Spiral bevel gears are the default choice for right-angle drives that must carry high torque smoothly. Their teeth are cut along a spiral, so contact starts at one end of the flank and rolls across it rather than striking the whole tooth at once. This gives a higher contact ratio, lower impact and noise, and better load sharing than straight bevel gears. In compact transmissions, a single high-ratio stage is attractive because it removes a stage, shaft, bearings, and housing volume. The cost is a pinion with very few teeth.
A representative case is a constant-height spiral bevel gear pair with \(z_1=4\) and \(z_2=41\). At a \(90^\circ\) shaft angle it provides a ratio of 10.25:1. The pinion pitch diameter is only 18 mm, while the wheel is 184.5 mm. The design has to keep a four-tooth pinion from undercutting, keep its tip from becoming knife-edged, maintain continuous contact, and avoid concentrating load on the wheel tip. The route combines radial and tangential profile shift, off-pitch-cone meshing, a circular-arc-modified pinion cutter, and curved-surface synthesis of the machine settings.
Spiral bevel gears use curved teeth to obtain gradual engagement and a higher contact ratio than straight bevel gears.
Why a Four-Tooth Pinion Is Difficult
Four teeth are below the conventional undercut limit for a \(20^\circ\) pressure angle. A standard full-depth pinion would have its root cut away; even with shift, the tooth is short, thin at the tip, and flexible. The wheel is more than ten times larger in diameter, so the two members differ greatly in stiffness and thermal mass. If the contact path is placed symmetrically around the pitch cone, the pinion carries a disproportionate share of sliding and the wheel tip becomes vulnerable.
The design uses the equivalent cylindrical gear at the mid-face to apply established profile-shift rules to the bevel pair. The virtual tooth numbers \(z_{v1}\) and \(z_{v2}\) govern undercut and tip thickness. A positive radial shift on the pinion thickens the root and raises the addendum; a negative shift on the wheel lowers its addendum. A tangential shift changes tooth thickness independently. The combination is chosen to balance root strength, tip thickness, and contact ratio.
Off-pitch-cone meshing is the bevel analogue of off-node meshing in cylindrical gears. The contact zone is deliberately placed on one side of the pitch cone. The geometric marker is that the equivalent tip circle of the wheel lies inside its pitch circle, so the wheel has a negative addendum. In this design, the wheel addendum is \(-0.1039\) mm and its tip diameter is slightly smaller than its pitch diameter. The pinion has a large positive addendum of \(5.2997\) mm. The total tooth height remains the same on both members, about \(5.845\) mm, because the wheel’s negative addendum is offset by a larger dedendum. The intended benefit is that the relative sliding velocity keeps the same sign over the active flank, so the friction force does not reverse within the contact cycle. That can reduce scuffing and wear reversal.
The design is bounded by four geometric conditions. The first makes the off-pitch-cone coefficient negative:
$$ \lambda=\frac{0.5d_{va2}-0.5d_{v2}}{m_{mt}}<0 $$
Here \(d_{va2}\) and \(d_{v2}\) are the equivalent tip and pitch diameters of the wheel, and \(m_{mt}\) is the mid-face transverse module. The second condition keeps both tip thicknesses above a minimum:
$$ s_{a1},\,s_{a2}\ge 0.25\,m_{mt} $$
The third prevents undercut:
$$ x_1\ge h_{at}^{*}\left(1-\frac{z_{v1}}{z_{v,\min}}\right),\quad x_2\ge h_{at}^{*}\left(1-\frac{z_{v2}}{z_{v,\min}}\right),\quad z_{v,\min}=\frac{2h_{at}^{*}}{\sin^2\alpha_{vt}} $$
where \(h_{at}^{*}\) is the addendum coefficient and \(\alpha_{vt}\) is the transverse pressure angle. The fourth ensures continuous, smooth motion:
$$ \varepsilon_o=\sqrt{\varepsilon_\alpha^2+\varepsilon_\beta^2}\ge 1.5 $$
where \(\varepsilon_\alpha\) is the profile contact ratio and \(\varepsilon_\beta\) is the overlap ratio. These conditions are not independent. Increasing the pinion shift helps undercut and root strength but reduces tip thickness and can lower the contact ratio; moving the contact farther from the pitch cone changes the sliding and the load distribution. The geometry in Table 1 is a compromise that satisfies all four.
