6-AXIS LOAD CELL / DESIGN & CALIBRATION

Six components.
A calibrated matrix.

A six axis force transducer (also called a six-axis load cell or 6dof transducer) resolves Fx, Fy, Fz, Mx, My and Mz. Calibration relates coupled electrical channels to a wrench in a defined reference frame.

From bridge signals to N and Nm

A monolithic elastic body strains under applied load. Its bridge channels usually respond to more than one component. A calibration matrix converts the zero-corrected signal vector into the six physical components.

w = C · (s − s₀)
w = [Fx, Fy, Fz, Mx, My, Mz]ᵀ

Confirm whether decoupling occurs inside the sensor, in an external acquisition unit or in host software. Electronic integration is a packaging decision, not proof of lower measurement uncertainty.

Validate independent loads and combinations

1. Define the frame. Record axes, positive signs, mounting origin and calibration temperature.
2. Load the complete range. Apply traceable force and moment references in positive and negative directions, with repeat cycles.
3. Fit and verify separately. Use independent validation points to report nonlinearity, hysteresis, repeatability and cross-axis residuals.
4. Check simultaneous loads. Validate representative application combinations and temperature points, not only isolated axes.
5. Document the chain. Record serial number, matrix revision, electronics, cable, firmware, filter setting and uncertainty budget.

A tool offset changes the moment requirement

With a vector r from the target reference origin to the force application point, the added moment is r × F. A 100 N perpendicular force acting 150 mm away produces 15 Nm. State the origin and sign convention before combining this term with measured moments.

For a planar foot contact at the sensor origin, an illustrative convention gives x = −My/Fz and y = Mx/Fz. This relation requires sufficient Fz, a stated ground plane and proper frame transforms; it is not a full gait stability controller.

Ask for resolution at the working bandwidth

ADC bit depth does not determine usable force resolution. Request per-channel noise statistics, filter bandwidth, effective output rate and temperature drift. Keep resonant frequency, usable bandwidth and digital sample rate distinct.