the inner trajectory followed by 1X to 3X harmonics. However, the damaged frequency of bearing amplitude is still low and covered by the frequency of other components. The envelope method can extract the impact with very low energy and give rise to damaged frequency amplitude on the bearing. Thus, the identification of damaged bearings will be very clear with the emergence of frequencies from broken bearings in the outer and inner tracks followed by 1X to 3X harmonics.