Geometry of the 4:41 Pair
Table 1 lists the resulting geometry. The pitch cone angles add to \(90^\circ\), as required. The face width is 26 mm, about 28% of the outer cone distance. The spiral angle is \(36^\circ\), which provides overlap; the total contact ratio is 1.6961, above the 1.5 limit. The pinion addendum is larger than the module, while the wheel addendum is negative. The wheel tip diameter is 184.4798 mm, just below its pitch diameter of 184.5 mm. These values are the geometric signature of the off-pitch-cone design.
| Parameter | Pinion | Wheel |
|---|---|---|
| Number of teeth \(z\) | 4 | 41 |
| Shaft angle \(\Sigma\) | \(90^\circ\) | \(90^\circ\) |
| Normal pressure angle \(\alpha_n\) | \(20^\circ\) | \(20^\circ\) |
| Spiral angle \(\beta\) | \(36^\circ\) | \(36^\circ\) |
| Outer transverse module \(m_{et}\) (mm) | 4.5 | 4.5 |
| Face width \(b\) (mm) | 26 | 26 |
| Outer cone distance \(R_e\) (mm) | 92.688 | 92.688 |
| Height coefficient \(k_d\) | 1.66 | 1.66 |
| Clearance coefficient \(c^*\) | 0.125 | 0.125 |
| Tangential modification coefficient \(x_t\) | 0.27 | -0.27 |
| Pitch cone angle \(\delta\) (°) | 5.5722 | 84.4278 |
| Pitch diameter \(d\) (mm) | 18 | 184.5 |
| Tip diameter \(d_a\) (mm) | 28.5492 | 184.4798 |
| Addendum \(h_a\) (mm) | 5.2997 | -0.1039 |
| Dedendum \(h_f\) (mm) | 0.5456 | 5.9491 |
| Total contact ratio \(\varepsilon_o\) | 1.6961 | 1.6961 |
Generating the Tooth Surfaces
The wheel is cut by the formate method. Its pitch cone angle is \(84.43^\circ\), so the wheel is nearly flat; generating would require a large cradle motion and offers little benefit. In formate cutting, the cutter profile is copied onto the wheel, and the machine has four setup parameters: radial cutter position \(S_{r2}\), angular cutter position \(q_2\), bed position \(X_{b2}\), and axial wheel position correction \(X_{d2}\). The cutter is a rotating cone with nominal radius \(r_0\) and profile angle \(\alpha_2\). The wheel surface is obtained by transforming the cutter surface into the wheel coordinate system.
The pinion is cut by the single-side generating method. Each flank is generated separately: the cutter head and the workpiece roll in a timed relationship, and the cutter position and roll ratio are set for one side, then for the other. This allows independent control of the convex and concave flanks, which is necessary because the two sides need different ease-off targets.
Because the wheel is not generated, its flank curvature is essentially fixed by the cutter profile. A formate wheel can therefore leave the pair with too little relative curvature, producing a narrow contact pattern or edge contact. To restore curvature, the pinion cutter is modified with a circular-arc edge. The arc radius \(R\) and the base point \(u_0\) add a controlled curvature term to the generated pinion surface. In the axial section, the arc satisfies
$$ \sin\phi=\frac{-u_0\cos\alpha_1+R\sin\alpha_1+u_1\cos\alpha_1}{R} $$
where \(u_1,\theta_1\) are surface parameters, \(r_{c1}\) is the theoretical tip radius, and \(\alpha_1\) is the profile angle. The second set of unknowns is the machine setting: radial position \(S_{r1}\), angular position \(q_1\), vertical wheel position \(E_{m1}\), axial wheel position \(X_{d1}\), roll ratio \(m_{21}\), cutter tip radius \(r_{c1}\), arc radius \(R\), and the surface coordinates \(u_1,\theta_1\) at the reference point. The blank mounting angle \(\gamma_{m1}\) and the profile angle \(\alpha_1\) are chosen in advance; the bed position follows from \(X_{b1}=-X_{d1}\sin\gamma_1-h_{fm1}\), where \(h_{fm1}\) is the mid-face dedendum of the pinion.
