Spiral bevel gears carry motion between intersecting shafts and are valued in aircraft, automotive, metallurgical and drilling transmissions because curved teeth engage gradually, run quietly and tolerate high load. When the same parts are case-hardened to raise wear resistance and service life, the shaping operation becomes the bottleneck: cutting a free-form flank in hard steel loads a slender cutter, and the resulting flank accuracy, surface integrity and tool cost are difficult to control simultaneously. Electrochemical machining (ECM) offers a way around the hardness barrier, but only if the cathode tool — the shaped electrode that is copied into the workpiece — is designed with enough geometric fidelity. The work described here addresses that tool-design problem for a Gleason-format spiral bevel gear.
The machining problem: hardened flanks and a worn cutter
Conventional production of spiral bevel gears relies on generating (rolling) motions executed by cutter heads with complex blade geometry. Two-sided milling of a hardened flank has three consequences that compound as the material gets harder. First, the cutter and the workpiece both deflect under load, so the tooth form deviates from the theoretical conjugate surface. Second, flank wear progresses along the blade, which shifts the effective cutting edge and drifts the tooth thickness over a production run. Third, the process is slow and the tooling is expensive, since a cutter head is dedicated to a narrow range of gear geometry.
ECM removes these constraints by changing the material-removal mechanism. The workpiece is made anodic, the shaped cathode is fed towards it, and the electrolyte in the inter-electrode gap carries the current that dissolves the metal closest to the tool. Hardness, toughness and prior heat treatment have essentially no influence on the dissolution rate; what matters is the local current density, which is controlled by the gap width and the electrolyte conductivity. There is no mechanical contact, so the cathode does not wear and no cutting force deforms the part. The surface produced is a dissolution surface rather than a cut surface, which is advantageous on hardened, previously ground or carburized flanks.
The price of this mechanism is that accuracy is transferred almost entirely into tool design and gap control. Material is removed wherever current can reach, so stray dissolution at tooth edges, an unevenly distributed gap field, or a poorly flushed cavity all show up directly as profile error. Cathode design practice is well developed for smooth, quasi-planar surfaces such as turbine blade profiles; spatial flank surfaces with strongly varying curvature and narrow flow passages are less well covered. The design task therefore reduces to two coupled problems: generating an accurate cathode surface from the conjugate tooth geometry, and delivering electrolyte uniformly enough that the gap field stays stable during the feed.
Geometry: a Gleason-format gear and its offset tooth surface
The reference part is a Gleason-system spiral bevel gear whose design parameters are listed in Table 1. The addendum coefficient (0.85) and clearance coefficient (0.188) are deliberately chosen so that the meshing pair retains backlash: a tight, zero-clearance design risks thermal expansion of the hardened teeth closing the contact and jamming the transmission.
| Parameter | Value |
|---|---|
| Large-end module (mm) | 2.5 |
| Number of teeth | 25 |
| Pressure angle (°) | 20 |
| Clearance coefficient | 0.188 |
| Addendum coefficient | 0.85 |
| Spiral angle (°) | 35 |
Because the flank is a spatial surface that cannot be described by a single analytic patch, the model is built from discrete points. A coordinate frame OgXgYgZg is placed at the intersection of the gear end face and the reference axis, tooth-profile curves are drawn in that frame, and the flank is then reconstructed from sampled points.
The sampling scheme used for the surface is a point-offset method: a series of reference points is taken on the tooth flank, each point is displaced along the local surface normal, and the displaced points form an offset curve. Four such curves, distributed along the spiral direction, are sufficient to define the surface once the corresponding conjugate-surface equations are solved for each point. The result is a closed geometric description of the flank from which a matching cathode surface can be derived by the same offsetting logic, with the offset magnitude set by the process rather than chosen arbitrarily.
