Experimental Piezoelectric System Identification

In this research article, you will learn about experimental system identification of natural frequencies of vibration of a piezoelectric film element including a detailed introduction to several factors that often confound application of theories to the real world: 1) additional response data induced by signal measurement, 2) signal harmonics, and 3) signals induced by the power supply. Signals are read and processed from a piezoelectric element configured as a cantilever, which is bent by a motor and cam assembly. Due to the piezoelectric effect, the strain created by the mechanical displacement generates charges in the piezoelectric material, which is translated to a voltage reading with a charge amplifier circuit. The effects of reference resistance and capacitance and the time constant of the circuit were investigated using a National Instruments myDAQ. The myDAQ oscilloscope effectively displayed time response, but spectral data was suspect. Especially since system identification (ID) largely comprises identification of the natural frequency, it is preferred to not modify the signal being measured (as is the case with the oscilloscope). Furthermore, improved spectral plots were seen with increased supply voltage (not always a good thing); therefore buffers were investigated next. The buffer provided improved spectral data, but the buffer output did whatever was necessary to the signal to make the voltages at the inputs be equal (again, modifying the signal). Using op-amps in the buffer configuration resulted in pretty spectral plots, but contained “ghost” resonances, while using the op-amps in a two-stage charge amplifier configuration suppressed the “ghost resonances”. In all cases, taking measurements at the output of the charge amplifier was superior to taking measurements at the voltage amplifier. A two-stage amplification configuration provided on-the-order-of triple voltage signal (peak minus offset) amplification. Several of the cases investigated provided good signal amplification with very legible spectral data plots.