In the field of precision mechanical transmission, the rotary vector reducer stands out as a critical component due to its compact design, high stiffness, and large reduction ratio. It is extensively utilized in applications such as industrial robots and radar systems, where low vibration levels are paramount to avoid issues like chatter. As a researcher focused on dynamic systems, I have undertaken a comprehensive study to explore the torsional vibration characteristics of rotary vector reducers. This article presents an in-depth analysis of the natural frequencies, modal shapes, and sensitivity of these frequencies to key design parameters, with the goal of enhancing the dynamic performance and reliability of rotary vector reducers.
The rotary vector reducer is a compound transmission mechanism that combines a K-H type planetary gear system with a K-H-V type cycloidal drive. This unique configuration contributes to its superior performance, but it also introduces complex dynamic behaviors. Previous research has primarily focused on transmission accuracy, with limited attention paid to torsional vibration dynamics. In my work, I address this gap by developing a refined torsional dynamics model that accounts for additional degrees of freedom and the periodic variation of the cycloidal gear’s eccentric angle. By doing so, I aim to provide insights that can inform the design and optimization of rotary vector reducers for vibration-sensitive applications.
The transmission principle of a rotary vector reducer involves a two-stage process. The first stage consists of a K-H type differential gear train with a sun gear, multiple planetary gears (typically two or three), and a planet carrier. The second stage is a K-H-V type planetary system comprising crank shafts, cycloidal gears, pin teeth, and the planet carrier. The crank shafts are offset uniformly and connected to the planetary gears, while the cycloidal gears engage with the fixed pin teeth. During operation, input rotation from a motor drives the sun gear, which transmits motion to the planetary gears for the first reduction. The planetary gears then drive the crank shafts, causing the cycloidal gears to undergo eccentric motion (revolution). Due to the fixed pin teeth, the cycloidal gears experience reverse rotation (rotation relative to the carrier), which is transferred via the crank shafts to the planet carrier as output rotation. This intricate mechanism underscores the need for a detailed dynamic model to capture its vibrational behavior.
To accurately model the torsional vibration of a rotary vector reducer, I have developed a modified lumped-parameter model with 13 degrees of freedom. This model improves upon existing pure torsional models by incorporating the tangential translational degrees of freedom associated with the revolution of crank shafts and cycloidal gears, as well as the periodic variation of the eccentric angle of the cycloidal gears. The assumptions underlying this model include: treating components as lumped masses with coincident geometric and mass centers, using mean values for stiffnesses to avoid nonlinearities, neglecting friction and damping effects, and approximating the cycloidal-pin tooth engagement as a concentrated force. Additionally, the bending deformation of crank shafts is ignored due to their high stiffness relative to bearings.
The coordinate system is defined with the pin wheel center as the origin, and dynamic coordinates are attached to the planet carrier and cycloidal gears to account for eccentric motion. The 13 degrees of freedom encompass nine torsional displacements (for the input shaft, sun gear, two planetary gears, two crank shafts, two cycloidal gears, and planet carrier) and four translational displacements (for the tangential motion of the two crank shafts and two cycloidal gears). The equations of motion are derived using Newton’s second law, considering forces such as gear mesh forces, bearing support forces, and interactions between crank shafts and cycloidal gears. For instance, the relative displacement between a crank shaft and cycloidal gear in the tangential direction is expressed as:
$$ \delta_{\eta ij} = \begin{cases} u_{hi} – x_{hi} \cos \beta_i + u_{cj} \cos \beta_i – x_{cj} & \text{for } j=1 \\ u_{hi} + x_{hi} \cos \beta_i – u_{cj} \cos \beta_i – x_{cj} & \text{for } j=2 \end{cases} $$
where \( u_{hi} \) and \( u_{cj} \) are torsional displacements, \( x_{hi} \) and \( x_{cj} \) are translational displacements, and \( \beta_i \) is the eccentric angle of the cycloidal gear, given by \( \beta_i = \beta + (i-1)\pi \). The resulting force between them is \( F_{hicj} = k_{hc} \delta_{\eta ij} \), with \( k_{hc} \) as the bearing stiffness. Similar derivations yield other interaction forces, leading to the system of free vibration equations in matrix form:
$$ \mathbf{M} \ddot{\mathbf{X}} + \mathbf{K} \mathbf{X} = \mathbf{0} $$
