Strain-Gauge Torque-Transducer Zero-Drift in Student-Built Torsion Test Rigs for Engineering Education: A Dead-Weight Calibration Linking Wheatstone-Bridge Thermal Offset to Yield-Strength Measurement Error

Student-built torsion test rigs are a staple of undergraduate engineering laboratories, yet the strain-gauge torque transducers at their heart are rarely calibrated beyond a single room-temperature dead-weight check. The model treats the bridge output as a sum of mechanical strain, thermal offset and random noise, and it derives the bias in inferred yield torque as a function of the offset voltage, the bridge excitation, the gauge factor and the specimen geometry. A sensitivity analysis over plausible component tolerances shows that a zero drift of 5 microvolts per volt, which is typical of uncalibrated student instrumentation, produces a yield-strength bias of approximately 1.2 per cent for a mild-steel specimen, rising to 3.4 per cent when the drift is 15 microvolts per volt. The paper then specifies a dead-weight calibration procedure that uses a known torque arm and masses to estimate the offset and its temperature coefficient, and it derives the number of calibration points needed to bound the residual bias below 0.5 per cent at 95 per cent confidence. The contribution is a traceable chain from bridge-level thermal offset to a defensible bound on yield-strength measurement error, together with a calibration design that engineering educators can implement without specialised equipment. The analysis is theoretical and does not report experimental data; it provides the error budget and calibration specification that a laboratory implementation would require.