In our study of worm gear drives, the roller enveloping hourglass worm gear represents a significant advancement in transmission technology. Unlike conventional worm gear pairs that rely on sliding friction between tooth surfaces, this innovative design incorporates rollers as the worm wheel teeth, effectively transforming sliding friction into rolling friction. This fundamental change greatly reduces friction coefficients and improves transmission efficiency. The worm gear drive’s unique geometry, where the hourglass worm is generated by enveloping the roller surfaces, results in multiple tooth contacts and high load capacity. However, during actual assembly, inevitable errors such as center distance deviation, axial misalignment of the worm, axial displacement of the worm wheel, and shaft angle errors can significantly affect the contact pattern and potentially cause interference or even seizure. Our research aims to systematically analyze how these assembly errors influence the contact interference of the roller enveloping hourglass worm gear drive, providing quantitative evaluation indices and numerical calculation methods.
To begin our analysis, we first establish the theoretical meshing geometry of the worm gear in the absence of errors. We define coordinate systems as follows: a fixed spatial frame σm attached to the worm wheel and σn attached to the worm, with respective unit vectors (im, jm, km) and (in, jn, kn). The worm wheel rotates about its axis z1 with angular velocity ω1, while the hourglass worm rotates about its axis z2 with ω2. The angular displacements are φ1 and φ2, satisfying φ1/φ2 = i12, where i12 = Z2/Z1 is the transmission ratio, Z1 is the number of worm starts, and Z2 is the number of roller teeth. The center distance between the two axes is denoted by a.
The transformation from the worm wheel coordinate system σ1 (attached to the worm wheel) to the worm coordinate system σ2 is given by:
$$
\left( x_2, y_2, z_2, 1 \right)^T = M_{n2} M_{mn} M_{1m} \left( x_1, y_1, z_1, 1 \right)^T
$$
where
$$
M_{n2} = \begin{bmatrix}
\cos\phi_2 & -\sin\phi_2 & 0 & 0 \\
\sin\phi_2 & \cos\phi_2 & 0 & 0 \\
0 & 0 & 1 & 0 \\
0 & 0 & 0 & 1
\end{bmatrix}, \quad
M_{1m} = \begin{bmatrix}
\cos\phi_1 & -\sin\phi_1 & 0 & 0 \\
\sin\phi_1 & \cos\phi_1 & 0 & 0 \\
0 & 0 & 1 & 0 \\
0 & 0 & 0 & 1
\end{bmatrix},
$$
$$
M_{mn} = \begin{bmatrix}
-1 & 0 & 0 & a \\
0 & 0 & 0 & 0 \\
0 & 1 & 1 & 0 \\
0 & 0 & 0 & 1
\end{bmatrix}.
$$
The roller tooth surface in the worm wheel coordinate system σ1 is defined by parameters u (along the roller axis) and θ (angular position around the roller axis), with roller radius r. The position vector is:
$$
\mathbf{r}_1 = u \, \mathbf{i}_1 – r \sin\theta \, \mathbf{j}_1 + r \cos\theta \, \mathbf{k}_1
$$
The unit normal vector of the roller surface is:
$$
\mathbf{n}_1 = 0 \, \mathbf{i}_1 – \sin\theta \, \mathbf{j}_1 + \cos\theta \, \mathbf{k}_1
$$
For two surfaces to remain in continuous contact, they must satisfy the meshing condition Φ = v(12) · n1 = 0, where v(12) is the relative velocity at the contact point. After coordinate transformation and vector operations, the meshing function for the roller enveloping hourglass worm gear drive is derived as:
$$
\Phi(u, \theta, \phi_1) = a \, i_{12} \cos\theta – u \sin\theta – u \, i_{12} \cos\phi_1 \cos\theta = 0
$$
The instantaneous contact lines on the roller surface are described by the simultaneous solution of the roller surface equation and the meshing equation. The theoretical contact line of this worm gear is a spatial cylindrical helical curve located near the middle plane of the roller, as determined by our numerical simulations.