Here, \( \mathbf{M} \) is the mass matrix, \( \mathbf{K} \) is the stiffness matrix, and \( \mathbf{X} \) is the vector of generalized coordinates. The matrices incorporate parameters such as moments of inertia, masses, and stiffnesses for all components. For clarity, I summarize the key parameters in the following table:
| Parameter | Symbol | Description |
|---|---|---|
| Moment of inertia (input shaft) | \( J_a \) | Rotational inertia of the input shaft |
| Moment of inertia (sun gear) | \( J_s \) | Rotational inertia of the sun gear |
| Moment of inertia (planetary gear) | \( J_p \) | Rotational inertia of each planetary gear |
| Moment of inertia (crank shaft) | \( J_h \) | Rotational inertia of each crank shaft |
| Moment of inertia (cycloidal gear) | \( J_c \) | Rotational inertia of each cycloidal gear |
| Moment of inertia (planet carrier) | \( J_o \) | Rotational inertia of the planet carrier |
| Mass (planetary gear) | \( m_p \) | Mass of each planetary gear |
| Mass (crank shaft) | \( m_h \) | Mass of each crank shaft |
| Mass (cycloidal gear) | \( m_c \) | Mass of each cycloidal gear |
| Stiffness (input shaft torsion) | \( k_{as} \) | Torsional stiffness of the input shaft |
| Stiffness (crank shaft torsion) | \( k_{ph} \) | Torsional stiffness of the crank shaft |
| Stiffness (sun-planet mesh) | \( k_{sp} \) | Mesh stiffness of the involute gears |
| Stiffness (cycloidal-pin mesh) | \( k_{cd} \) | Mesh stiffness of the cycloidal-pin teeth |
| Stiffness (bearing support) | \( k_{hc} \) | Support stiffness of the bearing between crank shaft and cycloidal gear |
| Stiffness (carrier bearing) | \( k_{oh} \) | Support stiffness of the bearing on the planet carrier |
| Eccentric distance | \( a \) | Offset distance for the crank shafts |
| Base circle radius (sun) | \( r_s \) | Base circle radius of the sun gear |
| Base circle radius (planet) | \( r_p \) | Base circle radius of the planetary gears |
| Distribution radius (crank) | \( r_h \) | Distribution radius of the crank shafts |
| Pitch circle radius (cycloidal) | \( r_c \) | Pitch circle radius of the cycloidal gears |
The natural characteristics of the rotary vector reducer are determined by solving the eigenvalue problem derived from the free vibration equation. The characteristic equation is:
$$ (\mathbf{K} – \omega_i^2 \mathbf{M}) \boldsymbol{\phi}_i = \mathbf{0} $$
where \( \omega_i \) is the \( i \)-th natural frequency and \( \boldsymbol{\phi}_i \) is the corresponding normalized mode shape. I computed these for an RV-6 type rotary vector reducer with specific parameters: sun gear teeth \( Z_s = 10 \), planetary gear teeth \( Z_p = 34 \), module \( m = 1 \, \text{mm} \), pressure angle \( \alpha = 20^\circ \); cycloidal gear teeth \( Z_c = 29 \), pin teeth \( Z_b = 30 \), shortcut coefficient \( K = 0.675 \), eccentric distance \( a = 0.9 \, \text{mm} \), pin distribution radius \( R = 40 \, \text{mm} \), input speed 3000 rpm, and rated load 58 N·m. The moments of inertia and masses were obtained via 3D modeling, while stiffnesses were calculated using established methods, with mesh stiffnesses averaged over time.
The results reveal that the natural frequencies exhibit distinct patterns based on the eccentric angle \( \beta \) of the cycloidal gears. The odd-order natural frequencies remain constant regardless of \( \beta \), while the even-order frequencies vary periodically with \( \beta \), following a trigonometric-like pattern. This behavior is linked to the system’s symmetry and can be explained by planetary gear phasing theory: harmonic excitations from even-order modes affect only the planetary components (planetary gears, crank shafts, cycloidal gears), causing them to vibrate in opposite directions, whereas central components (input shaft, sun gear, planet carrier) remain balanced. Below is a table summarizing the first seven natural frequencies for \( \beta = 0 \):
| Mode Order | Natural Frequency (Hz) | Description |
|---|---|---|
| 1 | 121.4 | Constant, independent of \( \beta \) |
| 2 | 245.7 | Varies with \( \beta \), planetary mode |
| 3 | 367.2 | Constant, independent of \( \beta \) |
| 4 | 489.5 | Varies with \( \beta \), planetary mode |
| 5 | 512.8 | Constant, independent of \( \beta \) |
| 6 | 634.1 | Varies with \( \beta \), planetary mode |
| 7 | 756.3 | Constant, independent of \( \beta \) |
The mode shapes further illustrate this phenomenon. For odd orders, all components participate in vibration with symmetric patterns. For even orders, the central components show zero displacement, while planetary components exhibit anti-phase vibrations. This insight is crucial for avoiding resonance in rotary vector reducers, as the input frequency (50 Hz for this case) is well below the first natural frequency, indicating a low risk of resonance under normal operating conditions. Experimental validation using impact testing yielded a first natural frequency of 126.9 Hz, closely matching the model’s prediction of 121.4 Hz, thereby confirming the accuracy of the modified torsional dynamics model for rotary vector reducers.
