Nutation drives convert a wobbling motion into a slow output rotation through a bevel gear pair whose axes are inclined by a small nutation angle. The arrangement is compact and can produce a large reduction ratio in a single stage, which makes it attractive for robot wrist joints and other space-constrained precision transmissions. The load-carrying behavior of the drive is governed by the meshing contact of its gear pair. In the design considered here, that pair consists of an external and an internal double circular-arc spiral bevel gear.
Double circular-arc gears differ from involute gears in that the normal tooth profile is composed of convex and concave circular arcs. In a spiral bevel version, this profile is swept along a helical tooth trace, producing a tooth surface that can contact on either arc. The result is a contact process that may involve one or two active zones on the same tooth. The accompanying changes in contact area, contact stress, and load sharing are not obvious from the nominal geometry alone.
The relevant performance indicators are the contact path along the face width, the peak contact stress, the contact ratio, and the load distribution among simultaneous tooth pairs. Installation errors are equally important because a small axial offset or a deviation in nutation angle can move the contact zone toward the tooth edges, where the available area is small and the local stress rises. A quasi-static finite element contact model is used to resolve these quantities for a seven-tooth-pair sector of the gear pair, with tooth surfaces generated from an imaginary crown gear.
Geometry and meshing of the gear pair
The external and internal bevel gears are generated from an imaginary planar crown gear that is conjugate to both members. The crown-gear surface is defined by a normal profile and a tooth trace. The normal profile follows the double circular-arc standard profile of GB/T 12759—1991, represented by eight circular segments. In the profile coordinate system \(S_n(x_n,y_n,z_n)\), with \(x_n\) on the symmetry line and \(y_n\) on the reference line, each segment is written as
$$
\mathbf r_{ni}
=
\begin{pmatrix}
r_i\sin\alpha_i + E_i\\
r_i\cos\alpha_i + F_i\\
0
\end{pmatrix},
\qquad i=1,\ldots,8
$$
where \(r_i\) is the arc radius, \(\alpha_i\) the angular parameter, and \((E_i,F_i)\) the arc-center coordinates for segment \(i\). The profile is swept along a helical tooth trace with spiral angle \(\beta\). For the convex arc \(j=1\) and the concave arc \(j=2\), the trace is expressed as
$$
x’_p = x_p + \sigma(\rho_j-l_j)\sin(\theta-\Delta\theta_j+\beta),
\qquad
y’_p = y_p – \sigma(\rho_j-l_j)\cos(\theta-\Delta\theta_j+\beta)
$$
where \(\rho_j\) and \(l_j\) are the arc radius and center offset, \(\theta\) is the trace parameter, \(\Delta\theta_j\) is the angular offset of the arc, and \(\sigma=-1\) for the left flank and \(\sigma=+1\) for the right flank. The crown-gear surface \(\mathbf r_c\) is then transformed to the external gear \(w\) and the internal gear \(n\):
$$
\mathbf r_w = \mathbf M_w^c \mathbf r_c,
\qquad
\mathbf r_n = \mathbf M_n^c \mathbf r_c
$$
The matrices \(\mathbf M_w^c\) and \(\mathbf M_n^c\) contain the coordinate transformations that incorporate the 5° nutation angle and the pitch-cone geometry, so the same crown-gear surface produces the mating external and internal tooth surfaces.
| Parameter | External gear | Internal gear |
|---|---|---|
| Teeth | 26 | 28 |
| Toe modulus (mm) | 2 | 2 |
| Nutation angle (°) | 5 | 5 |
| Spiral angle (°) | 25 | 25 |
| Pitch cone angle (°) | 47.19 | 127.81 |
| Cone distance (mm) | 60 | 60 |
| Tooth width (mm) | 8 | 8 |
| Addendum (mm) | 1.8 | 1.8 |
| Dedendum (mm) | 2.15 | 2.15 |
The two pitch-cone angles sum to 175°, consistent with the 5° nutation angle between the axes. Using the computed surface points, the tooth flanks are built as solid models and assembled into the external–internal gear pair. The curved tooth trace distributes the contact along the face width, while the double circular-arc profile creates the possibility of simultaneous contact on the convex and concave arcs.