When assembly errors are introduced, the transformation between the fixed frames σm and σn must be modified. We consider four types of errors: center distance error Δa, worm axial error ΔL2, worm wheel axial error ΔL1, and shaft angle error ΔΣ. The modified transformation matrix becomes:
$$
M_{mn}^{error} = \begin{bmatrix}
-1 & 0 & 0 & a + \Delta a \\
0 & \sin\Delta\Sigma & \cos\Delta\Sigma & -\Delta L_2 \\
0 & \cos\Delta\Sigma & -\sin\Delta\Sigma & \Delta L_1 \\
0 & 0 & 0 & 1
\end{bmatrix}
$$
Assuming the worm wheel remains in its ideal position (all errors are considered as relative displacements of the worm with respect to the wheel), the worm tooth surface under errors can be expressed by converting the roller surface points through the erroneous transformation. The resulting worm surface equation is:
$$
\mathbf{r}’_2(u, \theta, \phi_1) = x’_2 \mathbf{i}_2 + y’_2 \mathbf{j}_2 + z’_2 \mathbf{k}_2, \quad \text{with } \Phi(u, \theta, \phi_1)=0
$$
where the coordinates are complicated functions of the errors and motion parameters. For each mesh position, we can calculate the distance between any point on the worm surface and the corresponding roller axis. If this distance is less than the roller radius r, interference occurs.
To quantitatively evaluate the interference, we propose two types of interference cross-section shapes: triangular and quadrilateral, depending on how the worm surface intrudes into the roller body. For a triangular interference (three vertices), the area is computed using Heron’s formula:
$$
S_i = \sqrt{P(P-a_T)(P-b_T)(P-c_T)}
$$
where a_T, b_T, c_T are side lengths and P is the semiperimeter. For a quadrilateral interference (four vertices), we split it into two triangles and sum their areas:
$$
S_i = S_{\Delta123} + S_{\Delta134}
$$
The vertices are obtained by solving the condition that the distance from the roller axis equals the roller radius r, combined with the worm surface equation. We developed a numerical iterative procedure using MATLAB to find these intersection points for a given error case. The flow involves sweeping through the meshing angle φ1 within its working range and identifying the boundary points where the worm surface touches the roller surface exactly.
To validate our model and investigate the sensitivity of each error, we performed an example analysis on a specific worm gear pair with the following parameters:
| Parameter | Value |
|---|---|
| Center distance a (mm) | 80 |
| Transmission ratio i12 | 20 |
| Roller radius r (mm) | 7 |
| Roller parameter u range (mm) | [56, 68] |
| Worm wheel rotation φ1 range (°) | [-40, 40] |
First, we examined the theoretical contact without errors. The contact lines on the roller surface are spiral-shaped and concentrated in the mid-plane region. This pattern ensures smooth meshing under ideal conditions.
Next, we introduced each error component individually and computed the total interference area (sum of areas on both tooth flanks of five roller teeth pairs). The results are summarized in the following tables.
| Δa (mm) | Tooth 1 | Tooth 2 | Center tooth | Tooth 4 | Tooth 5 |
|---|---|---|---|---|---|
| -0.10 | 2.14 | 1.35 | 0.00 | 1.35 | 2.14 |
| -0.05 | 0.98 | 0.62 | 0.00 | 0.62 | 0.98 |
| 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 |
| 0.05 | 0.98 | 0.62 | 0.00 | 0.62 | 0.98 |
| 0.10 | 2.14 | 1.35 | 0.00 | 1.35 | 2.14 |
As seen in Table 2, the interference areas for different teeth are symmetric with respect to zero error. The outer teeth (tooth 1 and 5) experience larger interference than inner teeth. The center tooth remains unaffected under pure center distance error because of symmetry.
| ΔL₂ (mm) | Tooth 1 | Tooth 2 | Center tooth | Tooth 4 | Tooth 5 |
|---|---|---|---|---|---|
| 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 |
| 0.02 | 0.56 | 0.55 | 0.54 | 0.55 | 0.56 |
| 0.04 | 1.04 | 1.03 | 1.02 | 1.03 | 1.04 |
| 0.06 | 1.47 | 1.46 | 1.45 | 1.46 | 1.47 |
| 0.08 | 1.85 | 1.84 | 1.83 | 1.84 | 1.85 |
| 0.10 | 2.19 | 2.18 | 2.17 | 2.18 | 2.19 |
Table 3 shows that worm axial error affects all teeth almost equally. The interference area increases with error magnitude, but the growth rate gradually decreases.
