The transmission precision and operational smoothness of a rotary vector reducer are paramount for advanced applications such as industrial robotics. At the heart of its two-stage design lies the cycloidal-pin wheel transmission, where the presence and magnitude of meshing backlash become a critical performance-defining factor. While traditional analytical formulas provide a foundational estimate for initial backlash, they often fall short of capturing the dynamic realities of the meshing process. This article, from my research perspective, delves into the limitations of conventional methods and establishes a comprehensive framework for calculating the accurate meshing backlash at any rotational position during the operation of the rotary vector reducer.
Re-examining the Foundations: Limitations of Traditional Backlash Calculation
For a standard, unmodified cycloidal drive, the tooth profiles are theoretically conjugate, leading to multi-tooth contact and near-zero operational backlash. In practice, modifications such as equidistant (profile shift) and radial (pin circle radius change) adjustments are applied to the cycloidal wheel to optimize load distribution and manufacturing tolerance. It is for these modified profiles that backlash calculations become essential. The prevailing analytical formula for the initial backlash \(d_i\) at the i-th pin position is given by:
$$d_i = \Delta r_{rp} \left(1 – \frac{\sin \varphi_i}{\sqrt{1 + k_1^2 – 2k_1 \cos \varphi_i}}\right) + \frac{\Delta r_p \left(1 – k_1 \cos \varphi_i – \sqrt{1 – k_1^2} \sin \varphi_i\right)}{\sqrt{1 + k_1^2 – 2k_1 \cos \varphi_i}}$$
where \(\varphi_i\) is the angular position of the i-th pin relative to the crank arm, \(k_1 = a z_p / (r_p + \Delta r_p)\) is the cycloid coefficient, \(a\) is the eccentricity, \(z_p\) is the number of pins, \(r_p\) is the pin circle radius, \(\Delta r_p\) is the radial modification, and \(\Delta r_{rp}\) is the equidistant modification.
My analysis reveals several fundamental limitations inherent in this traditional approach:
- Inaccurate Initial Contact Point Assumption: The formula implicitly assumes the initial meshing occurs at a specific angle \(\varphi_0 = \arccos(k_1)\). However, for a modified profile, the true initial contact point shifts. Using Tooth Contact Analysis (TCA), I determined that the actual initial meshing angle can deviate from this theoretical value, leading to a miscalculation of the entire backlash distribution from the outset.
- Static Node Assumption: The derivation relies on the principle that the common normal at the contact point always passes through a fixed instantaneous center of rotation (the node) on the crank arm. This holds true for perfect conjugate action. For modified profiles, which are no longer perfectly conjugate, the contact point and its corresponding common normal migrate during rotation. Consequently, the node’s position on the crank arm is not fixed but oscillates periodically with the crank’s rotation. Assuming a static node introduces error.
- Limited to Initial Position: The formula calculates backlash only for the initial assembly state (\(\phi_2 = 0\)). It cannot predict how the backlash at each tooth pair evolves as the input crankshaft rotates—a dynamic essential for understanding transmission error and torsional stiffness under motion.
- Lack of Generality: The equation is derived specifically for combined equidistant and radial modifications. Modern rotary vector reducer manufacturing often employs more complex modifications (e.g., parabolic, circular arc) to optimize performance. The traditional formula is not applicable to these advanced tooth forms.
The following table summarizes the key parameters for a sample rotary vector reducer cycloidal drive used in my comparative analysis:
| Parameter | Symbol | Value |
|---|---|---|
| Number of Pins | \(z_p\) | 20 |
| Number of Cycloid Teeth | \(z_c\) | 19 |
| Eccentricity | \(a\) | 1.0 mm |
| Pin Circle Radius | \(r_p\) | 25.145 mm |
| Pin Radius | \(r_{rp}\) | 2.2 mm |
| Radial Modification | \(\Delta r_p\) | -0.1 mm |
| Equidistant Modification | \(\Delta r_{rp}\) | -0.05 mm |
Defining and Pursuing the “Accurate Meshing Backlash”
To overcome these limitations, I propose a new definition and a corresponding computational methodology. The accurate meshing backlash is defined as the shortest distance between the theoretical pin tooth flank and the modified cycloid tooth flank, measured along the direction of the common normal line at the actual instantaneous contact point. This definition physically represents the gap through which one tooth could move before contacting its mate under no load, respecting the true kinematic constraints.
The cornerstone of calculating this backlash is a precise Tooth Contact Analysis (TCA) model. The model considers the spatial relationship between the fixed pin gear and the moving cycloidal wheel, which undergoes both revolution (eccentric motion) and rotation.
