2. Ways to avoid failure

The best way to solve the problem of bearing failure is to avoid failure. This can be achieved by considering key performance characteristics during the selection process. These characteristics include noise, starting and running torque, stiffness, non-repetitive runout, and radial and axial play. 
Torque requirements are determined by lubricant, cage, ring quality (roundness of bends and surface finish), and whether seals or shields are used. The viscosity of the lubricant must be chosen carefully, as unsuitable lubricants can generate excessive torque, especially in small bearings. In addition, the noise characteristics of different lubricants are also different. For example, grease produces a little more noise than oil. Therefore, lubricants should be selected according to different uses. 
During bearing rotation, if there is a random eccentricity between the inner and outer rings, a non-repetitive runout (NRR) very similar to cam motion occurs. The dimensional error of the cage and the eccentricity of the bearing ring and the ball can cause NRR. Unlike repetitive runout, NRR has no way to compensate.
In the industry, bearings of different types and accuracy grades are generally selected according to specific applications. For example, when minimum runout is required, the non-repetitive runout of the bearing cannot exceed 0.3 microns. Likewise, machine tool spindles can only tolerate minimal runout to ensure cutting accuracy. Therefore, bearings with less non-repetitive runout should be used in machine tool applications. 
In many industrial products, contamination is inevitable, so seals or shields are often used to protect bearings from dust or dirt. However, due to the movement of the inner and outer rings of the bearing, the sealing of the bearing cannot be perfect, so the leakage and pollution of lubricating oil is always an unsolved problem. 
Once the bearing is contaminated, the lubricant will deteriorate and the operating noise will also become louder. If the bearing overheats, it will seize. When contamination is between the balls and the bearing rings, it acts like abrasive particles between metal surfaces, causing bearing wear. Using seals and shields to keep out dirt is one way to control contamination. 
Noise is an indicator that reflects the quality of bearings. The performance of bearings can be represented by different noise levels.
Noise analysis is performed with an Anderson meter, which is used in bearing production to control quality and to analyze failed bearings. A sensor is attached to the outer ring of the bearing, while the inner ring rotates on the mandrel at 1800 r/min. The unit of measurement noise is anderon. That is, the bearing displacement expressed in um/rad. 
Based on experience, observers can discern tiny defects from sound. For example, dust produces an irregular crackling sound; ball scratches produce a continuous pop, which is the most difficult to identify; inner ring damage usually produces a continuous high frequency noise, while outer ring damage produces a Intermittent sound. 
Bearing defects can be further identified by their frequency characteristics. Usually bearing defects are divided into three bands: low, medium and high. Defects can also be identified by the number of irregularities per bearing revolution. 
Low frequency noise is the result of irregular changes in the long waveband. Such irregular changes can occur 1.6 to 10 times per revolution of the bearing, and they are caused by various interferences (such as dimples on the raceway of the bearing ring). Perceivable pitting is a manufacturing defect that is formed during the manufacturing process by clamping the multi-jaw chuck too tightly. 
The characteristic of intermediate frequency noise is that the irregular changes occur 10 to 60 times per rotation of the bearing. This defect is caused by vibrations that occur during the grinding process of the bearing rings and balls. High-frequency irregular changes occur 60 to 300 times per rotation of the bearing, which indicates that there are dense vibration marks or large areas of roughness on the bearing. 
By classifying a bearing by its noise characteristics, the user can determine the noise level of the bearing in addition to the ABEC standard used by most manufacturers. The ABEC standard only defines dimensional tolerances such as bore, outer diameter, runout, etc. As the ABEC level increases (from 3 to 9), the tolerance becomes progressively smaller. However, the ABEC grade does not reflect other bearing characteristics, such as bearing ring quality, roughness, noise, etc. Therefore, the division of noise levels contributes to the improvement of industry standards.