This project is an example of a uniquely challenging sensor system project. The goal was to determine the maximum strain, maximum loading rate, and frequency content of the strain response in the cable based closure system for an alpine ski boot. The shape of the ski boot, flexible cable, need for live testing during a ski run, and environmental factors all presented challenges. Additionally, the extremely tight timeline demanded the use of piezo strain sensors, which are not ideal for this application. A custom algorithm and some clever test procedures were used to extend the linear range of the sensors and preserve as much of the low frequency content of the response as possible. This project is a good example of expertise in sensor system design, data acquisition, signal analysis and filtering, LTI system theory, and Matlab/Python.
SENSOR SYSTEM DESIGN
The image above shows a CAD model of the sensor assembly. The piezo strain sensor used for this study was the PCB 708B02. The sensor is 0.6 inches long so it is quite small and requires special handling. Although the sensor comes fully characterized and packaged on delivery, the custom mounting system shown above was required in order to mount the sensor inline with the ski boot cable. Additional calibration and characterization were required due to the unique mounting configuration and loading pattern for this application.
Skiing involves both low frequency loading (due to turning) and high frequency loading (due to vibration and small changes in the snow surface) making a piezo sensor less than ideal for this application. Project time constraints required use of this sensor so a custom procedure was created to preserve DC strain response when initially tightening the ski boot. Additionally, a full LTI system characterization was performed in a laboratory setting. Using the results from the laboratory, the proper data filters, and a custom algorithm, the linear range of the sensor was extended from 2.5 Hz down to 0.25 Hz. This allowed the required data to be gathered with a worst case error better than 15% and satisfied the project timeline. The image above shows a comparison between an Instron pull tester, foil strain gauge, and the improved PCB 708B02 piezo strain sensor.
Additionally, a custom data acquisition system had to be created in order to enable the sensor to be run while a test subject was skiing. Up to 4 sensors could be mounted on the person at any time so the DAQ had to handle multiple sensors and needed to be reliable in a sub freezing, wet, and jarring (riding in a backpack) environment. With only 3 weeks to prepare for the test there was no margin for error.