Curved-Surface Synthesis of the Pinion Cutter and Machine Settings
The synthesis starts from the conjugate pinion. Given the wheel surface and the prescribed relative motion, the conjugate pinion is the surface that would mesh with the wheel without clearance. It is obtained from the meshing condition \(\mathbf{n}_m\cdot\mathbf{v}_m^{(12)}=0\), where \(\mathbf{n}_m\) is the normal of the generating surface and \(\mathbf{v}_m^{(12)}\) is the relative velocity. The resulting pinion would have line contact in theory; in practice it would be sensitive to misalignment and would not provide a controlled contact pattern.
The modified pinion differs from the conjugate pinion by a prescribed ease-off surface. At the reference point the two surfaces are tangent; away from it, the normal separation grows according to a paraboloid:
$$ z_d=\frac{1}{2}k_a x_d^2+\frac{1}{2}k_b y_d^2 $$
where \(x_d,y_d\) are coordinates along the principal directions of the ease-off surface, and \(k_a,k_b\) are its principal curvatures. The paraboloid sets the size, shape, and orientation of the contact ellipse and the shape of the transmission-error curve.
The design target is a contact ellipse with semi-major axis \(a=0.4b_w\) and semi-minor axis \(b=0.5h\), tilted by \(20^\circ\) to the tooth direction, where \(b_w\) is the face width and \(h\) is the tooth height. The prescribed depth is \(0.012\) mm. This is twice the \(0.006\) mm thickness of the marking compound used in rolling inspection, so the contact pattern remains visible, and it is consistent with a transmission-error step of about \(0.006\) mm at the tooth-pair transition. The principal curvatures follow:
$$ k_a=\frac{8z_d}{a^2},\quad k_b=\frac{8z_d}{b^2} $$
The ease-off curvatures are then rotated into the longitudinal and transverse directions, combined with the conjugate surface curvatures, and converted to the principal curvatures and directions of the target pinion surface.
The synthesis matches the target and the actual machined surface at the reference point. The conditions are coincidence of position, normal, principal direction, and the two principal curvatures. The unknowns are the nine pinion parameters. The system is solved by nonlinear constrained optimization. Table 2 lists the result for the 4:41 pair.
| Parameter | Convex flank | Concave flank |
|---|---|---|
| Cutter profile angle \(\alpha_1\) (°) | -16.0 | 24.0 |
| Radial cutter position \(S_{r1}\) (mm) | 42.9147 | 46.7355 |
| Angular cutter position \(q_1\) (°) | 65.2822 | 64.6142 |
| Blank mounting angle \(\gamma_{m1}\) (°) | 5.5722 | 5.5722 |
| Bed position \(X_{b1}\) (mm) | -0.4497 | -0.5839 |
| Axial wheel position \(X_{d1}\) (mm) | -0.9876 | 0.3946 |
| Roll ratio \(m_{21}\) | 10.6656 | 9.9004 |
| Cutter tip radius \(r_{c1}\) (mm) | 58.5734 | 55.3288 |
| Cutter arc radius \(R\) (mm) | 32.0 | 80.0 |
| Vertical wheel position \(E_{m1}\) (mm) | 0 | 0 |
The two flanks are not symmetric. The convex flank uses a cutter profile angle of \(-16.0^\circ\), a tip radius of 58.5734 mm, and an arc radius of 32.0 mm; the concave flank uses \(24.0^\circ\), 55.3288 mm, and 80.0 mm. The roll ratios, 10.6656 and 9.9004, are close to the nominal 10.25:1 but not identical, because they define the generating motion rather than the gear ratio. The blank mounting angle equals the pinion pitch cone angle, \(5.5722^\circ\). The vertical wheel position is zero on both sides, and the bed and axial corrections are below 1 mm.
