Offline PD testing re-measures a circuit, not just a transformer: coupling capacitance, measuring impedance and calibration decide the picocoulomb number you write down.
A PRPD plot’s horizontal axis is phase, not time, and that makes the comparison across three phases the cheapest discriminator an analyst has. What each pattern means and what it costs to read them.
A partial discharge alarm threshold copied from one pC figure compares quantities that were never comparable. How to baseline a unit and alarm on trend instead.
Partial discharge detection methods compared by physics, not marketing: what UHF, HFCT, AE and RF each sense, where they mount, and what each one cannot hear.
Transformer fault diagnosis means choosing a method whose failure edge you know. Ratios, the Duval triangle and ML classifiers each stop answering somewhere.
Online vs offline partial discharge testing is not a sensitivity contest. One is pC-calibrated and brief, the other relative and continuous, and they answer different questions.
Bushing partial discharge monitoring begins at the flange tap. The condenser core is a coupling capacitor, and the tap is the port it was always meant to be read from.
Partial discharge source location turns a detected pulse into a position. Why it needs nanosecond timing and multiple UHF sensors, and where it fails.
OLTC dissolved gas analysis is a discipline apart from main-tank DGA. Why IEEE C57.139-2015 governs tap-changer oil, and which readings are normal.
DGA gas ratios turn a nine-gas reading into a fault type. How the Rogers, Doernenburg and IEC 60599 schemes work, where they fail, and what machine learning adds.