Cathode profile from the equilibrium gap
The cathode is not a negative copy of the finished tooth. It must be smaller by the gap that exists when the process reaches steady state, because dissolution stops where the tool has approached to the equilibrium distance for the given feed rate. Under one-dimensional, constant-feed conditions the equilibrium gap follows the standard balance:
$$\delta_b=\frac{\eta\,\omega\,\kappa\,U_e}{v_f}$$
where η is the current efficiency, ω the electrochemical volume equivalent of the workpiece material (mm³ A⁻¹ s⁻¹), κ the electrolyte conductivity (S mm⁻¹), Ue the effective inter-electrode voltage (V) and vf the cathode feed rate (mm s⁻¹). The cathode surface is then obtained by offsetting the target flank along its normal by the local gap:
$$\mathbf{r}_c(u,v)=\mathbf{r}_w(u,v)+\delta(u,v)\,\mathbf{n}(u,v)$$
with rw the workpiece surface, n its unit normal and δ the local gap, which in practice varies over the flank because curvature, flow conditions and current density vary. Getting that variation right is what separates a usable cathode from a geometric curiosity.
For the gear of Table 1, twenty points arranged as four rows by five columns were evaluated on the conjugate surface, giving the cathode-surface coordinates listed in Table 2. The points were then fitted into a continuous tool surface. Rather than interpolating the sampled points directly, the surface was adjusted by constrained numerical optimisation: sequential quadratic programming was used to solve the conjugate-point problem, and a global optimisation routine in the Isight environment was used to refine the whole-flank fit so that the accumulated deviation across all twenty points, not just the four sampled curves, was minimised. The resulting head profile forms the working face of the tool.
| Point | Curve 1 | Curve 2 | Curve 3 | Curve 4 | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| X | Y | Z | X | Y | Z | X | Y | Z | X | Y | Z | |
| 1 | 67.124 | 5.235 | −20.762 | 66.123 | 5.348 | −17.326 | 65.121 | 5.683 | −14.075 | 64.091 | 5.947 | −11.739 |
| 2 | 65.452 | 8.471 | −19.437 | 64.436 | 8.321 | −16.462 | 63.386 | 8.210 | −13.537 | 62.275 | 7.896 | −10.627 |
| 3 | 63.271 | 11.552 | −18.668 | 62.238 | 11.241 | −15.294 | 61.247 | 11.147 | −12.662 | 60.570 | 10.837 | −9.826 |
| 4 | 61.876 | 14.387 | −18.371 | 60.851 | 14.271 | −15.104 | 59.790 | 14.204 | −12.461 | 58.742 | 13.931 | −9.540 |
| 5 | 59.793 | 17.269 | −17.652 | 58.695 | 17.130 | −14.753 | 57.631 | 17.143 | −11.472 | 56.591 | 16.824 | −8.285 |
Flow path, sealing and workholding
The tool body is built from four functional groups: the profiled head, the shank, the internal cavity and the clamping interface. The head and shank are made as separable parts so that a worn or redesigned profile can be replaced without remaking the whole assembly; six threaded holes in the shank fix the tool to the machine spindle interface, and a matching set in the base fixes the assembly to the machine table.
Flushing is handled by a two-level manifold inside the cavity. Electrolyte enters a cylindrical main channel, where a throttling (flow-restricting) surface in the upper half of the channel builds back pressure and evens out the supply. Four transfer holes in the channel wall then pass the fluid into four distribution channels, from which it leaves through outlet holes at a controlled velocity and fills the inter-electrode space. After machining, the spent electrolyte is drawn back through a return port rather than dumped, so that it can be filtered and reused. The staged main-channel/distribution-channel arrangement is what prevents the jet from one outlet from starving the gap at another location — the usual cause of localised overcutting on a curved flank.
Two further details support stable operation. A transparent glass shroud encloses the tool body, with two insulating sealing rings of different diameters at the top and bottom, providing both electrical isolation from the machine frame and visual access to the flow during setup. The workpiece is held on a square base fitted with a locating boss, which fixes the gear in the correct angular and axial position, and a conductive plate that raises the workpiece to the anodic potential. Together with the six base fixing holes, this completes a cathode assembly that can be mounted, aligned and sealed as a single unit.
