My instrument produces a phase-resolved discharge plot. What features distinguish the common defect types?
PRPD (Phase-Resolved Partial Discharge) plots pulse magnitude and pulse count against the phase angle of the supply waveform, so the defect signature appears as a repeatable shape rather than a single number.
Two clusters 180° apart, similar amplitude: Internal discharge in a void within the groundwall insulation. Amplitude rises smoothly with voltage and the pattern is symmetrical because the void sees both half-cycles equally.
Strongly asymmetric cluster at one peak: Corona from a sharp point or a poor connection at the endwinding. Corona is polarity-dependent, so one half-cycle dominates.
Broad, diffuse cloud near the zero crossing: Surface tracking across contaminated or moist endwinding insulation. This is the pattern that most often precedes a flashover and deserves priority.
Pulses concentrated in the slot region / high count at low magnitude: Slot discharge from a loose bar or lost conductive slot coating - confirmed by comparing with Stator Slot Couplers against terminal couplers.
On VFD-fed machines: The phase reference must be taken from the drive output, not the supply side. A reference taken upstream of the inverter smears the pattern into an unreadable band because the PWM carrier shifts the effective phase.
Always read the trend: Absolute amplitude alone is weak evidence. The same pattern growing over months is strong evidence.