Abstract

A broadband digital harmonic vector voltmeter proposed previously and studied theoretically by the authors was implemented using a special-purpose, random sampling strategy, to avoid the bandwidth limitations due to the finite conversion time of the sample-and-hold and analog-to-digital-conversion (S/H-ADC) devices. The experimental results have shown that the bandwidth of the instrument is not limited by the finite conversion time of S/H-ADC devices, since good accuracy can be achieved even when the average sampling frequency is much lower than the signal bandwidth. The amplitude and phase uncertainty, with sinusoidal test signals up to 1 MHz and an average sampling rate of 10 kHz, was found to be lower than 3% and 0.03 rad, respectively. For more careful testing of the broadband performance of our instrument, we also carried out two-frequency, variable order harmonic measurements, which showed good accuracy (amplitude error less than 1.5% and phase error less than 0.03 rad) with harmonics up to 300 kHz. Reasonable accuracy (i.e., sufficient to correctly reconstruct the actual signal waveform) was also found with a highly distorted square-wave signal