Standing in front of a massive power transformer at three in the morning with a fresh set of Sweep Frequency Response Analysis curves that do not match the factory test reports is a stressful experience. A deviation in the curves can indicate that the transformer suffered severe mechanical damage during transport or a recent through-fault. However, more often than not, the culprit behind a bad curve is poor testing technique, incorrect grounding, or high-frequency interference on the substation floor.
Understanding the Three Frequency Bands
An SFRA scan injects low-voltage signals across a wide frequency range to analyze the complex RLC network of the transformer windings. The lower frequency range, typically below two kilohertz, is dominated by the core magnetization inductance and can be affected by residual magnetization. The medium frequency range, between two kilohertz and one hundred kilohertz, represents the main winding structure and is highly sensitive to radial or axial winding movement. The high frequency range, above one hundred kilohertz, is dominated by the tap changer configuration, internal leads, and bushing connections.
Eliminating Measurement Noise and Bad Grounding
Before you write up a non-conformance report for winding displacement, you must verify the physical integrity of your test connections. The most common cause of high-frequency curve mismatch is the quality of the shield grounding. Ensure that you are using flat, wide copper braids for grounding the coaxial shields to the bushing flanges rather than thin, round wires. The contact surface must be completely free of paint, rust, and moisture to ensure a low-impedance connection that prevents noise from corrupting the high-frequency measurements.
Interpreting Deviation on the Substation Floor
Compare your current test results with the factory benchmark signature using the logarithmic correlation coefficients outlined in IEC 60076-18. If the low-frequency peaks match perfectly but the mid-frequency range shows distinct frequency shifts, you may be looking at real winding deformation. If the discrepancies are isolated to the high-frequency zone, always suspect your bushing connections first. Clean the bushing terminals with a wire brush, re-tighten the clamps, and run the sweep again before making any final diagnostics.