Mesh Simulation: Ease-Off, Contact Path, and Transmission Error
With the machine settings fixed, the ease-off surface of the actual pinion can be computed. The difference between the conjugate pinion and the machined pinion is plotted in micrometres. The computed surfaces follow the prescribed parabolic gradient: zero at the reference point, growing toward the edges. This confirms that the cutter modification and machine settings reproduce the intended ease-off.
On each ease-off contour, the lowest point along a contact line is a contact trace point. Connecting these points gives the contact path. For the concave flank, the path runs from the heel root to the toe tip; for the convex flank, from the toe root to the heel tip. The heel is the outer end of the tooth and the toe is the inner end. Both are inner-diagonal paths. This is a conventional target for spiral bevel gears because it keeps the contact away from the tooth edges and spreads the load along the flank.
The transmission error is obtained by plotting the ease-off depth at the trace points against pinion rotation. It is parabolic, with a smooth minimum near the design point. A parabolic TE curve means the motion error changes gradually rather than jumping at tooth entry or exit, which helps reduce impact and noise excitation.
A three-dimensional model of the pinion, built from the machine settings and assembled with the wheel in UG, shows the instantaneous contact ellipse. The contact area is elliptical, consistent with the predicted contact path. The simulation and the subsequent cutting trial are therefore consistent at the level of contact geometry.
Cutting and Rolling Inspection
The pair was cut on a GH-35 mechanical gear-cutting machine. The synthesis parameters were converted into machine settings: drum eccentric angle and cradle angle, as listed in Table 3. The wheel and the two pinion flanks use different settings because the wheel is formate-cut and the pinion is generated one flank at a time. The large difference in cradle angle between the wheel (\(109.1121^\circ\)) and the pinion (\(197.116^\circ\) and \(201.220^\circ\)) reflects the different cutting methods, not a change in geometry.
| Setting | Wheel | Pinion convex | Pinion concave |
|---|---|---|---|
| Drum eccentric angle (°) | 51.6030 | 51.5979 | 51.0318 |
| Cradle angle (°) | 109.1121 | 197.116 | 201.220 |
After cutting, the pair was run in a rolling inspection with marking compound. Several iterations of tooth matching and contact adjustment were needed before the final pattern. The result is an elliptical contact patch on the wheel, oriented along the inner diagonal, with no contact at the tooth tip. The pinion and wheel tips show no distortion. The pattern agrees with the simulated instantaneous contact ellipse.
This trial validates the geometric design and the cutter-modification method for the 4:41 pair. It shows that a four-tooth pinion can be cut with the calculated settings and that the resulting pair rolls with the intended contact pattern.
What the Trial Establishes and Where the Method Fits
The validation is geometric. A rolling inspection with marking compound shows the contact pattern under light load; it does not measure loaded contact, transmission error under torque, efficiency, noise, or fatigue life. Those are separate tests. The significance of the result is that the manufacturing chain—geometry, cutter shape, machine settings, and CNC program—can produce the intended tooth form in a very high-ratio spiral bevel pair.
The method is local: it matches curvature at a reference point and then checks the whole surface by ease-off simulation. It is not a global optimization of load distribution. For another ratio or size, the ease-off target and the nine machine parameters must be recalculated. The four constraints set the feasible window; as the pinion tooth count falls, that window narrows quickly.
Practical applications are compact right-angle reducers where a single stage can replace two. The 4:41 case is an extreme example; less extreme ratios should have more margin in tip thickness and contact ratio. The same workflow—equivalent-gear geometry, off-pitch-cone design, arc-modified cutter, and curved-surface synthesis—is applicable, but the numerical values in the tables are specific to this prototype.