| ΔL₁ (mm) | Tooth 1 | Tooth 2 | Center tooth | Tooth 4 | Tooth 5 |
|---|---|---|---|---|---|
| 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 |
| 0.02 | 0.08 | 0.08 | 0.08 | 0.08 | 0.08 |
| 0.04 | 0.16 | 0.16 | 0.16 | 0.16 | 0.16 |
| 0.06 | 0.23 | 0.23 | 0.23 | 0.23 | 0.23 |
| 0.08 | 0.31 | 0.31 | 0.31 | 0.31 | 0.31 |
| 0.10 | 0.38 | 0.38 | 0.38 | 0.38 | 0.38 |
Worm wheel axial error has the least influence among the four error types, as indicated by the small area values in Table 4. This is because the engagement along the worm wheel axis direction is minimal.
| ΔΣ (°) | Tooth 1 | Tooth 2 | Center tooth | Tooth 4 | Tooth 5 |
|---|---|---|---|---|---|
| -0.5 | 2.86 | 1.52 | 0.00 | 1.52 | 2.86 |
| -0.3 | 1.12 | 0.59 | 0.00 | 0.59 | 1.12 |
| -0.1 | 0.13 | 0.07 | 0.00 | 0.07 | 0.13 |
| 0.0 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 |
| 0.1 | 0.13 | 0.07 | 0.00 | 0.07 | 0.13 |
| 0.3 | 1.12 | 0.59 | 0.00 | 0.59 | 1.12 |
| 0.5 | 2.86 | 1.52 | 0.00 | 1.52 | 2.86 |
Shaft angle error, though its absolute magnitude is small (within 0.5°), causes the most rapid increase in interference area, especially on the outer teeth. This highlights the importance of controlling shaft alignment within tight tolerances.
A typical roller enveloping hourglass worm gear drive assembly is shown in the image below, which illustrates the compact structure and the roller-based tooth design that distinguishes this worm gear from conventional ones.
To compare the relative impact of different error types, we normalized the errors to the same numerical range (0 to 0.1 mm for linear errors, 0 to 0.1° for angle error) and plotted the total interference area (sum over all teeth). The following observations were made:
- Worm axial error produced the largest interference area per unit error magnitude. A worm axial shift of 0.1 mm resulted in an interference area of about 10.9 mm² (sum of five teeth).
- Center distance error had a moderate effect, with a total area of about 6.6 mm² at 0.1 mm.
- Worm wheel axial error had the smallest effect, with only about 1.9 mm² total at 0.1 mm.
- Shaft angle error, when converted to an equivalent linear displacement at the contact zone, showed very high sensitivity; even 0.1° caused an interference area comparable to 0.1 mm worm axial error. However, because the angle error is measured in degrees, direct comparison is not linear; we recommend keeping shaft angle error within ±0.5°.
These quantitative findings provide clear guidance for assembly tolerances of the roller enveloping hourglass worm gear drive. The worm axial positioning is the most critical factor; therefore, proper thrust bearing arrangements and preload adjustments are essential. Center distance tolerance should be monitored but is less sensitive. Worm wheel axial clearance has minimal impact, so it can be relaxed. Shaft angle alignment, however, must be carefully controlled during installation to avoid excessive edge loading.
In conclusion, our study established a comprehensive interference analysis model for the roller enveloping hourglass worm gear drive considering four typical assembly errors. We derived the mathematical expressions for the tooth surfaces under errors, proposed quantitative interference area evaluation indices (triangular and quadrilateral cross-sections), and implemented a numerical solution algorithm. Through parametric analysis on a representative worm gear pair, we determined the sensitivity ranking: worm axial error > center distance error > shaft angle error (when limited to small angles) > worm wheel axial error. The theoretical contact lines are spatial helical curves near the middle plane. These results offer valuable insights for the design, manufacturing, and assembly of high-performance worm gear drives, enabling engineers to prioritize error control and improve the reliability and efficiency of the transmission system.