Finite element contact model
To keep the contact region away from truncated boundaries while retaining a manageable model, seven tooth pairs were retained, and the middle fourth pair was used for detailed contact readings. The external gear was fully constrained, and the internal gear was allowed to rotate only about its own axis. A torque of 120 N·m was applied to the internal gear and increased linearly from zero to the full value to aid convergence. The tooth surfaces were treated as frictional contact with a coefficient of 0.1.
The gear bodies were assigned a density of 7,850 kg/m³, an elastic modulus of \(2.1\times10^5\) MPa, and a Poisson ratio of 0.3. Free tetrahedral meshing produced 187,370 elements and 278,426 nodes. The analysis was quasi-static: inertia, lubrication, and thermal effects were not represented, so the results describe the elastic contact response under a slowly applied load. Throughout the discussion, heel denotes the large end and toe the small end of the bevel tooth.
Contact evolution during a mesh cycle
At the start of a mesh cycle, contact is confined to the convex arc and enters at the heel. The contact zone moves along the face width toward the toe, and the contact stress rises as the force transmitted by the pair increases. When the convex contact approaches the toe, the concave arc begins to engage at the heel. Both arcs then share the load, and the total contact area reaches its maximum. Within this double-zone phase, the stress passes through a local minimum when the area is largest.
After the convex contact ends, the tooth carries load only on the concave arc. The contact zone continues toward the toe, and just before exit the stress reaches its maximum for the cycle. The critical location is therefore the concave arc near the small end.
| Phase | Active arc or arcs | Contact location | Stress trend |
|---|---|---|---|
| I | Convex | Enters at the heel and moves toward the toe | Rises with the transmitted force |
| II | Convex and concave | Convex near the toe; concave at the heel; area largest | Local minimum at the largest area |
| III | Concave | Moves toward the toe | Maximum near exit |
The single-zone–double-zone–single-zone sequence reflects the S-shaped profile and the tapered face width. Because the double-zone phase spreads the load over both arcs, it is the most favorable part of the cycle for load carrying. The final single-zone phase is the least favorable: the taper reduces the available area, and the load is no longer shared between arcs. The concave toe is therefore the likely site of wear and fatigue initiation, and the contact pattern near the toe deserves particular attention in geometry design and inspection.
Load effects on contact ratio and load sharing
The contact ratio is defined as the ratio of the duration of one mesh cycle for a single tooth pair, \(\Delta T\), to the interval between the entries of two adjacent pairs, \(\Delta t\). In a quasi-static representation this is equivalent to the ratio of the driving rotation over one cycle, \(\Delta\varphi\), to the driving rotation between successive entries, \(\Delta\lambda\):
$$
\varepsilon = \frac{\Delta T}{\Delta t} = \frac{\Delta\varphi}{\Delta\lambda}
$$
Values above unity indicate that more than one pair is engaged on average. The load distribution coefficient describes how the total contact force is divided at a given instant:
$$
C = \frac{F_j}{\sum_{j=1}^{n} F_j}
$$
where \(F_j\) is the contact force on tooth \(j\), and \(n\) is the number of teeth in contact at that instant. The contact ratio and the peak load distribution coefficient were evaluated at three torque levels; the results are summarized below.
| Torque (N·m) | Contact ratio | Peak load distribution coefficient |
|---|---|---|
| 50 | 1.923 | 1.000 |
| 120 | 2.307 | 0.859 |
| 150 | 2.538 | 0.788 |
Over the 50–150 N·m range, the contact ratio rises by about 32%, from 1.923 to 2.538, while the peak load distribution coefficient falls by about 21%, from 1.000 to 0.788. The mechanism is elastic deformation: a higher torque increases tooth deflection, so adjacent pairs enter contact earlier and remain engaged longer. The load is divided among more teeth, and the peak share of any single tooth decreases.
At 50 N·m the peak coefficient reaches unity, meaning that at one instant in the cycle the instrumented tooth carries the entire contact load. At 150 N·m the peak is below unity, so at least two teeth share the load even at the most uneven instant. For the fourth tooth, the variation of \(C\) over a cycle follows the contact-force curve, with the peak occurring in the phase where both convex and concave arcs are engaged and the tooth carries its largest share. Increasing the torque does not change this qualitative pattern; it lowers the peak.
