Most series arcs lead to electrical fires that cannot be interrupted by circuit breakers, because the arc’s current is outside the breaker’s operating range. In this paper, experiments were conducted on the detection and identification of series arcs to prevent electrical fires. Plugs and outlets specified in KS C 8305 were deteriorated to replicate arc faults commonly found in fields. The characteristics of series arcs resulting from various types of loose connections were determined by analyzing the frequency spectra and phase distributions of detected arc pulses. The results showed that the simulated arc defects used in this study were more similar to actual arc phenomena than the existing arc generator specified in UL 1699. In addition, loose connections, such as wire-wire, terminal-wire, and outlet-plug, can be identified by phases of 0°, 180° and 360°, respectively. These phases can be detected by a band pass filter with a frequency range of 5~10 MHz, which can be used as the trip signal for circuit breakers.

This paper dealt with the frequency component analysis of acoustic signals produced by corona and series-arc discharges as a diagnostic technique for closed-switchboards. Corona and series-arc discharge were simulated by a needle-plane electrode and an arc generator specified in UL1699, respectively. Acoustic signal was detected by a wideband acoustic sensor with a frequency bandwidth of 4 Hz∼100 kHz (-3 dB). We analyzed frequency spectrums of the acoustic signals detected in various discharge conditions. The results showed that acoustic signals mainly exist in ranges from 30 kHz to 60 kHz. From the experimental results, an acoustic detection system which consists of a constant current power supply (CCP), a low noise amplifier (LNA) and a band pass filter was designed and fabricated. The CCP separates the signal component from the DC source of acoustic sensor, and the LNA has a gain of 40 dB in ranges of 280 Hz∼320 kHz. The high and the low cut-off frequency are 30 kHz and 60 kHz, respectively. We could detect corona and series-arc discharges without any interference by the acoustic detection system, and the best frequency is considered in ranges of 30 kHz∼60 kHz.