Observed failure mode
Installed problem statement
Competing bandwidth and foot-strike requirements pushed the assembly toward preload migration and technician-dependent builds because support order and clamp sequence were not yet repeatable.
Back to case-study index
Housing and stack-order review summary
Humanoid knee concept balancing bandwidth, shock tolerance, and low-volume build repeatability
Engineering response
Corrective design move
The assembly sequence was reordered around bearing support and gearbox pilot engagement, and a verification set was defined for backlash, drag, torsional wind-up, and thermal rise.
Review inputs
Foot-strike energy assumptions, reversal-speed targets, and bandwidth requirements.
Assembly order, preload strategy, and cable-routing disturbance points.
Pilot-run verification data for backlash, drag, torsional wind-up, and thermal rise.
Expected outputs
Assembly-order revision linked to bearing support and gearbox pilot engagement.
Verification set for pilot builds with pass-fail criteria.
Housing watch list covering local flex, support span, and impact-energy concentration.
Engineering interventions
Compared rigid and compliant stack options against settling behavior, tooth-load protection, and tolerance sensitivity rather than selecting for peak stiffness alone.
Shifted the assembly sequence so bearing support, gearbox pilot engagement, and cable routing were established before final clamp-up of the outer structure.
Defined a verification set covering backlash window, torsional wind-up, no-load drag, and thermal rise after a representative duty block.
Critical watch items
Final clamp-up occurring before support interfaces are stabilized.
Impact energy concentrating at one thin housing wall or one tooth mesh.
Pilot-run instructions relying on technician-specific tuning.
| Review dimension | Observed result |
|---|---|
| Primary failure mode | Competing precision and impact-survival requirements |
| Critical change | Assembly-order revision with verification set definition |
| Pilot effect | Less technician dependence and clearer pass-fail criteria at bring-up |
Design lessons
Bandwidth targets lose value if the assembly cannot absorb field shock without distorting the support structure.
Low-volume repeatability improves when the verification set is defined alongside the mechanical architecture, not after hardware arrives.
Use this review package