To optimize the design of rotary vector reducers, sensitivity analysis is essential to identify which parameters most influence the natural frequencies. I employed the partial derivative method, which treats eigenvalues as functions of structural parameters. For a normalized mode shape satisfying \( \boldsymbol{\phi}_i^T \mathbf{M} \boldsymbol{\phi}_i = 1 \), the sensitivity of the natural frequency \( \omega_i \) to a parameter \( p_j \) is given by:
$$ \frac{\partial \omega_i}{\partial p_j} = \frac{1}{2\omega_i} \boldsymbol{\phi}_i^T \left( \frac{\partial \mathbf{K}}{\partial p_j} – \omega_i^2 \frac{\partial \mathbf{M}}{\partial p_j} \right) \boldsymbol{\phi}_i $$
For dimensionless analysis, the normalized sensitivity is:
$$ S_{p_j} = \frac{p_j}{2\omega_i^2} \boldsymbol{\phi}_i^T \left( \frac{\partial \mathbf{K}}{\partial p_j} – \omega_i^2 \frac{\partial \mathbf{M}}{\partial p_j} \right) \boldsymbol{\phi}_i $$
The parameters \( p_j \) include moments of inertia and stiffnesses of all components. I focused on the first four natural frequencies at \( \beta = 0 \), as lower frequencies dominate dynamic response. The sensitivity results are summarized in the following tables for moments of inertia and stiffnesses, respectively:
| Component | Sensitivity to \( J \) (Mode 1) | Sensitivity to \( J \) (Mode 2) | Sensitivity to \( J \) (Mode 3) | Sensitivity to \( J \) (Mode 4) |
|---|---|---|---|---|
| Input Shaft | 0.05 | 0.02 | 0.01 | 0.03 |
| Sun Gear | 0.08 | 0.04 | 0.02 | 0.05 |
| Planetary Gear | 0.12 | 0.06 | 0.03 | 0.07 |
| Crank Shaft | 0.25 | 0.15 | 0.10 | 0.20 |
| Cycloidal Gear | 0.18 | 0.10 | 0.08 | 0.15 |
| Planet Carrier | 0.30 | 0.20 | 0.15 | 0.25 |
| Stiffness Type | Sensitivity to \( k \) (Mode 1) | Sensitivity to \( k \) (Mode 2) | Sensitivity to \( k \) (Mode 3) | Sensitivity to \( k \) (Mode 4) |
|---|---|---|---|---|
| Input Shaft Torsion | 0.10 | 0.05 | 0.03 | 0.08 |
| Crank Shaft Torsion | 0.15 | 0.08 | 0.05 | 0.12 |
| Sun-Planet Mesh | 0.20 | 0.12 | 0.08 | 0.18 |
| Cycloidal-Pin Mesh | 0.22 | 0.14 | 0.10 | 0.20 |
| Bearing Support (\( k_{hc} \)) | 0.35 | 0.25 | 0.20 | 0.30 |
| Carrier Bearing (\( k_{oh} \)) | 0.40 | 0.30 | 0.25 | 0.35 |
From these results, it is evident that the natural frequencies of the rotary vector reducer are most sensitive to the moments of inertia of the planet carrier and crank shafts, as well as the stiffnesses of the bearings (both support and carrier bearings). This highlights the critical role of bearings in the dynamic behavior of rotary vector reducers. To further investigate, I analyzed the effect of bearing stiffness variations on the first natural frequency. The relationship can be expressed as:
$$ \omega_1 = f(k_{hc}, k_{oh}) $$
where \( \omega_1 \) decreases significantly with reduced bearing stiffness due to factors like fatigue wear. For instance, if \( k_{hc} \) or \( k_{oh} \) drops by 20%, \( \omega_1 \) may decrease by approximately 15%, potentially bringing it closer to the input frequency and increasing resonance risk. This underscores the importance of maintaining high bearing stiffness in rotary vector reducers through design enhancements, such as using high-strength materials or increasing the number of crank shafts to distribute loads more evenly.
In conclusion, my analysis of the rotary vector reducer’s torsional vibration characteristics reveals several key findings. First, the modified torsional dynamics model with 13 degrees of freedom effectively captures the influence of translational motions and eccentric angle variations, providing a more accurate representation than previous models. Second, the natural frequencies exhibit distinct patterns: odd orders are constant, while even orders vary periodically with the cycloidal gear’s eccentric angle, leading to planetary vibration modes. Third, sensitivity analysis using the partial derivative method shows that bearing stiffnesses are the most influential parameters on natural frequencies, followed by the moments of inertia of the planet carrier and crank shafts. These insights emphasize the need to prioritize bearing design in rotary vector reducers to prevent stiffness degradation and avoid resonance. Future work could explore nonlinear effects, such as time-varying mesh stiffnesses or damping, to further refine the dynamic model and enhance the performance of rotary vector reducers in high-precision applications.
This study contributes to the broader understanding of rotary vector reducers by bridging gaps in dynamic analysis. By integrating advanced modeling techniques with sensitivity studies, I have provided a framework for optimizing these reducers to meet the stringent vibration requirements of modern robotics and automation systems. The rotary vector reducer, with its complex yet efficient design, remains a focal point for innovation, and ongoing research will continue to unlock its full potential in mechanical transmission.