The surface of the i-th pin in its local coordinate system \(S_1\) is given by:
$$\mathbf{r}_1^{(i)}(\beta) = \begin{bmatrix}
-r_{rp}\sin\beta \cos(\frac{2\pi i}{z_p}) – \sin(\frac{2\pi i}{z_p})(r_{rp}\cos\beta + r_p) \\
-r_{rp}\sin\beta \sin(\frac{2\pi i}{z_p}) + \cos(\frac{2\pi i}{z_p})(r_{rp}\cos\beta + r_p) \\
b_p \\
1
\end{bmatrix}$$
where \(\beta\) is the pin surface parameter.
The modified cycloid tooth profile in its local coordinate system \(S_2\) is represented by a general form:
$$\mathbf{r}_2^{(c)}(\alpha) = [x_c(\alpha), y_c(\alpha), b_c, 1]^T$$
The specific expressions for \(x_c(\alpha)\) and \(y_c(\alpha)\) incorporate the modification parameters \(\Delta r_p\) and \(\Delta r_{rp}\), as well as the basic cycloid generation parameter \(\alpha\).
Using homogeneous coordinate transformation matrices \(\mathbf{M}_{f1}(\phi_1)\) and \(\mathbf{M}_{f2}(\phi_2)\), these surfaces are expressed in a global fixed coordinate system \(S_f\):
$$\mathbf{r}_f^{(1)}(\beta, \phi_1) = \mathbf{M}_{f1}(\phi_1) \mathbf{r}_1^{(i)}(\beta)$$
$$\mathbf{r}_f^{(2)}(\alpha, \phi_2) = \mathbf{M}_{f2}(\phi_2) \mathbf{r}_2^{(c)}(\alpha)$$
where \(\phi_1\) and \(\phi_2\) are the rotation angles of the pin gear and cycloid wheel, respectively. In the rotary vector reducer, the pin gear is fixed (\(\phi_1=0\)), and \(\phi_2\) is the output rotation.
The condition for continuous contact is the equality of position vectors and surface normals at the contact point in \(S_f\):
$$
\mathbf{r}_f^{(1)}(\beta, \phi_1) = \mathbf{r}_f^{(2)}(\alpha, \phi_2)
$$
$$
\mathbf{n}_f^{(1)}(\beta, \phi_1) = \mathbf{n}_f^{(2)}(\alpha, \phi_2)
$$
This TCA system contains three independent scalar equations (for the 2D case). Given two parameters (e.g., \(\phi_2\) and the pin number \(i\)), the system can be solved for the remaining three unknowns (\(\phi_1\), \(\alpha\), \(\beta\)), uniquely identifying the contact point.
Computational Framework for Accurate Backlash
The procedure for calculating the accurate meshing backlash at any crank angle involves two major phases: determining the instantaneous contact condition and then computing the gap along the common normal for all non-contacting teeth.
Phase 1: Determining the Instantaneous Contact Pair
- For a given input crank angle (which defines the cycloid wheel’s phase \(\phi_2\)), the TCA equations are solved iteratively for each potential pin tooth \(i\).
- The solution identifies the specific pin tooth \(m\) for which a valid, physically realizable contact point exists (i.e., the contact point lies within the active tooth flank boundaries). This pin \(m\) is the load-bearing tooth at that instant. The corresponding parameters \(\alpha_m\) and \(\beta_m\) define the exact contact point \(C\) on both profiles.
Phase 2: Calculating Backlash for All Teeth
For the load-bearing tooth \(m\), the backlash is defined as zero. For every other pin tooth \(i \neq m\), the backlash \(d_i\) is calculated as follows:
- Find the Common Normal Line from the Contact Point: Calculate the unit normal vector \(\mathbf{n}_c\) at the contact point \(C\) on the cycloid profile (transformed to the global frame). The equation of the common normal line \(L_i\) passing through \(C\) is:
$$L_i: (\mathbf{r} – \mathbf{r}_c) \times \mathbf{n}_c = 0$$
where \(\mathbf{r}_c\) is the position vector of \(C\). - Find the Intersection of the Normal with the Cycloid Profile: Theoretically, for the conjugated pair \(m\), this line also passes through the center of pin \(m\). For a non-conjugated pair \(i\), we find the point \(B_i\) on the cycloid profile (parameter \(\alpha_i\)) where the line \(L_i\) intersects it. This involves solving a nonlinear equation:
$$(\mathbf{r}_2(\alpha_i) – \mathbf{r}_c) \times \mathbf{n}_c = 0$$ - Calculate the Minimum Distance: The line \(L_i\) also passes through the center of pin \(i\), \(A_i(x_{pi}, y_{pi})\). The accurate meshing backlash \(d_i\) is the distance from \(B_i\) to \(A_i\), minus the pin radius \(r_{rp}\):
$$d_i = \| \mathbf{A}_i – \mathbf{B}_i \| – r_{rp}$$
This distance is measured along the common normal direction, satisfying our physical definition.
