Optimal filtering of constant velocity torque data.

Area of Science:

• Biomechanics
• Kinesiology
• Biomedical Engineering

Background:

• In vivo dynamometric data processing requires accurate filtering.
• Common analog smoothing methods may not be optimal for Cybex II dynamometers.
• Inappropriate filtering can lead to significant errors in torque measurements.

Purpose of the Study:

• To implement an optimal filtering strategy for in vivo dynamometric data.
• To appraise the validity of commonly accepted analog smoothing methods.
• To establish reliable methods for predicting optimal digital filtering for Cybex II dynamometers.

Main Methods:

• Utilized an inert gravitational model to assess filtering requirements.
• Analyzed torque and angular displacement data from Cybex II dynamometers at various speeds and loads.
• Employed spectral analysis and optimized low-pass digital filtering.
• Developed third-order least-squares polynomials to predict optimal cutoff frequencies.

Main Results:

• Identified two main sources of data contamination: servomechanism control system oscillations and Cybex motor oscillations.
• Determined that the optimal cutoff frequency is a function of the percentage of maximum tachometer feedback voltage (% Vfb max).
• Established distinct polynomial equations for different Cybex II machines to predict optimal filtering.
• Demonstrated that inappropriate filtering, especially over-smoothing, can mask inertial artifacts and induce errors, particularly in peak torque measurements.

Conclusions:

• Optimal digital filtering is essential for valid processing of dynamometric data.
• The use of Cybex recorder damping settings is discouraged due to potential inaccuracies.
• The developed polynomial functions provide a reliable method for determining optimal filtering parameters for Cybex II dynamometers across different settings.