High sensitivity brush arcing monitor for a dynamoelectric machine

Abstract

High noise immunity instrumentation for monitoring brush arcing in a dynamoelectric machine is disclosed. In a preferred form the instrumentation includes a stabilized variable threshold discriminator operative in conjunction with a feedback network and a non-linear high frequency amplifier to provide a discriminator threshold which increases with increasing levels of background noise so that only the arc indicative portions of the signal are passed. A substantial increase in the signal to noise ratio is obtained. The feedback network includes a relatively long time constant integrator and buffer amplifier. Further, to provide an alarm upon the occurrence of brush arcing, an alarm network is provided which includes a comparator for continuously comparing the arc indicative signal with a reference signal which is automatically adjusted to compensate for changes in the average level of the signal from the non-linear high frequency amplifier. This compensation provides for automatically resetting the alarm comparator. This is achieved by providing a second integrator having a time constant somewhat longer than the first integrator. By appropriately selecting the relative time constants of the first and second integrators, resolution of the alarm network is narrowed to a unit time basis, thus quantifying the measurement to provide an indication of the condition of the brushes.