The entire computational workflow is summarized below, applicable for both initial and any arbitrary rotational position analysis.
| Step | Action | Key Equations/Methods |
|---|---|---|
| 1 | Define system parameters and tooth profiles. | \(z_p, z_c, a, r_p, r_{rp}, \Delta r_p, \Delta r_{rp}\), \(\mathbf{r}_1^{(i)}(\beta)\), \(\mathbf{r}_2^{(c)}(\alpha)\). |
| 2 | Specify the operating condition (crank/cycloid angle \(\phi_2\)). | Input angle \(\phi_2\). |
| 3 | Solve TCA for all pins to find the contact pair \(m\). | Solve \(\mathbf{r}_f^{(1)}=\mathbf{r}_f^{(2)}\), \(\mathbf{n}_f^{(1)}=\mathbf{n}_f^{(2)}\) for each \(i\). |
| 4 | For contact pair \(m\), set backlash \(d_m = 0\). Extract contact point \(C\) and normal \(\mathbf{n}_c\). | Obtain \(\mathbf{r}_c, \mathbf{n}_c\). |
| 5 | For each non-contact pin \(i \neq m\): a) Locate pin center \(A_i\). b) Find intersection \(B_i\) of line \((C, \mathbf{n}_c)\) with cycloid profile. c) Compute backlash \(d_i\). |
a) \(A_i\) from geometry. b) Solve for \(\alpha_i\). c) \(d_i = \|A_i – B_i\| – r_{rp}\). |
| 6 | Output the accurate backlash distribution \(d_i\) for the specified angle \(\phi_2\). | Array \(d_i\). |
Comparative Results and Implications for the Rotary Vector Reducer
Applying this methodology to the sample rotary vector reducer parameters yields insightful comparisons. First, the initial contact angle \(\varphi_0\) was found to be approximately 35.970°, a deviation of about 2.88% from the traditional formula’s assumption of \(\arccos(k_1) \approx 37.007°\). This shift directly affects the calculated backlash values.
The table below compares the initial backlash distribution calculated by the traditional formula and the proposed accurate method for half of the pins (due to symmetry in engagement):
| Pin Number (i) | Traditional Backlash (mm) | Accurate Backlash (mm) | Deviation (mm) |
|---|---|---|---|
| 1 | 0.010703 | 0.005553 | -0.005150 |
| 2 | 0.000019 | ~0.000000 | ~-0.000019 |
| 3 | 0.003596 | 0.004228 | +0.000632 |
| 4 | 0.011342 | 0.011913 | +0.000571 |
| 5 | 0.020163 | 0.020567 | +0.000404 |
| 6 | 0.028749 | 0.028998 | +0.000249 |
| 7 | 0.036394 | 0.036524 | +0.000130 |
| 8 | 0.042709 | +0.000052 | |
| 9 | 0.047252 | 0.047266 | +0.000014 |
| 10 | 0.049997 | 0.050014 | +0.000017 |
While the overall trend is similar, significant deviations exist, particularly for the first few teeth. More importantly, the proposed method’s power is its ability to generate such a table for any input angle \(\phi_2\), revealing how the backlash distribution changes dynamically. This dynamic mapping is crucial for predicting the transmission error and nonlinear stiffness of the rotary vector reducer under varying load and position, factors that directly impact robotic positioning accuracy and vibration.
Conclusion: Towards a More Realistic Design and Analysis Tool
In conclusion, the traditional formula for calculating backlash in cycloidal drives of a rotary vector reducer, while useful for initial estimates, is built upon assumptions that do not hold for modified, non-conjugate profiles in motion. The node is not static, and the contact condition evolves with rotation. The methodology I have presented—defining backlash as the gap along the instantaneous common normal and calculating it via a rigorous TCA model—provides a superior framework for accurate meshing backlash analysis.
This approach offers several key advantages for the design and performance evaluation of high-precision rotary vector reducers:
- Dynamic Analysis: It computes backlash for any input angle, enabling the study of transmission error spectra.
- Generality: It is applicable to any tooth profile modification (equidistant, radial, parabolic, etc.), making it compatible with modern grinding techniques.
- Physical Fidelity: It respects the true kinematic constraints by using the actual, moving common normal direction.
- Foundation for Advanced Studies: The precise backlash data serves as critical input for nonlinear dynamic modeling, contact force analysis, and lifetime prediction of the rotary vector reducer.
By adopting this more precise computational perspective, designers and engineers can better optimize cycloidal tooth modifications, minimize undesirable effects of backlash, and ultimately enhance the performance and reliability of the critically important rotary vector reducer in advanced mechanical systems.