The performance limits of robotic joints are largely determined by two types of bearings: cross-roller bearings and angular contact ball bearings. If the preloading process is not executed correctly, even the highest precision ratings specified on the engineering drawings will be effectively nullified during assembly. This article draws on engineering practice to clarify the principles of preloading for these two types of bearings, the methods for achieving it, and techniques for on-site verification.
I. The Essence of Preloading: Operating Bearings with Negative Clearance
1.1 What Is Preloading?
Rolling-element bearings typically come with positive internal clearance when they leave the factory. Preloading involves applying a load to the bearing during assembly to eliminate this clearance, causing elastic deformation between the rolling elements and the raceways of the rings, even in the absence of an external load. At this point, the measured clearance becomes zero or even negative; the inner and outer rings are pressed firmly against the rolling elements, leaving the bearing in a "pre-stressed" or "tightly supported" state.
Bearing manufacturers refer to this condition as "negative clearance" or "preloading." It is important to note that preloading is not merely about "squeezing" the bearing tighter-the contact between the rolling elements and the raceways is elastic. In essence, preloading utilizes a controlled, initial elastic deformation to minimize additional deformation when the bearing is subjected to external loads during operation.
1.2 Why Is Preloading Necessary?
The requirements for bearings in robotic joints can be summarized into four points, which correspond to the four primary benefits of preloading:
Increased Rigidity. The contact between rolling elements and raceways can be modeled as a spring. When clearance exists, only a few rolling elements within the load zone participate in supporting the load. Preloading expands the load zone and increases the number of load-bearing elements, significantly reducing deformation under a given external load. For instance, in ball bearings, preloading can increase rigidity by two to three times compared to a non-preloaded state. Robotic joints require a support platform that offers "instant response to force with near-zero deformation"; without preloading, achieving high repeatability is impossible.
Elimination of Backlash (Hysteresis). Negative clearance eliminates backlash in the direction of rotation, ensuring that positional errors at the end of a robotic arm are not further amplified at the joints. For bearings located at the output of a speed reducer, this factor directly determines whether the overall system backlash remains within the design specifications.
Prevention of slippage. Under light-load or high-speed operating conditions, the inertial force of the rolling elements may exceed the frictional force at the raceway contact points. This can cause the rolling elements to slip against the raceways, leading to wear and noise. Preloading ensures a sufficient frictional margin to drive the rolling elements, suppressing both gyroscopic and spin-induced slippage-which is precisely why angular contact ball bearings operating at high speeds must be preloaded.
Vibration suppression and wear compensation. Preloading improves the damping characteristics of a bearing, helping to suppress axial vibration and resonance-induced noise. Furthermore, if wear occurs during long-term operation-which would otherwise increase clearance-the initial preload can provide a degree of compensation.
1.3 The Trade-offs of Preloading
More preload is not necessarily better. Excessive preload leads to increased frictional torque, reduced transmission efficiency, higher heat generation, the breakdown of the lubricating film, and a significant reduction in service life. Conversely, insufficient preload results in inadequate rigidity, backlash (hysteresis) at the joints, increased operating noise and vibration, and precision drift. For high-precision, light-load applications like robotics, an optimal preload exists within a narrow window: it must be sufficient to eliminate clearance while maintaining a minimal, stable load-straying outside this range leads to various operational issues.
II. Crossed Roller Bearings: Preloading During Assembly
2.1 Structure and Load-Bearing Characteristics
In a crossed roller bearing, cylindrical rollers are arranged in a V-shaped raceway at 90° angles, with the axes of adjacent rollers positioned perpendicular to one another; this configuration creates line contact. This design allows a single bearing to simultaneously support radial loads, bidirectional axial loads, and overturning moments. Furthermore, its thin cross-section and lightweight nature make it ideally suited for the complex loading conditions found in the shoulder and waist joints of robots, as well as at the output stages of RV and harmonic speed reducers. Spacers (or cages) within the raceway guide the adjacent rollers, ensuring they maintain their crossed orientation without shifting out of position. Cross-sectional view of a crossed roller bearing (with rollers arranged in a 90° crisscross pattern).
Because the rolling elements are constrained in all four directions, crossed roller bearings are often described as bearings whose rotational precision is "achieved through preload." The preload design directly determines the upper limit of joint rigidity and the level of backlash.
2.2 Two Approaches to Preload
Preloading for crossed roller bearings is categorized into radial preload and axial preload.
Radial preload is achieved by inducing the expansion of the inner ring through the axial displacement of a tapered bore. A typical application is the use of tapered-bore bearings in machine tool spindles; however, this method is rarely used for crossed roller bearings themselves. In robotics, the standard approach is axial preload: a split-ring design-where either the inner or outer ring is divided into two halves-allows the two halves to be locked together axially during assembly. This forces the rollers into the V-shaped raceways, creating "negative clearance" (interference). The amount of preload is controlled by the axial displacement or the tightening torque applied; therefore, for the end-user, the preload is established during assembly on the workbench, rather than being fixed at the factory.
External appearance and internal structure of a crossed roller bearing (featuring solid/integral rings, spacers, and seals).
Alternatively, there are designs with solid (integral) rings, such as the THK RBU series. For these bearings, the preload is set during the manufacturing phase by grinding the end faces. There is no room for adjustment in the field; during assembly, the components simply need to be tightened to a specified torque, and the adequacy of the preload is verified by spot-checking the rotational torque. When selecting a bearing, it is crucial to confirm whether the model is "adjustable" or "fixed," as the assembly procedures for each are entirely different.
2.3 Achieving the Correct Preload in the Field
The adjustment process for split-ring designs follows two key principles: the tightening sequence and the verification method.
Regarding the tightening sequence: first, tighten all bolts in a diagonal pattern until they make contact (snug-tight), then tighten them to the target torque in 2–3 incremental steps to ensure even loading. After each incremental application of force, rotate the bearing at least one full revolution and observe any changes in the rotational resistance torque-this is the primary criterion for determining whether the preload is correctly set.
Crossed roller bearings for robot joints (flanged type with mounting holes, designed for direct attachment to the housing)
Flanged crossed roller bearings with mounting holes-commonly used in robot joints-are secured directly to the housing via bolts. This configuration is a classic example of a "preload-upon-assembly" design. There are three common methods for verifying the preload level:
Rotational resistance torque method: The most widely used approach. Since the preload level corresponds directly to the starting frictional torque, it can be measured using a torque gauge or assessed by manual "feel." If the resistance torque falls within the range specified by the manufacturer, the preload is considered correct.
Axial displacement method: This involves first calibrating the relationship between the tightening travel (displacement) and the resulting preload force, then setting the preload based on a specific displacement value. This method is well-suited for mass production.
Temperature rise method: If the bearing housing exhibits an unusually high temperature rise during the break-in (run-in) period, it often indicates that the preload is too high; in such cases, the preload should be reduced by one increment and readjusted.