The improvement in load sharing should not be read as a general benefit of higher torque. The absolute contact stress and contact force still rise with load. The contact ratio and load distribution coefficient describe how the load is distributed, not whether the stress level is acceptable. The practical value of the trend is that the transmission becomes smoother relative to its own load level because the load is spread over more teeth.
Installation error sensitivity
Two rigid-body errors were considered: an axial offset \(\Delta C\) of the internal gear along its center axis, and a nutation-angle error \(\Delta\alpha\). The sign convention is defined by the relative motion of the two gears: positive \(\Delta C\) separates them, while negative \(\Delta C\) brings them together; positive \(\Delta\alpha\) brings them together, while negative \(\Delta\alpha\) separates them. Cases of \(\pm0.05\) mm and \(\pm0.05^\circ\) were analyzed at 120 N·m.
Positive offset shifts both the convex and concave contact points toward the transition zone between the arcs. Negative offset shifts the convex contact toward the addendum and the concave contact toward the dedendum. Positive nutation-angle error produces the same contact shift as negative offset: convex toward the addendum and concave toward the dedendum. Negative nutation-angle error produces the same shift as positive offset: both arcs toward the transition zone.
Negative offset and positive nutation-angle error increase the maximum contact stress and contact force; positive offset and negative nutation-angle error decrease them. The common feature of the unfavorable cases is that the contact zone approaches the tooth edge. The edge truncates the contact patch, reduces the effective area, and raises the local pressure. The favorable cases keep the contact away from the edge and allow a larger area to share the load.
Positive offset and negative nutation-angle error give a flatter load-distribution curve and a lower peak. Negative offset and positive nutation-angle error give a steeper curve and a substantially higher peak. The same edge-contact mechanism applies: when the contact is pushed to the edge, the affected tooth pair becomes the dominant load path, so its share rises.
| Error | Contact shift | Peak stress and force | Peak load share | Assessment |
|---|---|---|---|---|
| \(+\Delta C\) | Both arcs toward the transition zone | Lower | Lower | Less severe |
| \(-\Delta C\) | Convex toward addendum; concave toward dedendum | Higher | Higher | Edge contact; avoid |
| \(+\Delta\alpha\) | Convex toward addendum; concave toward dedendum | Higher | Higher | Edge contact; avoid |
| \(-\Delta\alpha\) | Both arcs toward the transition zone | Lower | Lower | Less severe |
Within the metrics examined, negative offset and positive nutation-angle error are the critical conditions. They push the contact toward the tooth edges, raise the peak stress, and concentrate the load on fewer teeth. Positive offset and negative nutation-angle error reduce the peak stress and flatten the load distribution, but they shift the contact toward the transition zone. The present model does not evaluate whether that shift is beneficial for transmission error or vibration, so the opposite side of the tolerance band should not be treated as unconditionally preferable. The clear result is that the negative-offset/positive-nutation side is the more severe side for load carrying.
Design implications and model limits
The contact cycle should be checked at the maximum torque, because both the contact ratio and the load-sharing coefficient change with load. The concave toe is the critical stress location, and the loaded contact pattern should be kept away from the toe edge under the worst combination of torque and installation error. Tolerances on axial offset and nutation angle should be tighter on the negative-offset and positive-nutation side, where edge contact and load concentration are most pronounced.
The results are quasi-static and assume ideal elastic bodies with a constant friction coefficient. They do not include dynamic mesh forces, lubrication film, thermal distortion, manufacturing flank deviations, or elastic deflection of the shafts and housing. The installation errors are represented as two rigid-body parameters, so combined errors and the effect of reverse torque are outside the scope. The stress values are best used comparatively, for example to rank load cases or error directions, rather than as absolute design allowable values without experimental calibration.
The contact behavior of double circular-arc spiral bevel gears in a nutation drive is therefore characterized by a three-stage cycle, a toe-critical stress location, a contact ratio that increases with load, and a load-sharing peak that decreases with load. Negative axial offset and positive nutation-angle error move the contact toward the edges and are the conditions to control. These trends provide a quantitative basis for tooth geometry and assembly tolerance decisions, within the stated quasi-static and ideal-geometry scope.