Constant velocity dynamometer: an appraisal using mechanical loading.

Measurement characteristics of two constant velocity dynamometers (Cybex II, Lumex, Inc., NY) were evaluated at 11 selected speeds (0 to 5 rad X s-1) under various conditions of inert gravitational loading. A linear accelerometer was affixed to the lever arm. Speed settings were standardized percentages of each machine's maximum tachometer feedback voltage (0 to 100% Vfb max). Regression of Cybex transducer-observed voltages (V0) on true, i.e., expected, torques (Me) revealed that, for any static loading condition or angle, the measurement system was essentially linear and valid, although greater variance was evident for smaller torques. However, different calibrations, i.e., regression lines, were required for different speed settings. The error in predicting Me from V0 increased substantially with increased % Vfb max. Methodologies which assume constant velocity dynamic loading induce questionable, often invalid, results and inferences. Dynamic V0 represents both gravitational and inertial components. Considerable fluctuations in lever arm instantaneous velocity were ascribed to the combined influence of load and % Vfb max upon the servomotor. After computing inertia-corrected Me, calibration variability and prediction error were markedly reduced. Unexplained variability may be due to poor proximity of the Cybex transducer to the force application point. We concluded that: Cybex II should be calibrated only statically; use of an optimal filter will avoid false estimates of joint torque; inertial corrections must be applied in order to minimize potentially serious errors due to system angular acceleration; and reported muscle function relationships might be usefully re-assessed in view of these findings.