Machining trials on quenched 20CrMnTi
Validation used quenched 20CrMnTi gear blanks. The electrolyte was a mixed nitrate–chloride solution of 25% NaNO3 and 10% NaCl, held at (30 ± 5) °C; the chloride component raises conductivity and dissolution rate, while the nitrate component favours the passivating, more uniform removal that keeps stray attack on the unexposed flanks low. Tooth flanks were machined with the assembled cathode and then measured on a coordinate measuring machine, from which tooth-profile traces were plotted over a common evaluation length La; the profile-form deviation H1 of the ECM gear remained below that of a generating-machined reference over the same length.
Surface finish was evaluated along four traces with five measurement points each; Tables 3 and 4 give the roughness values for the ECM gear and for a generating-machined comparison gear.
| Trace | Point 1 | Point 2 | Point 3 | Point 4 | Point 5 | Mean |
|---|---|---|---|---|---|---|
| 1 | 1.25 | 0.97 | 1.24 | 1.13 | 0.88 | 1.09 |
| 2 | 0.62 | 0.81 | 1.30 | 1.46 | 0.85 | 1.01 |
| 3 | 1.27 | 0.92 | 1.35 | 0.96 | 0.57 | 1.01 |
| 4 | 1.14 | 0.48 | 1.47 | 0.71 | 0.61 | 0.88 |
| Overall | 20 measurements | 1.00 | ||||
| Trace | Point 1 | Point 2 | Point 3 | Point 4 | Point 5 | Mean |
|---|---|---|---|---|---|---|
| 1 | 1.59 | 0.89 | 1.69 | 1.28 | 0.92 | 1.27 |
| 2 | 1.34 | 1.14 | 0.83 | 1.50 | 1.65 | 1.29 |
| 3 | 1.83 | 1.58 | 1.75 | 1.07 | 0.97 | 1.44 |
| 4 | 1.72 | 1.68 | 1.39 | 0.90 | 1.24 | 1.39 |
| Overall | 20 measurements | 1.35 | ||||
Averaged over all twenty measurement points, the electrolytically machined flanks reach Ra ≈ 1.00 µm against ≈ 1.35 µm for generating machining, a reduction of roughly 26%, with the single best and worst values also improving (0.48 µm versus 0.83 µm at the low end, 1.47 µm versus 1.83 µm at the high end). The spread within each data set is comparable, which matters: the improvement comes from a lower central level rather than from a few favourable points.
Design implications and open issues
Three features of this design carry most of the performance. The offset-sampling route ties the cathode surface to the conjugate geometry through a single, auditable parameter — the local gap — so a change in feed rate or electrolyte conductivity can be absorbed by re-offsetting rather than by remodelling the tooth. The optimisation step corrects the surface globally instead of forcing it through twenty measured points. And the staged internal manifold keeps the flow distribution stable enough that the gap field does not drift during a pass, which is the mechanism by which a geometrically correct cathode usually fails in practice.
The reported evidence should be read within its scope. It covers one gear specification (module 2.5, 25 teeth, 35° spiral angle) and a comparison based on four roughness traces and profile-form traces taken on a coordinate measuring machine; full-flank scanning, gap-field measurement and a quantified deviation budget for the whole tooth would be needed before claiming flank-wide accuracy at a specified tolerance grade. The process also carries costs that are not visible in a surface-finish comparison: chloride-containing electrolytes demand corrosion protection for the machine and waste treatment for the spent liquor, and current density at tooth edges must be managed to avoid stray dissolution where the flank meets the root and the toe. For production, the natural niche for this cathode concept is finishing or semi-finishing of already-hardened bevel gears, where cutter wear and hard-steel cutting forces dominate, rather than bulk stock removal.