Stepover Rules the Trade-Off in CNC Finishing of Prosthetic Foot Molds

Researchers have mapped out a practical process window for CNC milling of the steel molds used to make prosthetic feet, and their analysis points to a single parameter that dominates both machining time and surface quality. In a study published in Discover Mechanical Engineering, Dhimas Yogie Nugraha and Wahyu Dwi Lestari optimized computer numerical control (CNC) ball end milling of a two-piece prosthetic foot mold made from AISI 1020 low-carbon steel. Their central finding is stark: the stepover, the lateral spacing between adjacent cutter passes, contributed 52.9869 percent and 52.7903 percent of the variation in the estimated machining time of the upper and lower mold halves respectively, and a striking 81.7726 percent of the variation in surface roughness.

The motivation is a familiar conflict in mold manufacturing. Productivity-oriented cutting parameters can shorten cycle time, but they deteriorate the mold surface, and a rough cavity surface translates into more manual polishing before the mold can produce rubber- or polymer-based prosthetic components. Mold surface quality governs the dimensional accuracy, surface texture, demolding behavior, and repeatability of the final product, so the researchers set out to balance two responses simultaneously: the estimated machining time for each mold half and the arithmetic mean surface roughness, Ra, measured after finishing.

Roughing conditions were held constant across all 27 runs so that observed changes could be attributed to the finishing stage alone. Machining times for the upper and lower mold halves were estimated in Mastercam 2025 under identical machine definition, tool library, feed settings, post-processing logic, and simulation settings, yielding responses labeled Tupper and Tlower. Surface roughness was measured experimentally with a contact stylus profilometer on representative specimens machined from the same AISI 1020 steel with the same 10 mm carbide ball end mill and flood cooling used in the design.

The machining platform was a three-axis SMTCL BRIO MILLER 8 CNC milling machine with a Siemens Sinumerik 808D control system. The two-piece mold geometry contained freeform cavity features corresponding to the anatomical shape of a prosthetic foot, and the two halves carried different cavity orientations and toolpath lengths. That geometric difference showed up directly in the data: for every parameter combination, the upper mold half required a longer estimated execution time than the lower half, with Tupper ranging from 12.333 to 24.796 hours and Tlower from 11.506 to 21.233 hours. Because roughing was constant, this variation was attributed mainly to the finishing toolpath conditions.

Signal-to-noise ratio analysis under the smaller-is-better criterion and analysis of variance at a 95 percent confidence level quantified the factor effects. For machining time, the fastest condition came at the largest stepover, highest feed rate, and deepest axial cut: s = 0.45 mm, f = 800 mm/min, and ap = 0.40 mm, producing Tupper = 12.333 hours and Tlower = 11.506 hours. Stepover and feed rate were both statistically significant with p-values of 0.0000, contributing 52.9869 and 38.8540 percent respectively for the upper half, and 52.7903 and 39.0719 percent for the lower half. Axial depth of cut was negligible in both cases, contributing only 0.0001 percent with p-values near 1, because the finishing-focused window and controlled remaining stock meant that varying axial engagement changed neither the number of cutter passes nor the programmed toolpath length.

Surface roughness told the opposite story. The best finish, Ra = 1.101 μm, occurred at the smallest stepover, lowest feed rate, and shallowest cut: s = 0.15 mm, f = 300 mm/min, and ap = 0.20 mm, but that combination demanded the longest machining times, 24.796 hours for the upper half and 21.233 hours for the lower. Stepover was the only statistically significant factor for roughness, with p = 0.000 and its 81.7726 percent contribution reflecting the scallop-height mechanism of ball end milling: wider lateral spacing between adjacent passes leaves a taller residual cusp of uncut material between toolpaths, roughening the surface even as it shortens the cycle. Feed rate contributed just 2.5026 percent and was not significant at p = 0.228, and axial depth of cut contributed 0.0405 percent at p = 0.975.

To resolve the conflict, the authors applied a desirability function with equal response weights and equal importance assigned to Tupper, Tlower, and Ra, transforming each response on a smaller-is-better basis with boundaries drawn from the observed dataset. The best compromise emerged at s = 0.30 mm, f = 800 mm/min, and ap = 0.30 mm, with an overall desirability of 0.8768. This setting produced Tupper = 13.221 hours, Tlower = 12.197 hours, and Ra = 2.368 μm. Compared with the fastest condition, it increased machining time by only 7.2 percent for the upper half and 6.0 percent for the lower half, yet cut surface roughness by 58.2 percent, from 5.660 μm down to 2.368 μm. The intermediate stepover curbed the excessive scallop formation of the 0.45 mm setting while avoiding the long cycle of the 0.15 mm setting, and the high feed rate preserved productivity.

The authors are careful about what these numbers do and do not mean. The machining times are CAM-estimated toolpath execution times, not complete shop-floor production times; they exclude workpiece setup, tool change, machine acceleration and deceleration behavior, and operator-dependent delays. The roughness values come from representative specimens designed to reproduce the local ball end tool contact and inclined-surface finishing condition of the freeform cavity, not from full roughness mapping of the actual mold, where local curvature, surface inclination, cutter contact point, and tool engagement vary continuously. Because the three profilometer traces per run were repeated measurements on the same specimen rather than independent machining replicates, the ANOVA results are interpreted as factor contributions within the controlled design rather than a full assessment of run-to-run variability.

The team also frames the recommended Ra = 2.368 μm as a pre-polishing figure rather than a universal functional limit, noting that no single universally accepted Ra threshold exists for prosthetic foot molds and that polishing or superfinishing remains available when smoother surfaces are required. They recommend the s = 0.30 mm, f = 800 mm/min, ap = 0.30 mm window specifically for AISI 1020 prosthetic-foot mold finishing with a 10 mm carbide ball end mill under flood cooling on the investigated machine-tool-material system, and they call for confirmation through full-scale mold machining, dimensional inspection, demolding trials, tool wear evaluation, and independent machining replicates before generalizing to other molds, materials, or CNC systems. Within those limits, the framework offers manufacturers a quantitative alternative to trial-and-error: a way to estimate the trade-off between cycle time and surface quality before committing to a parameter set, with stepover as the principal dial that balances toolpath density against residual surface texture in freeform mold finishing.

Subject of Research: Optimization of CNC ball end milling parameters for finishing AISI 1020 prosthetic foot molds

Article Title: Process parameter optimization for CNC ball end milling of AISI 1020 prosthetic foot molds

Article References: Nugraha, D. Y., & Lestari, W. D. (2026). Process parameter optimization for CNC ball end milling of AISI 1020 prosthetic foot molds. Discover Mechanical Engineering, 5(1), Article 193. https://doi.org/10.1007/s44245-026-00387-5

Image Credits: AI Generated

DOI: 10.1007/s44245-026-00387-5

Keywords: CNC milling, ball end milling, prosthetic foot mold, AISI 1020 steel, stepover, surface roughness, Taguchi L27, desirability function, ANOVA, Mastercam, machining time optimization, mold finishing